Archive for July, 2019

the KMT primary

July 4, 2019

I started this post last week after the first KMT presidential debate. It was supposed to be a debate summary. However, I don’t seem to have the time to write an exhaustive debate analysis, and I have an important point about polling that I want to throw out there. So I’ll just publish this without much debate analysis.


It’s hard to know where to start with this field since there are so many different angles. I guess I’ll start with Han Kuo-yu. The Kaohsiung mayor has seemingly been in the lead all year. In a previous post, I suggested that Han draws support from three different types of voters: orthodox KMT supporters, the uncommitted and not-very ideological centrist voters, and voters fed up with the establishment from both parties.

The deep blue voters are head over heels for Han. He is a mainlander who came up through the KMT Huang Fu-hsing (military) party branch, so maybe it isn’t all that surprising that they love him so much. I think one thing that is often overlooked with Han is just how orthodox his views about Taiwan and China are. We often think of him as a populist, throwing bombs from the outside. He has offended some KMT insiders, such as former president Ma, but his offenses are stylistic rather than ideological. On policy, his views fit in quite comfortably with standard KMT views. Economically, he wants to develop by integrating Taiwan’s economy into China. This is not fundamentally different from Ma’s policy. Of course, Han packages it differently. He stresses how lower-class people can benefit, such as farmers selling agricultural products or people in the tourist industry catering to Chinese visitors. Ma viewed the problem from a more abstract macroeconomic angle. Nevertheless, they are both pushing similar policy agendas. Han also isn’t really challenging the 92 Consensus, though he doesn’t always phrase his ideas in those terms. In other words, Han is different from someone like Trump. Republicans have to accept some policies they don’t like about Trump (notably trade policy and, to a lesser extent, immigration policy) to get the things they like (taxes, judges, and executive power). They also have to accept his style, which most of them dislike. KMT members don’t have to make any policy compromises with Han. They only have to accept his style, which isn’t nearly as outrageous of a departure as Trump’s. This is a small price to pay for winning. Let’s not forget that Han performed a miracle last year, winning what was widely regarded as an unwinnable race. To put it even more dramatically, after 2016 it looked as if the KMT message was obsolete. To win back power, the KMT was going to have to make some painful changes to some of its most cherished positions. Instead, Han emerged and managed to sell an almost unaltered version of their old-time religion to some of the traditionally hostile voters in the country. Moreover, he created the wave that carried the KMT along with it in many other important races all over the island. “We can win elections without doing anything differently? Sign me up!”

So it makes perfect sense to me that the deep blues adore Han. What is somewhat harder to understand is why he is so attractive outside that group. After all, it is the ability to go beyond deep blue votes that enables him to win (and thus so enthralling to the deep blues). He has been embraced by the KMT local factions in a way that is nearly unthinkable for a mainlander (who speaks only lousy Taiwanese) from the Taipei area who came up through the military party branch. The obvious answer is that he married well. His wife’s (Lee Chia-fen) family is deeply enmeshed in Yunlin factional politics. Her family has held a seat in the county assembly since 1986. Her father had it for three terms, then her brother held it for three terms, and now she is on her third term. Her father started out as a factional opponent of current Yunlin political godfather Chang Jung-wei. In Chang’s first big stab at power, the 1994 Yunlin speaker election, her father supported the other side. (That was a wild affair. Both sides bribed heavily and then took their purchased assembly-elect members on foreign trips to make sure the other side wouldn’t poach them, returning only on the day of the election. About 2000 police surrounded the county assembly so that gangsters wouldn’t be able to physically intimidate the members. The first ballot was tied, and Chang’s opponent won when one of Chang’s votes was ruled invalid because it was (inadvertently or maybe not so inadvertently) splashed with betelnut juice.) Four years later, Chang won the 1998 speaker election. In that election, Lee’s father switched sides, the Lee family has been allied with Chang ever since. There are at least two important consequences of this history. First, Han has been in close contact with local faction politicians for his entire thirty-year marriage. He has learned how to speak their language and be comfortable with their culture. As a family member, he has been trusted and socialized in ways that very few mainlander politicians can claim. In short, he is not an outsider. Second and more specifically, his ties with Chang Jung-wei go back two decades. He is not just a recent ally of convenience. This is a long-term relationship. Chang is one of the most influential faction leaders. If he vouches for Han, it carries some weight. It has been fascinating to watch the local factions switch their allegiance over the past year. A year ago, Wang Jin-pyng and Wu Den-yi were considered the leaders of this part of the KMT. Wang campaigned hard for Han in the mayoral race, bringing the Kaohsiung factions into the fold. Now however, Wang and Wu seem to have lost their leadership positions. All those local factions seem more responsive to Han than to them.

I’m not entirely convinced by the story I just told about Han becoming part of the factional family. I can’t quite explain why, but it just seems too easy to me. I suspect (without any evidence) that something more substantial is going on behind the scenes. Political scientists have traditionally understood local factions as a network of hierarchical patron-client relationships, with the top-level patron as the KMT. The KMT distributed resources to its clients, which they in turn distributed down their networks. However, many of the old sources of goodies, such as the farmers association credit unions, township budgets, and irrigation associations, have dried up. Why has Han emerged as such a powerful leader? The sinister explanation is that he is the connection to a new top-level patron distributing resources. We know that China is trying to penetrate Taiwanese society in exactly this sort of way. Admittedly, I have no evidence for this suspicion, but it seems to me a more convincing explanation of the factions’ sudden rush to embrace such an unlikely figure.

These two pillars of Han’s support, the deep blues and the local factions, are pretty solid. That might seem unexpected, since I just argued that one of the sources of Han’s support is the idea that he can win. One might expect that bad polls would puncture that bubble, and all of his support would evaporate. I do not expect his support to be so tenuous. The deep blue voters have seen him deliver a miracle once, and I suspect they will be reluctant to abandon him without definitive evidence in the form of a losing election. After all, switching to another candidate probably means accepting some (painful) adjustments to China policy. They are all in on him. Likewise, many of the local factions have made serious commitments to the Han campaign; they are also heavily invested in Han. Moreover, there is a big payoff to being a core supporter of a winner. Many will take their chances at that prize rather than becoming peripheral supporters of another candidate who (momentarily) appears to have a slightly higher chance of winning.

Outside these two pillars, however, Han’s position seems to be gradually eroding. His lackluster performance as mayor has hurt him. He doesn’t seem to be able to manage the Kaohsiung city government, so voters might wonder if he is up to the task of overseeing the much more complex and challenging central government. Moreover, at the same time he seems a bit overmatched by the task of governing Kaohsiung, he is grasping at even more power by running for the presidency. Many voters who thought he was a different type of politician may be having second thoughts. He is also hurt whenever he deals with or comments on China. Ordinary voters mostly overlooked his position on China, but they were not happy with his lack of support for the Hong Kong demonstrations or his startling reluctance to criticize One Country Two Systems. Han’s support among ordinary centrist voters – and most are not directly controlled by local factions – seems to be waning. His support among disillusioned voters has probably declined even more. His unfavorable ratings have steadily crept upward over the past six months, and he is now one of the more disliked politicians in Taiwan. He appears to have a high floor and a low ceiling.


When Terry Gou jumped into the race, I expected him to run on a platform of integrating Taiwan’s economy with China’s and sweeping aside any political obstacles (such as disagreements about One China) that might get in the way of that goal. After all, it was widely rumored that Gou had been recruited by former president Ma as a way to block Han from getting the nomination, so it seemed reasonable to expect that Gou would share Ma’s vision. Moreover, as the head of Hon Hai, the largest private employer in China, Gou has a strong incentive to ensure that Taiwan and China enjoy smooth relations.

It hasn’t turned out that way. Gou has positioned himself as a defender of the ROC who is not at all interested in unification. Rather, he has staked out a position in the center of the independence – unification spectrum that might be described as maintaining the status quo indefinitely. The DPP currently also wants to maintain the status quo, but Gou and the DPP have different ideas of what the status quo is. The DPP says that the status quo is de facto independence, while Gou believes that the status quo involves a sovereign ROC that is not an independent Taiwan. In this, he is most similar to Lee Teng-hui’s position in the 1990s.

Upon entering the race, Gou called for two countries. This was probably not a sophisticated and carefully considered position, since he backtracked fairly quickly in face of the outcry from orthodox KMT voices. Rather, it probably reflects a pragmatic business approach in which he has never really had to parse his words with the fanatical precision that politicians do. Gou retreated to a version of the 92 Consensus, but he is not all the way back to Ma Ying-jeou’s 92 Consensus, much less to the slightly more China-friendly version most of the KMT is holding today. During the first presidential debate, Gou flatly stated that the 92 Consensus must be one China, each side with its own interpretation. Moreover, the second part of the formula – each side with its own interpretation – was the more important part of that formula. Without that second clause, Gou insisted that there can be no consensus. If Gou is elected and insists on this, it might effectively torpedo the utility of the 92 Consensus for lubricating official interactions with the PRC since the PRC has doggedly ignored the existence of the second clause. To them, the 92 Consensus is only One China. The KMT has ignored the PRC’s redefinition, waving it away as the PRC’s interpretation. After all, they say, each side has its own interpretation. Gou is suggesting that the 92 Consensus requires the PRC to admit that the ROC interpretation exists and has some degree of legitimacy.

Eric Chu is the third KMT candidate. Chu is professional, respected, has thought through various policy positions, and – unlike Han or Gou — generally ready to be president. KMT voters seem to be uninterested. He continues to lag in a clear third place in almost all polls.

Here are my quick debate impressions of the first two debates. In the first debate (which was the most important, both because it was about sovereignty and because people pay more attention to the first debate), Han seemed nervous. It seemed like he was trying to look and act like what he thought a presidential candidate was supposed to look and act like, but he didn’t know exactly what that was. He spoke in vague and mostly unsatisfying platitudes. The second debate was on social issues, and Han was clearly much more at home. He still didn’t have any concrete policies, but he was quite comfortable complaining about the current state of society. Gou is also clearly an amateur. Neither one of his debate performances inspired much confidence. However, I think he did much better in the next day’s newspapers than on stage. In each debate, he put forward a few ideas that dominated coverage. In the first, it was his statements about the nature of the ROC. At one point, he addressed fears that the PRC would use his company as a hostage by saying he could withdraw from China any time: Who’s afraid of who? In the second debate, he promised to pay all costs for children under the age of six. Chu was polished and prepared; he was the only person who looked anything near presidential. He did well in the first debate, both by subjective impressions and by Google searches. On China, he took a position between Han and Gou, saying that Taiwan’s democracy is non-negotiable and that Taiwan shouldn’t be afraid to offend China by standing up for democracy. The polls don’t show him making much headway, though. There were two minor candidates. Former Taipei county magistrate Chou Hsi-wei is running as the representative from 1982. He promised to unify China under the ROC. In his closing statement, he talked about how wonderful it would be when the country became an international superpower! I was disappointed that he didn’t go all the way and use the old slogan, “unify China under the three principles of the people.” Chang Ya-chung, an extreme unification ideologue, is running as what DPP supporters might call “the surrender candidate.” Chang argues that war would be so horrifyingly catastrophic that Taiwan must do anything to avoid it. Naturally, this means moving quickly toward a political settlement with China. Chang also promised that he would strictly prohibit government officials from openly advocating Taiwan independence. [Note: I am a government employee, so let me just politely say to him, “Fuck off.”]


Let’s go to the polls. Depending on which question you ask, the polls are either wildly different or remarkably consistent. If you are interested in inter-party politics, the polls are all over the place. In the course of two weeks, major polls have shown Tsai way ahead of the field and way behind the other candidates. Ko is either in the high 20s, in or close to first place, or struggling to maintain 20% and clearly in third place. The craziest outlier (at least I think this one is the outlier) is a TVBS poll showing Tsai at 35-37% in the three-way races and 45-50% in head to head matchups with the three KMT contenders. She is at least 8% ahead of her closest competitor in all of those matchups. Remember, TVBS polls usually show a lean to the KMT. TVBS seems to publish at least one head-scratcher every year. I suppose this is evidence that they aren’t herding (adjusting their results to keep them in line with other polling results), which is commendable. In contrast, United Daily News and Apple Daily polls published this week both show Tsai languishing in the low 20s in three-way races and losing all the head-to-head matchups by a considerable margin. The polls are just all over the place, and I have no idea who is ahead and who is trailing.

KMT Tsai Ko DK
Apple 6/19 Han 32.8 27.1 24.3 15.8
Apple 6/19 Gou 27.2 25.4 24.2 23.2
Apple 6/19 Chu 26.8 26.6 25.5 21.1
Apple 6/19 Han 39.1 40.3 20.6
Apple 6/19 Gou 40 32 28
Apple 6/19 Chu (+0.2%) ?
Apple 6/26 Han 33.7 26.7 22.9 16.7
Apple 6/26 Gou 27.3 25.2 21 26.5
Apple 6/26 Chu 26.2 26.6 25.1 22.1
Apple 6/26 Han 38.8 40.6 20.6
Apple 6/26 Gou 36.9 33.8 29.3
Apple 6/26 Chu 35.6 37.3 27.1
Apple 7/3 Han 35.8 24 23.1 17.1
Apple 7/3 Gou 32.5 20.1 21.4 26
Apple 7/3 Chu 29.5 22.8 22.6 25.1
Apple 7/3 Han 42.1 33.9 24
Apple 7/3 Gou 43.8 26.1 30.1
Apple 7/3 Chu 39.3 30.9 29.8
TVBS 6/22 Han 29 37 20 14
TVBS 6/22 Gou 24 35 21 20
TVBS 6/22 Chu 21 36 23 20
TVBS 6/22 Han 36 50 14
TVBS 6/22 Gou 35 45 20
TVBS 6/22 Chu 33 48 19
UDN 7/1 Han 35 22 26 17
UDN 7/1 Gou 31 19 24 26
UDN 7/1 Chu 25 21 28 26
UDN 7/1 Han 43 38 19
UDN 7/1 Gou 45 30 25
UDN 7/1 Chu 42 33 25


However, right now it isn’t that important how Tsai or Ko are doing. The immediate question at hand is the KMT nomination, and the polls are stunningly consistent on the state of that race. Let me state my conclusion first: Han is a prohibitive favorite to win the polling primary and, thus, the KMT nomination.

The single-most important fact is that, in every poll, Han does better than Gou in the three-way race. Forget Tsai and Ko for a minute and just focus on Han and Gou. Han’s support is always 3-5% higher than Gou’s. For example, in the UDN poll, it is irrelevant that Han beats Tsai by 13% while Gou beats her by 12%. What matters is that Han gets 35% and Gou gets 31%, so Han leads Gou by 4%. This is not how we usually look at polls, but this is how the KMT will calculate its interparty question, which accounts for 85% of the KMT’s nomination decision. Again, in five polls with wildly different inter-party outcomes over the past two weeks, Han always beats Gou by a consistent margin of 3-5%. This pattern has also emerged in every other poll I’ve seen over the past month or two, though Han’s lead has sometimes been bigger.

The other 15% of the nomination is determined by an intra-party comparison. Here, Han and Gou are more tightly matched.

Han Gou Chu Chang Chou
Apple 6/19 27.9 25.6 11.4 0.0 1.2
Apple 6/26 29.0 21.4 13.4 0.3 1.0
Apple 7/3 29.6 28.4 12.3 0.4 0.4
TVBS 6/22 27 29 19 0.4 1
UDN 7/1 30 29 15 1 1

This section only accounts for 15% of the total score, so Gou would have to win this question by a huge margin to make up for his deficit in the other section. For example, if Han wins the first section by 3%, Gou would have to win this question by at least 17%. In other words, the first question is decisive, and this one only matters if they are tied in that first question.

This all implies that Han has enjoyed a consistent, though not overwhelming lead. Apple has helpfully calculated the nomination scores based on its poll results. The race has tightened up in the last week, but Han is still clearly in the lead.

Han Gou Chu
Apple 6/19 38.6 32.6 28.8
Apple 6/26 39.7 31.6 28.7
Apple 7/3 37.4 34.3 28.3

Yet I am arguing that Han is a prohibitive favorite to win. There is one more factor to consider. In the DPP polling primary, Tsai had been leading slightly in the final polls, but she ended up with a comfortable victory in the polling primary. Tsai was favored by DPP supporters by about 2 to 1 over Lai. This group was almost certainly overrepresented in the actual results because DPP supporters mobilized themselves. They sat by their phones at home, stayed out of the shower during polling hours, answered unknown numbers instead of ignoring them, and once they answered the phone they didn’t hang up on the pollsters. My guess was that they were over-represented by about 50%, and this expanded Tsai’s margin over Lai from about 2% to about 8%. The same effect will occur in the KMT’s primary. KMT supporters will be much more motivated than everyone else to answer their phones. Moreover, since the KMT is only calling landlines (unlike the DPP which also called cell phones), mobilization to stay home is even more important. This is a big advantage for Han. Most polls don’t publish preferences broken down by party support, but the few results I have seen all tell the same story. KMT supporters overwhelmingly prefer Han. For example, the Apple 6/19 poll published the following:

Chu Han Gou No response
overall 11.4 27.9 25.6 35.1
Pan blue 8.7 52.7 28.0 10.6
Neither 7.9 19.5 23.7 48.9
Pan green 18.1 6.2 26.6 49.1
No response 1.9 9.7 5.3 83.1

Among blue camp supporters, Han crushes Gou by 52.7 to 28.0, roughly 2 to 1. Gou’s support comes mostly from neutral and green respondents. In the general election, that would make Gou a strong candidate; in the polling primary it is a disaster for him. The KMT primary looks eerily like a mirror image of the DPP primary.

The media is making a big deal out of how close this race is. However, I expect Han to win the polling primary by roughly 10%.

Before moving on, I want to pause for a minute to think about an alternate world. What if the KMT had decided to use a head to head matchup with Tsai instead of the three-way race for the (85%) interparty question? Gou does much better in the two-way question; he is usually tied or leading by slim margin. Han’s advantage among blue voters would probably still swing the polling primary his way, but it would not be a sure thing. By choosing the three-way race, the KMT basically rigged the outcome in favor of Han. I used the word “rigged” on purpose, because the KMT has been enthusiastic about applying it to the DPP for exactly the same choice. I’m not sure Gou’s team had any polling experts, but I am damn sure that Han’s team did. They knew exactly what they were doing when they chose this question, and I suspect KMT chair Wu Den-yi and his crew were in on it as well.


The fact that I think Han will cruise to victory does not imply that I think he is the stronger candidate. In fact, I think the data pretty clearly show that the KMT is more likely to win the general election if they nominate Gou. [Note: I am looking at current data. I’m not considering gaffes, international developments, scandals, or anything else that might happen between now and January.]

There are a couple of numbers that I think are significant. First, in the head-to-head races, Gou always does better than Han, relative to Tsai. Let me edit that big table from above:

KMT Tsai DK Margin
Apple 6/19 Han 39.1 40.3 20.6  -1.2
Apple 6/19 Gou 40 32 28  +8.0
Apple 6/26 Han 38.8 40.6 20.6  -1.8
Apple 6/26 Gou 36.9 33.8 29.3  +3.1
Apple 7/3 Han 42.1 33.9 24  +8.2
Apple 7/3 Gou 43.8 26.1 30.1  +17.7
TVBS 6/22 Han 36 50 14  -14
TVBS 6/22 Gou 35 45 20  -10
UDN 7/1 Han 43 38 19  +5
UDN 7/1 Gou 45 30 25  +15

Gou is always 5-10% better than Han. That is a pretty significant difference. The easiest way to see what is happening is to eyeball the difference between the three-way and two-way races. That is, what happens to Ko’s support if he doesn’t run? When Gou (or Chu) is the KMT candidate, about half of Ko’s support goes to the KMT and about half goes to the DPP. However, if Han is the KMT candidate, about two-thirds of Ko’s support goes to Tsai and only one-third goes to Han. I think we are starting to see the effects of Han’s increasingly high disapproval. People who like him absolutely love him, but (the increasing number of) people who don’t like him are fairly unlikely to vote for him.

Another thing to look at in the above table is the “don’t know” column. When Han is in the race, voters mostly know who they will vote for. When Gou is in the race, the number of undecided voters goes up by 6-10%. That is, there are a lot of people who still haven’t made up their minds about Gou.

I believe this is not simply because Gou is a newcomer to politics and unfamiliar to many voters. Rather, this is a direct result of Gou’s positioning on sovereignty. Gou has placed himself much nearer the center of the political spectrum than Tsai, so there are a lot of light blue or neutral voters who are unhappy when Han plays footsie with China and who are intrigued when Gou criticizes the Red Media or One Country, Two Systems. In a general election, Gou can appeal to a much bigger audience that Han (assuming he can hold the deep blues).

To use Donovan Smith’s vivid formulation, Han has a high floor but a low ceiling. In a three-way race in which Ko maintains his support, Han’s high floor might be sufficient or even an advantage. However, if Ko doesn’t run or if his support fades, Han’s low ceiling could be a big problem for the KMT.