Today’s quake and 921

As you’re probably aware, there was a big earthquake in Taiwan today. Most of the news stories mention that it’s the biggest earthquake Taiwan has experienced in 25 years. At first glance, this one looks almost as big as that one. The September 21, 1999 earthquake was a 7.3 and this one is being reported by local media as 7.2. (The international media is reporting this one at 7.4, but I also remember the 921 earthquake was initially reported as a 7.6.) I was here on 921, and that was one of the most terrifying experiences of my life. (To answer the musical question, I will never forget the 21^st night of September.) So I want to make a few comments about how that experience was different from this earthquake.

The 921 quake was much more destructive. There were two things that are not reflected in that 7.3 number. First, it lasted a long time. Earthquakes generally have some softer periods has some harder periods, but the really intense shaking only lasts 10 or 20 seconds. The intense part of the 921 earthquake lasted well over a minute. Second, most earthquakes shake side to side. The 921 quake shook up and down, which is a much more devastating movement. Initial reports are that four people have died in this earthquake. More than 2000 people died in the 921 quake. We are seeing pictures of the same few buildings that are precariously leaning over. In 1999, there were hundreds, maybe thousands of buildings that were structurally ruined and had to be torn down. Actually, many of them didn’t need to be torn down because the earthquake completely flattened them.

My personal experience was also very different. Both times I was sleeping, and the earthquake woke me up. In the 921 quake, I was on the third floor of a fairly old four-story building, the kind you see all over the place in Taiwan. I had a few 3-foot high bookshelves, and they slammed violently against the wall repeatedly. It was the only time I have ever been genuinely afraid that the building might collapse. I had post-traumatic stress every aftershock for the next few weeks. Out of instinct, I went across the street to the NCCU campus and my unofficial home at the Election Study Center. About 10 of us gathered there and spent the next few hours reassuring each other that we would be OK. At that time, the ESC was still in an old, dilapidated building, so we had to run outside into the rain every time there was an aftershock. The power had gone out completely but one person had a battery powered transistor radio (wow, that sounds ancient now!) and said that there had been a report of an explosion at the distillery in Puli (Nantou). I thought, holy cow, this earthquake was so big that they felt our earthquake 200 kilometers away! Later, I learned that we had felt their earthquake, not the other way around. It was a massive quake.

This morning there was some significant shaking, but none of my 6-foot bookcases banged against the walls. Nothing fell off our shelves, and we didn’t have any damage. My wife and I didn’t get out of bed, and when the shaking stopped, we went back to sleep. Our building has a frame of steel girders, not the old-fashioned rebar and cement, so my experience this time might be different because I was in a sturdier building. That said, I never felt anxious, much less afraid, this time.

After the 921 earthquake, the power went out all over Taiwan. One of the electric towers sending power from central Taiwan to the north slid off a mountain, and it took several weeks to repair it. All the available electricity was reserved for the highest priority sites, such as hospitals and military bases. In late September, it was still too hot to sleep without at least an electric fan until 2:00 or 3:00 AM, so everyone was groggy and grumpy for the next few weeks. This time, my electricity didn’t go out at all. Some people lost power, but I don’t think anybody lost power for very long.

This is a political blog, so you’re probably wondering what kind of political impact this earthquake will have. I confidently predict that it will have almost no impact at all. This earthquake took place three months after the presidential election. The 921 earthquake took place six months before a presidential election. There were a lot of people who thought that the earthquake would be the defining issue of that election. After all, it upended hundreds of thousands of people’s lives. In Nantou, county magistrate Peng Pai-hsien 彭百顯 turned his back on a career of opposition politics to endorse the KMT presidential candidate Lien Chan. Lien was still the premier at the time, and Nantou desperately needed relief. In Taipei, the city government could post a deputy mayor at the single collapsed building; Nantou didn’t have enough county government employees to post a civil servant of any rank at every collapsed building. Nantou was also among the poorest county governments. The legislature quickly passed an emergency relief act, and Peng needed to be sure Nantou would get its fair share. He may have also made a political calculation that the earthquake would be the turning point in Lien’s presidential campaign. Even though Lien’s polling numbers were miserable, usually under 10% (Soong routinely had 40% and Chen usually had around 20%), there were many people who just could not fathom the KMT losing the presidency. Surely, they could mobilize enough votes with their vaunted political machine. The earthquake provided an ideal opportunity for Premier Lien to hand out lots and lots of cash in the crucial battlegrounds of central Taiwan. It didn’t work out that way. Six months later, the earthquake had faded into the background as a political issue. If you look at the election returns, is hard to discern which areas were hit hardest by the earthquakes and which areas were relatively spared. Surveys also failed to identify the earthquake as an important factor. If that traumatic experience had almost no electoral impact, it’s hard to imagine that this one will matter very much with the next selection more than 2 ½ years away.

[Edit: As a reader points out, Lien was NOT premier at the time. He stepped down from that post in 1997. Vincent Siew (Hsiao Wan-chang 蕭萬長 was the premier. Lien was VP. Still, he could have used the opportunity to go to all kinds of places and represent the government handing out cash. He didn’t do that at all.]

Another difference is in the international media coverage. Because China has become such an inhospitable place for international media, there are now many correspondents based here in Taiwan. They were right on the scene and able to report this story quite thoroughly. In 1999, there were very few international reporters based here in Taiwan. If my memory is correct, CNN flew Mike Chinoy in from Hong Kong. He went from the airport straight to downtown Taipei, reported on the single building that had collapsed there, and turned around and went back to Hong Kong. CNN completely missed the main story, which was in central Taiwan. That’s what the international news environment was like a generation ago; this sloppy reporting wasn’t unusual. The quality of international news coverage is much, much better now.

A final consideration is that Taiwan is more prepared to deal with this kind of natural disaster now than it was 25 years ago. In the aftermath of that disaster, building codes were rewritten and more stringently enforced. One of the worst stories from that quake involved an elementary school that collapsed because some of the pillars were filled with empty salad oil tin cans instead of solid concrete. More prosaically, there were a lot of buildings that collapsed between the second and third floors. When you build with rebar, you are supposed to overlap the rods so that there is not a weak point. However, rebar is expensive so many builders ignored this requirement. The standard rebar piece is about two floors high, so there was a weak point between the second and third floor which the earthquake mercilessly exposed. A lot of the shoddy housing stock from the 1960s and 1970s has now been replaced with higher quality buildings.

Let’s end this piece by going back to electricity. Someone just sent me a very interesting article. About 300,000 households lost power this time compared to 6.5 million in 1999. Still, today’s earthquake did knock several generators and transformers offline, accounting for 3.2 million kilowatts, more than 10% of the total supply at the time. Why didn’t this overload the system and lead to a more widespread system failure? This article argues solar power came to the rescue. At 7:50 (just before the earthquake), solar power accounted for 2.9 million kilowatts (10.8% of the total supply). At 8:00 AM (2 minutes after the earthquake), that figure had risen to 3.3 million kilowatts (12.2%). At 8:10, solar power provided 3.7 million kilowatts (13.2%). At 8:40, it was 4.8 million kilowatts (13.2%). By 11:40 (when most of the damage had been repaired and the regular sources were coming back online), solar power provided 8.4 million kilowatts (25.8%). In short, the government’s drive to transition to green energy has had the added bonus of making the grid more resilient!

How close were the KMT and DPP to a majority?

I haven’t had time to dig through the election results in much detail, but I did notice something that was a little bit unexpected to me. Even though the KMT won more seats, I think the DPP was actually closer to winning an outright majority.

That isn’t to say that the DPP didn’t have a terrible night in legislative elections. It did. Compared to four years ago, the DPP lost 14 seats. Actually, since they won a by-election in Nantou 2 but lost that seat on January 13^th, it was really a loss of 15 seats.

Yes, I know the DPP legally won 61 four years ago, and the KMT legally won 52 this time. But it’s silly to exclude the candidates who were KMT and DPP representatives in everything but name. Four years ago, the DPP didn’t nominate anybody to run against Chen Po-wei, Chao Cheng-yu, Su Chen-ching, and Freddy Lim, and Tsai Ing-wen urged her supporters to vote for them. In the legislature, they were reliable DPP allies. There is a big difference between having 61 and 65 members in your caucus when you need 57 for a majority, and we lived in the world in which the DPP had a comfortable majority. Likewise, in the new legislature, Chen Chao-ming and Kao Chin Su-mei are better thought of as KMT members than as true independents.

So how did the KMT win a three seat plurality?

They made the most out of their party list votes. In fact, the DPP won a few more votes than the KMT. For several hours on election night, my TV was estimating the DPP would win 14 list seats and the KMT only 12. However, the KMT did well enough with the late reporting votes to narrow that gap just enough so that both parties ended up winning 13 seats.

Taiwan’s electoral system is a majoritarian system. In most cases, the legislative majority is decided by the 79 nominal tier seats. The 34 party list seats really only make a difference if the nominal tier seats are roughly tied. In the four previous elections under this system, one party won the nominal tier seats by enough to ensure a legislative majority. This year was the exception.

Four years ago, the DPP won the 79 nominal tier seats 52 to 27 margin. This year, the KMT won those 79 seats 41 to 38. Obviously, the KMT must have won a lot more votes, the DPP must have lost a lot more votes, or both. Let’s look at the results:

According to the official results, the two main parties got almost exactly the same vote shares as four years ago. How could the results be so wildly different? Um, this is another example of why looking at official party totals is stupid. These numbers ignore all the other candidates who ran with the two main parties but not under the party labels. In every district, it was clear who the KMT-approved candidate was and who the DPP-approved candidate was, whether or not they were official nominees of those two parties. Let’s look at the results again, but this time let’s look at the votes for the candidates the parties were actually supporting.

This makes more sense. Four years ago, DPP-supported candidates beat KMT-supported candidates by 7.5% nationally. It’s not very hard to imagine how that translates into a 52 to 27 seat victory.

But wait, even using this more reasonable way of looking at the votes, this year DPP supported candidates still got 1.7% more votes than KMT-supported candidates. How did the KMT end up with more seats?

There are three answers. The first one will not surprise many long-term observers of Taiwan elections. The KMT did better in indigenous seats. Taiwan is one of the few countries in the world that over-represents a disadvantaged minority group. About 2.2% of the electorate has indigenous status, but they are guaranteed 5.3% of the total seats. In practical terms, this means that it takes fewer votes to win an indigenous seed than a regular district seat. Indigenous voters have long been a reliable KMT constituency, and the KMT won 4 of those 6 seats this year.

Second, the KMT did very well in the smallest districts. If you rank all the districts by the number of eligible voters, the KMT won 5 of the 6 smallest districts and 9 of the 11 smallest districts. I’m not making a conspiracy theory argument here. Other than the fact that Lienchiang is probably too small to merit its own seat, there isn’t really anything unfair about district sizes, at least from a partisan perspective.* for balance, I should probably also add that the KMT also won four of the five biggest districts.

[* From a normative perspective, it is a travesty that some of the districts are so much bigger than the others, even within the same city. For example, New Taipei 1, the country’s biggest district, has 387,634 eligible voters, while New Taipei 6 only has 222,016. Voters in Tamsui and Linkou should be furious that their vote is only about 60% as important as someone who lives in Banqiao.]

Third, the KMT did very well in close races. There were 11 races that were decided by 4% or less, and the KMT won 9 of those, including all three under 2%.

Here’s one way to sum up the accumulated effect of all these factors:

The 41 nominal tier KMT winners got an average of 84,798 votes.

The 38 nominal tier DPP winners got an average of 96,265 votes.

All the seats count the same, but holy guacamole that is a huge difference.

So how far away were the two main parties from an outright majority? To put it another way, how much more would the presidential candidate had needed to pull in the required extra seats on his coattails? If Hou had gotten 34.5% instead of 33.5% (and we assume that all the district candidates and the party list also got that extra 1%), would that have been enough?

The KMT won 54 seats, so they only needed three more. However, they didn’t have many close misses. Their third closest miss (after New Taipei 6 and Kaohsiung 3) was New Taipei 10, which they lost by just over 4%. Anything less than 4% probably wouldn’t have been enough, though an extra 4% on the party list votes might have gotten the KMT an extra list seat, depending on who you take the 4% away from.

For the purposes of this exercise, let’s estimate that Hou would have needed to close to 38% to deliver a legislative majority instead of the 33.5% that he actually got.

What about the DPP? They only won 51 seats, so they needed to find 6 more. It’s not difficult at all. They lost seven seats by under 3% (Taoyuan 2, New Taipei 7, Yunlin 2, Taoyuan 3, Taoyuan 6, Taichung 4, Nantou 2), and if you add 3% to their party list vote they would certainly win another seat there.

If Lai had been able to win 43% instead of 40.1%, he would probably have a majority in the legislature.

Again, I’m not suggesting there’s anything wrong or illegitimate about the results. I’m merely pointing out that the KMT maxed out its potential seats with the number of votes it won. This isn’t a proportional system; that’s what they’re supposed to try to do. They did win 54 seats, and those all count. They will probably win a lot of battles in the legislature precisely because they made their electoral votes count so effectively. The only thing I’m suggesting is that perhaps their underlying electoral performance wasn’t as strong as you might assume when you hear that they are the biggest party in the legislature.

The election results and what’s next

There were no big winners and no big losers in yesterday’s election. All three parties got something, but none of them got everything they wanted or hoped for. The DPP kept the presidency, but it got far fewer votes than the previous two elections and lost its majority in the legislature. The KMT didn’t win the presidency, but it did come in second. More importantly, it won a lot of legislative seats and is now the biggest party in the legislature. However, it didn’t win an outright majority. The TPP came in third in the presidential race, but it was a strong third. The TPP held on to almost all of its potential votes. We won’t know for sure about strategic voting until we get some survey results, but it looks like there wasn’t much strategic voting at all. However, the TPP’s biggest win was in the legislative election. They didn’t win any districts, though all their district candidates ran far ahead of my expectations. But what really matters is that their eight party list seats put them in a pivotal position in the legislature. Neither major party has a majority without the cooperation of the TPP. At the end of the night, none of the parties was celebrating wildly or sobbing distraughtly.

The KMT’s election results might be the most interesting. They have now failed to break 40% in three consecutive presidential elections. That’s not good. They will rationalize this defeat by talking about how Ko split the anti-DPP vote. However, they should be asking themselves why they were unable to defend that vote. A successful political party goes out and wins votes; it doesn’t just wait for the other party to mess up. Protest votes are not a foundation for sustained success.

Nevertheless, votes against the other major party are definitely useful in the short term. The KMT did quite well in legislative races this time, winning 14 more seats than four years ago. However, because they didn’t win an outright majority, they didn’t actually increase their power by that much. It reminds me somewhat of the DPP in 2009 and 2010, when they won a lot more votes than in the previous local election cycle. This improved result was enough to keep Tsai Ing-wen in office as party chair. But they didn’t actually win much more power; the only local government to change hands was Ilan, which is not exactly the biggest or most important local governments in the country. The KMT will have more power in the legislature, but not as much as some people might think. The next speaker will probably be from the KMT, but the TPP could probably veto Han Kuo-yu if it wanted to. The KMT will win more committee seats and committee convener seats. However, the speaker is not nearly as powerful in Taiwan as in some other countries, and the KMT will not be able to unilaterally arrange the agenda. Even if there is a working KMT-TPP coalition, the legislative rules give minority parties some power, so the DPP caucus will have a bit of influence. The KMT was hardly an afterthought in the current legislature, and the DPP will not be powerless in the next legislature.

Ko Wen-je didn’t win the presidential election. His 26.5% was more than many people expected, but it still put him in 3^rd place. Nevertheless, Ko is one of the winners in this election. He delivered nearly all of his presidential vote to his legislative party list. The 22.1% party list votes earned the TPP 8 seats. That makes the TPP the smallest party in the next legislature, but those eight seats will have a massive impact on how the next four years unfold. There are 113 seats in the legislature so a majority requires 57. The KMT has 54 seats (including the two independents who are reliable KMT votes), and the DPP has 51. That means the TPP holds the balance of power with its eight votes.

The DPP held on to the presidency. It wasn’t a pretty victory, but it was a victory. Since the presidency is the most powerful office in the ROC political structure, DPP values and programs will continue to guide most government decisions for the next four years. However, this is yet another election in which the DPP hasn’t done very well. I don’t know what percentage they want in the legislative district races, but 40% in the presidential race and 36% in the party list section, well, it’s not outstanding. Then you think about their miserable performance in the 2022 and 2018 local elections and the 33% they got in the party list section in 2020, and that’s a lot of lousy results. Perhaps Tsai Ing-wen’s two smashing victories and the two accompanying district legislative elections were the exceptions, not the standard DPP vote. The KMT is not the only party that needs to do some soul searching.

So how is the government going to work in this new world?

Let’s start with the biggest question: Who will control the executive branch?

We need to review some constitutional rules. Taiwan is not a presidential system; rather it is a semi-presidential system. The premier, not the president, appoints cabinet ministers and chairs cabinet meetings. That is, the premier runs the executive branch and is directly responsible for the day-to-day operations of the government. That doesn’t mean the president is impotent. Since 1997, the president has the power to directly appoint the premier, and no vote of investiture from the legislature is necessary. As a result, the premier has usually been a close ally of the president with the responsibility of actually implementing the president’s political vision. However, it doesn’t have to be this way. The constitution also gives the legislature the right to pass a no confidence motion against the premier. If the no confidence vote passes, the premier must resign. However, if this happens, the president has the option to either appoint a new premier more acceptable to the legislature or dissolve the legislature and call for new elections.

During Taiwan’s first experience with divided government under president Chen (2000-2008), the legislature never exercised its right to vote no confidence even though all his premiers (except for the first, who lasted less than six months) were DPP stalwarts. These were not unity governments or grand coalitions. The majority KMT-PFP coalition never voted no confidence against the DPP premiers. This was the case even though Chen and the DPP became increasingly unpopular during his second term. During Chen’s first term, the majority coalition might have feared Chen would dissolve the legislature and they might lose their majority. By midway through his second term, Chen and the DPP were unpopular enough that this probably wasn’t much of a deterrent. The KMT-PFP might have decided that it was better to just wait until the end of his term and go into the presidential and legislative elections with a clear message rather than trying to gain control of the executive branch and risking being held partly responsible for the chaos. And it’s also possible that, even though the majority coalition parties weren’t worried as a group, individual legislators did not want to face the costs and risks of running another campaign. For district legislators, the costs of running an extra campaign were probably a bigger factor than the risk of losing. For list legislators, running for reelection might not be expensive, but there was no guarantee that they would be on the next party list. In the aftermath of the Chen presidency, it seemed pretty clear to most observers (including me) that the no confidence vote was an empty threat. If the legislature didn’t use it then, when would it use it? No matter the balance of power in the legislature, it seemed pretty clear that the president would dominate the executive branch.

I’ve been thinking about this for the past few weeks and wondering whether that conclusion was too strong. If president Lai (presumably elected with a very weak plurality) were faced with a single party KMT majority (which presumably would have gotten a much higher percentage of the vote in legislative elections then Lai did in the presidential elections), would the KMT demand a premier from their party and control of the executive branch rather than face a full four years out of power? Would they automatically vote no confidence for any DPP premier appointed by Lai and risk the possibility of new legislative elections? I’m not sure the KMT wouldn’t try. And I’m not sure president Lai wouldn’t capitulate in this scenario. It’s not unthinkable that we could end up with something like the French cohabitation scenario, with the president controlling international relations and national security and the legislative majority party in charge of almost all domestic policies.

However, the actual result, with the TPP holding the balance of power in the legislature, makes this very unlikely. The TPP should be terrified by any possibility of an early election. For one thing, they wouldn’t be running with Ko Wen-je on the ballot pulling votes in for the rest of the party. The legislative candidates would be on their own, and that might not work out so well. For another thing, they can’t do any better than they are right now. Right now, they hold the pivotal position in the legislature. If there were a new election, one of the other parties might end up with a single party majority, and that would make the TPP irrelevant. Finally, they should probably want to avoid going into an election too closely identified with one of the big parties. There are a lot of examples from Europe that when a coalition of a big party and a small party face an election, the small party usually gets slaughtered. In the cohabitation scenario, the KMT would enjoy most of the spoils of office, and the TPP could probably look forward to a disastrous next election.

So I think it’s highly likely that Lai will appoint a premier of his choosing, the KMT and TPP will complain that it’s a terrible choice, and they won’t do anything about it. Lai will probably control the executive branch.

OK, but will Lai be able to pursue his agenda?

The short answer is: somewhat. But I know you want a longer, nerdier answer.

The national budget is the single most important piece of legislation every year, and the rules empower the executive branch. The executive branch writes the budget, and the legislature cannot increase the budget for individual items or shift money from one item to another item. The only thing the legislature can do is cut the budget of individual items. It is politically difficult to slash the budget too much for one item. And of course, the executive strategically puts in money that can be cut so that the legislature can claim it was doing oversight when it passes 98% of the original budget request. The other thing that empowers the executive branch is that if the legislature does not pass the budget by the deadline, the previous year’s budget automatically goes into effect. The legislature cannot blackmail the executive with the nuclear option of shutting down the government by not passing a budget.

This means that all existing programs are basically protected. However, new programs or spending increases in an existing program are a different matter. The legislature must agree to fund those, and they might not do so.

The international community is particularly interested in the question of arms sales. Arms sales are usually funded with special budgets, and special budgets are, by definition, new spending. That means that the legislature does indeed have the power to hold up arms sales. In fact, this happened during the Chen presidency. However, the international context is different today than it was 20 years ago. The threat from China is more ominous and pressing. It might be harder to run the risk of being branded as soft on national security.

Other major projects funded with special budgets might be politically impossible. For example, the huge Forward-Looking Infrastructure project that Tsai passed over fierce KMT objections in 2017 would probably be dead on arrival in the new legislature.

OK, but can the legislature control Lai through its power to make new laws or amend existing laws?

If the legislature passes a law, the executive branch has to obey that. And the legislature can pass laws. It’s a messy process, and the minority party has plenty of opportunities to gum up the works, but a cohesive majority can eventually pass things that it wants to pass. The president only has a very weak veto that can be overridden by an absolute majority. This isn’t nothing. There was a case in the Chen presidency in which the vote to override a veto was 109 to 103, but since there were 225 members an absolute majority required 113 votes, so the veto held. However, in most cases the legislative majority will prevail, and the executive branch has to accept it.

The thing is, you can’t write a new law for every little decision. There just isn’t enough time on the legislative calendar. Most decisions are made in the executive branch by appointed officials. The legislature can win big battles, but it has to choose what its priorities will be. Overall, legislature will be much more successful trying to block new initiatives then in trying to reshape existing institutions.

Overall, Lai will be able to run the government, but he will not be able to pursue an ambitious new agenda unless the KMT and/or TPP agree to it. And if both those parties decide that they want to make his life hell in the hopes that he will be a one term president, these next four years might be rocky.

A few quick thoughts on the election eve rallies

I suspect not many people will read this post since the election results will be coming out in a few hours.

As you probably know, Ma Ying-jeou was not invited to speak or even be present at the final KMT rally. That’s just so bizarre to me, Ma not being welcome at a KMT event. He has played such a strange role in this last month, being the motivating force behind the blue white attempted joint ticket and then totally failing to push for it at the actual meetings. And now, giving this tone deaf interview to DW just a few days before the election. I’m still trying to figure out what to think about these events and Ma’s place in the current KMT.

Other than that, the KMT had a pretty successful rally. The crowd looked a little small on YouTube, but the drone pictures in today’s newspaper showed that there were almost as many people in another section that the YouTube video didn’t capture.

The TPP had a rollicking event in front of the presidential building. They had a big, boisterous crowd. If you were thinking that Ko’s candidacy was hopeless and you should vote strategically for someone else, this event might have changed your mind. At the very least, you came away thinking that Ko’s core supporters weren’t about to abandon him.

The DPP rally also seemed pretty enthusiastic. It’s hard to estimate crowd sizes since the spaces were such different shapes and it’s hard to tell how densely people are packed in. However, if I had to give you an answer, I might say that the DPP event was the smallest.

Bi-khim Hsiao give a barnburner of a speech, and Su and Tsai were pretty good. However, the presidential candidate who was the closer, and Lai’s speech was awful. It was completely disorganized, and never made a case for why you should vote for him for president. He spent the first few minutes telling people to go out and vote. That’s what you do at the end of the speech, not the beginning. About 10 minutes in, we finally started talking about substantial issues, that is to say, it was about 8 minutes too late. And he didn’t stick to that topic very long. Then he went back to standard “everybody go out and vote” rhetoric. I couldn’t believe this was his election eve message. I know he can do better; his speech on Thursday night was much, much better.

About that Thursday night speech in front of the presidential building: it reminded me of something I’ve heard before. He started his speech by talking about the history and significance of Ketagalan Blvd. Presidents Tsai gave a similar speech four years ago on election eve. I went back and rewatched that speech just to make sure I wasn’t imagining the similarities. It turned out that the similarities were limited. Tsai’s speech was brilliant. It might be the best outdoor speech I’ve ever heard her give. Watch the other people behind her on the stage react. Some of the young people are crying at times. And Su Tseng-chang, who for my money is the best outdoor speaker Taiwan’s democracy has ever seen, starts grinning early on and is absolutely beaming by the end. Game recognizes game. Lai’s version was a pale imitation. It was like me trying to retell a comedy routine that I heard when I was a teenager. It just isn’t the same thing at all.

To me, the fact that even Lai’s better speech couldn’t hold a candle to Tsai’s closing message four years ago is a vivid illustration of the difference in quality between these two. Remember, Lai is supposed to be a pretty good speaker, and Tsai is not a natural outdoor orator. But even in this area, there’s no question who is better.

A few interesting district races

I have barely written anything at all this year about the legislative races. So I feel I should write a little something on this topic before Election Day.

The DPP has won majorities in the last two general elections. There are 113 seats in the legislature, so you need 57 for a majority. In 2016, the DPP won 68 seats, and its ally, the NPP, won an additional five seats. That’s a comfortable majority. In 2020 they didn’t do quite so well, winning 61 seats. By this time, the NPP was not really an ally anymore, so we can’t count their three seats in with the DPP bloc. 61 seats is a working majority, but it isn’t comfortable. Fortunately for the DPP, they managed to stay fairly unified, so they were able to pass most of their agenda.

The common thread in those two elections was that the DPP won the presidency by a large margin. In 2016 Tsai got 56% of the vote; in 2020 it was 57%. Not everyone who voted for Tsai also voted for the DPP district legislative candidate, but a lot did. Tsai pulled several legislative candidates into office on her sizable coattails. [Aside: I think of Tsai as more of a northern politician and Lai as more of a southern politician, but let’s take a minute to note that the DPP swept every seat in the South in both of Tsai’s victories.]

No one is expecting Lai to win by such a significant margin. Even if he does win, most people think that he will get under 50%. Most of the legislative races are basically one-on-one contests, so most winners will need to get a majority. The key will be how TPP voters break in the legislative races. The TPP has not nominated a candidate in most districts, so most Ko supporters will have to choose between a KMT and DPP legislative candidate. Four years ago, most TPP party list voters supported the DPP district candidate. This year it seems likely that more of them will go for the KMT candidate. In short, there probably aren’t going to be enough votes for many of the DPP candidates.

I’m going to present three types of seeds. First, I’ll who did some districts that I think are very likely to change parties. Second, I’ll go through a few districts that I think are more likely to change hands than to stay with their current party. Finally, I’ll look at two districts that I find particularly interesting.

Very Likely to Change Hands

Taitung County

This is a traditional KMT stronghold, but the DPP’s Liu Chao-hao 劉櫂豪 has somehow managed to win it twice in a row. However, he lost the nomination polling primary to longtime rival Lie Kuen-cheng 賴坤成. Liu did not accept defeat well; he is running as an independent. It was already a difficult district for the DPP, and now it is nearly impossible. The KMT will win this seat.

Taipei 4

Four years ago, incumbent KMT legislator Lee Yen-hsiu 李彥秀 lost this seat in a very close race to the DPP nominee Kao Chia-yu 高嘉瑜. This year is a rematch, with Lee trying to regain her former seat. Kao has had a fairly stormy term. A lot of green voters have never forgiven her for playing footsie with Ko during the 2018 mayoral campaign, and she’s made more headlines for being cute than for substantive achievements. Lee meanwhile, won a seat on the city council with an overwhelming vote, and looks very, very strong.

New Taipei 8

At least one member of the extended Chang-Chiang-Chao 張江趙 family has held a seat in the legislature from Zhonghe since 1992. Years ago, I wrote about factional politics in this district, and this clan has been at the center of it for decades. Zhonghe is traditionally a very blue area, with lots of military villages and mainlanders. This is slowly changing as the military villages are being rebuilt and more and more outsiders are moving into the area, but Zhonghe remains more blue than green. The DPP has somehow won this seat in the last two elections, a fact that I attribute to Tsai’s large coattails and their local candidate, Chiang Yung-chang 江永昌. However, he is not running for reelection. Instead of a candidate with deep local ties and a long record of winning elections here, the DPP is nominating an outsider who has never won anything. Justin Wu Cheng 吳崢 lost races for the Taipei City Council in 2018 and 2022, the first time representing the NPP and the second time as an independent. He is young, handsome, and very well spoken, and he is about to get steamrolled by the KMT nominee. Chang Chih-lun’s 張智倫 father is Chang Ching-chung張慶忠, who served three terms in the legislature. (Yes, that guy. He played a major role in sparking off the sunflower movement.) His mother Chen Chin-ting陳錦錠 has served 8 terms in the Taipei County / New Taipei City council. I will be shocked if this race is even close.

The KMT is a near lock to win those three races. The next set of districts are ones that the incumbent party will probably lose, but it still has a fighting chance.

Somewhat likely to change parties

Keelung

The incumbent DPP legislator Tsai Shih-ying isn’t running for reelection. Keelung leans blue, but Tsai managed to win two terms through a combination of president Tsai’s coattails, lousy KMT candidates, and third party candidates who took KMT votes. This year the DPP is nominating their third choice. Their first choice was for Tsai to run for reelection, but he announced early on that he wouldn’t do that. Then the person who won the nomination had to withdraw due to a scandal. So they turned to Cheng Wen-ting 鄭文婷, a city councilor. She seems like a fairly generic DPP politician. The KMT candidate Lin Pei-hsiang林沛祥, on the other hand, has a stunning pedigree. His father, Lin Shui-mu 林水木, was mayor from 1989 to 1997. His mother, Hsu Shao-ping 徐少萍, served 6 terms in the legislature. On paper, this seems like an easy KMT victory. The only reason I’m not putting it up with the previous three races is that this isn’t the first time Lin Pei-hsiang has dipped his toe into politics. He has tried several times before, failing to win the KMT nomination each time. Given his family background, losing at that stage seems impossible. I wonder if there’s something about him that turns people off.

Taichung 5

Four years ago, when the DPP won this race, it was perhaps the biggest upset in the country. This district had reliably voted blue in nearly every previous election; it was usually the bluest part of the city. The longtime legislator Lu Hsiu-yen 盧秀燕 had just been elected mayor, and she was replaced in the by-election with an easy victory by another KMT stalwart Shen Chih-hui 沈智慧. Chuang Ching-cheng莊競程 had no previous electoral experience, so his victory over Shen in 2020 came completely out of the blue (pun intended 😊). Chuang hasn’t been a particularly notable legislator; I can’t say much either positive or negative about his tenure in office. Maybe he’s been doing a great job at constituency service. To be honest, I can’t tell you much about his opponent 黃健豪either. Still, until proven otherwise, I’m going to assume that the default choice for this district is KMT.

Other districts that might flip:

Taichung 2 (incumbent party: DPP)

Taoyuan 6 (independent, but cooperates with DPP)

Taipei 5 (DPP)

New Taipei 1 (KMT)

New Taipei 12 (DPP)

Taoyuan 4 (KMT)

Changhua 1 (DPP)

Changhua 3 (KMT)

I could go on listing seats, because unlike in many other countries, not many of the seats in Taiwan are actually “safe.” Since Taiwan started using this new electoral system in 2008, most of the 73 district seats have changed parties at one point or another. I tallied it up a few years ago, and I don’t remember the exact result, but it was something like only 20 or 25 seats had never changed parties. And many of those saw at least one very close election.

The two most interesting districts

There are two districts I find more interesting than all the others.

Hsinchu City

The media hasn’t really singled out a particular race as the glamour race this year, but I would choose Hsinchu City. The incumbent is KMT legislator Cheng Cheng-chien鄭正鈐, and he is running for reelection. I consider him to be a very weak incumbent. He ran in both 2016 and in 2020. The first time, in a three-way race, he got 36.5% and lost. The second time, in another 3-way race, he got 37.0%, but this was enough to win. It’s just not a very impressive record, especially considering Hsinchu is traditionally a fairly blue city. The DPP has nominated Lin Chih-chieh林志潔, a professor at a local university. My wife thinks she’s fairly charismatic, but I’m a little more skeptical. This isn’t the resume of your average electoral juggernaut.

In recent years, third parties have done quite well in Hsinchu. The third party in both the previous legislative elections was the NPP, which garnered 16.6% in 2016 and 28.6% in 2020. And let’s not forget that the TPP won the mayoral race in 2022 with 45.0%, though much of that probably came from strategic voting by KMT supporters. I don’t think either the NPP or TPP actually has that much support in Hsinchu. I think they are drawing from the same block of voters, and those voters are anti-establishment rather than pro-NPP or pro-TPP. What makes this year particularly compelling is that both the NPP and TPP are running credible candidates. The NPP’s nominee is Handy Chiu Hsien-chih邱顯智. He was the candidate here in 2016, he is currently a legislator on the party list, and he has been the NPP party chair. He should have some votes. However, the anti- establishment bloc of voters might swing heavily to the other candidate in the race. Ko Mei-lan 柯美蘭 is officially running as an independent candidate rather than as a TPP nominee, but she’s not fooling anyone. Her brother is kind of a big deal in that party: some former surgeon named Ko Wen-je.

So you have a weak incumbent, a political novice from the other major party, one of the few credible politicians left in the dying NPP, and a presidential candidate’s sister. Cheng and Lin will probably both break 30%, Chiu will probably come in fourth place, and Ko is a complete wild card. I haven’t got a clue who’s going to win this race.

Nantou 1

It pains me to write about this race for two reasons. This race isn’t a great advertisement for Taiwanese democracy, and that hurts me because this is one of my home districts. This district has two large towns, Puli 埔里and Caotun 草屯, and I lived in Caotun for three years in the early 90s.

The DPP nominee Tsai Ming-hsuan 蔡銘軒 is a county counselor, and I’m sure he’s a nice fellow, but I’m really not interested in him at all. This race is all about the KMT incumbent, Ma Wen-chun馬文君. Her father was a member of the Provincial Assembly, and her younger brother was elected to the National Assembly, but she has had a far better political career than either of them.

The first time I saw her was in 1994, when I went to the Nantou County assembly to watch a session from the visitors’ gallery. Everything was in Taiwanese, so I didn’t understand a word that was being said. In a sea of drab middle-aged men, she was a gorgeous young (still in her 20s, just a few years older than me) woman who confidently grilled government officials. Let’s just say she stood out from the crowd.

Ma served several terms in the county assembly, and then she was elected mayor of Puli twice. When president Ma tapped Wu Den-yi (the most famous politician to come out of Caotun) to be premier in 2009, Ma Wen-chun won his seat in the legislature in the ensuing by-election. She has been reelected three times, and none of the elections were that close. Caotun usually votes similarly in presidential and legislative elections, that is, slightly blue. Puli, however, usually gives her much more support than it gives the KMT presidential candidate. Perhaps because it’s further inland and more isolated, people in Puli seem to have a much stronger local identity. Historically, they have voted very strongly for local candidates from all parties. With her blue leaning district, solid record of constituency service, and personal popularity in Puli, Ma should be a shoe-in for reelection in 2024.

But I wouldn’t be writing about this race if it were that simple.

Ma is facing two serious attacks. First, she is accused of illegally building a mansion for herself on public land. Charges of corruption of this sort are not great for your career, but they are not that unusual and they’re usually survivable. Several other candidates are being accused of similar things in this election cycle alone. (It does make the KMT’s barrage of attacks on William Lai’s far more modest house look a bit hypocritical.) Second, and far more seriously, she has been accused leaking information about Taiwan’s domestic submarine program to a foreign government. She allegedly obtained this classified information in a closed-door hearing in the legislature’s Foreign Affairs and National Defense Committee. She has denied any wrongdoing, and the KMT has said it’s all a smear campaign. For the record, she was an outspoken opponent of the submarine program, and she tried several times to cut its budget.

I can’t remember it any previous case of a legislator simultaneously facing accusations of corruption and violations of national security. This is unprecedented. I don’t have any idea how the voters will take this. These are serious accusations, and they might throw her out of office. Or they might circle the wagons, insisting that their local hero is being unfairly attacked. I’m not happy that we have the opportunity to find out.

How China Shapes Taiwan’s Elections

I assume that readers of frozen garlic are more sophisticated about Taiwan than readers of the Economist, so this won’t be very surprising to anybody here, especially coming from me.

You’ll note that I didn’t mention the TPP at all. This was partly because of the space limitations, and partly because I don’t think they’ll still matter in 10 years.

Most importantly, I guess I didn’t shave before that picture was taken. I don’t usually think of myself as having a goatee. But it seems that’s the feature they decided to emphasize when they turned the photo into stylistic picture. That, and the bags under my eyes. I guess the Economist is dedicated to the unvarnished truth, even when I would prefer a little airbrushing to make things less accurate.

batto-2024-economist-op-edDownload

Lai explains independence

[ I’m a week behind on this post. A lot of other things have gotten in the way, and I haven’t been able to get it done. However, I feel strongly that this is an important topic I needed to address at some point.]

During the presidential debate on December 30^th, most of the discussion was, in one way or another, about how to deal with China. All three candidates made statements that I did not expect based on their previous rhetoric. However, since Lai is the front-runner and a lot of the international discussion has been about him, I’m going to focus on two exchanges that shed some light on where he stands.

Most of the international discussion about Lai starts with his statement in 2017 when he was the premier that he is a pragmatic Taiwan independence worker. As a presidential candidate, he has not renounced this statement. Rather, he has tried to put it into a context that is more palatable to the general population and the international community. The most important part of this strategy is declaring that he will firmly follow the path established by president Tsai.

During the debate, Ko Wen-je challenged him explicitly on his statement being a pragmatic Taiwan independence worker.

(Note: translations are all mine. These are not word for word; I left out the non-essential parts.)

You say you’re a pragmatic Taiwan independence worker. The United States has said clearly that they do not support Taiwan independence. Heavyweight American scholars have also called on you to rescind the Taiwan independence plank in the DPP platform. As the DPP party chair, how are you going to handle this?

In the policy forum the other day, you said ECFA is not the right path for Taiwan to follow. Why didn’t you abolish ECFA when you where the premier? Now you’re the vice president; why don’t you announce we will stop implementing it immediately?

It’s only during elections that you don’t dare to talk about Taiwan independence. An honorable man would be consistent about this. You are a fake supporter of Taiwan independence and a real fraudster to get votes.

(Aside: it’s a little strange that Ko, not Hou, is the one who asked this question. Ko has spent so much time telling us he is above “ideology,” and here he is plunging right into the debate.)

(Aside2: Congratulations to Bonnie, Jessica, and Tom for breaking into domestic political arguments. In other places, I’ve even seen them equating your argument with the official American position. No one here takes anything I say in Frozen Garlic so seriously.)

Lai’s response included a definition of Taiwan independence and clarified what he meant by being a pragmatic worker for that.

The so-called American or American scholars’ opposition to Taiwan independence is a full discourse. They oppose China using military force to change the status quo in the Taiwan straits. At the same time, they don’t support Taiwan formally declaring independence. The goal is to preserve the status quo in the Taiwan straits. You are purposely misrepresenting and twisting this position and using it for your own propaganda.

In my opening statement, I said I will lead the country to maintain the status quo. Protect the country, stabilize cross straits, and promote peace in the Indo-Pacific region. That is my mission.

When president Tsai took office, she promised not to go back to the old path of conflict. That is, she would honor all the agreements made with China by presidents Lee. Chen, and Ma. There is no question about whether Taiwan will continue to honor ECFA; the only question is whether China will use ECFA for economic coercion or to try to influence Taiwan’s elections.

What is the definition of Taiwan independence? Taiwan sovereignty belongs to 23 million people. It does not belong to the PRC. The ROC and the PRC are not subordinate to each other. This is the definition of Taiwan independence. I am a pragmatic worker. My goal, as I said in my opening statement, protect Taiwan, construct the country, develop the economy, and take care of the people. These are the tasks for a pragmatic worker. That is my sincere answer to you.

This is a solid answer. The DPP’s “constitutional order” froze the Taiwan independence plank way back in 1999 with the Resolution on Taiwan’s Future, which asserted that Taiwan is already sovereign and independent and that any changes to this status quo must be ratified by a popular referendum of the 23 million citizens. That is, the DPP’s official stance is that there is no need for an independence referendum or declaration. Lai could have said he is a worker for maintaining the status quo, but that doesn’t sound quite as sexy.

So if the task of a Taiwan independence worker is not to try to get referendum or a formal declaration, what is it? Lai gives a very bland answer: protect Taiwan, construct the country, develop the economy, and take care of the people. The only one of these that could be controversial is “construct the country” (建設國家  jianshe guojia), which probably means from building more railroads and water reservoirs but could also be interpreted as establishing a new culture. If it’s the former, these are all pretty mundane tasks.

In short, Lai is telling us not to freak out about the phrase “pragmatic Taiwan independence worker.” It just means he will try to maintain the status quo and try to do all the normal things that politicians do to create a better society.

I doubt he changed anyone’s mind, but he presented the argument in a competent fashion. Maybe some people looking for reassurance will be reassured.

However, this is not the response that caused controversy. That came earlier in the debate, in response to a question from the representative of the China Times newspaper.

Remember, the China Times is a rabidly pro-unification outlet. It’s not so much deep blue as purple bleeding into red. So Lai should have expected that the question was going to be aimed at trying to entice him into saying something controversial.

Here is the question:

I have three questions.

First, do you support the ROC constitution? Please answer yes or no.

Second, the ROC constitution defines ROC as the existing territory of China, and the amendments indicate that both Taiwan and mainland China are part of the ROC. Do you agree that mainland China is part of the ROC territory?

Third, we can turn the question around. Some people say, that under the ROC’s one country two systems, the country is divided into mainland China and Taiwan-Penghu-Kinmen-Matsu. Do you agree with this? (Note: I think she meant to say the PRC’s one country two systems.)

Please tell us your true feelings about these questions, and talk about your vision for cross straits relations.

If he wasn’t already on guard, her invitation to express his “true feelings” should have been an extra reminder that this was not a friendly question.

There is an easy answer to this question: “Like president Tsai, I support the ROC constitutional order. The constitutional order includes the 1946 constitution, all the amendments to the constitution, the various laws that implement the constitution, all the judicial decisions that interpret the constitution, and the actual behavior of bureaucrats and other people. Together, these constitute the legal framework that governs our country, and I wholeheartedly support this constitutional order.” Then he could have pivoted and spent the rest of his time attacking one of the other candidates for something they said earlier. Lai did not go down this road. He needed to give his own answer with his own twist and a little bit of elaboration.

The constitution is the country’s fundamental law. It is also the guarantor of the people’s human rights. When we talk about the constitution, the first topic should be guaranteeing human rights. To guarantee human rights, you must prioritize a liberal democratic constitutional system.

If you look at the constitution established in 1946, you will be disappointed. During Lee Teng-hui’s term in office, the constitution was amended six times. What we are now operating under is the system established by these amendments, which is a system centered on developments in Taiwan.

If the constitution or other laws can unify the people, of course we use them. But if you want to use them to resolve cross straits issues, I’m afraid that is a different level.

China, the People’s Republic of China, has never accepted the existence of the ROC. China has three clear principles that it has stated internationally. First, there is only one China. Second, both sides of the Taiwan Strait belong to China. Third, the People’s Republic of China is the only legitimate government of China

If we here in this country take the ROC as our protective sacred mountain, will that bring peace or disaster? We have to think about it.

Taiwan has already moved from an authoritarian and dictatorial era into the light of open democracy. We must cherish this; we cannot go backwards. We must move bravely forward.

When I am the president, I will forge a consensus in Tawan to create a path to move forward together and both protect the country and develop the economy for the benefit of all the people.

The other candidates and the media seized on a quick sound bite: Lai’s statement that the ROC would bring disaster. How could he say such a thing? This was clearly evidence that Lai hates the ROC!

His explanation after the debate hardly helped. Lai clarified that, in his haste, he misspoke. What he meant to say was the “ROC constitution,” but he left out the word “constitution.” That doesn’t seem any better. In the original statement, he seems to be saying that, just by existing, the ROC is inviting disaster. In the revised statement, he seems to be saying that, by being a constitutional democracy, Taiwan is inviting disaster. Talk about blaming the victim!

I think I understand what Lai was trying to say, but he managed to be both not clear enough and too clear.

Article four of the 1946 constitution says, “The territory of the Republic of China according to its existing national boundaries shall not be altered except by resolution of the National Assembly.” This is the basis for the argument that the ROC constitution covers both Taiwan and mainland China. However, the constitution has been amended seven times since 1991. As part of the final round of amendments in 2005, they combined all the changes into a cleaned-up document, which is called “additional articles.” While the 1946 constitution officially still remains in effect; in practice, most of it has been superseded by the 12 additional articles. The additional articles make a distinction between the “free area of the ROC” and “mainland China.” All political power is derived from democratic mandates from citizens in the free areas. That is, the legitimacy of the current ROC regime comes exclusively from Taiwan (and the other islands governed by the current ROC government). And since this distinction between the free area of the ROC and mainland China was ratified by the National Assembly, the national boundaries have arguably already been altered to only include Taiwan. From there, it follows quite naturally that Taiwan, which is formally named the Republic of China, is sovereign and independent, and that the ROC and PRC are not subordinate to each other.

I think Lai was trying to argue that Article 4 – which he thinks is obsolete – is effectively One China. It would tie Taiwan back to China and compromise the basis of Taiwan’s sovereignty and independence. That is, the 1946 constitution, which is the version that the China Times seemed to be asking him to pledge loyalty to, would turn Taiwan into a domestic PRC question like Hong Kong. This would both invite the PRC to “resolve” this purely domestic problem, and simultaneously deprive any international actors a valid rationale for helping Taiwan defend itself. This would be a disaster for Taiwan. At least that’s what I think Lai was trying to say.

Of course, this is not the only way to understand the current ROC constitution. A more literal reading of the additional articles suggests that, if there are free areas of the ROC there must also be unfree areas. Additional article 11 makes this clear, saying “Rights and obligations between the people of the Chinese mainland area and those of the free area, and the disposition of other related affairs may be specified by law.” Moreover, the National Assembly never explicitly changed the national borders. Arguably, the current ROC constitutional order does indeed imply a One China principle

This ambiguity is why Tsai’s ROC constitutional order discourse has been so powerful. She can slide back and forth between different interpretations as necessary. She can claim that there is a thing called ROC, Taiwan, in which the ROC and Taiwan are (roughly) equivalent, that ROC Taiwan is sovereign and independent, and that the ROC and PRC are not subordinate to each other. At the same time, she can claim to be upholding the status quo, since there has never been a formal declaration changing the national boundaries or rescinding 1946 constitution. She hasn’t accepted One China, but, by championing the ROC constitutional order, she has deliberately left a loose thread hanging conspicuously out there.

To my knowledge, Tsai has never given a speech thoroughly clarifying exactly what the constitutional order means, implies, or obliges the president to do. I’ll never forget her presentation at CSIS in Washington DC in 2015 when she explained that she would respect the constitutional order in order to maintain the status quo. During the Q&A section, several people asked her to clarify what she meant and whether she would continue to honor the 92 consensus. Each time she calmly replied that the answers to those questions were very clear in her prepared speech (note: they weren’t), and the questioner should reread the transcript more carefully. Then she moved on to the next question, resisting the temptation to say anything further. It was a very disciplined performance.

The presence of loose threads like this one an important reason she’s been so successful in building stronger relationships with other democracies. She has demonstrated the discipline not to cut that thread, which reassures other countries she won’t take rash action on other matters as well. She’s not going to plunge the world into crisis with some ill-advised off-the-cuff remark.

Here, Lai revealed that he doesn’t have the same discipline as Tsai. He couldn’t resist the temptation to explain why respecting the constitution is compatible with Taiwan independence. These are the types of clarifications that legislators, pundits, and scholars can make, but presidents, cabinet ministers, and diplomas probably should leave unsaid.

I was also taken aback by a couple other points Lai made about constitutional democracy. The primary purpose of a constitution is to define how the government will be formed and operate. Human rights are a by-product of a healthy democratic constitution, not the starting point. He also implied that, if they don’t unify the people, it’s acceptable to ignore the constitution and laws.

In short, I thought this is answer was awful. I still think that Lai is sincere when he says he wants to follow the path laid out by president Tsai, but I’m significantly less confident that he has the capacity to actually do that.

the final polls

The polling blackout starts in a few hours, so let me say a few things really quickly.

There have been quite a few polls released in the last two days. My favorite poll, the Formosa poll, is not one of them. I guess they decided not to do polling over the weekend or on the New Year’s Day holiday. Here are the most recent presidential polling results which I shamelessly took from the Wikipedia page on presidential polls. I’m not making any comments about the quality of individual polls. Some of them are pretty good, others are a little more dodgy. Still, taken together, they tell a fairly consistent story.

	date of survey	method	sample size	Lai	Hou	Ko	DK/NR
				Hsiao	Jaw	Wu
				DPP	KMT	TPP
TVBS	1月1日	市話+手機	1,281	33%	30%	22%	15%
ETtoday	12月31日至1月1日	手機簡訊	1,557	38.90%	35.80%	22.40%	2.80%
鏡新聞（大地）	12月30日至12月31日	市話+手機	1,099	35.60%	24.10%	24.20%	16.20%
ETtoday	12月30日至12月31日	手機簡訊	1,295	35.40%	33.40%	22.10%	9.10%
TVBS	12月30日	市話+手機	1,318	33%	30%	24%	13%
震傳媒（山水）	12月28日至12月29日	市話+手機	1,200	31.50%	28.40%	25.20%	14.90%
聯合報	12月26日至12月30日	市話+手機	1,215	32%	27%	21%	20%
鋒燦傳媒	12月26日至12月29日	市話+網路	2,405	35.80%	27.30%	22.10%	14.80%
榮泰創數據	12月26日至12月29日	手機簡訊	10,437	34.70%	31.40%	30.70%	3.20%
影響力數據顧問（QuickseeK）	12月26日至12月29日	市話+手機	1,285	33.70%	23.70%	27.70%	14.80%
菱傳媒（皮爾森）	12月25日至12月29日	網路	12,409	36.89%	32.83%	28.64%	1.63%
ETtoday	12月27日至12月28日	手機簡訊	1,740	36.60%	33.80%	22.20%	7.40%
TVBS	12月22日至12月28日	市話+手機	1,074	37%	33%	22%	9%
三立電視（山水）	12月26日至12月27日	市話+手機	1,095	30.90%	27.90%	23.80%	17.50%

All of them show Lai leading. In most of them, Lai’s lead is under five points. However, there are a couple in which he leads by double digits. In all but one poll, Hou is in second place. His lead over Ko is usually more than five points, and in some polls it’s in double digits. However, there are a few in which the two are very close and one in which Ko leads Hou.

In 2016 and in 2020, Tsai led the final polls by large margins, often by more than 20 points. Of course, there is always uncertainty in democratic elections, but we could go into Election Day fairly confident about who would win, and the main question was how big her victory would be. This year is not like that.

The polls suggest that Lai is the most likely to win, but it is no sure thing. Likewise, the polls suggest that Ko will finish last, but that’s also not written in stone. Hou is most likely to finish second, but I wouldn’t bet my house against him either winning or coming in last place. This is one of those elections where we have to wait until they count the votes. The crystal ball is pretty cloudy.

And remember, there are still 10 days to go, and we won’t have any polling during that period. Lots of crazy things can happen in that time to make all this previous polling irrelevant.

Presidential debate — in English!!

I was starting to write a post about the candidates positions towards cross straits or international relations as expressed in the various residential policy forums and the presidential debate. They don’t always say the same things to the domestic audience as they do to the international media.

Then I stumbled across this video, which is spectacular. TaiwanPlus has done a simultaneous translation of the entire presidential debate. Simultaneous translation is really, really hard, and they have done a pretty good job at it. There are a few places where I would translate things differently, but that’s because I’ve had time to think about what to say. Overall, this is a fabulous production. (If you want to see the debate in the original language, here is the video.)

I might still write a more complete post about this debate, but for now I’ll just say a few points.

The most surprising thing about this debate is how much of it is focused on China. In the previous policy forums, where they took turns giving three 10 minute speeches, they spent a lot more time on other domestic issues and attacking each others’ integrity. In this debate, where they had to answer questions, almost all the questions were about China. Even their opening and closing remarks had more content about China than their previous statements in the policy forums.

The exchanges between Lai and Hou highlight how differently they understand 92 consensus and interaction with China. Lai repeatedly equates the 92 consensus with one country two systems, and Hou repeatedly rejects the idea that the two are at all connected.

On a lighter note, as a political scientist, it’s almost comical to hear Hou and Ko opine about wanting a parliamentary system. They both seem to think that one of the key elements of a parliamentary system is that the president should face questions in the legislature. Um, that would increase the president’s power, not decrease it. If the president directly faces the legislature, where does that leave the premier? In the thousands of pages I’ve read on parliamentary systems, I don’t ever remember reading about this particular question. Also, it’s bizarre to hear Hou say that Tsai’s decision not to hold a formal press conference for a few years is evidence of DPP tyranny (an attack he has made many times). Yes, that was the hallmark of the authoritarian government in Taiwan in the 1950s through the 1980s: CKS and CCK didn’t hold enough formal press conferences!

If you are interested in hearing how the three candidates explain themselves to the domestic audience, it’s worth watching this entire video.

Only a few days until the blackout, and this…

The last few weeks, I have been thinking that the polling blackout was two weeks long and so it should start today, December 30^th. Fortunately, my brain is like Swiss cheese with lots of holes in it. The polling blackout is only 10 days long and doesn’t start until 12:01 AM on January 3rd, so I still have a few days to write about the current state of the polls.

There are a few interesting points from recent polls to discuss.

The headline development is from the Formosa daily tracking polls. Recall that each one of these combines the previous three days of interviews, so every 4^th day there is completely new data. On the 86^th poll to the 96^th poll (basically all through mid-December) Hou was usually within three or four points of Lai. This is what the trend line looked like after the 96^th poll, which included data from the 20^th to the 22^nd (Wednesday to Friday). It looked like a very close race, and certainly one that could go either way especially if Ko’s supporters turned to Hou.

The monthly Formosa National Affairs Survey was conducted during this period (December 20-21), and it has some very good results for the KMT. On party ID, the KMT was within 2.1% of the DPP (DPP 30.8%, KMT 28.7%, TPP 13.5%), the closest the two parties have been since the heady days of the Han Wave in early 2019. On the question of feelings (Do you have good feelings or bad feelings toward the KMT/DPP?), the DPP was 10.1 points underwater (35.4% good, 45.5% bad), its worst showing in a year. The KMT, on the other hand, was nearly even (43.0% good, 43.3% bad). This was the first time since Formosa began conducting this poll in December 2018 that the KMT had reached 40% good feelings! On the question of how respondents intended to vote, Lai still led by 4.3% (Lai 38.2, Hou 33.9, Ko 16.1%), but the underlying numbers were very encouraging for the KMT.

It was a similar story in the legislative races in the December Formosa poll. In the district races 32.6% said they would vote for the DPP candidate, 31.7% said they would support the KMT candidate, and 8.1% said they would opt for the TPP candidate. Given that the KMT will probably do better in the smallest districts where it takes fewer votes to win a seat, these numbers aren’t too bad for the KMT. The party list numbers are even better for them. KMT has 33.8%, the DPP has 30.9%, and the TPP has 13.8%. All the other parties are less than 2%. There hasn’t been a lot of polling on the legislative races, but the ones I have seen look fairly similar to this or maybe a tad better for the KMT.

Then something happened. Starting on Christmas Day, the polls look different. In the 97^th poll, Lai’s lead increased from 4.9% to 9.5%. Most of this change was from a drop in Hou’s numbers, from 32.6% to 29.4%.

My first instinct was that this was simply an unlucky polling day for Hou, and the effect would disappear in three days. These kinds of statistical blips happen all the time in polling, and the daily tracking polls are great for observing them. In landline polls, the problem is usually due to a sudden change in 20 to 29 year olds. This is the hardest group to reach in landline polls, so they’re always underrepresented. The standard way to remedy this is to weight them heavily. A young respondent might count as much as two (or maybe even more) cases in the weighted data. This sort of heavy weighting makes this group vulnerable to big swings: a few more or a few less respondents expressing support for you might change your overall support level quite a bit. For example, in the 74^th poll, Ko’s support among 20 to 29 year olds was 22.6%, and his overall support was 17.8%. In the 76^th poll (Formosa did not publish a report on the 75^th poll), Ko’s support among 20 to 29 year olds soared to 40.7%, and his overall support rose to 23.1%. In the real world, Ko’s support among young voters probably didn’t vote double in just a few days. He probably just had a very lucky polling day where several more young respondents expressed support than might be expected. Lo and behold, his support only remained around 25% for three days, and then it went back down when that really good day exited the data. The same type of thing happened in the opposite direction a few days later when Ko went from 17.7% in the 81^st poll to 14.9% in the 82^nd poll. His support among young voters dropped from 28.9% to 21.2% overnight. Again, three days later his numbers bounced back.

However, Hou’s bad Christmas Day wasn’t due to 20 to 29 year olds. He only had a slight decrease, from 20.3% to 18.0%. His bigger drops were among the 30 to 39 (-4.3%) and 40 to 49 (-5.3%) groups. These age groups are typically well represented in the raw data, so this probably wasn’t an effect of heavy weighting exaggerating a small difference.

And unlike those earlier examples, this one did not disappear after three days. We are now in day 5 of Hou’s lower numbers (poll #101). Normally I’d like to see this for a full week or more, but we don’t have that much time. The blackout starts on Wednesday, so there won’t be many more polls to look at (and I won’t be able to discuss them). We will probably enter the blackout period with the Formosa polls showing Lai with a significant lead and Ko a bit closer to Hou for second place than he was a week ago.

This change is not merely about the individual candidates. It is also reflected in the party ID numbers. If we compare the five polls before Christmas and the five polls since Christmas, you can see the clear difference in party ID and presidential support. If this change in party ID holds, it could have significant implications for the legislative races.

Average Party ID in five polls before and after Christmas

	DPP	KMT	TPP
#92-#96	27.9	25.7	13.3
#97-#101	30.9	22.5	14.0

Average support for presidential candidates in five polls before and after Christmas

	Lai	Hou	Ko
#92-#96	36.3	32.2	17.9
#97-#101	39.5	29.0	17.7

What drove these changes? As usual, the answer is: I don’t know. There isn’t an obvious answer. I’ve heard several people speculate about what might be the cause, though I don’t find any of these answers particularly compelling. There have been three presidential and one vice presidential policy forums since December 20^th. Perhaps the DPP candidates persuaded more voters than the KMT candidates that they were on the right track or ready to take office. I watched all these forums, and I didn’t see any obviously devastating attacks or critical mistakes. Mostly they just said what you expected them to say. I guess it’s possible that a lot of voters weren’t listening to the rest of the campaign and are hearing this all for the first time. But I find it difficult to believe that large masses of people who haven’t been paying attention suddenly decided to tune into a pretty boring policy forum. Another possibility is that DPP attacks on Hou’s property dealings are starting to have some effect. Alternatively, maybe there is a backlash against KMT attacks on Lai’s (quite modest) family property being an illegal structure. Another person suggested that there is a rift between Hou and Han Kuo-yu, and the Han Fans are angry at Hou. Again, I’m not particularly convinced by any of these arguments, but I don’t have anything better to offer.

I’ve been focused on Formosa polls. The Formosa polls have driven a lot of the political discussion here because they come out nearly every day and also because Tai Li-an is a very respected pollster. Still, there are other polls out there. I want to look at two recent ones that I think illustrate interesting points.

			Lai	Hou	Ko
CNEWS 匯流	Landlines only	1005	40.0	27.7	18.9
CNEWS 匯流	Cell phones + landlines	2034	35.9	27.6	25.5
Credere Media 信傳媒	Cell phones only	1066	29.5	22.7	27.8

CNEWS reports two types of results. For one, they tell us what their results were with just landline calls. And it turns out that landline-only calls from CNEWS look quite a lot like landline-only calls from Formosa. However, CNEWS doesn’t consider that to be their main result. Their headline result is the survey that does half cell phones and half landlines. This result shows Lai to be ahead by a smaller amount, but, more importantly, it shows Hou and Ko are almost tied. This difference between the two types of sampling fits nicely with the Credere Media results. This is the only survey I know of that relies entirely on cell phones. Not coincidentally, this is the only recent survey in which Ko is ahead of Hou. In fact, he’s almost even with Lai. If you were to combine the Credere Media cell phone survey with one of the Formosa landline surveys, you might get something that looks a lot like the CNEWS mixed sample. (Note: Hou’s support is roughly the same in this survey when the cell phones are added. However, in all of CNEWS’s previous surveys, Hou’s support was higher in the landline-only version than in the combined version.) (Note2: I don’t know anything about either one of these media outlets or their survey companies, so I can’t tell you whether they’re well respected in the industry or not. However, the fact that their surveys results line up so nicely with each other and with Formosa gives me some degree of confidence.)

The point here is that cell phone results are different from landline results. These are both valid sub-populations, so an accurate poll should blend the two methods. However, no one knows quite exactly how to blend them. Should it be half and half, 70% landline and 30% cell phones, or something else? We don’t really know. Moreover, the answer is probably different depending on whether you are trying to estimate the full population or predict the outcome of the election. It’s quite possible that people who answer landlines turn out at a higher rate than those who answer cell phones.

The Formosa polls are probably underestimating Ko. However, we really don’t have any idea how big that effect is. He might still be far behind in 3^rd place, or he might be very close to or even ahead of Hou.

Finally, just for fun, let’s look at a survey the KMT released today. Remember this is a party poll, not a media survey, so we don’t expect it to be neutral or even accurate. The KMT released this poll for political propaganda purposes.

			Lai	Hou	Ko
KMT	Landlines only	3258	30.2	29.2	14.3
KMT	Cell phones + landlines	5013	27.2	25.6	22.4

I think the KMT wanted to tell us two things. First, regardless of whether it is a landline poll or a mixed cell phone and landline poll, the race between Lai and Hou is very close. Second, using either methodology, Hou is always ahead of Ko. The implication is that if Ko supporters vote strategically for Hou, Lai can be beaten.

There are a couple problems with this, though. For one thing, this poll was taken over a very long time, from December 15^th to December 28^th. As we have seen, Hou did much better over the first part of that period than in more recent days. In other words, these results might be out of date.

More importantly, these results don’t give me confidence that Hou is actually in second place. The mixed sample is not half and half like most mixed samples. It is 2/3 landlines and only 1/3 cell phones. Yet, 1/3 cell phones is enough to almost make up the entire gap between Hou and Ko. Hou leads by 14.9% in the landline sample, but only 3.2% in the mixed sample. If it had been a half and half sample, Ko would probably be ahead of Hou. As I stated above, we don’t know what the proper mixture should be, but Hou is not unambiguously in second place. If I were a Ko supporter looking at this poll, I would definitely NOT vote strategically. Even the KMT propaganda poll suggests that there is a very real possibility that Ko is in second place!