Reforming the Recall Law

The recall vote against Kaohsiung city councilor Huang Chieh failed miserably today. Under the current law, there are two ways a recall can fail. This one failed both ways. At least 25% of the eligible voters must vote for the recall, and only 19.0% actually did so. Moreover, 10,000 more people voted against the recall than voted for it. This was a decisive repudiation of the recall, and one might wonder why it even got this far. However, perhaps the real takeaway is that the current recall law is so permissive that even someone as popular as Huang Chieh is vulnerable.

This marks the fourth major recall vote since the new law was passed in November 2016. (There have also been many recalls for grassroots offices.) Two have passed, and two have failed. This new law has proven to be a disaster, and it should be modified to make recalls much more difficult.

	25% threshold	yes	no	Pass?
Huang Kuo-chang	63888	48693	21748	no
Han Kuo-yu	574996	939090	25051	yes
Wang Hao-yu	81940	84582	7128	yes
Huang Chieh	72892	55261	65391	no

Let’s think about the rationale for fixed terms in political offices. On the one hand, you want politicians to be responsive to public opinion, so holding regularly schedules elections is a good idea. On the other hand, you don’t want a constant election. You have a contest, there is a winner, and the matter is settled for a period of time. We can all return to normal life. A recall is an emergency mechanism to be used in extreme circumstances when the elected official has violated public trust in some dramatic way such that there is overwhelming opposition to that person staying in office – including among large numbers of people who supported them in the first place. It isn’t a mechanism to overturn an electoral result you don’t like the first time someone loses a small sliver of their support.

What would be an appropriate time to recall someone? Remember former legislator Lin Yi-shih, who was guilty of sensational corruption? They fished garbage bags of cash out of the fish pond in front of his home in front of cameras from all the TV news stations. “014” (which sounds like his name) became nationally recognized shorthand for corruption. Lin was already out of office by that point, but imagine he had still been in office and prosecutors said that it would take several years to go through all the legal hurdles to remove him from office. Lin’s actions were a major breach of trust with his voters, since almost none of them voted for him with the expectation that he would engage in bribery or embezzlement. It certainly wasn’t one of the things he said he would do in his campaign slogans. Voters would certainly have been justified in recalling him. Alternatively, imagine a DPP legislator elected in a deep green district in Tainan who announced a month after taking office that she had decided to quit the DPP and join the KMT. This would mean opposing almost everything the voters had supported and supporting almost everything they had opposed. Again, this breach of trust would justify a recall.

What have the four recall targets done to violate trust with their voters? Have they fundamentally betrayed the ideals they presented to their voters in their original campaigns? The campaign against Huang Kuo-chang was fueled by people opposed to gay marriage. Huang had always supported this, so his actions inside the legislature should not have surprised anyone. The reasons given for recalling Wang Hao-yu and Huang Chieh are similarly lousy. They have acted in ways entirely consistent with how they presented themselves during their campaigns. Their recalls are fueled almost entirely by revenge for Han Kuo-yu’s recall. Han is the only one for whom there is even a plausible case. During the 2018 campaign, Han promised to put politics aside and concentrate 100% on economics. I don’t recall if he explicitly promised not to run for president before the election, but he certainly did make that promise repeatedly in the six months before he changed his mind and decided to run. Because he was running for president, he also was effectively an absentee mayor after promising to work hard to solve problems. On the other hand, being so nationally popular that your party demands you run for president is not generally seen as a breach of trust with the people who elected you. Han retained quite a lot of support in Kaohsiung. He got 611,000 votes in Kaohsiung in the presidential election, which is to say that about two out of every three people who supported him in 2018 still supported him in 2020 (for a higher office). In a telephone poll conducted just after he was recalled, 37% expressed satisfaction with his performance as mayor against 49% who were dissatisfied. That’s not great, but it’s also not a disaster. Did he deserve to be recalled? In my mind, probably not. At best, it’s a borderline case.

The current law makes it very easy to recall officeholders. This law has some roots in the Sunflower movement. After that upheaval, activists tried to recall several KMT legislators, including Lin Hung-chih, Wu Yu-sheng, Chang Ching-chung, and Tsai Cheng-yuan. Under the old law, the signature requirements for recalls were much more difficult, and Tsai Cheng-yuan was the only one who they actually managed to put on the ballot. However, the threshold for successfully recalling someone was also much higher – turnout had to be at least 50% of eligible voters. Tsai ignored the recall, and they fell far short of the required threshold. We should note that all of the recall targets lost their seats in the 2016 election. However, there is no evidence that any of them were overwhelmingly unpopular among the people who originally elected them. Still, the activists were convinced that their recalls SHOULD have succeeded, so the logical conclusion was that the threshold was too high. Many of those activists would end up in the New Power Party, which strongly supported the 2016 revision.

The other major force supporting the new law was the Taiwan independence wing of the DPP. They had never been interested in recalls, but they have always been interested in referendums. Prior to the revision, referendums failed for the same reason that recalls did. Turnout had to be 50%, and the side against the referendum simply wouldn’t vote. That meant that the side supporting the measure had to supply all 50%, and this had always been an insurmountable barrier. Ok, but what do recalls have to do with referendums? ROC political theory puts them in the same basket. In his Three Principles of the People, Sun Yat-sen said that people have the right to “election, recall, initiative, and referendum” 選舉罷免創制複決 and this phrase is the title of Chapter 12 of the ROC constitution. For most of ROC history, these “sacred” rights were ignored or grossly violated, but people (ironically, including people who would vomit at the idea that they were following ROC ideology) have internalized the idea that they are somehow related. It wasn’t much of a stretch to adopt the same thresholds for referendum and recall. The 2016 reform was based on the idea that the problem was that the people against the proposal were killing it by “unfairly” not turning out to vote, so they simply cut the threshold in half. Now, all proponents have to do is produce more “yes” votes than “no” votes and supply “yes” votes equal to 25% of the eligible voters.

This new law has fundamentally changed the strategy of recalls. In the past, officeholders could generally just ignore recalls. With the lower threshold, every recall has to be taken seriously. It is now much easier to recall someone than to elect someone.

Turnout in Taiwanese elections is usually 60-75%. In a single-seat race, if the winner gets a majority, that usually comes to 35-45% of eligible voters. Critically, the loser might also get 25-35% of eligible voters. That is, the losing side might not have enough support to win a general election, but they might have enough support to recall the winner from office.

Of course, it is difficult to turn out all your potential supporters. Mobilization is hard. Campaigns are enormously complex and expensive, and they take lots of time and energy. This is especially true in a recall campaign when the entire society hasn’t been building toward the excitement of a national election for several months. Just because the opposition theoretically has enough potential support to successfully recall the winner doesn’t mean they can actually produce it.

Politicians facing a recall have two choices: ignore it or fight it. If you ignore the recall, you are betting that the other side can’t mobilize 25% of eligible voters against you. You also are declining to divert precious resources – time, energy, manpower, money – away from your normal political priorities toward this unwanted political fight. However, since the other side usually has a pool of at least 25%, so you have to be confident that they are not competent enough to mobilize those votes.

Fighting – trying to mobilize your own supporters so that “no” votes outnumber “yes” votes – is also an unattractive option. As noted, mobilization is expensive. You’d almost certainly rather spend your time and money pursuing the normal duties of your office. Moreover, it’s nearly impossible to mobilize your side without also helping to mobilize the other side. When you put up ads or hold marches, you might excite your own supporters to go out to vote. However, you are also reminding people from the other side that there is an upcoming recall vote. The more you mobilize, the more you help them get to that 25% threshold. This means that you really have to think you can defeat the recall the other way, by getting more “no” votes. This is usually also problematic. Most recall targets are at least a little unpopular – that’s usually why they were targeted in the first place. Even if they have a partisan advantage in their district, their party supporters might not enthusiastic about turning out. Everyone is busy and has other things to do. If the opposition is enthused (as oppositions usually are) and voters are ambivalent about the incumbent (as they often are), it is entirely plausible that a recall could succeed. It’s also entirely plausible that the incumbent would win the seat again in a hypothetical general election held the next day.

Wang Hao-yu and Han Kuo-yu both chose to ignore the recall. Their choice was easy since both of them faced districts in which the other party had a clear partisan advantage. There was simply no way they were going to mobilize enough “no” votes to defeat the recall. Their only hope was that the “yes” side wouldn’t be able to get to 25%. They both lost this bet. Huang Chieh chose to fight. Her district has a clear green majority, so she could expect that there were enough potential “no” votes to defeat the recall if she was able to effectively mobilize. (The recall campaign also made it clear that she is generally well liked in her district.) Huang Kuo-chang came from a tossup district. He didn’t really make a clear choice. He said he was fighting and tried to mobilize “no” votes. However, he didn’t really have much mobilization capacity, and his turnout was pathetic. Fortunately for him, the other side wasn’t much better. There were far more “yes” votes than “no” votes, but they fell far short of the 25% threshold so the recall failed.

What we see is that strategies are largely determined by the nature of the district. You would hope that recalls would cross party lines since everyone could agree that the lousy politician had fundamentally betrayed the public trust, but that isn’t what has happened so far. The permissive rules encourage partisan recalls against politicians who don’t really deserve this political harassment.

Han Kuo-yu’s case illustrates another problem with the current law. When an incumbent chooses to ignore the recall, there is no constructive role for his supporters. They might think the recall is disgusting, but they are discouraged from mobilizing “no” votes. After all, their best bet is low turnout. Instead, they will be tempted to try to depress turnout. In a normal election, campaigns encourage everyone to vote. They can’t be sure everyone hearing their message will actually vote for their preferred candidate – the ballot is secret – but they can assume that most people listening to them will vote the “right” way. This produces a virtuous circle in which everyone encourages voting and works to ensure that all of their supporters can vote. Recalls are different. If only one side is mobilizing, opponents can assume that everyone voting is on the other side. They have a strong incentive to set up as many barriers, formal or informal, to voting as possible. In Han’s recall, the city government tried to limit the number of polling places to discourage voting. Han’s side certainly wouldn’t have minded if a shortage of poll workers caused long lines. The Central Election Commission eventually stepped in to ensure that there would be an adequate number of adequately staffed polling places, but the incentives against good election administration are embedded in the recall law. There were also persistent rumors that Han supporters – even organized crime gangs – would monitor who voted and enforce various types of punishments. Nothing eventually came of these rumors, but a healthy democracy should not want to encourage this sort of thing. The mere idea of voter intimidation erodes trust in the system. It would be far better to give opponents a constructive role.

The current law is bad enough for single-seat districts, but it is indefensible in multi-seat districts such as the ones city councilors are elected in. The basic premise of this electoral system is that voters vote FOR someone, not AGAINST someone. If a district has ten seats, each voter gets one vote and the ten candidates with the most votes win seats. If 7% of voters love the Bewildered Alpaca Party and vote to elect a Bewildered Alpaca candidate, it doesn’t matter how the other 93% feel about her. They can vote to elect their own damn candidates. The system is designed to empower relatively small slices of the electorate. Both Wang Hao-yu (Green Party) and Huang Chieh (New Power Party) were elected as small party candidates. (They have both since left those parties, but their party switching did not seem to motivate the recalls against them.)

However, while the election law is semi-proportional, the recall law is majoritarian. To pass a recall, you need 25% of the eligible voters. To defeat a recall, you need more “no” votes than the other side can muster. Suddenly, it doesn’t matter much that the Bewildered Alpaca Party has 7% support. What matters is the other 93%. There is basically no way the Bewildered Alpaca Party can defeat this recall. This law is designed to crush small parties.

The current recall law should not be used in multi-seat districts.

A Proposal

The recall law should be modified to make recalls much, much more difficult. The goal should be that recalls will only succeed in extreme cases. Most recalls should fail.

I propose that recall targets should be able to defeat the recall by any one of three ways:

1. If the number of “yes” votes is less than 35% of eligible voters, the recall fails.
2. If the number of “no” votes is greater than the number of votes the first loser got won in the general election, the recall vote fails.
3. If the number of “yes” vote is not greater than the number of “no” votes times one plus the number of seats in the district [“no” * (m+1)], the recall fails.

Why 35%? Since most general elections have a turnout somewhere around 70%, half of 70% (ie: a majority of normal turnout) seems like a reasonable threshold. It will be difficult, but not impossible, to reach this threshold in a recall. Alternatively, this threshold could be set at half the turnout in the district in the previous general election. However, this would produce a different threshold in every district and every election. 35% is arbitrary, but it is consistent and easy to understand.

In Han Kuo-yu’s recall, there was a possibility that Han could be recalled by as few as 574,996 votes. Recall that 892,545 people voted for Han in 2018. His side complained that 574,996 people shouldn’t be able to overrule 892,545 people, and they had a good point. Perhaps the recall threshold should have been 892,546. However, there are two problems with this argument. First, Han didn’t need that many votes to win the 2018 election. He beat his main opponent by roughly 150,000 votes. Why should he be penalized for winning by so much? A more reasonable threshold would be 742,240, one more than Chen Chi-mai won. Second, the premise of a recall is that the incumbent doesn’t still have the same support as they enjoyed in the general election. If they can still demonstrate they do, in fact, have that much support, they should not be recalled. Instead of looking at the “yes” votes, we should look at the “no” votes. If Han could have turned out 742,240 “no” votes, it shouldn’t matter how many people voted “yes.”

Rule #3 is more complicated. Another basic idea is that the “yes” side has the burden of demonstrating an “overwhelming” level of support for the recall, but the meaning of “overwhelming” has different meanings in different contexts. Another basic idea is that opponents of the recall should be given a way to constructively oppose the recall within the system. Their votes should be crucial.

In normal elections, winning by one vote is sufficient. Recalls are extraordinary. Winning by one vote should not be sufficient. In a single-seat district (m=1), the recall would require at least twice as many “yes” votes as “no” votes. Most incumbents could easily manage to defend their seat under this rule. You would have to be extremely unpopular to fail to mobilize enough “no” votes to make it impossible to for other side to get twice as many “yes” votes. Of course, that’s the goal. Only extremely unpopular incumbents need worry.

This rule also makes recall possible in multi-seat districts. Theoretically, if you can get 1/(m+1) percent of the valid votes in your district, you are guaranteed to win a seat in this electoral system. Recall the Bewildered Alpaca Party. Their 7% isn’t theoretically large enough to guarantee victory in a general election, but in practice it will almost always be enough. In their ten-seat district, the other side needs to produce eleven times as many “no” votes as “yes” votes to recall their incumbent. Suddenly, the 7% of the electorate who are Bewildered Alpaca supporters can hope to defend their seat. The other 93% might really dislike them, but as long as the BAP can mobilize they have a reasonable shot at setting the threshold high enough to stay in office. This also puts the onus on the pro-recall side to produce a strong turnout to show that the incumbent really is horribly unpopular.

Let’s look at Wang Hao-yu’s recall. 84,582 voted to recall him, just a hair over the 81,940 threshold. Only 7,128 people showed up to vote “no,” but of course there wasn’t really any incentive for opponents to turn out. I think the 84,582 “yes” votes are pretty unconvincing in and of itself, especially considering how blue Chungli is. More importantly, it’s pretty unlikely that more than a handful of those 84,582 people voted for Wang in the general election. Why should their dislike of him matter now? There are eleven seats in this district, so my rule would require “yes” to twelve times as large as “no.” Even with the anemic turnout, Wang’s “no” vote is sufficient to (barely) defeat the recall. That is, he would have been able to defeat this frivolous recall while barely lifting a finger.

Wait, there’s more. Imagine that the “yes” side had produced a more impressive result. After all, Wang was elected from a tiny party; it isn’t implausible that most voters actively dislike him. In the general election, about 170,000 people voted for candidates other than Wang. What if 170,000 people had voted to recall him? By Rule #3, he could defeat that by getting 14,200 “no” votes. He got over 16,000 votes in the general election, but getting 14,000 votes in a recall is probably a lot harder than getting 16,000 votes in a recall. Rule #2 gives him another way to defeat the recall. The first loser in his district, the one who came in 12^th, got 9508 votes. As long as Wang could mobilize 9509 “no” votes, he could defeat the recall and keep his seat. I’m confident he could have managed that. In other words, Wang was probably popular enough that his seat should have been impregnable.

One more case. Let’s think about former president Chen Shui-bian. Chen was elected president in 2000 in a three-way race. The KMT was unable to coordinate its support on a single person, so Chen won with only 39% of the vote. Under the current rules,* that election wouldn’t have settled anything. We’d have had a recall effort start the day after the election, the KMT would have overturned the election, there would have been a by-election, there would have been no guarantee of a different outcome, but there would have been very bitter feelings all around. It would have been a disaster for democracy. You just shouldn’t get a do-over. You get one chance to settle on a candidate, and you have to live with the results of that decision. We don’t want to live in a world of perpetual campaigns.

[*OK, technically not. There is a separate law for presidential elections, and they did not change the recall provisions for that one. This is an illustration.]

My proposed rules would help deal with this problem. Rule #3 says that a recall would have to produce twice as many “yes” votes as “no” votes. If Chen could maintain his 39% support, he would be able to defeat a recall. That is, even in a three-way race, a recall wouldn’t be inevitable. What if it had been a four-way or five-way race and Chen had won with only 31%? Rule #2 says that as long as the incumbent can mobilize as many “no” votes as the top loser got, the recall is defeated. This might require the incumbent to get some support from one of the smaller parties, but it would be doable. That is, there would be a path for the incumbent to defeat the recall as long as they weren’t horribly unpopular.

We want elections to resolve political conflicts, at least temporarily. We don’t want to live in a world of perpetual campaigns. The current recall law should be significant modified to make recalls much more difficult.