Gerrymandering in Taiwan?

So we all know that the 2008 Legislative Yuan elections were a bloodbath for the DPP.  With the new MMM system and the overall 55-40 advantage for the blue camp over the green camp, the DPP ended up with only 27 of 113 (24%) seats.  Most of us assumed that the DPP made a huge mistake in agreeing to the new system.

In fact, it goes further.  Aborigines’ seats typically all go to blue camp.  In fact, in all the different sorts of national elections (legislative, national assembly, provincial assembly), the DPP has only won one seat.  (Chen Ying 陳瑩 won a legislative seat in 2004.  Her father was an old KMT politician, and she relied heavily on his networks.  Her election did not signify a new partisan regime.)  Depending on how and who you county, aborigines make up 1.5-2.0% or so of the total population.  Since they are acknowledged as a disadvantaged group with a special place in Taiwan, they have always been overrepresented in the legislature.  In the 1998-2008 legislature, aborigines had 8 of 225 (3.6%).  Now they have 6 of 113 (5.3%).  I don’t know why the DPP agreed to this.

According to the rules, each county gets at least one seat.  The four smallest counties, Taidong, Penghu, Jinmen, and Lianjiang, had 262,016 eligible voters in 2008, or 1.6% of the total SMD electorate.  In the previous legislature, they had 4 of the 168 SNTV seats (ie, 225 total minus the 8 aborigines’ seats and the 49 party list seats).  Now they have 4 of the 73 SMD seats.  That’s an increase from 2.4% to 5.5%.  In case you’re wondering, these four seats have always been solid blue.  (Well, until last month’s shocking by-election in Taidong, that is.  But more on that later.)

So according to the rules, roughly 3% of the population that is reliably blue elects 11% of the seats in the new system, up from about 6% in the old system.  That sounds like a solid case of malapportionment to me.[1] It also sounds like a significant systemic advantage for the KMT.  If the DPP wants to govern, it should need to get quite a bit more than 50% to overcome this malapportionment.

Just for fun, I took the 2008 presidential vote and cut it into the 73 SMD districts.  Since the KMT won 58-42, not much different from the blue camp’s victory in the party list (55-40), I didn’t expect a much different outcome.  In fact, it’s basically the same.  The DPP actually won 13 seats in 2008, but the green camp led the blue camp in the party list tier in 15 districts.  For the presidential election, the green camp led in 16 districts.

What about the 2004 vote?  If we wanted to know what a DPP majority would look like, this is the best we’ve got.  Recall that the DPP won the 2004 election by a whopping 50.11-49.89%.  In other words, it was basically tied.  If we assume that on such a 50-50 vote, the KMT and DPP would each get 17 list seats, and we further assume that the KMT would sweep the six aboriginal seats, the DPP would need to win the 73 districts by a 40-33 margin.  This seems unlikely, given a 50-50 vote, but that’s exactly what you get.  Chen’s vote was higher than Lian’s vote in 40 of the 73 districts.  In other words, this little exercise produces a one seat DPP advantage, 57-56.

I was absolutely shocked by this result.  Somehow a system that looks skewed in favor of the KMT is actually quite fair.  A 50-50 election produces a 50-50 legislature.  There must be something offsetting the malapportionment.

In my previous post, I pointed to 16 districts that were very close to the overall national average in the 2008 party list vote.  Chen won almost all of these districts (by very small margins).  Ten of the twelve districts in Taichung and Changhua (in Central Taiwan) are among these 16 bellwether districts, and Chen swept them all.  In other words, the DPP has an advantage in that a large group of districts tip its way slightly before the party gets to 50% overall.  This is what offsets the malapportionment.

I had never done this exercise before, but I’m sure that both the KMT and DPP did it before agreeing to the new system.  No one wonders why the KMT agreed.  They expected to benefit from it because they didn’t expect the DPP would ever get to 50% again.  In 2008 at least, they were clearly right.  The question is why the DPP agreed, and now we have another piece of the answer.  If they can get to 50% they can win power.  This system is not as skewed as it looks.

Oh, before we conclude, let’s revisit that malapportionment.  Now that the DPP has won Taidong (albeit in highly favorable conditions unlikely to hold in the future), we probably have to stop thinking of that seat as an automatic KMT victory.  Penghu also deserves a second look.  While the DPP hasn’t won anything there since 1993 and blue incumbent seems quite well-entrenched, this is not an unwinnable district for the green camp.  According to the 2008 party list vote, this is a 52-40 district.  Last December, the DPP candidate for county executive only lost by 600 votes, or about 1%.  So really there are only 8 overrepresented seats that the KMT seems sure to win, not 10.  Even the malapportionment isn’t as severe as it appears at first glance.

---

[1] If you’re really a nerd and need to know these things, I figured the Samuels-Snyder Malapportionment Index to be .0704, which makes the Taiwanese legislature a moderate, though not a severe, case of malapportionment.

Significance of the By-Elections

A month ago, the DPP swept three by-elections, picking up seats previously held by the KMT.  In a little more that a week, four more seats will be up for grabs, and the media is frothing with expectations of another DPP sweep.  I’ll have more to say about the individual races, but here I’d like to look at where these districts fit into the political landscape.

The Legislative Yuan has 113 seats.  34 are elected by party lists, 6 are elected by aborigines in two three-seat SNTV districts, and the other 73 are elected in single member districts by the plurality rule.  Technically, this is an MMM (mixed-member majoritarian) system, though with the nominal tier including both SMD and SNTV seats.  It was used for the first time to elect the current legislature in January 2008.

Most analysts assume that the party vote is a better indication of the underlying partisan structure of each district than the nominal vote, which is polluted by personal factors.  Taiwanese parties are commonly sorted into two big camps.  The blue camp is centered around the KMT and includes the PFP and New Party, while the green camp is centered around the DPP and includes the TSU.  These days, the small parties are gasping for life, so we have almost returned to the pure KMT/DPP two-party system that prevailed prior to the New Party’s splintering away from the KMT in August 1993.  At any rate, the best way to look at each legislative district is to sort them by party votes, aggregated into the blue and green camps.  Overall, the blue camp held a 55.2%-40.4% advantage.  However, the election was held at the nadir of green camp popularity, and things would certainly be much different if another election were held today.  My guess is that the blue advantage would by 50-45 or so.

Sorting the 73 districts by their blue camp party list vote, we get the following

			blue	green
1	Jinmen	金門縣	90.7	4.6
2	Lianjiang	連江縣	86.1	8.9
3	Hualian	花蓮縣	73.0	22.8
4	Miaoli 2	苗栗縣第二選區	72.5	21.3
5	Taipei County 9	臺北縣第九選區	70.4	26.2
6	Hsinchu County	新竹縣	70.1	23.1
7	Taidong	臺東縣	69.2	25.4
8	Taipei City 8	臺北市第八選區	68.7	27.2
9	Taipei County 11	臺北縣第十一選區	68.3	28.3
10	Taoyuan 5	桃園縣第五選區	65.4	29.4
11	Jilong	基隆市	65.0	30.7
12	Taipei City 6	臺北市第六選區	64.9	31.0
13	Taoyuan 6	桃園縣第六選區	64.2	32.0
14	Taoyuan 3	桃園縣第三選區	63.7	31.8
15	Taipei City 7	臺北市第七選區	63.1	33.2
16	Taipei County 8	臺北縣第八選區	63.0	33.5
17	Nantou 1	南投縣第一選區	62.8	31.9
18	Taipei City 4	臺北市第四選區	61.7	34.6
19	Miaoli 1	苗栗縣第一選區	60.2	34.7
20	Hsinchu City	新竹市	59.7	35.2
21	Taichung City 1	臺中市第一選區	59.3	36.5
22	Taoyuan 4	桃園縣第四選區	59.3	36.9
23	Taichung City 2	臺中市第二選區	59.2	36.9
24	Taipei County 12	臺北縣第十二選區	58.9	36.5
25	Taipei City 3	臺北市第三選區	58.8	37.6
26	Taoyuan 1	桃園縣第一選區	58.6	37.5
27	Taipei County 1	臺北縣第一選區	58.2	37.7
28	Taipei City 5	臺北市第五選區	57.9	38.6
29	Nantou 2	南投縣第二選區	57.6	37.4
30	Taipei County 10	臺北縣第十選區	57.0	39.3
31	Changhua 2	彰化縣第二選區	56.7	39.0
32	Taipei City 1	臺北市第一選區	56.6	39.4
33	Taichung County 2	臺中縣第二選區	56.5	38.3
34	Kaohsiung City 1	高雄市第一選區	55.6	41.1
35	Taichung County 4	臺中縣第四選區	55.6	39.4
36	Taoyuan 2	桃園縣第二選區	55.3	39.2
37	Changhua 4	彰化縣第四選區	55.1	39.9
38	Taipei County 7	臺北縣第七選區	55.1	41.0
39	Taichung City 3	臺中市第三選區	54.9	41.2
40	Taichung County 1	臺中縣第一選區	54.8	40.4
41	Taichung County 3	臺中縣第三選區	54.5	40.8
42	Taichung County 5	臺中縣第五選區	54.2	41.2
43	Taipei County 6	臺北縣第六選區	53.9	42.2
44	Changhua 3	彰化縣第三選區	53.5	40.5
45	Changhua 1	彰化縣第一選區	53.2	40.2
46	Taipei County 5	臺北縣第五選區	53.0	43.2
47	Penghu	澎湖縣	52.6	40.0
48	Taipei County 4	臺北縣第四選區	52.0	44.3
49	Pingdong 2	屏東縣第二選區	51.8	44.6
50	Kaohsiung County 4	高雄縣第四選區	51.4	45.3
51	Kaohsiung City 4	高雄市第四選區	49.1	47.6
52	Chiayi City	嘉義市	49.1	47.1
53	Ilan	宜蘭縣	49.0	46.6
54	Tainan City 2	臺南市第二選區	48.7	47.9
55	Taipei City 2	臺北市第二選區	48.6	47.6
56	Taipei County 2	臺北縣第二選區	48.5	47.7
57	Kaohsiung City 3	高雄市第三選區	48.1	48.5
58	Taipei County 3	臺北縣第三選區	47.7	48.8
59	Yunlin 1	雲林縣第一選區	47.5	47.0
60	Yunlin 2	雲林縣第二選區	47.2	46.6
61	Kaohsiung County 2	高雄縣第二選區	46.8	48.2
62	Pingdong 3	屏東縣第三選區	46.7	48.9
63	Kaohsiung City 2	高雄市第二選區	46.4	50.6
64	Kaohsiung City 5	高雄市第五選區	46.0	50.6
65	Tainan County 3	臺南縣第三選區	45.1	51.2
66	Kaohsiung County 1	高雄縣第一選區	45.0	47.9
67	Chiayi County 1	嘉義縣第一選區	44.6	49.6
68	Tainan City 1	臺南市第一選區	44.6	52.0
69	Pingdong 1	屏東縣第一選區	44.2	51.4
70	Chiayi County 2	嘉義縣第二選區	43.1	52.0
71	Kaohsiung County 3	高雄縣第三選區	42.9	52.3
72	Tainan County 1	臺南縣第一選區	39.8	55.9
73	Tainan County 2	臺南縣第二選區	39.4	56.3

The DPP won 13 districts in the 2008 election; they are marked with green English names.  The other 60 were won by the blue camp.  I want to draw your attention to the 16 districts in the middle of the table marked in red.  These are median districts.  There are 29 districts that lean more toward the blue camp, and 28 that lean more toward the green camp.  Whoever wins these 16 districts will win a majority.  (Actually, since the blue camp regularly sweeps all the aborigines’ seats, the green camp probably has to win 40 SMD seats to win a majority.  As such, they would need to win three-fourths of this group.)  In 2008, these 16 districts were all basically 55%-40% in favor of the blue camp.  Not surprisingly, the blue camp easily swept all 16 seats.

Recall that the national party vote was also 55-40, and that the gap would almost certainly be smaller today.  So just how close would these bellwether districts be?  By-elections shed some light on the situation.

The three districts that had by-elections last month and the four that will hold them later this month are highlighted in pink.  Two of the bellwether districts, Taichung County 2 and Taoyuan 3 held elections last month, and both were won quite easily by the DPP.  Does this mean that the DPP has gained enough popularity that it would sweep the 16 battleground districts today?  And the DPP even won Taidong, a place that it had never even sniffed a victory before.  (Well, never prior to its stunningly close loss in December’s county executive race.)  Not so fast there, hoss.  The KMT’s national popularity clearly has slipped, but there are also three other important factors that must be taken into account

First, the DPP had a clear advantage in candidate quality in Taichung and Taoyuan, and arguably also in Taidong.  In Taichung and Taoyuan, both DPP candidates had previously served in the legislature, having won seats in the old SNTV system.  Both ran for re-election in 2008 and lost, but both put up a good showing.  In Taichung 3, Jian Zhaodong 簡肇棟 faced a 13.7% (54.5-40.8) deficit on the party list, but only lost by 10.0% (55.0-45.0).   Similarly, Guo Rongzong 郭榮宗in Taoyuan 2 faced a 16.1% partisan deficit but only lost by 9.6%.  In Taidong County, the DPP’s candidate Lai Kuncheng 賴坤成 has run in numerous elections since 1992 and has generally acquitted himself well.  He even won a term as mayor of Taidong City, which has about half the votes in Taidong County.  So all three of the DPP’s nominees were battle-tested and formidable.

The KMT, by contrast, did not field a great roster of candidates.  I don’t think Chen Liling 陳麗玲 in Taoyuan had any electoral experience.  She had previously worked in the county government on environmental policy, which makes her a nice person and a terrible candidate.  Yu Wenqin 余文欽in Taichung was a little better.  He was mayor of Taiping City 太平市, which had about half the votes in the legislative district.  Township mayors are like major league baseball minor leaguers.  That’s where many stars come from, but not all minor leaguers are ready for the big leagues just yet.  At any rate, Yu had much less of a proven track record than Jian.  In Taidong County, things are a bit murkier.  The KMT’s candidate was Kuang Lizhen 鄺麗珍, who had been the county executive.  She had not won the previous race in 2005, her husband had.  However, he was removed from office for “irregular” behavior.  He had appointed her as his deputy, so when he lost the office, she stepped into it.  Both of them were reportedly pious devotees of the clientilistic arts, but perhaps they were a bit heavy-handed.  The KMT did not renominate her for county executive in 2009, instead turning to the incumbent legislator, Huang Jianting黃建庭.  However, in order to get her to step aside, they agreed to nominate her for the legislative seat that would open when Huang took office.  Now, it is easy in retrospect to suggest that Kuang was a weak candidate, but candidates with illicit money flying around them tend to do quite well in Taiwanese elections.  Certainly, her family had proven its ability to assemble a winning campaign team in the past.  In this case, though, I would have to agree with the conventional wisdom and argue that she was a big drag on the KMT effort.  So give the DPP a big edge in candidate quality.

The second factor is scandal.  As just mentioned, in Taidong Kuang and her husband had to deal with swirling (and credible) rumors of corruption.  In both Taichung and Taoyuan, the seats were vacant because the KMT incumbent had been convicted of buying votes in the 2008 election and had been stripped of his seat.  So both Jian and Guo benefited from a sympathy vote, having lost to a cheater.  (I have never quite understood this psychology.  The very voters who accepted the money in the first election are supposed to be the ones who cast the sympathy votes in the second election.  But if they shouldn’t need a court conviction to tell them someone was buying votes, they took the cash!  My imaginative deficiencies notwithstanding, this sympathy vote is widely held to be a real and have a significant effect on the outcome.)

Third, turnout was very low.  In the general election, turnout was just under 60%.  In this by-election, turnout was just under 40%.  We political scientists still have no idea what to make of turnout, but I’m damned sure it’s important in producing a result like this.  If turnout had been higher, I doubt the results would have been the same.  I just don’t know how they would have been different.

Where does this leave us?  I tend to think that the Taidong election result is probably just a one-time freak occurrence.  Kuang and her husband were corrupt, heavy-handed, and bad at governance (one poll named her the worst county executive in Taiwan).  The voters were simply punishing a terrible incumbent.  (Yes, I know she wasn’t technically an incumbent since she was running for a different office.  I also realize that the county executive race in December produced a nearly identical outcome with a much cleaner KMT candidate, except that the aboriginal vote swung the outcome the other way in that race.)  I just don’t think this election suddenly produced a sea-change in the electorate.  They’ll get over it and go back to voting for blue candidates next time.  If I were Lai Kuncheng, I wouldn’t get too comfortable in Taipei.

I attach much more significance to the two elections in Taichung and Taoyuan.  While the DPP won both handily (55-45 in Taichung, 58-40 in Taoyuan), I don’t think that the DPP is so strong right now that it would easily roll through the 16 bellwether districts if another general election were held today.  However, these districts are certainly not 55-40 anymore.  They are probably around 50-45 and may even be even.

Is this a stunning development?  Why yes, it is.  2012 is more than two years away, and it looks to me like a lot of KMT incumbents will view that date with dread while a lot of DPP challengers will await it eagerly.