KMT party chair election, revisited

Wu Den-yi was elected KMT chair about a month ago. At the time, one of the popular theories about his win was that it represented a victory of the Taiwan-oriented local factions over the orthodox Chinese nationalist wing. (Or, if you prefer, the Taiwanese wing defeated the Mainlander wing.) In this line of thought, Wu was inheriting the support previously won by Lee Teng-hui, Wang Jin-pyng, and Huang Min-hui. The unspoken implication was that native Taiwanese Wu would lead the KMT in a more localist direction, perhaps even becoming another Lee Teng-hui.

I’ve never been too enamored with this discourse, but I keep talking with smart people who believe it is more or less what happened. I see Wu as a firm believer in the orthodox KMT catechism. He may not be as extreme as Hung Hsiu-chu, but all of his statements and actions over the past four decades seem to me to indicate someone who is quite comfortable with the direction established by Lien Chan and Ma Ying-jeou. That is, he should be acceptable to both wings of the party. I think what happened in the chair election is that KMT members – who want to return to power – simply chose the strongest leader.

So what if I’m wrong? What if Wu was elected because the local factions mobilized to support him? What would that look like? One notable difference between the KMT chair elections in 2016 and 2017 was that there were about 50% more eligible voters and valid votes in the 2017 election. Many people have speculated that this was the result of local factions signing up new party members in support of Wu. If so, we should see a clear pattern. There should be far more new voters in central and southern Taiwan, where the local factions are strongest. Moreover, if Wu inherited and built on Huang Min-hui’s 2016 support, the increase should be greatest in places where more new people signed up for KMT membership.

Let’s look at the results of the 2016 and 2017 KMT party chair elections. The KMT tallied results for individual ballot boxes, but I can only find the full results aggregated up to the city and county level:

 

2016 KMT party chair election

		陳學聖	李新	黃敏惠	洪秀柱
		Chen	Lee	Huang	Hung
合計	139558	6784	7604	46341	78829
.
台北市	12802	756	901	2990	8155
新北市	16694	723	916	4131	10924
基隆市	1931	121	136	504	1170
宜蘭縣	2845	139	138	1110	1458
桃園市	10745	1597	787	1698	6663
新竹縣	3378	153	191	1389	1645
新竹市	1944	74	112	485	1273
苗栗縣	5204	216	265	1796	2927
台中市	11238	548	751	3484	6455
彰化縣	8074	249	325	4217	3283
南投縣	4038	159	210	1905	1764
雲林縣	4354	148	188	2627	1391
嘉義縣	3842	47	92	2765	938
嘉義市	2678	27	62	1748	841
台南市	11102	316	561	3895	6330
高雄市	15996	632	1048	4956	9360
屏東縣	6358	197	370	2808	2983
花蓮縣	3420	189	243	795	2193
台東縣	2738	121	117	1315	1185
澎湖縣	1367	86	79	361	841
金門縣	1606	74	23	132	1377
連江縣	445	33	14	65	333
海外黨部	6759	179	75	1165	5340

 

And here is the 2017 election:

	valid	Hung	Han	Pan	Hau	Chan	Wu
	有效票	洪秀柱	韓國瑜	潘維剛	郝龍斌	詹啟賢	吳敦義
合計	272704	53065	16161	2437	44301	12332	144408
.
台北市	26887	5209	1689	248	6250	1338	12153
新北市	28684	6486	1658	240	4544	984	14772
基隆市	4537	461	217	33	1586	156	2084
宜蘭縣	6055	1244	302	63	749	180	3517
桃園市	18372	4001	998	132	4067	458	8716
新竹縣	7192	955	400	70	1413	346	4008
新竹市	5253	1576	355	78	696	212	2336
苗栗縣	9671	1641	693	100	1579	445	5213
台中市	22588	3934	1121	151	4035	707	12640
彰化縣	18808	2566	889	172	2770	1002	11409
南投縣	8566	879	234	31	577	179	6666
雲林縣	8765	1062	1476	95	1390	288	4454
嘉義縣	5038	898	198	19	524	391	3008
嘉義市	4810	1078	267	63	817	675	1910
台南市	20535	4588	1262	178	3124	1882	9501
高雄市	36623	6657	2239	389	4645	1695	20998
屏東縣	14798	2377	667	108	1418	476	9752
花蓮縣	9645	2681	690	156	1424	318	4376
台東縣	5100	810	255	42	1193	114	2686
澎湖縣	2711	768	124	32	546	302	939
金門縣	2382	747	148	20	448	91	928
連江縣	574	124	53	1	97	26	273
海外黨部	5110	2323	226	16	409	67	2069

 

You will notice right away that the total number of valid votes nearly doubled, increasing by 133,146. At the same time, the number of votes won by the (supposed) representative of local factions (Huang in 2016, Wu in 2017) increased by 98,067. It seems plausible that these two shifts are related.

98,068 divided by 133,146 is .74. A reasonable interpretation is the pre-existing party members voted basically as they had in 2016, but 74% of the new party members voted for Wu. However, once you start looking at individual cities and counties, things start to break down. We expect Wu’s mobilization efforts to be most effective in central and southern Taiwan, where the local factions supposedly went all out to mobilize new party members for Wu. Assuming Wu’s increase came entirely from new members, he only won 8% of the new members in Chiayi City and 20% in Chiayi County. Those results can perhaps be explained away because Huang was from Chiayi, so they might have already mobilized for her in 2016. However, if you accept the hometown effect for Chiayi, you also have to discount the high ratio in Nantou, since that is Wu’s home. Throughout the rest of the region, the ratio does not differ markedly from the national average; if anything it is slightly lower. At any rate, Wu’s supposed share of new voters is lower in all of central and southern Taiwan (excepting Nantou) than in New Taipei (.89) and Taoyuan (.92). These are not the supposed loci of local factions in Taiwan.

		Increase	Increase
		Wu-Huang	Valid	ratio
合計		98067	133146	0.74
.
台北市	Taipei	9163	14085	0.65
新北市	New Taipei	10641	11990	0.89
基隆市	Keelung	1580	2606	0.61
宜蘭縣	Yilan	2407	3210	0.75
桃園市	Taoyuan	7018	7627	0.92
新竹縣	Hsinchu Cnty	2619	3814	0.69
新竹市	Hsinchu City	1851	3309	0.56
苗栗縣	Miaoli	3417	4467	0.76
台中市	Taichung	9156	11350	0.81
彰化縣	Changhua	7192	10734	0.67
南投縣	Nantou	4761	4528	1.05
雲林縣	Yunlin	1827	4411	0.41
嘉義縣	Chiayi Cnty	243	1196	0.20
嘉義市	Chiayi City	162	2132	0.08
台南市	Tainan	5606	9433	0.59
高雄市	Kaohsiung	16042	20627	0.78
屏東縣	Pingtung	6944	8440	0.82
花蓮縣	Hualien	3581	6225	0.58
台東縣	Taitung	1371	2362	0.58
澎湖縣	Penghu	578	1344	0.43
金門縣	Kinmen	796	776	1.03
連江縣	Lienchiang	208	129	1.61
海外黨部	Overseas	904	-1649	-0.55

 

Maybe I’m thinking of this wrong. Maybe the point is that the growth in new KMT voters was much higher in central and southern Taiwan. The valid votes grew by 95% from 2016 to 2017. In 2016, Huang Min-hui won 33.2% of the votes, while Wu Den-yi won 53.0% in 2017, for an increase of 19.7%. If it was mobilization, these two numbers should move together. For example, valid votes increased by 129% while Wu beat Huang by 26.4%. Both of these numbers are larger than the national average, and Kaohisung is in the south. The problem is that we don’t see similar numbers throughout the rest of center and south. For example, in Changhau valid votes increased substantially, by 133%. However, Wu only bested Huang by 8.4%. All those extra voters didn’t seem to be going to Wu. In Tainan, valid votes only grew by 85% and Wu only outperformed Huang by 11.2%. In fact, some of Wu’s best areas were in the north. Wu outperformed Huang by 31.6% in Taoyuan and 26.8% in New Taipei, but neither one of these places had a particularly large increase in new voters. If you stare really hard and long at this table, you might convince yourself that you see a pattern. However, you are probably hallucinating. The correlation between the two columns is 0.05, just about as close to zero as you will ever see.

		% increase	Vote share
		Valid votes	Wu-Huang
合計		95	19.7
.
台北市	Taipei	110	21.8
新北市	New Taipei	72	26.8
基隆市	Keelung	135	19.8
宜蘭縣	Yilan	113	19.1
桃園市	Taoyuan	71	31.6
新竹縣	Hsinchu Cnty	113	14.6
新竹市	Hsinchu City	170	19.5
苗栗縣	Miaoli	86	19.4
台中市	Taichung	101	25.0
彰化縣	Changhua	133	8.4
南投縣	Nantou	112	30.6
雲林縣	Yunlin	101	-9.5
嘉義縣	Chiayi Cnty	31	-12.3
嘉義市	Chiayi City	80	-25.6
台南市	Tainan	85	11.2
高雄市	Kaohsiung	129	26.4
屏東縣	Pingtung	133	21.7
花蓮縣	Hualien	182	22.1
台東縣	Taitung	86	4.6
澎湖縣	Penghu	98	8.2
金門縣	Kinmen	48	30.7
連江縣	Lienchiang	29	33.0
海外黨部	Overseas	-24	23.3

In the end, there just isn’t any compelling evidence for the idea that local factions elected Wu chair by mobilizing tons of new voters for him. Heck, there isn’t evidence that anyone mobilized new voters for Wu.

I think the increase in new KMT voters is related to party morale, not to the KMT party chair election. Morale was at a nadir in the aftermath of the 2016 wipeout, and lots of party members let their membership lapse. As morale has recovered (slightly), some of those party members have drifted back (and paid their dues). The turnout rate was also markedly higher this time. However, the number of eligible voters and valid votes are far below the levels of 2005, when the winner was widely expected to become the next president.

	Valid votes	Eligible voters	turnout
2005	518324	1033854	50.2
2016	139558	337351	41.6
2017	272704	476147	58.1

At any rate, I think the evidence suggests that Wu Den-yi was elected by a fairly broad base of support within the KMT rather than by any specific group such as local factions or Taiwan nationalists. Admittedly, there is a limit to what we can see with crude data like this, so maybe it is best to state my conclusion in the negative. I don’t see any clear evidence for the local faction mobilization thesis.