A great Ketagalan Media piece from December 2015 on how the Legislative Yuan elections work raised the possibility of redistricting to address imbalances in the number of residents in each of the 73 constituency seats around the country. The most populous district, Hsinchu County, had 396,492 registered voters as of early 2016, whereas as Lienchiang County (more commonly known as Matsu) had just 9,921. The common perception is that these imbalances, whether by accident or design, favour the Kuomintang. But is this true?

To examine this question I looked at the results from all three elections for which the current boundaries have been in place, and assigned each constituency to one of four categories. Those which have been won by either the KMT (or allies) or the DPP (or allies) in all three elections were designated “Deep Blue” and “Deep Green” respectively. Those in which two out of three had gone to the KMT side are “Light Blue”, and “Light Green” seats were won by the DPP side in two out of three elections. 2008 was a KMT landslide, 2012 was a narrower Pan-Blue win, and 2016 was of course a DPP landslide. This is admittedly a pretty crude method based on a small sample size, but it has the advantage of being quick and easy to measure.

Just eyeballing the list of constituencies (ordered from most to least number of eligible voters) it looks reasonably balanced:

But let’s take a closer look. The mean number of electors across the 73 constituencies is 250,755, which could be considered an “ideal” district. The Blue mean (Deep and Light Blue seats) is 251,729 while the combined Green mean is 249,189 – both very close to the “ideal” number. However, if we consider just the safe seats, Deep Blue constituencies average 240,279 electors, while the mean in Deep Green districts is 266,477; 10.9% larger. These safe seats may constitute a better indication of possible gerrymandering.

A more thorough analysis would take into account winning margins and include results from other races (presidential, regional and local) in determining the “blueness” or “greenness” of each constituency, but you would need to control for registered Aboriginal voters (who don’t vote in the geographical constituencies). Also, the scale of the anti-KMT landslide in 2016 swept districts that were previously considered safe blue seats, like Hualien, Taichung 3, and Taipei 1.

Bearing in mind the limited dataset I’ve used, I think the answer to the title question (do the districts favour the KMT?) is “possibly, a little”. This shouldn’t be a surprise, as the KMT was in the driving seat when these districts were drawn up in 2005. However, the effect is not that large, and the overall numbers are distorted by tiny Lienchiang, which is probably the safest blue seat in the country.

Towards Redistricting?

There are two significant rules in place for determining constituency boundaries. Firstly, each county and municipality* must have at least one representative each. Secondly, and I am unsure whether this is actually written down anywhere, constituencies cannot cross county or municipality borders. The first rule in particular explains why voters in Taiwan’s small offshore counties are overrepresented. What would redistricting look like if both of these rules were dropped? That’s something for a possible future post.

If you want to play with the numbers yourself, you can download my spreadsheet below. Original data was sourced from the Central Election Commission.

2008-16 LY district electors and results

*Municipality here includes the nine cities which are governed independently of Taiwan’s county system.