I was thinking about the re-election incentive[1] for a paper I am planning to write, and I realized I’ve never put together numbers on how many people run for re-election. So I put together a dataset of all legislators from the 2nd to 6th terms (1993-2007), looking at whether they ran for and/or won re-election. Each legislator in each term counts as a separate case (ie: there are five separate cases for Wang Jinping since he had to decide whether to contest re-election each time). Also, this data set includes everyone who served in the legislature, including people who won by-elections and people who were substituted in for party list legislators who quit or were kicked out of their party (and therefore forfeited their seats). A legislator is considered to seek re-election if s/he is a candidate in the next general election. So here are some cuts of the data.
First, the big picture. About a fourth of legislators don’t run for re-election, and another fifth run and lose. So three-fourths of legislators seek re-election, and a fourth of those who do seek re-election lose. This looks like the re-election incentive is very strong. Of the quarter who do not seek re-election, probably a very high percentage of them would have liked to be re-elected but simply saw that it was not very likely and chose not to fight. Many of them sought but did not get nominations and chose not to run.[2] Another chunk of them sought election to other offices, generally county executives. It is hard to know exactly what percentage of the retiring legislators were forced out and what percentage retired willingly, but my guess is that at least half were unwilling. All in all, re-election looks very desirable.
It also looks very hard. Lots and lots of incumbents lose. This is nothing like the 90% re-election rates in the USA or Japan. In Taiwan, losing is a very realistic possibility for everyone every time. (In fact, I wrote a paper a couple of years ago in which I demonstrated that strategic voting makes the most popular candidates the most vulnerable. There are no safe seats in Taiwan’s SNTV elections.) If re-election is desirable and difficult, then it seems safe to assume legislators will respond to the re-election incentive in predictable ways. (Hooray!)
count | % | |
Did not run for re-election | 276 | 26.7 |
Ran and lost | 225 | 21.8 |
Ran and won | 531 | 51.5 |
Total | 1032 | 100.0 |
We can break down the data in other ways. There are big differences between district and list legislators. 80.4% of legislators elected in districts ran for re-election, while only 50.4% of list legislators did. However, there wasn’t much difference in their winning percentage, given the decision to enter the race.
Didn’t run | Ran and lost | Ran and won | N | |
District legislators | 19.6% | 23.5% | 56.9% | 786 |
List legislators | 49.6% | 16.3% | 34.1% | 246 |
You might wonder about party differences. Fantastic! Here’s a breakdown by party affiliation (at the time of the original re-election.)
Didn’t run | Ran and lost | Ran and won | N | |
KMT | 26.4 | 15.5 | 58.1 | 458 |
DPP | 28.4 | 24.7 | 46.9 | 373 |
NP | 18.6 | 46.5 | 34.9 | 43 |
PFP | 21.7 | 22.9 | 55.4 | 83 |
TSU | 33.3 | 41.7 | 25.0 | 24 |
IND | 29.4 | 25.5 | 45.1 | 51 |
It looks like politicians from all parties seek re-election at about the same rate, and they all have a strong possibility of losing, especially those from small parties.
But wait a minute, what about legislators who are elected as member of one party, don’t get re-nominated, and run as independents in the general election. Don’t they lose at a higher rate? Perhaps they are the ones driving these re-election rates down. Ok, let’s consider people to be KMT only if they won the original election as a KMT member AND contested re-election as a KMT member. (Since they have to have a party affiliation at time two, this rules out all the people who didn’t run for re-election.)
Ran and lost | Ran and won | N | |
KMT | 19.8 | 80.2 | 308 |
DPP | 31.9 | 68.1 | 254 |
NP | 50.0 | 50.0 | 30 |
PFP | 30.0 | 70.0 | 40 |
TSU | 50.0 | 50.0 | 12 |
IND | 42.3 | 57.7 | 26 |
Changed parties | 45.3 | 54.7 | 86 |
Party switchers don’t do well as expected, but after we get rid of them, KMT members still lose once in five times and DPP candidates lose once every three times.
Why are those DPP re-election rates so low? One big reason is that we are including incumbents from the 6th term who ran for re-election in the new mixed member system. As we all know, the DPP was massacred in that election. However, if we confine our data to the old SNTV/closed list system, the difference between the two big parties is not as marked.
KMT | DPP | |||
term | Ran, Won% | N | Ran, Won% | N |
2nd | 81.2 | 69 | 70.0 | 40 |
3rd | 90.2 | 61 | 79.1 | 43 |
4th | 60.8 | 74 | 88.2 | 51 |
5th | 91.3 | 46 | 74.2 | 62 |
6th | 84.5 | 58 | 34.5 | 58 |
Total | 80.2 | 308 | 68.1 | 254 |
One of the most well-known lessons from the American context is that if you want to beat an incumbent, you had better do it the first time they come up for re-election. After they win the first re-election, they are basically bulletproof. That is not true here.
Number of terms served | Didn’t run | Ran and lost | Ran and won | N |
1 | 26.0 | 21.0 | 53.0 | 523 |
2 | 26.0 | 24.9 | 49.1 | 281 |
3 | 29.6 | 20.8 | 49.6 | 125 |
4 | 34.0 | 22.0 | 44.0 | 50 |
5 | 16.0 | 16.0 | 68.0 | 25 |
6 | 30.8 | 23.1 | 46.2 | 13 |
7 | 28.6 | 14.3 | 57.1 | 7 |
8 | 50.0 | 0 | 50.0 | 4 |
9 | 0 | 0 | 100.0 | 2 |
10 | 50.0 | 0 | 50.0 | 2 |
total | 26.7 | 21.8 | 51.5 | 1032 |
It is striking just how little this table changes as legislators gain seniority. The number of cases decreases in an almost perfect pattern: each row has half the cases of the previous row.
The final thing I did was to look at people who switched districts. For example Hong Qichang ran in Taichung City in 1992, the party list in 1995, Taipei City South in 1998, and the party list again in 2001, 2004, and 2007 (he originally was elected to the legislature in 1989 from Tainan City, but that isn’t in this data set). So in my data set, the first three re-elections are coded as district changes while the last two are not. Somewhat surprisingly to me, people who changed districts didn’t do a whole lot worse than people who stayed in the same place.
Ran and lost | Ran and won | N | |
Same district | 28.7 | 71.3 | 624 |
Changed districts | 34.8 | 65.2 | 132 |
total | 29.8 | 70.2 | 756 |
I don’t think this means that politicians can run anywhere and have about the same probability of winning. Rather, legislators are very strategic about changing districts. The ones who do actually change districts have good reason to believe they might find success in the new districts. The ones who can’t foresee anything good happening from a change simply stay put.
All in all, this was a fun exercise for me. More importantly, I managed to convince myself that the re-election motive should be very powerful in Taiwanese elections. No matter how you cut the data, legislators want re-election and have difficulty in securing it. If American legislators are, as Gary Jacobson famously put it, “running scared,” then Taiwanese legislators must be positively terrified by the possibility of losing.
[1] One prominent tradition in political science suggests that we can understand quite a lot about why legislators act the way they do if we assume that they are entirely motivated by re-election.
[2] The only group I have good data for on nomination contests is the DPP in the 6th term. Of the 39 DPP legislators who did not run for re-election in 2008, 10 sought a nomination but lost out in the telephone survey stage.