All the media coverage is on the presidential race, even though we all know that it is almost certain that Tsai Ing-wen will win. Again and again, we are reminded that the real fight is for the legislature, but then no one seems to have much to say about that election. I have also been guilty of this, for the simple reason that I don’t have much data to pore through. Without data, I’m spinning my wheels, just like everyone else. Still, I thought I should give it another shot.
Rather than directly predict what is going to happen, I’m going to make a conditional prediction. If you tell me how much the DPP legislative candidates will beat the KMT legislative candidates by in the national aggregate vote, I’ll try to tell you how many district seats each side will get. To put it another way, the polls say that Tsai is winning the presidential race by 20% or so over Chu. We all assume that the KMT district legislative candidates will run ahead of the KMT presidential candidate, but we don’t really know how much better they will do. If they run an average of 5% ahead (and the DPP district candidates run an average of 5% behind Tsai), that would imply that the DPP candidates beat the KMT candidates by a national average of about 10%. What kind of seat distribution would that 10% margin imply?
I’m using a very crude model with lots of assumptions built in. Most of them are untenable. So don’t get too caught up in the individual results. The bigger point is that if the DPP beats the KMT by X%, there has to be an excess vote margin somewhere. Maybe it will be more concentrated in a particular district than I assume, but it can’t all be concentrated in Tainan. If X is larger, the DPP will inevitably win more seats.
I’m basing this model loosely (read: from memory) on a regression model I published from the 2012 election. In that model, I found that DPP district candidates won about 95% of Tsai’s presidential votes. Incumbents running for re-election did a little better, and challengers facing KMT incumbents did a little worse. The KMT model was nearly a mirror image, with the exception that I combined Ma’s and Soong’s votes to create the presidential baseline. So let’s start there. A first expectation is that each party will get 95% of the 2012 presidential vote. Incumbents running for re-election get a bonus of 3%, while challengers facing incumbents get a penalty of 2%.
Right away, you will see several assumptions built in. This assumes that we are still in the old blue vs green world, and each camp will have one candidate. I’m just ignoring three-way races, and I’m essentially assuming that people like Lee Ching-yuan, the blue camp city councilor running in Taipei 8 in alliance with the DPP against the KMT incumbent, are equivalent to DPP nominees.
When you total up those initial expectations, you get a national aggregate vote of 52.4% for the KMT and 43.4% for the DPP. (Note that these are both slightly less than the actual 54.4-45.6% presidential result. The 9.0% gap is slightly larger than the 8.8% presidential gap because the KMT has a lot more incumbents. It is also significantly larger than the official gap of 3.7% between KMT and DPP candidates in 2012, though those numbers don’t include independents such as Yen Ching-piao, who were actually blue or green candidates.) What kind of seat distribution does that 9% gap imply? Based on these initial expectations, the KMT would win 49 seats, while the DPP would win 24.
Ok, no one expects the KMT to have a 9% national advantage. So let’s slowly scale that back. I’m going to subtract 1% from the KMT vote and add 1% to the DPP vote, and we will see how many seats that shifts. Then I’ll do it again and again, and we will try to see when the balance of power switches from blue to green.
Note that there is another crazy assumption built into this. I am assuming that the relative strength of the parties in each district will remain the same as in 2012. In other words, the DPP’s vote share will increase by exactly the same in every district. Once again, I understand that is ridiculous; the DPP vote share will almost certainly increase more in Taoyuan or Changhua than in Taipei or Lienchiang. However, for simplicity’s sake (and because I have no idea how to allow the model to vary increases by district), I’m just ignoring it.
With all these caveats out of the way, here are the results:
|-5||43||30||Changhua 4, Taichung 7, Pingtung 2, Penghu, New Taipei 4|
|-3||41||32||Taichung 6, New Taipei 6|
|+1||37||36||Changhua 3, Changhua 1, New Taipei 5|
|+3||35||38||Taichung 3, Taichung 8|
|+5||33||40||New Taipei 7, Taichung 2|
|+7||31||42||New Taipei 10, Taoyuan 2|
|+9||29||44||Nantou 2, Changhua 2|
|+11||25||48||Taipei 1, Taoyuan 1, New Taipei 1, New Taipei 12|
|+13||21||52||Taipei 5, Taipei 3, Hsinchu City, Taoyuan 4|
|+15||17||56||Taichung 5, Taichung 4, Nantou 1, Taipei 4|
|+17||15||58||Miaoli 1, Keelung|
|+21||12||61||Taoyuan 6, New Taipei 8, Taipei 7|
|+23||10||63||Taoyuan 3, Taoyuan 5|
|+29||6||67||Hsinchu County, New Taipei 11|
Remember, because the KMT dominates the indigenous seats, the DPP needs to win more than half of the district seats. It is possible that the DPP will win an indigenous seat this year, but I think that it will be easier to win the 40th district seat than the first indigenous seat. That is, I think that Chen Shih-kai has a better chance of beating Yen Kuan-heng in Taichung 2 than Chen Ying does of winning a lowlands indigenous seat. If this is correct, the (pan-) DPP needs to win 40 district seats to win an overall majority in the legislature. (39 could plausibly suffice if the DPP has at least a 5% advantage over the KMT in the party list vote.)
In this model, the DPP wins that 40th seat when their advantage in the aggregate vote reaches 5%. In other words, if the (pan-) DPP wins the district vote by roughly 50-45%, they will win a majority in the legislature.
My guess is that right now the DPP is easily ahead by 5%. In fact, I have been guessing that the gap is in the low double digits. If this is correct, it could spell disaster for the KMT. When the gap is this large, we start moving from battleground districts into districts that should be reliably blue. Moreover, there are a lot of districts closely bunched in this range. The KMT is predicted to win 29 seats if the gap is +9. When the gap reaches +15, the KMT is all the way down to 17 seats. This is roughly the mirror image of 2008, when the KMT won the aggregate vote by about 15% and the DPP was left with only 13 district seats. That 6% swing (actually only 3% each way) from 52-43 to 55-40 is the difference between a small DPP majority and a DPP landslide. In the past few days, Eric Chu has publicly set his goal for the legislature at 40. Sometimes this seems to mean 40 overall seats, while other times it seems to mean 40%. To get 40 seats, the KMT district candidates will need roughly 41-42%. 40% probably won’t do it.
Another thing this table is good for is thinking about general expectations. A few days ago, someone asked me if Freddy Lim was going to win a seat for the New Power Party in Taipei 5. I’m not terribly optimistic. Look at the seats around Taipei 5 in the table. Do you think that the DPP will be competitive in most of them? Perhaps they might, if the KMT really is so far behind. However, remember that Taipei is historically much more stable than the rest of Taiwan. My gut feeling is that the DPP will likely win Hsinchu City, Taoyuan 4, Taichung 4, Nantou 1, Keelung City, and maybe even Miaoli 1 before it wins Taipei 5. Sweeping those is a tall order, which suggests to me that Freddy is fighting a steep uphill battle. If he wins, the KMT will be in full disaster mode.