The effects of SNTV and quasi-SNTV

In SNTV, voters cast a single vote for a candidate in a multimember district and votes are non-transferable. Each party may put forward more than one candidate in each district. Seats are allocated by the plurality rule until all seats are filled (Carter and Farrell, Reference Carter, Farrell, LeDuc, Niemi and Norris2010; Farrell, Reference Farrell2011). The advantages of this system are that it tends to produce a more proportional outcome and improve the representation of small parties and minority representation than a single-member plurality system (Wang, Reference Wang2008; Farrell, Reference Farrell2011). However, SNTV often suffers from the problem of super-proportionality, especially in districts of large magnitude. Election results in Japan (till 1993) and Taiwan (through 2004) demonstrate that large parties are likely to commit errors of overnomination (the party's candidates exceed the MAXS,Footnote ⁵ and the party ultimately gets fewer seats than the MAXS), undernomination (it could have won more seats by nominating more candidates), and misallocation (the party's candidates are equal to the MAXS, but the party obtains fewer than the MAXS), while small parties usually can benefit from this and obtains seats with small vote shares. (Lijphart et al., Reference Lijphart, Pintor, Sone, Grofman and Lijphart1984; Hsieh, Reference Hsieh1992; Wang, Reference Wang2008). In particular, when one of the large parties commits errors (loses seats), other large parties are likely to benefit from this and become over-represented (advantage ratio > 1), especially when small parties are not viable (Cox and Niou, Reference Cox and Niou1994; Wang, Reference Wang2008).Footnote ⁶ In this sense, SNTV is likely to be super-proportional, especially in a given district of large magnitude.

Cox and Shugart (Reference Cox and Shugart1995) have conducted extensive investigations into the effects of quasi-SNTV. According to their case study of Colombian quasi-SNTV prior to 2006, it resembles SNTV in Japan in that there are multiple lists under the same party label and votes cannot be transferred from one list to another.Footnote ⁷ Multiple lists under one party label is a result of weak party discipline. The absence of vote pooling in SNTV, just as in Colombian quasi-SNTV, makes it more difficult for a party to control its candidates. While the maximum of district magnitudes is six in Japan, district magnitudes in Colombia range from 1 to 29. Similar to Japan's SNTV, the larger the district magnitude, the more errors parties commit in Colombia. Errors rates of major Colombian parties are even higher than those of Japan since weak party discipline makes Colombian parties unable to control their increasing party lists (Cox and Shugart, Reference Cox and Shugart1995; Pachon and Shugart, Reference Pachon and Shugart2010). Although the quasi-SNTV emerges because of weak party discipline, Colombian parties became familiar with the quasi-SNTV and managed to win more seats by putting forward multiple lists from 1990 to 1998. Nearly all advantage ratios of major parties (Liberal Party and Conservative Party) are over 1, and those of small parties or independents are below 1. Clientelism enabled the large parties to attract voters to support particular candidates and thus made it easier for parties to allocate votes (Shugart et al., Reference Shugart, Moreno, Fajardo, Welna and Gallon2007). It is evident that large parties in Columbia are over-represented as well. The super-proportionality of Colombian quasi-SNTV substantiates that it is very similar to SNTV.

The absence of vote pooling in SNTV or quasi-SNTV has contributed to personalized electoral machines since the competition is among individuals in SNTV and those heading each list in quasi-SNTV (Cox and Shugart, Reference Cox and Shugart1995; Cox, Reference Cox1997; Pachon and Shugart, Reference Pachon and Shugart2010). There are few incentives for candidates to be loyal to their parties because the weak party discipline in SNTV enables candidates to cultivate their votes through their personal networks (koenkai in Japan) (Baerwald, Reference Baerwald1986; Hrebenar, Reference Hrebenar1986; Curtis, Reference Curtis1988; Shugart et al., Reference Shugart, Moreno, Fajardo, Welna and Gallon2007). In Taiwan's SNTV, both marginally elected and so-called national celebrities switched their partisanship or became independent in order to win reelection since they can win by their strong clientelistic networks (Chang and Tang, Reference Chang and Yen-chen2015). Scholars of Hong Kong politics such as Ngok Ma and Ivan Choy also claim that quasi-SNTV in Hong Kong is quite similar to SNTV in that many lists run in a district are under the same party label (Ma and Choy, Reference Ma and Choy2003a). Also, the study of campaign advertising in the 2004 and 2012 Legislative Council elections in Hong Kong also demonstrates that there is more advertising focusing on a candidate's image than on policies or issues (Niu, Reference Niu2008, Reference Niu2013). In this sense, quasi-SNTV can promote candidate-centered campaigns and party-switching happens frequently as well in Hong Kong's quasi-SNTV, such as Chin-shek Lau and Albert Wai-yip Chan in 2004, Albert Wai-yip Chan and Audrey Eu in 2008, and Cyd Ho and Emily Lau in 2012.Footnote ⁸ In particular, Kwong (Reference Kwong2010) also discovers patron–client networks facilitate candidates’ personal support in Hong Kong's Legislative Council elections.

Although the absence of vote-pooling is the major mutual characteristic between SNTV and quasi-SNTV, votes can be pooled in quasi-SNTV once there are excess votes in one list. Take the DAB in 2008 for example. The DAB put forward one list in New Territories East and got 102,434 votes. Though votes could not be transferred to the independent Scarlett Oi-lan Pong, whom the pro-Beijing united front endorsed, votes could be pooled across the candidates on its list (Cheng, Reference Cheng2010). The total votes allowed the 1st seat (Kong-wah Lau) to win by quota and the 2nd seat (Gary Hak-kan Chan) to win by remainder. Nevertheless, votes are wasted on excessively popular candidates in SNTV while other candidates under the same party label have too few votes to be elected. The example of the 1998 Legislative Council election in the South District of Taipei City can demonstrate this. While the DPP's top-ranked candidate Fu-hsiung Shen got 86,900 votes, his colleague Tien-fu Huang obtained only 28,179 votes and was defeated only by a margin of 2,324 votes (Yu, Reference Yu1999). The excessive votes in SNTV cannot be pooled and this is the distinction between quasi-SNTV and SNTV. Since the seats are more expensive in vote-pooling in LRPR, seats are cheaper while putting forward multiple lists (quasi-SNTV). This probably explains why the DAB ran multiple lists in three districts in 2012.

This article intends to explore how quasi-SNTV in Hong Kong takes place, whether the DP and DAB put forward multiple lists or cooperate with their respective alliances or not, and the aftermath of multiple lists and coordination or failure of coordination. In order to maximize seats in one district, parties may cooperate with fellow parties or independent candidates in the same camp to have one list or one party can have two or more lists. Except in 2004, the pro-democracy camp lacks an overriding authority to coordinate with different parties or candidates in the same camp. In contrast, the pro-China camp seems less divided than its rivals because it adopts the electoral strategy of a united front (Cheng, Reference Cheng2010). The quasi-SNTV in Hong Kong features the strategic coordination of the two political camps, which is different from that in Colombia.

Political/social cleavages in Hong Kong

The main political cleavage in Hong Kong is pro-China vs. pro-democracy. The major distinction between these two sides is their attitude toward political reform (especially the issue of universal suffrage) (Cheng, Reference Cheng2010). The pro-democracy camp appeals for rapid and radical political reform while the pro-China camp emphasizes the importance of political stability and advocates for political reforms to proceed gradually (Yip and Yeung, Reference Yip and Yeung2014). The emergence of this political cleavage can be traced to Hong Kong's unique political background and development. Therefore, it is important to understand the evolution of Hong Kong's political parties because this is directly related to the political cleavages.

The strategic coordination of the parties in Legislative Council elections

The direct election in Hong Kong's Legislative Council (or LegCo) election uses a closed party-list LRPR with Hare Quota. Given the similarities between quasi-SNTV and SNTV, either the DP or the DAB puts forward multiple lists in a given district. In particular, the DP has observed vote division strategies under SNTV by political parties in Taiwan. There are two main vote division strategies in Taiwan's Legislative Yuan elections. The first is the ‘vote responsibility zone’ of the Kuomingtang (KMT); the second is the vote division strategy preferred by the Democratic Progressive Party (DPP) and the New Party (NP) (Choy, Reference Choy, Hsin-chi, Siu-kai and Wang2002; Ma and Choy, Reference Ma and Choy2003a).

The KMT, which had more resources and consolidated local and social networks, usually divided a geographical constituency into several small zones and assigned different candidates to different ‘spheres of influence’ so that they could campaign and mobilize in their assigned district. In contrast to the KMT, the DPP and the NP had fewer resources and were not as well organized (Choy, Reference Choy, Hsin-chi, Siu-kai and Wang2002). Therefore, the DPP and the NP adopted different vote division strategies from the KMT. In this special vote division strategy, which was innovated for Taiwan's Legislative Council elections in 1995, the DPP and NP canvassed their supporters to equalize votes through the media.Footnote ¹⁶ This vote division strategy was shown to be very successful in 1995 because it helped to reach high rates of optimal performance. However, the vote division strategy of the DPP did not really work well in the 1998 Legislative elections in the South District of Taipei City because the division strategy was usually difficult to put into practice. The success or failure of the strategy depended on many factors, such as the changing of candidates, strengths of different candidates, factional struggles, the adherence of voters to the party's vote division strategy, and other parties’ strategies (Yu, Reference Yu1999).

The DP decided to adopt the KMT's ‘vote responsibility zone’ strategy since it began splitting lists in 2000. Occasionally, some pro-democracy marginal candidates may ask voters to vote strategically; however, this has actually shown to be ineffective since there are relatively few voters (only 7% at most) who engage in strategic voting (Choy, Reference Choy, Hsin-chi, Siu-kai and Wang2002; Ma and Choy, Reference Ma and Choy2003a).Footnote ¹⁷ In 2004, the DP cooperated with some of their pro-democracy colleagues to run in the LegCo elections by either coordinating around one list (e.g. the Diamond list in New Territories East in 2004) or putting forward multiple lists. Similar to the pro-democracy camp, the DAB has split two lists (both are under the combined label of DAB/HKFTU) in Kowloon East starting in 2004.

Table 3 shows errors under quasi-SNTV the parties committed by district magnitudes in five districts. First, the actual and hypothetical results of the two parties are calculated if they were to put forward multiple lists in each district to determine whether multiple lists can help to win extra seats. If any of them has engaged in vote-pooling (put one list in a given district only), then it is not counted since I assume that votes are not pooled and seats which are won by remainder are the cheapest. The strategic coordination around a single list either by a party or even by a camp (e.g. the democratic camp's Diamond list in NTE in 2004) makes votes pooled and seats the most expensive. In this sense, the party or the camp cannot commit errors. The next step is to calculate the MAXS, error rates, and the optimal performance rates for each party.Footnote ¹⁸ The 2012 LegCo election in New Territories West is an example of the calculation in Table 4. Each row shows the lists, actual vote share and seats the list got; the total votes for the DP and the DAB from column 4 to 7 are the hypothetical allocation if they ran one list. Since each election presents a different scenario with respect to coordination, the situations of each Legislative election are outlined briefly before hypothetical and actual results are calculated and errors rates and optimal performance rates of both are analyzed.

Table 3.

The quasi-SNTV and errors by district magnitudes

Notes: N = district magnitudes; U = undernomination, O = overnomiation, OS = overnomination but safe, OP = optimal performance, – = not applicable.

Table 4.

The example of actual and hypothetical results of the DP and the DAB in New Territories West in 2012

The error and optimal performance rates under quasi-SNTV

Table 5 shows the error and optimal performance rates of the DP and the DAB. Lists which are not engaged in vote pooling (multiple lists in one party label-Quasi-SNTV) are counted as the party's strategies; lists engaged in vote pooling either under any party label or a camp's coordination around a list (e.g. the ‘Diamond List’ in New Territories East in 2004) are not included since a single list under one party label is normal under LRPR globally and seats become expensive. As Table 5 shows, multiple lists do not always maximize seats for the two parties. If coordination is not done well either at the nomination level or the campaign level, the situation becomes even worse than coordination around a single list (e.g. see Table 4: the DP in New Territories West in 2012). The information in Table 3 also substantiates that parties are less likely to commit different types of errors when the district magnitudes are small, such as in Kowloon East and Kowloon West. In contrast, the error rates are higher when district magnitude increases such as in New Territories East and New Territories West (Cox and Niou, Reference Cox and Niou1994; Cox and Shugart, Reference Cox and Shugart1995; Wang, Reference Wang2008).

Table 5.

The error and optimal performance rates of the DP and the DAB

Notes: 1. U = Undernomination, Over = Overnomination, OP = optimal performance.

2. Overnomination but safe is not counted as an error as in Wang's definition since the MAXS is still reached.

The DP's error rate is 37.5%, and the optimal performance rate is only 25% (‘overnomination but safe’ is neither counted as an error nor optimal performance); the DAB's error rate is 25%, while their optimal performance rate is 75%. It is evident that the strategic coordination of the DAB is better than that of the DP. In particular, the pro-China camp is more internally cohesive, which helps the strategic coordination succeed. The DP's failure can be attributed to its internal conflicts and the lack of coordination among pro-democracy parties and candidates in the nomination and campaign stages. Take 2012 for example, the DP committed errors of overnomination in the nomination stage in New Territories West and in New Territories East. In the campaign stage, the Civic Party and the DP suffered from attacks by their radical colleagues in the LSD, People Power, and New Democrats (Yip and Yeung, Reference Yip and Yeung2014). In addition, the Civic Party asked the DP supporters to vote for them which caused the pro-democracy camp as a whole to lose some seats (Cheng, Reference Cheng2010).Footnote ²³ This situation is especially apparent in the 2012 LegCo election, as is evident in Table 4.

Although multiple lists of the DAB starting from 2004 are the result of some candidates’ insistence (Yuen Han Chan in Kowloon East), actually the DAB has fewer internal conflicts than the DP. In addition to the DAB's long-term community service endeavors, their success can also be attributed to the pro-China camp's united-front machinery. This canvassing machinery enables the pro-China camp to allocate votes equally and maximize seats (Cheng, Reference Cheng2010). From the experiences of the DP and the DAB, quasi-SNTV is more effective whenever the political camp is more internally coherent and there is an overriding authority to boost coordination in both the nomination and campaign levels.