Tzu-chiao Su Chinese Culture University, Taiwan

The Effect of Electoral System and Election Timing on Party System and Government Type: a Cross-Country Study of Presidential and Semi-presidential Democracies Tzu-chiao Su Chinese Culture University, Taiwan I. Introduction The parliamentary party system in a democratic country is mainly affected by the country s electoral system for the parliament. The widely known Duverger s law maintains that the single-member district plurality system (first-post-the-post system, FPTP) tends to result in a two-party system, whereas proportional representation tends to result in a multiparty system. However, in democracies, elected officials at the central government level are not limited to members of parliament (MPs) but may include a president. Democracies that use the parliamentary system only feature s for MPs and have no direct presidential s. Although the party system in a parliament is shaped by parliamentary s, in democracies that use the presidential or semi-presidential system, in which both MPs and the president are directly elected, the parliamentary party system is affected not only by the parliamentary electoral system directly but also indirectly by the presidential electoral system. In other words, the electoral systems for parliament and president together shape the party system in the parliament. In addition to the influence from the parliamentary and presidential electoral systems, a party system is affected by the timing at which the two types of s are held. In countries that use the presidential or semi-presidential system, the parliamentary party system shaped by the parliamentary and presidential electoral systems and timing result in various government types. In a two-party system, typically one party obtains a majority of seats in parliament. If the president belongs to this party, then the government type is a unified government. Otherwise, the government type is a divided government. When the parliamentary party system is a multiparty system, rarely can one party obtain a majority of seats; thus, a divided government is often the result. Therefore, in a democracy that uses the presidential or semi-presidential system, its government type is shaped by the parliamentary electoral system, the presidential electoral system, and timing. This study explored the following topics. (1) In democracies using the presidential or semi-presidential system, in addition to the direct influence of the 1

parliamentary electoral system, whether the presidential electoral system (such as the plurality system or the absolute majority system) or timing (such as honeymoon s, midterm s, counter-honeymoon s, and simultaneous s) has a greater influence on the parliamentary party system. (2) What government types are formed under the parliamentary party system that is shaped by the parliamentary electoral system, presidential electoral system, and timing. This study conducted preliminary research on democracies that use the presidential or semi-presidential system. From a broad perspective of world electoral systems, this study investigated the possible influences of the parliamentary electoral system, presidential electoral system, and timing on the parliamentary party system and government type. II. Literature review The consensus is that the parliamentary electoral system shapes the parliamentary party system. Furthermore, different combinations of parliamentary party systems and constitutional systems result in different government types (Fig. 1). For example, the single-member district plurality system tends to result in a two-party system in the parliament, and the two-party system and parliamentary system combination often results in one-party cabinets. Proportional representation tends to result in a multiparty system, and a multiparty system combined with the presidential system typically results in divided government. However, this conventional view is too simplistic. As already stated, in democracies that use the presidential or semi-presidential system, the parliamentary party system is affected not only by the parliamentary electoral system but also by the presidential electoral system and timing. For example, if the president is elected by a plurality, small parties are more effectively suppressed than they would be were the president elected by an absolute majority, resulting in fewer parties represented in parliament. As for timing, midterm s are more favorable to small parties than other types of. Thus, they result in more parties represented in parliament. Election timing affects not only the parliamentary party system but also the government type. For example, midterm s tend to lead to a divided government, whereas simultaneous s and honeymoon s tend to result in a unified government. The following section describes the effect of the parliamentary electoral system and the presidential electoral system on the party system and the influence of timing on the party system and government type, respectively. 2

Parliamentary electoral system Parliamentary party system Government type Figure 1 Constitutional system Relationship among parliamentary electoral system, parliamentary party system, constitutional system, and government type A. Influence of the parliamentary electoral system and the presidential electoral system on the party system Duverger s law explains how a parliamentary electoral system can influence a country s party system. According to Duverger (1954: 224 226; 1986: 70), the single-member district plurality system tends to result in a two-party system, whereas proportional representation tends to result in a multiparty system. In addition, parliamentary electoral systems that use the two-round system tend to result in multiparty alliances. Although the word law has caused dispute and criticism, and although real-life cases do not always fit Duverger s law, the emphasis of Duverger s law, that is, that the parliamentary electoral system has a considerable effect on party system, is widely acknowledged. Scholars have suggested that whether a country has direct presidential s and the type of direct presidential affects the party system. First, in countries where the president is directly elected by the people, the number of parties in parliament is less than that in countries without presidents or countries whose presidents are not directly elected. In other words, holding direct presidential s reduces the number of parties in parliament (Epstein, 1967; Lijphart, 1994: 131; Cox, 1997). Second, studies have revealed that, in presidential s, the effect of the plurality system in reducing the number of parties in the parliament is stronger than that of the two-round system, where an absolute majority is required for a presidential candidate to win. In other words, presidential s that use the plurality system result in fewer parliamentary parties (Shugart and Carey, 1992: 221-225; Jones, 1994: 41-57). In addition to the separate influences of the parliamentary and presidential electoral systems on the parliamentary party system, scholars have reported on their combined influence. Su and Wang (2012) reviewed democracies and discovered that, among countries that use proportional representation in parliamentary s, using the plurality system for presidential s substantially reduces the effective number of parties in parliament compared with countries that use the two-round system. Therefore, countries that adopt the plurality system for presidential s 3

have a smaller effective number of parties in parliament than those that use the two-round system. The implication is that although a parliamentary electoral system may use proportional representation and small parties thus have more room to exist and a multiparty system can more easily be formed, the survival of these parties is also affected by the adopted presidential electoral system. In summary, whether the parliamentary party system is two-party or multiparty is mainly affected by the parliamentary electoral system. Although the presidential electoral system does not have a direct influence on whether the parliamentary party system becomes two-party or multiparty, it still reduces the number of parties represented in parliament. This oppressive effect is particularly prominent in countries where the parliamentary electoral system is proportional representation. B. Influence of timing on the party system and government type In countries using the presidential or semi-presidential system and holding both parliamentary and direct presidential s, the influence of direct presidential s on party system and the influence of presidential and parliamentary timings on party system can be observed. Election timing can be categorized as one of the following four types: (1) a simultaneous refers to a presidential that is held on the same day as the parliamentary ; (2) a honeymoon refers to a parliamentary that is held soon after the presidential ; (3) a counter-honeymoon refers to a parliamentary that is held shortly before the presidential, and (4) a midterm refers to a parliamentary that is held a long time before or after the presidential. Election timing plays a critical role in government type formation. Simultaneous s and honeymoon s are more likely to result in a unified government (i.e., the president s party is the same as that of a majority of MPs). The reason is that in simultaneous and honeymoon s, a presidential candidate who is projected to win (or who is already president-elect) may result in the coattail effect, whereby voters who vote (voted) for the leading presidential candidate (president-elect) in the presidential vote for the same candidate s (president-elect s) party in the parliamentary, resulting in a president and parliamentary majority of the same political party. By contrast, midterm s tend to result in a divided government (i.e., the party to which the president belongs differs from that of the parliamentary majority). This is because midterm s are held when a president has been in office for some time, and disputes over his or her administration have arisen. Midterm s provide voters with the opportunity to register their discontent with the president s administration; thus, they tend to vote for parliamentary candidates of opposite parties. Consequently, parliamentary candidates from the president s party 4

tend to lose the, resulting in the political party of the parliamentary majority differing from that of the president. In counter-honeymoon s, the coattail effect is weaker than that in simultaneous or honeymoon s. Therefore, when a counter-honeymoon is held, a divided government is more likely to occur than in a simultaneous or honeymoon. However, if in a counter-honeymoon, the timing of the parliamentary is close to that of the presidential that follows, the chance of a unified government is relatively large. In summary, the coattail effect that a presidential candidate (or president-elect) exerts is strongest in simultaneous and honeymoon s, followed by counter-honeymoon s, and finally midterm s. By observing the effect of timing on government type, we can infer the effect of timing on party system. Because a presidential is an for a single position, it is undoubtedly more favorable to large parties. The stronger the coattail effect of a presidential candidate or president-elect, the more easily a unified government can be formed and the more beneficial the effect is for large parties; consequently, the effective number of parties in parliament becomes smaller. In summary, a strong coattail effect results in a smaller effective number of parties in parliament. The strength of the coattail effect is, as stated before, strongest for simultaneous and honeymoon s, followed by counter-honeymoon s, and finally midterm s. The effective numbers of parties in parliament is largest after midterm s, followed by counter-honeymoon s, and smallest after simultaneous and honeymoon s. III. Research framework On the basis of the aforementioned research, this study assumed that in countries using the presidential or semi-presidential system, the relationship among the variables, parliamentary electoral system, presidential electoral system, timing, parliamentary party system, and government type, is as shown in Fig 2. The independent variables (viz., parliamentary electoral system, presidential electoral system, and timing) all influence parliamentary party system (mediator), whereas the different combinations of parliamentary party system and presidential and semi-presidential systems result in different government types (dependent variable). In addition, timing (independent variable) directly influences government type (dependent variable). In summary, on the basis of the analysis framework comprising the aforementioned variables, this study analyzed how parliamentary electoral system, presidential electoral system, and timing affect the parliamentary party system and government type in democracies that use the presidential or semi-presidential system. 5

Electoral timing Parliamentary electoral system Parliamentary party system Presidential electoral system Government type Constitutional Constitutional system Figure 2 Research Framework The countries analyzed in this study were those using either the presidential or semi-presidential system and that had held at least two consecutive democratic presidential s and parliamentary s between the end of World War II (1945) and the end of 2014. The term democratic used in this study refers to s held under democratic politics. This study adopted the Freedom House score for each country to determine whether it had implemented democratic politics. The research objects of this study were countries using the presidential or semi-presidential systems that had held at least two consecutive parliamentary s and two presidential s during the duration when they were scored 2.5 or below by Freedom House. After the objects of this study were selected, they were divided into two categories on the basis of their presidential electoral system: the plurality system and absolute majority system (i.e., the two-round system). On the basis of their parliamentary electoral system, they were divided into four categories: single-member district plurality system (FPTP), proportional representation system (PR), mixed-member majoritarian system (MMM), and mixed-member proportional system (MMP).1 Thus, as Table 1 shows, 2 4 = 8 combinations of presidential electoral system and parliamentary electoral system exist.2 1 2 Regarding parliamentary electoral systems, in addition to these four major systems, other systems also exist, for example the two-round system (used in France), the instant-runoff voting system (used in Australia), the single transferable vote (such Ireland and Malta), and the single nontransferable vote (such as Vanuatu). However, because few countries have adopted these systems, this study did not include them for analysis. In Table 1, the cells representing the combination of the compensatory system for the parliamentary electoral system and the plurality system and absolute majority system for the presidential electoral system are empty. In fact, countries that use the compensatory system and absolute majority system are Venezuela and Bolivia. Niger is the only country that uses the compensatory system and plurality system. However, they are not listed in Table 1 because they are insufficiently democratic to meet the screening standard for democratic countries of this study. 6

Table 1 Presidential and parliamentary Electoral Systems in the presidential and semi-presidential democracies all over the world Presidential electoral system parliamentary electoral system Single-member district plurality system (FPTP) Proportional representation system (PR) Mixed-member majoritarian system (MMM) Mixed-member proportional system (MMP) Two-round system (majority system) Presidentialism:Ghana Semi-presidentialism : Mongolia (premier-presidentialism) Presidentialism: Argentina 4, Benin, Brazil, Chile, Colombia, Cyprus, El Salavador, Guatemala, Uruguay Semi-presidentialism: Premier-presidentialism: Austria, Cape Verde, Croatia, East Timor, Finland, Poland, Czech Republic, Portugal, Romania, Sao Tome and Principe, Slovakia, Slovenia, President-parliamentarism: Serbia, Namibia, Peru Semi-Presidentialism: Bulgaria, Georgia, Lithuania (premier-presidentialism) None Plurality system Presidentialism:USA 3 Presidentialism:Argentina, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Honduras, Paraguay, Venezuela Semi-presidentialism:Iceland (premier presidentialism) Presidentialism:Mexico, Philippines, South Korea Semi-presidentialism:Taiwan (president-parliamentarism) None In the research framework of this study, parliamentary party system is the mediator. In particular, this study focused on the degree of party fractionalization in parliament, for which the number of parties serve as an indicator. To calculate the effective number of parliamentary parties, this study adopted the equation proposed by Laakso and Taagepera (1979: 3 27): 2 N = 1 / ΣP i where P i is each party s proportion of all seats in parliament, and N is the effective number of parties in the parliament. The party seat rates for each country at each parliamentary used in this study were obtained from the following three databases: Election Resources on the 3 The United States presidential uses the Electoral College system. However, the presidential still uses the popular vote. Except for Maine and Nebraska, the entire electoral delegation of every other state is awarded to the candidate with the highest popular vote. Therefore, the system is based on the spirit of the plurality; thus, this study included the United States as a country that uses the plurality. 4 In this table, among countries that use plurality or two-round system, some repetitions are shown (such as Argentina, Brazil, Chile, and Colombia). It was because those countries adopted different presidential electoral systems in different periods, and their systems all changed from plurality to the two-round system. 7

Internet, ACE The Electoral Knowledge Network, and Adam Carr s Election Archive. For the definition of timing, this study used the 1-year interval to define honeymoon, counter-honeymoon, and midterm s. In other words, honeymoon and counter-honeymoon s are parliamentary s held 1 year after or before a presidential, respectively; whereas a midterm is a parliamentary held more than 1 year before or after a presidential. To calculate the effective number of parties in parliament, the unit adopted in this study was not the number of countries but the number of s. Take a group with countries A, B, and C as an example. If countries A, B, and C have held parliamentary s 4, 5, and 6 times, respectively; then the average effective number of parties in those parliaments is 4.0, 5.0, and 6.0, respectively. Subsequently, when this study calculated the average effective number of parties in parliament, we did not simply add up the average effective number of parties in each country s parliament and then divide them by the number of countries: (4.0 + 5.0 + 6.0)/3. Rather, we added up the total number of s multiplied by the effective number of parties in each country and then divided them by the total number of s in all countries: [4.0 4 + 5.0 5 + 6.0 6]/(4 + 5 + 6) = 5.13 effective number of parties. The rest of this study is presented is follows: Sections IV and V investigate democracies that use the presidential or semi-presidential system to discuss the combined influence of the presidential and parliamentary electoral systems and the timing on the parliamentary party system and to analyze the government types resulting from various combinations of parliamentary party system and constitutional system (presidentialism and semi-presidentialism). As illustrated in Fig. 2, Section IV discusses the relationships among the four variables in the research framework, the four in the gray block: presidential electoral system, parliamentary electoral system, electing timing, and parliamentary party system. Section V studies the relationships among the variables in the dashed line block: parliamentary party system, constitutional system, and government type. The discussion on them includes the direct influence of timing on government type. In summary, this study conducted preliminary research on the relationships among the variables in the research framework to discuss the causes of different party systems and government types in democracies using the presidential and semi-presidential systems. IV. Integrated influence of the presidential and parliamentary electoral systems and timing on party system A. Research hypotheses Studies have reported that the parliamentary electoral system directly affects the 8

parliamentary party system, whereas the presidential electoral system indirectly affects the parliamentary party system. The present study investigated whether timing also affects the parliamentary party system in democracies using the presidential or semi-presidential system. The general view on timing s effect on the parliamentary party system is that midterm s result in many parties, whereas honeymoon s and simultaneous s result in few parties; counter-honeymoon s result in a number of parties between those of the other three types. This study investigated whether this general view can be verified empirically. Moreover, to determine whether parliamentary party system is influenced by timing, this study investigated which of the two indirect factors, presidential electoral system or timing, had a greater effect. This study listed the democracies whose parliamentary s adopted proportional representation or the single-member district plurality system in the analysis framework in Table 2. Table 2 Combination of presidential electoral system and timing(analysis framework) timing honeymoon and counter-honeymoon midterm simultaneous electoral system two-round system I II III plurality system IV V VI When a parliamentary utilizes proportional representation, the parliamentary party system is generally directly affected by the parliamentary electoral system and results in a multiparty system. If timing has particular influences on the parliamentary electoral system, then the following three hypotheses will be verifiable in real-life experience. Hypothesis 1: In multiparty system countries whose parliamentary s use proportional representation, no matter whether the presidential electoral system uses a two-round system or the plurality system, the effective number of parliamentary parties resulting from midterm s is overall larger than that from counter-honeymoon s, whereas the effective number of parliamentary parties from counter-honeymoon s is larger than those from honeymoon s and simultaneous s. Table 2 uses the symbols I to VI to stand for combinations of different presidential electoral system and different timing. The relationship of the overall effective number of parties in each category is thus presented in Table 3: I and IV > II and V > III and IV. Hypothesis 2: In multiparty system countries whose parliamentary s use 9

proportional representation, if the presidential uses the two-round system, the effective number of parliamentary parties after a midterm should be overall larger than that after a counter-honeymoon, which is overall larger than those after honeymoon s or simultaneous s. Therefore, the relationship of the effective number of parliamentary parties is as shown in Table 3: I > II > III. Hypothesis 3: In multiparty system countries whose parliamentary s use proportional representation, if the presidential uses the plurality system, the effective number of parliamentary parties after a midterm should be overall larger than that after a counter-honeymoon, which is overall larger than that from honeymoon s or simultaneous s. Therefore, the relationship of the effective number of parties in parliament is as shown in Table 3: IV > V > VI. Table 3 Influence of the presidential and parliamentary electoral systems and timing on party system (hypothesis 1-3) Two-round system (I,Ⅱ,Ⅲ) Plurality system (IV,Ⅴ,Ⅵ) Midterm (I,IV) > > Counterhoneymoon (Ⅱ,Ⅴ) Honeymoon and simultaneous (Ⅲ,Ⅵ) Ⅰ Ⅱ Ⅲ > > > > IV Ⅴ Ⅵ If timing indeed affects the parliamentary party system and the presidential electoral system affects the parliamentary party system, then hypotheses 4 to 6 will be verifiable in real-life practice. Hypothesis 4: In multiparty system countries whose parliamentary s use proportional representation, no matter the timing, the effective number of parliamentary parties under the two-round presidential electoral system should overall be larger than that under the plurality system. As shown in Table 4, the relationship of the effective number of parliamentary parties is I, II, and III > IV, V, and VI. Hypothesis 5: In multiparty system countries whose parliamentary s use proportional representation, if the timing is midterm s, the effective number of parliamentary parties under the two-round presidential electoral system overall should be larger than that under the plurality system. As shown in Table 4, the relationship of the effective number of parliamentary parties is I > IV. Hypothesis 6: In multiparty system countries whose parliamentary s use proportional representation, if the timing is honeymoon s, the 10

effective number of parties in the parliament under the two-round presidential electoral system overall should be larger than that under the plurality system. As shown in Table 4, the relationship of the effective number of parliamentary parties is II > V. Hypothesis 7: In multiparty system countries whose parliamentary s use proportional representation, if the timing is honeymoon s or simultaneous s, the effective number of parties in the parliament under the two-round presidential electoral system should be overall larger than that under the plurality system. As shown in Table 4, the relationship of the effective number of parliamentary parties is III > VI. Table 4 Influence of the presidential and parliamentary electoral systems and timing on party system (hypothesis 4-6) Two-round system (I,Ⅱ,Ⅲ) > Plurality system (IV,Ⅴ,Ⅵ) Midterm (I,IV) Counterhoneymoon (Ⅱ,Ⅴ) Honeymoon and simultaneous (Ⅲ,Ⅵ) Ⅰ Ⅱ Ⅲ > > > IV Ⅴ Ⅵ The aforementioned hypotheses are the research hypotheses on presidential electoral system and timing s effect on parliamentary party system in countries whose parliamentary s use proportional representation. Similarly, the same analysis method can be applied to discuss presidential electoral system and timing s effect on parliamentary party system in countries whose parliamentary s use the single-member district plurality system. B. Empirical analysis 1. Verification of Hypothesis 1 This study examined all the democratic countries that use presidentialism and semi-presidentialism and proportional representation in their parliamentary s. The total number of midterm s that they have held is 115, and the average effective number of parties in parliament is 3.81. The total number of counter-honeymoon s that they have held is 44, and the average effective number of parties in parliament is 3.21. The total number of honeymoon s that 11

they have held is 14, and the average effective number of parties in parliament is 3.11. The total number of simultaneous s that they have held is 71, and the average effective number of parties in parliament is 2.99. The combined number of honeymoon s and simultaneous s is 85, and the average effective number of parties in parliament is 3.00. Therefore, the ranking of effective number of parties in parliament caused by differences in timing is midterm (3.81) > counter-honeymoon (3.21) > honeymoon and simultaneous (3.00). These findings verify Hypothesis 1. To verify whether the difference in the effective number of parties between the three groups of countries reached statistical significance, analysis of variance (ANOVA) was conducted. The results (Table 5) revealed statistically significant differences between the effective numbers of parties in the parliament. The Scheffe test revealed that among the countries that used proportional representation, the effective number of parties in the midterm subgroup was significantly larger than both that of the counter-honeymoon subgroup and that of the honeymoon and simultaneous s subgroup. Table 5 Election timing and effective numbers of parties in the parliament (ANOVA) Counter- Honeymoon and Midterm honeymoon simultaneous > > (1) (2) (3) Average effective 3.81 3.21 3.00 numbers of parties in the (n=115) (n=44) (n=85) parliament F-value=19.064 P<0.001*** ANOVA Scheffe test: (1)>(2) ; (1)>(3) 2. Verification of Hypothesis 2 This study examined the presidential and semi-presidential countries that use the two-round system in their presidential s and proportional representation in their parliamentary s. The number of midterm s that they have held is 78, and the average effective number of parliamentary parties is 3.83. The total number of counter-honeymoon s that they have held is 31, and the average effective number of parties in parliament is 3.17. The total number of honeymoon s that they have held is 12, and the average effective number of parties in parliament is 3.10. The total number of simultaneous s that they have held is 34, and the average effective number of parties in parliament is 3.10. The combined 12

number of honeymoon s and simultaneous s held is 46, and the average effective number of parties in parliament is 3.10. Therefore, the ranking of the effective number of parties in parliament at different timings is midterm (3.83) > counter-honeymoon (3.17) > honeymoon and simultaneous (3.10). These findings verify Hypothesis 2. To verify whether the difference in the effective number of parties between the three groups of countries reached statistical significance, ANOVA was conducted. The results (Table 6) revealed statistically significant differences in the effective numbers of parties in parliament. The Scheffe test revealed that among the countries that used proportional representation, the effective number of parties in the subgroup two-round system & midterm was significantly larger than both that of the two-round system & counter-honeymoon subgroup and that of the two-round system & honeymoon and simultaneous s subgroup. Table 6 Two-round presidential system, timing and effective numbers of parties in the parliament (ANOVA) Average effective numbers of parties in the parliament ANOVA Two-round system Two-round system & & Counter-honeymoon midterm (1) (2) 3.83 3.17 (n=78) (n=31) F-value=7.849 p<0.001*** Scheffe test: (1)>(2) ; (1)>(3) Two-round system & Honeymoon and simultaneous (3) 3.10 (n=46) 3. Verification of Hypothesis 3 This study examined the presidential and semi-presidential countries that use the proportional representation in their parliamentary s and plurality system in their presidential s. The total number of midterm s that these countries have held is 37; the average effective number of parties in parliament is 3.78. The total number of counter-honeymoon s that these countries have held is 13, and the average effective number of parties in its parliament is 3.21. Only Chile has held honeymoon s, twice, and the average effective number of parties in its parliament is 3.12. The total number of simultaneous s that they have held is 37, and the average effective number of parties in their parliaments is 2.88. The combined number of honeymoon s and simultaneous s held is 39, and 13

the average effective number of parties in parliament is 2.89. Therefore, the ranking of the effective numbers of parties in parliament for different timings is midterm s (3.78) > counter-honeymoon s (3.32) > honeymoon s and simultaneous s (2.89). These findings verify Hypothesis 3. To verify whether the difference in the effective number of parties between the three groups of countries reached statistical significance, ANOVA was conducted. The results in Table 7 reveal statistically significant differences in the effective number of parties in parliament. The Scheffe test revealed that among the countries that adopted proportional representation, the effective number of parties in the plurality & midterm subgroup was significantly larger than that of the plurality & honeymoon and simultaneous s subgroup. Table 7 Plurality presidential system, timing and effective numbers of parties in the parliament (ANOVA) Average effective numbers of parties in the parliament ANOVA Plurality system Plurality system & & midterm Counter-honeymoon (1) (2) 3.78 3.32 (n=37) (n=13) F-value=21.121 p<0.001*** Scheffe test: (1)>(3) Plurality system & Honeymoon and simultaneous (3) 2.89 (n=39) 4. Verification of Hypotheses 4, 5, 6, and 7 To facilitate observation and comparison, various data from earlier tables are integrated and presented in Table 8 to verify Hypotheses 4 to 7. For the verification of Hypothesis 4, Table 8 shows that among the countries adopting proportional representation for parliamentary s, the effective number of parties in parliament for countries using the two-round system as their presidential electoral system was 3.48, whereas that for countries using the plurality system as their presidential electoral system was 3.32. Because 3.48 > 3.32, the evidence supports Hypothesis 4. To verify whether the difference in the effective numbers of parties between the two groups of countries reached statistical significance, a t-test was conducted. The results in Table 9 reveal that the effective number of parties in the parliaments of the countries using the two-round system was nonsignificantly larger than that of the countries using the plurality. 14

Table 8 Effective number of parties in the democracies whose parliamentary electoral system is PR Election timing Honeymoon Simultaneous Counter- Midterm honeymoon (3.11) (2.99) Presidential (3.81) (3.21) 3.00 electoral system 3.10 3.10 Two-round system 3.83 3.17 (3.48) 3.10 Plurality system (3.32) 3.78 3.32 3.12 2.88 2.89 Table 9 Presidential electoral system and effective numbers of parties in the parliament (t-test) Two-round system Plurality system Average effective numbers of parties in the parliament t-test 3.48 (n=155) t=1.297 p>0.05 3.32 (n=89) Subsequently, Hypothesis 5 was verified. Among the countries that adopted proportional representation in their parliaments (Table 8), the effective number of parties in parliament for countries that held midterm parliamentary s and adopted the two-round system as their presidential electoral system was 3.83, whereas that for countries holding midterm parliamentary s and using the plurality system as their presidential electoral system was 3.78. Because 3.83 > 3.78, the evidence supports Hypothesis 5. To verify whether the difference in the effective numbers of parties between the two groups of countries reached statistical significance, a t-test was conducted. The results in Table 10 reveal that the effective number of parties in the parliaments of the group of countries holding midterm s and using the two-round system was nonsignificantly larger than that of the countries holding midterm s and using the plurality system. 15

Table 10 Midterm, presidential electoral system and effective numbers of parties in the parliament (t-test) Midterm & two-round system Average effective numbers of parties in the parliament t-test 3.83 (n=78) t=0.257 p>0.05 Midterm & plurality system 3.78 (n=37) Next, Hypothesis 6 was verified. Among the countries that adopted proportional representation in their parliaments (Table 8), the effective number of parties in parliament for countries that held counter-honeymoon parliamentary s and adopted the two-round system as their presidential electoral system was 3.17, whereas that for countries holding counter-honeymoon parliamentary s and using the plurality system for their presidential s was 3.32. Because 3.17 < 3.32, the evidence did not support Hypothesis 6. Because the evidence disapproved the hypothesis, the t-test results (in Table 11) of course failed to reveal the effective number of parties in the parliaments of the group of countries holding counter-honeymoon s and using the two-round system to be significantly larger than that of the countries holding counter-honeymoon s and using the plurality system. Table 11 Counter-honeymoon, presidential electoral system and effective numbers of parties in the parliament (t-test) Counter-honeymoon Counter-honeymoon & & two-round system plurality system Average effective numbers of parties in the parliament t-test 3.17 (n=31) t=-0.741 p>0.05 3.32 (n=13) Finally, Hypothesis 7 was verified. Among the countries that adopted proportional representation in their parliaments (Table 8), the effective number of parties in parliament for countries holding honeymoon s or simultaneous parliamentary s and using the two-round system for their presidential electoral system was 3.10, whereas that for countries holding honeymoon or simultaneous parliamentary s and using the plurality system for their presidential s was 2.89. Because 3.10 > 2.89, the evidence supports Hypothesis 7. However, the t-test results in Table 12 indicate that the effective number of parties in the 16

parliaments of the group of countries holding honeymoon and simultaneous s and using the two-round system was nonsignificantly larger than that of the countries holding honeymoon and simultaneous s and using the plurality system. Table 12 Honeymoon and simultaneous, presidential electoral system and effective numbers of parties in the parliament (t-test) Honeymoon and Honeymoon and simultaneous simultaneous & & two-round system plurality system Average effective numbers of parties in the parliament t-test 3.10 (n=46) t=1.232 p>0.05 2.89 (n=39) 5. Overall discovery The data in Table 8 reveal the influence of timing on parliamentary party systems. Among the countries whose parliamentary s used proportional representation, those who held midterm s had the largest effective number of parties, followed by those holding counter-honeymoon s, whereas those holding honeymoon and simultaneous s had the smallest effective number of parties. As Table 8 shows, 3.81 > 3.21 > 3.00. In addition, when all the countries who used proportional representation for parliamentary s were divided into two groups on the basis of whether their presidential s used the two-round system or the plurality system, the results showed that in both groups, countries holding midterm s had the highest effective number of parties in parliament, followed by those holding counter-honeymoon s, and then those holding honeymoon and simultaneous s. As Table 8 shows, 3.83 > 3.17 > 3.10 and 3.78 > 3.32 > 2.89, respectively. The rankings of the aforementioned numbers were completely in accordance with those of Hypotheses 1 to 3 presented in Table 3. In addition, ANOVA revealed that among the three groups of countries countries whose parliamentary s used proportional representation; countries whose parliamentary s used proportional representation and whose presidential s used the two-round system; and countries whose parliamentary s used proportional representation and whose presidential used the plurality system when their parliamentary s differed in timing, the effective numbers of their parties in parliament differed significantly. The results indicated that timing had a substantial influence on the parliamentary party system. The data in Table 8 also reflect the influence of presidential electoral system on parliamentary party system. Among countries whose parliamentary s used proportional representation, when the influence of timing was overlooked, the 17

effective number of parties of the parliaments in countries using the two-round system for presidential s is larger than that in countries using the plurality system for their presidential s. As Table 8 shows, 3.48 > 3.32. However, when the countries were divided into three groups based on timing midterm, counter-honeymoon, and honeymoon and simultaneous the phenomenon that the effective number of parties in the parliament is larger when the presidential uses the two-round system was not observed in all groups. As Table 8 shows, 3.83 > 3.78, 3.17 < 3.32, and 3.10 > 2.89, respectively. The aforementioned data were not completely in accordance with Hypotheses 5 7. When a t-test was utilized to examine the data, the difference in the effective number of parties in parliament between the three groups was nonsignificant. Therefore, although the presidential electoral system had its influence on parliamentary party system, the influence was not significant when timing was taken into consideration. Empirical evidence showed that compared with presidential electoral system, timing had a greater influence on party system. The aforementioned section discusses the influence of presidential electoral system and timing on the party systems in the countries whose parliamentary electoral system is proportional representation. In theory, the same analysis method can be applied to countries whose parliamentary electoral system is the single-member district plurality system. However, an investigation on democracies whose parliamentary electoral system used the single-member district plurality system revealed that most of these countries use parliamentarism, and they lacked presidents with real power elected directly by the people. Only three democracies (Mongolia, the United States, and Ghana) use the single-member district plurality system as their parliamentary electoral system and hold direct presidential s. Because of the scarcity of such cases, cross-analysis of the influence of presidential electoral systems and timing on party system in those countries cannot be conducted. This study applied the aforementioned analysis structure to the results from these three countries (Table 13), but no clear influence trends were revealed. Table 13 Effective number of parties in the democracies whose parliamentary electoral system is FPTP Election timing Midterm Presidential (1.93) electoral system Two-round system Mongolia (1.91) (1.73) Plurality system USA (1.95) (1.96) Counterhoneymoon None None Honeymoon None None Simultaneous (1.96) Ghana (2.09) USA (1.94) 18

V. Government types under different constitutional systems and party systems After discussing the influences of presidential electoral system, parliamentary electoral system, and timing on party system, this section discusses the three variables in the dashed line block in the analysis framework in Fig. 2, namely party system, constitutional system, and government type. In other words, we discuss what type of government is formed under the various combinations of parliamentary party system, presidentialism, and semi-presidentialism. A. Combinations of the presidential system and the two-party system In countries that use the presidential system, if the parliamentary party system is a two-party system, then one party in parliament generally controls a majority of seats. When the president is of the majority party, the government is a unified government; otherwise, it is a divided government. Simultaneous s or honeymoon s typically increase the chance of a unified government. By contrast, midterm s increase the chance of a divided government. In summary, the combination of the two-party system and presidential system usually leads to either a unified government or a divided government. Election timing also directly influences the chances of the occurrence of a unified or divided government. In all democracies, the presidential system and two-party system combination is extremely rare; the United States is the only existing case. B. Combination of the presidential system and the multiparty system In countries using the presidential system, if the parliamentary party system is a multiparty system, typically no party in parliament controls a majority of seats. Because the president belongs to a party that does not control a majority in parliament, he or she cannot control the majority of MPs. Thus, a divided government is the norm in countries using the presidential system with a multiparty system (Mainwaring, 1993). The combination of presidentialism and multiparty system is common in Latin America, such as in Argentina, Brazil, Chile, Colombia, Paraguay, Uruguay, Guatemala, the Dominican Republic, and Honduras. This study examined the direct influence of timing on government type. Simultaneous s and honeymoon s usually favor the party of the presidential candidate (or president-elect), which is typically one of the major parties. Regarding whether timing increases the chance of the president s party acquiring a majority of seats in parliament and thus results in a unified government, the present consensus in academia is that when a country has multiparty system and adopts presidentialism, a divided government is unavoidable. However, this view 19

overlooks the possible influence of timing, which therefore is worthy of further investigation. This study observed the five most recent parliamentary s of all the 15 democracies in the world that used the presidential system (i.e., countries using the presidential system and ranked 2.5 or below by Freedom House) and the multiparty system (i.e., whose effective number of parties in the parliament was 2.5 or above). As shown in Table 14, of the 75 s (5 times 15 countries), 46 were simultaneous s, 7 were counter-honeymoon s, 21 were midterm s, and 1 was a honeymoon. In those five most recent parliamentary s, nearly a quarter (23.9%) of simultaneous s resulted in a unified government, indicating that in democracies that used the presidential system, even when they used the multiparty system, they did not necessarily result in a divided government. Simultaneous presidential and parliamentary s might increase the chances of forming a unified government. As for whether honeymoon s also increased the opportunity to form a unified government, because the sample size was small (with only one among the 75 s observed in this study), no clear trend was revealed. Table 14 Past five parliamentary s resulting in unified government (the presidential democracies which have multiparty system) Counter-ho Simultaneo Midterm Honeymoon neymoon us Numbers of s 46 7 21 1 75(total) Numbers of parliamentary s resulting in unified government 11 1 0 0 12(total) Percentage 23.9% (11/46) 14.2% (1/7) 0% (0/21) 0% (0/1) C. Combinations of the semi-presidential system and the two-party system In countries using the semi-presidential system, if the parliamentary party system is a two-party system, the parliament typically has a party that controls a majority of seats. When parliamentary s are held simultaneously with the presidential or during the honeymoon period, more elected MPs tend to be from the same party as the president. In this situation, the president typically assigns a person from his party (which is also the party of a majority of MPs) to form a cabinet. The relationship between the president and parliament in this situation is one of unified government, and the relationship between the cabinet and parliament is one of a one-party cabinet (Fig. 3). For example, the government of Taiwan has had this configuration since 2008. 20

parliament (two-party system) unified government president One-party cabinet Figure 3 If the timing for a parliament is midterm, long before or after the presidential, more MPs tend to be from a different party from that of the president. If the constitutional system is premier presidentialism under the semi-presidential system, because the premier is, in reality, chosen by the parliamentary majority, the president typically assigns a person from the majority party to form the cabinet. In other words, the cabinet is typically formed by one party that controls a majority of seats. In this situation, the government type in terms of the relationship between the president and parliament is divided government, whereas the relationship between the parliament and cabinet is that of a one-party cabinet. The relationship between the president and cabinet is that of cohabitation (Fig. 4). For example, Mongolia has functioned under this situation since 2016. parliament (two-party system) Divided government president one-party cabinet Figure 4 However, when the president is of a different party from that of the parliamentary majority and when the constitutional system is president parliamentarism under the semi-presidential system, because the president has the power to appoint and dismiss the premier, if he or she insists on confronting the parliament, ignoring the majority party, and appointing a person from his or her party to form a cabinet, then the government type, in terms of the relationship between the president and parliament, is divided government. The relationship between the cabinet and parliament, because the cabinet members are not from the majority party of the parliament, is a minority cabinet (Fig. 5). In Taiwan, this type of government may be formed in the future if the president is of a different party from that of the parliament majority. 21

parliament (two-party system) Divided government president Minority government Figure 5 As with the situation discussed in the presidential system section earlier, timing may have a direct influence on government type in countries using the semi-presidential system. Simultaneous or honeymoon s typically increase the chances of a unified government and a one-party cabinet being formed (Fig. 3). By contrast, under the premier presidential system, midterm s increase the chances of a divided government, a one-party cabinet, and cohabitation (Fig. 4). Under the presidential parliamentary system, the chances of a divided government and a minority cabinet increase (Fig. 5). Overall, few democracies use the semi-presidential system and have two-party system simultaneously, with only three cases: Mongolia, Cape Verde, and Taiwan. D. Combination of the semi-presidential system and the multiparty system In the countries that uses the semi-presidential system, if the party system in parliament is multiparty, typically no party can win a majority of seats in parliament. When the president s party does not control a parliamentary majority, the relationship between the president and the parliament is that of divided government. In that situation, the cabinet type can be one of the following two: First, if the constitutional system is the premier presidential system, the president typically respects the alliance formed by various parties that can together control a majority of seats in parliament and appoints the person supported by the alliance as premier. In this situation, the government type is coalition cabinet in terms of relationship between the cabinet and parliament, and the relationship between the president and the cabinet is that of cohabitation (Fig. 6). Among democracies, the combination of the premier presidential system and multiparty system is common. Second, when the constitutional system is the presidential parliamentary system, the president may disregard the opinions of the parliamentary parties and appoint individuals from his party to form the cabinet. In this situation, the cabinet type, in terms of the relationship between the cabinet and parliament, is a minority cabinet (Fig. 7), such as that of present-day Serbia and Peru. 22