EXECUTIVE POWER
Executive power is the power to put into action that which the legislature has created. It encompasses not just a single official, such as a president, but also the cabinet and the various bureaucratic organs of the state. In this chapter we will first examine the basic choices considered by the framers, then move on to a broader description of executive power in contemporary democracies, and finally conclude with a list of differences between the United States and our other thirty democracies.
INSTITUTIONAL OPTIONS
As was noted in the previous chapter, the Philadelphia conventioneers focused most of their debate on the design and scope of the legislature. However, they also had to consider the shape of executive authority. They went into the convention with two models that they ended up rejecting based on experience. Their immediate experience under the Articles of Confederation was a legislature that lacked an actual executive. While there was a “president” of the Congress, it was a largely ceremonial role, and since one of the deficiencies of the Articles was its lack of executive power, this model was not to be emulated. Likewise they rejected the notion of a hereditary monarchy (although Alexander Hamilton did propose a plan that would have created an executive for life).
If we examine the actual proposals made at the convention (table 8.1), we see a number of possibilities: from a singular to a plural executive, and from one chosen by the legislature to one chosen by electors selected by the voters. There was also some debate in Philadelphia over the possibility of a popularly elected executive (Madison 1987: 368–69). In the end we know that the final selection was a chief executive wholly separate from the legislature, to be selected by an indirect process.
Table 8.1. Five versions of the US executive at the Philadelphia Convention
Of course, these options do not encompass all that a modern constitutional planner needs to consider. The remainder of this section examines the concepts and interactions needed to understand the role of the executive in a representative democracy. The first subsection addresses basic roles (head of state versus head of government; the second addresses the question of whether legislative-executive relations are based on hierarchical or transactional relationships; the third deals with the issue of cabinets; and the last, with executive election.
Basic Roles
The executive is the branch of government charged with two overarching tasks. First, the executive, in the function of head of state, is supposed to embody the nation as a whole and represent it in international affairs. These are, of course, primarily ceremonial functions; more important are the political and administrative roles of the executive, which are handled by the head of government. Heading the government means overseeing the day-to-day administration of the government and making key political choices regarding policy options. The head of government presides over the cabinet, which is comprised of the officials, known as ministers or secretaries, who head the various departments—for instance, defense, finance (treasury), transportation, and so on. Democracies vary in whether the roles of head of state and head of government are held by different officials or combined in a single person. The head of state may be a purely ceremonial figure, sometimes even a hereditary monarch. In this case of the ceremonial head of state, the role of head of government is always separate and is typically known as a prime minister. In other systems, notably the United States, the head of state—the president—is popularly elected, and also serves as head of government. As we shall see, still other countries have both an elected president, who serves as head of state but may be more than just ceremonial, and a prime minister to head the government. Table 8.2 details the heads of state and government for our thirty-one democracies. The table indicates that nine of our thirty-one democracies have a head of state called “president” who is not popularly elected. These include parliamentary-selected presidents who also serve as head of government, as in South Africa and Switzerland; heads of state selected by the parliament (the Czech Republic, Greece, Hungary, and Israel); and heads of state selected by electoral colleges that are made up of members of the national legislature plus regional delegates (Germany, India, and Italy).
In considering the ways in which executive power is structured in a given polity, one of the most important questions concerns the ways in which the executive works with the legislature. The reason this is so important is that in many systems—though not in the United States—executive authority originates from political parties represented within the executive and survives in office only so long as it maintains the support of a majority of legislators. Moreover, whatever the rules for the origination and survival of executive authority, nearly all major policy changes must be approved by the legislature, or may be vetoed by it. Thus we consider executive-legislative relations first among a series of variations in the origin and survival of executive authority.
Table 8.2. Head of state versus head of government in thirty-one democracies
Notes
a The monarch of Australia, Canada, and New Zealand are the same as that of England (currently Elizabeth II).
b Indicates presidency is not popularly elected. Note that this changed for the Czech Republic in 2012, but that is outside the scope of our study.
c The president of the Swiss Confederation is the presiding officer (but first among equals) of the seven-member Federal Council, and serves only a one-year term.
Executive and Legislatures: Transactional or Hierarchical Relations?
In considering matters of the constitutional design of executive authority, a critical characteristic is the way in which a given system juxtaposes hierarchical and transactional relations. A hierarchy results when one institution is constitutionally or otherwise subordinate to another, whereas transactions result when two coequal actors must share power in order to accomplish their objectives. All democracies entail mixes of hierarchical and transactional relations, but in some systems one or the other tendency is more dominant. In regards to this basic interaction, think back to figure 1.1, which diagrams the basic relationship we are discussing here. The United States Constitution represents the dominance of transactional relations, whereas the British political system, before 2010 (as we shall see) represents the most hierarchical of relations. Most other systems have some combination of the two types of relationships.
If we start with the basic dichotomy of presidential versus parliamentary systems, we can see that the constitutional relationship of the executive and legislature is fundamentally different in the two types. Figure 8.1 depicts the two systems in their purest forms, where solid lines indicate hierarchical relations with arrows running from principals to agent. Bolded lines represent transactional relationships, with arrows at either end noting the coequal nature of the authority of the actors bargaining over and concluding these transactions. Note that the hierarchy is depicted such that the principals are superior in a vertical relationship to their agents, while the transactions are depicted horizontally, because they occur between coequals. In any democracy, by definition, the voters are the ultimate principals, as they make their choices as to who their representative agents will be via the election process. In the presidential form of government that typifies the United States, voters have at least two choices to make in national elections, because they elect both a legislature and an executive. Of course, there may actually be three choices, because the legislature may be bicameral with both chambers elected, as in the United States, and as we discuss in the chapter on legislatures; however, we ignore bicameralism in the figure for simplicity. In this constitutional design, the elected president is both head of state and head of government. In a parliamentary system, on the other hand, the voters elect only a legislature (which again, may be bicameral). Executive authority then depends on the outcome of legislative elections, and the executive remains in office only so long as it enjoys the confidence of the majority in the legislature. That is, it remains in office until a majority of legislators votes it out, or until it is defeated in the next election.
Fig. 8.1. Delegation and hierarchical versus transactional relations in presidential and parliamentary democracy
Key: Hierarchical relations are depicted with solid lines, with the arrows showing the direction of authority from principal to agent. Transactional relations are shown as a bold two-headed arrow.
*Israel had a hybrid system (directly elected prime minister) from 1996 to 2003.
The relationship between the executive and legislature in the United States or other presidential democracies is transactional for two critical reasons. First, the executive and legislature have fixed terms, meaning that neither can dismiss the other when political disagreements arise. Second, both actors must cooperate to produce legislation, or else policy stasis results, with neither accomplishing its objectives (unless the objective is simply to say no to the other). In other words, the president has veto or other legislative powers so that he or she is, in a sense, another (unipersonal) “chamber” of the legislature, as well as the executive.
The relationship between the executive and legislature is hierarchical in the parliamentary system. What this means is that the executive—the prime minister and the cabinet that he or she heads—serves only so long as it enjoys the confidence of the majority in the parliament. If this majority is held by a single party, then we have the purest hierarchy that we can have in a democracy. In fact, while the cabinet is the agent of the majority party, the actual functioning of the political system is one characterized by executive dominance over the legislature. How can an agent dominate the principal to whom it should be subordinated? The answer lies in the dual role of the prime minister, who is not merely head of government but also head of the majority party. Because the party members in the parliament usually have no political interest in contradicting their own party leader on the most important legislative matters, prime ministers rarely suffer significant defeats in the parliament, lending an appearance of dominance: What they propose almost always passes.1 Dominance is further reinforced because a majority party would seldom exercise its vote of no confidence against an executive that is made up of its own leadership, for to do so would be to invite political instability and possible loss of power to the opposition at a new election.
When there is no majority party in the legislature of a parliamentary system, the hierarchy of executive dependence on parliamentary confidence remains. However, now the policy-making process is complemented by a large dose of transactional relationships. The transactions that are most important are generally not, however, between the executive and legislative institutions, as in the presidential system. Rather, they are between separate political parties that must share power in order to govern, because in the absence of a majority, no one party can govern alone. In the case of a coalition cabinet, two or more parties divvy up cabinet positions and transact over the policy program of the government, as we discuss in more detail later. In other cases, when there is no majority party, one or more parties may form a minority cabinet, in which the parties sharing cabinet positions do not control a majority in the parliament. In this case, there is a transactional relationship between the executive and legislature or, more specifically, with one or more parties in the parliament that otherwise might be tempted to join an opposition-led no-confidence motion to bring down the minority government. We return to the composition of cabinets later, but first we must consider other forms of government besides presidential and parliamentary.
Figure 8.1 shows that our thirty-one democracies include seven presidential systems and eighteen parliamentary systems. What about the others? These six are hybrids of various sorts. One is the Swiss system, in which the leaders of multiple parties in the parliament constitute a power-sharing executive, much as in parliamentary systems. However, there is no procedure by which the parliament may vote no confidence, meaning that there is a stronger separation of powers between executive and legislature than there is in a parliamentary system. There was also a hybrid in Israel during the time period of our analysis: From 1996 to 2003, the Israeli prime minister was directly elected (like the head of government in a presidential system),2 yet remained subject to parliamentary confidence. This experiment was widely considered to have been a mistake and was abandoned in favor of a return to the previous pure parliamentary system beginning in 2003 (Samuels and Shugart 2010: 179–90). The more common hybrid form is the semi-presidential type, which includes elements of both presidentialism and parliamentarism. Its structure of authority is depicted in figure 8.2, and is represented among our thirty-one democracies by Austria, Finland, France, Poland, and Portugal.3
Fig. 8.2. Delegation in semi-presidential systems
Key: Hierarchical relations are depicted with solid lines, with the arrows showing the direction of authority from principal to agent. Transactional relations are shown as a two-headed bold arrow.
In a semi-presidential system there is a popularly elected president, who serves as head of state. However, distinct from (pure) presidential systems like the United States, the president in a semi-presidential system is not constitutionally also the head of government. Instead there is a prime minister (or “premier”) who is subject to the confidence of the legislative majority. In a semi-presidential system, then, there is a hierarchical relationship between the head of government and the parliamentary majority—exactly as we described for a parliamentary system. However, the president and prime minister may exist in some form of transactional relationship, depending on the precise constitutional allocation of powers.4 France and Poland, as well as Finland until recently,5 offer examples of systems in which the president is a central political figure. If the president is the most important leader of his party, and this party in turn dominates the parliament, then the prime minister may be politically—even though not constitutionally—subordinated to the president. This was the case in France in nearly all cases from 1965 to 19866 and again since 2002 (as well as at times between these periods).
Some of these presidents, including in France, have the right, under certain conditions, to dissolve the parliament. However, this power is not as formidable as it may seem, because it simply returns to the ultimate principal, the voters, the power to decide what the new parliamentary majority will be. This worked to the president’s advantage in France in 1981, when, shortly after his own election, President François Mitterrand of the Socialist Party dissolved the conservative-dominated parliament, which the voters then replaced with a Socialist-dominated one, allowing Mitterrand to appoint a Socialist premier. However, a similar gambit failed miserably for Mitterrand’s successor, conservative Jacques Chirac, in 1997. Hoping to consolidate and extend his allies’ grip on the parliament, Chirac exercised his right of dissolution to hold an assembly election a year earlier than required. The voters responded by returning a majority for the left-wing bloc! As a result, Chirac was forced to accept the Socialist leader, Lionel Jospin, as premier. Periods such as these are known as cohabitation, defined as a president and premier from opposing parties, and the president’s party not represented in the cabinet (Elgie 2011: 59–60, Samuels and Shugart 2010: 44–46). This concept of cohabitation is utterly different from what in the United States is called divided government. Superficially they are the same: The president and the legislative majority (either House or Senate in the United States) are from different parties. However, in the United States, even with divided government, the president remains the head of government with freedom to control the cabinet and exercise executive full authority. By contrast, under cohabitation, the cabinet is under the control of the legislative majority and not of the president. Later on we will discuss the distinctiveness of situations in which the president and legislature are held by different parties, as well as the composition of cabinets.
Single-Party Cabinets versus Executive Power-Sharing
As we noted, the president in a presidential system is by definition the head of government and has a fixed term of office. In presidential systems, then, the cabinet’s origin depends on choices made by the voters, who choose the head of government, as well as choices regarding cabinet appointments made by the voter’s agent, the president. In parliamentary systems, on the other hand, the head of government is a prime minister who, along with the rest of the cabinet, serves only with the confidence of the majority in the legislative assembly. Thus the origin of the cabinet depends on the outcome of parliamentary elections and of transactions among leaders of different parties when none of them wins a majority. Similarly, the cabinet’s survival in power depends on not losing parliamentary confidence. In semi-presidential systems, the president typically initiates the selection of a premier, but the president’s choice, as well as whether the premier and cabinet remain in office, depends on the parliamentary majority.
In parliamentary or semi-presidential systems, if a single party holds a majority in the legislature, it almost always reserves all the seats in the cabinet for itself. If there is no single party with a majority, the result is either a coalition or minority cabinet, because two or more parties must transact in order to shepherd legislative proposals through the parliament as well as to maintain control of the executive.
As such, it is important to underscore that cabinet formation is fundamentally different in presidential systems than in parliamentary, or even semi-presidential, systems. While US presidents must gain legislative approval (as discussed in chapter 7) for appointments to the cabinet, and presidents in some other presidential systems (notably Brazil) negotiate cabinets with parties in the legislature, presidents do not depend on legislative support to remain in office, as do prime ministers. While presidents may choose to include members of other parties in their cabinet, such cabinet members are far less agents of their parties as is the case in parliamentary coalitions, and can always be dismissed by the president, whose term is fixed. Indeed, in the US case, the significance of party identification in the cabinet is so minimal that it is rarely discussed, and even when a cabinet member was identified with a party different from that of the president, it does not mean that the cabinet member has loyalties outside of his or her job within the administration. For example, President Barack Obama’s first secretary of defense was Robert Gates, who had held that same job under President George W. Bush. That Bush was a Republican and Obama a Democrat really had no special bearing on Gates’s role in that job.
In parliamentary systems, if no one party is in a position to form a government by itself, cabinets are formed as a result of transactions among parties. Often, two or more parties will form a coalition in which the leader of one of the transacting parties becomes prime minister, and leaders of some of the other parties that join the coalition will obtain other cabinet positions. Each of the parties that enter a coalition becomes a veto player, because the consent—or at least acquiescence—of each party is typically needed to enact a policy change. If a party determines collectively that it is unwilling to support some government policy, it can withdraw from the government, causing the need for a new coalition to form or perhaps resulting in early elections. However, a coalition partner will not just blithely bring down a government, because in doing so it risks losing influence over the next government that is formed or being blamed for creating a “crisis.” Most disagreements among coalition partners are resolved through interparty transactions well short of coalition breakdown.
Fig. 8.3. Delegation and creation of veto players in a parliamentary system
Key: Solid arrows indicate delegation relationships, with the direction of the arrow showing the flow of authority from principal to agent. Dashed arrows indicate accountability of agents to principals. The two-headed arrow shows transactional relationship.
This relationship is captured in figure 8.3, which details a theoretical legislature with four parties, A, B, C, and D. In the figure, we follow the chain-of-delegation argument that relates back to the discussion in chapter 1 and to figure 1.1 but adds layers of detail. Specifically, blocs of voters, acting as principals (since they are citizens who hold ultimate power in a democratic setting), delegate to their agents, politicians, who are members of specific political parties who win seats in the legislature. For simplicity, we will assume a unicameral legislature. So the process of forming a government is one of assembling a transaction among separate agents of different blocs of voters that can come together to control the one veto gate. Thus we can speak of a coalition as two or more veto players controlling the single veto gate.7 Each party is depicted on a left–right scale to simulate different policy positions, and each box is a different size to indicate seat shares. In this example, parties C and D jointly have a majority of seats, and thus transacted to form a government, by which we mean the following:
1. One of them (likely C, the largest party) provides the prime minister and some other ministers (secretaries, in US terms) of the cabinet, while the other party supplies other ministers (and perhaps a deputy PM).
2. The two parties compromise on policies, implementing some priority items of C and some of D, splitting the difference on other issues and agreeing to disagree on some issues that are especially contentious between the two parties. Usually the policies they agree to in their transaction are published in a coalition agreement, which becomes the program of the government.
3. Ongoing policy transactions take place between the parties, meaning primarily within the cabinet, where the ministers act as both specialists in the policy area (the “portfolio”) to which they have been assigned, and as agents of their respective party.
4. This transactional relationship makes parties C and D veto players, because both have to agree to policies for them to be passed by the one veto gate (in this example, a unicameral legislature).
Now, it should be noted that other coalitions are possible, for example: A, B, and C in a three-party cabinet. However, this would create a more complex transactional relationship because three veto players would have to be satisfied. Certainly the ideological space encompassed by the three parties might make such transactions more difficult than in a two-party arrangement. Further, party A, being the most left-leaning of our hypothetical parties, might find a three-way coalition to be less attractive than waiting for a future election when maybe A plus B could win a majority of seats and transact in a two-party coalition of their own (without C). Yet another possibility would be a minority government, in which the largest party, C, governs alone, meaning that it has all of the cabinet positions. In such a case, it would be able to remain in government by alternately transacting with the party to its left (B) or to its right (D). Which outcome prevails would depend on the preferences of each party and its leaders and which option gives it what it considers its best payoffs from the transactions.8 Box 8.1 has an example of two cases of coalition formation in parliamentary systems: Germany following two elections, 2005 and 2009.
. . .
BOX 8.1. COALITION EXAMPLE: GERMANY 2005 AND 2009
In Germany in 2005, the election resulted not only in no party having a majority, but also neither of the most likely coalitions of parties having a majority. Due to the proportional electoral system (specifically, Mixed-Member Proportional or MMP; see chapter 5) in Germany, and the country’s multiparty system, it is extremely unlikely for any single party to win a majority of seats. However, since the 1990s, there has been a general expectation that the country would be governed by either of two possible coalitions: a center-right bloc consisting of the Christian Democrats and Free Democrats, or a center-left formation of the Social Democrats and Greens.
Leading up to the 2005 election, the governing coalition consisted of the Social Democrats (SPD) and Greens. In the campaign, the two main options presented to voters were either a continuation of the incumbent coalition or a change to a center-right government. Table 8.3 shows the results of the September 2005 election. The SPD remained the largest party, with 222 seats. However, a majority required 308 seats, and even with the inclusion of the Greens’ 51 seats, this formation was far short of being able to govern. The alternative government would have consisted of the two conservative parties, the Christian Democratic Union (CDU) and the Christian Social Union (CSU),a together with the Free Democratic Party (FDP, also referred to as the Liberals). These parties had combined for 287 seats, and so this coalition also would be short of a majority.
With neither of the two expected coalitions having majority support in the Bundestag (first chamber of the parliament), several rounds of interparty bargaining ensued, during which attempts at various coalitions were made. In order to form a coalition, each party that would join must agree to compromise some of the program that it campaigned on. At the same time, it also will seek to get at least some of its key preferred policies included in any proposed coalition’s joint program of government, in order to be able to show benefits to the coalition for those who voted for the party. The parties will also bargain over how many ministerial positions each party will obtain and which ones. Parties typically seek to obtain the positions that most closely match the policy areas of greatest concern for the voters and interest groups that make up their support base in the electorate.
Table 8.3. Federal Republic of Germany election result, September 18, 2005
There were negotiations about bringing the Green Party into the center-right formation. This was nicknamed by the press as the “Jamaica Coalition” because the colors of the various parties are those of the Jamaican flag: black (Christian Democratic), yellow (Free Democratic), and Green. The negotiations on this potential coalition, which would have had 338 seats, were reported to have lasted only an hour and half before breaking down. The policy gap between the Greens and the other two parties was just too great for the parties to find common ground. (One commentator joked that no one could imagine Angela Merkel in dreadlocks—a reference to the Christian Democratic leader and the Greens’ support among younger and countercultural voters.)
Another possibility that was discussed was the “Traffic Light Coalition”: red (SPD), yellow, and green. This combination, with 334 seats, also was not attractive to the various parties, because the FDP’s policies are so distant from those of the other two. Theoretically, there could have been a “coalition of the left”: SPD, Greens, and the Left Party. These parties collectively had won 327 seats, and all are certainly left of center in an ideological sense. However, the Left Party consisted of the former Communist Party of East Germany, which made it anathema to many mainstream voters, along with a faction that had defected from the SPD over the latter’s enactment of reforms to the welfare state during the SPD-Green coalition government (1998–2005). For these reasons, the parties were politically incompatible despite superficial ideological affinities.
With none of these various combinations being viable, what finally resulted was a “Grand Coalition” consisting of the two big parties, the SPD and the CDU, plus the CSU. While the CDU/CSU block and the SPD normally oppose each other, they agreed to form a government together because each party preferred that possibility over taking in partners it and its voters saw as too extreme (Greens for the CDU/CSU, FDP or Left for the SPD). They also preferred the Grand Coalition over going to fresh elections, which would have been required had there been no government formed. Thus the Grand Coalition was no one’s first choice, but it was the most viable option.
A formal agreement was reached in November. The parties produced a 130-page coalition agreement in which they detailed a series of difficult policy reforms, including increasing the Value Added Tax, raising the retirement age, and many others. At a news conference, the SPD leader commented, “None of us was prepared for a grand coalition—none of you either. We learned to make compromises.”b Indeed, compromise is the essence of forming coalitions.
The Grand Coalition government was headed by Angela Merkel, the CDU leader, who assumed the position of chancellor, the equivalent of prime minister in the German case. The SPD, reflecting its status as the largest party, was able to bargain for a majority of the total number of cabinet ministries, after relenting on its initial demand for the chancellorship. The SPD also received several important posts, including the ministers of environment, finance, foreign affairs, health, and labor, as well as the position of vice chancellor. The CDU held other key positions, including defense, economics, and interior. Thus the transaction between the CDU/CSU and the SPD reflected the close balance of power between the two biggest parties in the parliament, with each party able to negotiate for itself key positions, while having to concede others to its partner.c
The government served the full four-year parliamentary term, and given its huge majority in the parliament, enacted several major policy reforms. In 2009, despite still being in government together, the CDU/CSU and the SPD each ran their separate campaigns, seeking an election outcome in which one could lead a government without the other. The CDU/CSU combined for 239 seats, and the FDP won 93; thus these parties were able to form a center-right coalition. Angela Merkel remained chancellor. This time, reflecting her party’s much stronger position vis-à-vis its coalition partner, the FDP, the two conservative parties held around two-thirds of the cabinet ministerial posts, including defense, environment, finance, interior, and labor. Nonetheless, the FDP also received some important posts in the cabinet, including economics, foreign affairs, and health, as well as the vice chancellorship. Earning these positions out of the transaction with the stronger CDU/CSU reflected its pivotal position as a veto player without whom the Christian Democrats could not form the center-right government that they preferred. The SPD, meanwhile, became the largest party of the parliamentary opposition, waiting to make its case to the voters in the next election.
NOTES
a. The CDU and CSU are separate party organizations, but they do not compete with one another in elections. The CSU contests elections only in Bavaria, and the CDU runs in all other states.
b. “New Government Pact Finally in Place,” Deutsche Welle, November 13, 2005, http://www.dw.de/new-government-pact-finally-in-place/a-1773778-1.
c. “Grand Coalition a Delicate Balancing Act,” Deutsche Welle, October 11, 2005, http://www.dw.de/grand-coalition-a-delicate-balancing-act/a-1737942-1.
. . .
As such, when there is no single party that can govern alone, the transactions in parliamentary systems are among the parties over the formation of the government (meaning its head, the prime minister, as well as the cabinet); transactions are ongoing over maintaining the government in office and compromising over policy—or alternatively, breaking it up early, likely leading to new elections. This stands in contrast to presidential systems, including the United States, where the government is headed by a president who has his or her own separately elected and fixed-term institution and is not dependent upon transactions among legislative parties. Presidents may choose to appoint cabinets containing members of parties other than their own, but the point is that this is their choice; unlike a prime minister, the president never has to transact in order to get into office, stay in office, or form a cabinet. Due to the separation of powers, transactions in presidential systems over policy enactment are between these separately elected institutions, each acting as a veto gate in the process and each sitting for a fixed term; policy in such systems can only be made when there is sufficient agreement across the institutions.
The question of whether cabinets are single-party majorities or are coalitions is crucial to parliamentary democracies, because it determines how many veto players must transact and agree to policy changes: the majority party alone or the two or more parties that together form a coalition. Minority governments are intermediate cases in which the party or parties in the cabinet do not collectively have the support of a majority in the parliament. Minority governments consisting of a single party sometimes are not greatly different in practice from single-party majority governments, because all members of the executive come from one party. However, given the hierarchical relationship between the parliamentary majority and the executive, which defines parliamentary democracy, a minority government is always subject to the threat of losing a no-confidence vote. The lower the probability of this happening—for instance, if the various opposition parties do not agree enough on policy to act together to replace the incumbent—the more the single-party minority government can resemble a single-party majority government.9 Box 8.2 offers some examples of minority governments in Canada and New Zealand.
Table 8.4 shows the prevalence of different types of cabinets in our thirty-one democracies. The table actually contains thirty-two entries, because we show New Zealand twice due to the change in their electoral system (see chapter 5) from plurality to a proportional system. As intended by those who promoted this electoral system change, the result was a substantial difference in the types of cabinets that formed, as we shall subsequently discuss further. The country cases are sorted in descending order by the frequency of coalition cabinets, and then within any group having the same percentage, in decreasing order of the frequency of minority governments. The table also shows the median number of parties in the cabinets of parliamentary democracies and the median of all coalitions.10
In scanning table 8.4, we see that the United States is in a distinct minority of our democracies that have had no coalitions during our time period. Only Argentina and Mexico, two other presidential systems, and the parliamentary systems of Australia, Canada, New Zealand (pre-reform,) Greece, and Spain have been without coalitions; in Spain, the norm is single-party minority governments,11 a format that has also occurred a few times in Canada (see box 8.2). Three other countries have had 20 percent or fewer of their cabinets be coalitions: Korea (presidential) and South Africa and the United Kingdom (parliamentary). In the South African case, the first cabinet upon the transition to democracy (1994) included the former ruling National Party for about two years, but all subsequent cabinets have consisted only of the party that has won a majority in all elections since 1994, the African National Congress. In the United Kingdom, all governments had been single-party majority until 2010, when no party won a majority in a House of Commons election and the Conservative and Liberal Democratic Parties formed a coalition.12
In eleven of our democracies, or 35 percent of them, all governments have been coalitions. These eleven include three of our presidential systems (Brazil, Chile, and Colombia), as well as several parliamentary systems. In all, we find that the median rate of coalitions is 80.9 percent across all countries and 71.4 percent in the parliamentary systems. In addition, we see that minority governments, while not especially common overall (22.2 percent is the median across the parliamentary systems), are very common in a few countries: 100 percent of governments in our time period in Denmark, between 80 and 90 percent in India, Spain, and Sweden, and more than 70 percent in both Belgium and post-reform New Zealand. The latter country is particularly significant, given that it is a case of “reengineering” (see chapter 5), having undergone a major change in its electoral system. Under the old system of plurality in single-seat districts, all governments were single-party majority.13 Under the Mixed-Member Proportional system first used in 1996, the percentages of both coalitions and minority government have gone from 0 to 71.4 percent. (There have been both majority and minority coalitions, as well as single-party minority cabinets; box 8.2 discusses some examples of New Zealand’s minority coalitions.)
In the column that shows the mean number of parties in the cabinet of parliamentary or semi-presidential systems, we see variation from 1.00 in the five nonpresidential cases that have had no coalitions, to more than 4.00 in Belgium, Finland, and Italy, and more than 5.00 in Israel; the median across the parliamentary cases is 1.97. The number of parties in a cabinet is a good guide to how many veto players there are at a given time, because normally all the coalition partners must agree to open the legislature’s veto gate and enact a policy. (If the coalition is also a minority government, then the number of veto players may be at least one more than the number of parties in the government). Thus this column of table 8.4 suggests highly complex policy-making transactions are required in several countries, but that the median for all parliamentary systems is not so complex: Around two parties as veto players is typical.
As for the hybrid systems, all of our semi-presidential systems (Austria, Finland, France, Poland, and Portugal) normally have coalition governments. Some of them, like Portugal, have occasional single-party majority or minority governments, and France has had several governments that were dominated by one party but that contained members from other parties. Switzerland, also a hybrid but without an elected president, has had only coalition governments. So did Israel during its brief hybrid phase. As we mentioned already, even some pure presidential systems, notably Brazil, Colombia, and Chile, have frequent coalitions. It is important to recall, however, that when there are coalitions in presidential systems, they are of a fundamentally different character from those in parliamentary or even semi-presidential systems, because they are always subordinate to the elected head of government, the president.14 Thus, the United States has an unusual cabinet format when compared to our other democracies in two senses: first, in having single-party governments, which table 8.4 has shown are not the norm; second, in having membership in those governments determined by the president, regardless of whether the president’s party has a majority in either house of Congress.15
Table 8.4. Types of cabinets in thirty-one democracies, 1990–2010
Notes: Presidential systems are in italics.
Data on number of parties in cabinet was unavailable for India and omitted for presidential systems.
a Semi-presidential systems.
b Other hybrids: Switzerland (neither executive elections nor cabinet responsibility to parliament) and Israel, 1996–2003 (directly elected prime minister); if hybrid period is removed from Israeli calculations, 10% of governments are minority and average number of parties in cabinet is 4.6.
c In Australia the Liberal and National Parties are considered a single party, as are the CDU and CSU in Germany.
Sources: For parliamentary and semi-presidential systems, authors calculations based on the Parliament and Government Composition Database (ParlGov), available at http://www.parlgov.org/. For presidential systems and other countries not in ParlGov (India, Israel, and South Africa), compiled by authors from various sources.
. . .
BOX 8.2. EXAMPLES OF MINORITY GOVERNMENTS: CANADA AND NEW ZEALAND
CANADA
Given that Canada has a parliamentary executive and a plurality electoral system for the first chamber of Parliament, it is not surprising that most of the country’s cabinets have been single-party majority governments. The term of Parliament is five years, but as in many parliamentary systems, elections can be called early (see chapter 7). Elections in Canada tend to happen about every four years. However, from 2004 to 2011 there was an unusual string of minority governments, as well as four general elections. The governments during this period always consisted of a single party, although one minority coalition was formally proposed but never took office. We can thus use the Canadian case as a window to understand how minority governments function.
As the 2004 election approached, the Liberal Party of Canada (LPC) was ruling in a single-party majority government. However, in the election its seat total was reduced to 135, several seats short of a majority in the 308-seat House of Commons. Paul Martin, the LPC leader, remained prime minister but now headed a minority cabinet. That is, all ministers in the cabinet were from the LPC, but because this party had less than a majority of seats, it could remain in office only so long as the other parties did not combine on a no-confidence vote against it. Two episodes during the tenure of this government demonstrate the dynamics of executive-legislative relations and interparty relations in a minority situation.
First, in May 2005, the third largest party in Parliament, the New Democratic Party (NDP), demanded some changes to the federal budget as the price of not joining with the rest of the opposition to defeat the government. In a parliamentary system, the vote on the government’s annual budget is by definition a matter of confidence: If the government is unable to obtain “supply” (that is, a majority voting for its spending proposals), it must resign. In order to obtain NDP support, Martin offered increased spending on health care and other social policy priorities of the NDP. This deal complete, the NDP voted with the government, which remained in office.a Thus, here we can see how a transaction between a single governing party and one party outside the government resulted in both a change in policy and the survival of a government that by itself lacked a majority.
Table 8.5. Canadian House of Commons election, 2004
In November 2005, the opposition parties all combined to bring down the Martin government, which had been rocked by a scandal involving some LPC campaign finance practices. In a no-confidence vote, every party in the House other than the LPC—the Conservatives, the Bloc Québécois (BQ), and the NDP—voted against the government. As a result, Martin resigned and early elections were called. In January 2006, the new elections again resulted in no party with a majority, only this time it was the Conservative Party of Canada (CPC) that emerged with the most seats. It formed a single-party minority cabinet with its leader, Stephen Harper, as prime minister.
A little more than two years later in October 2008, Harper decided to call early elections (strictly speaking, they are called by the governor general, but this is only a formality). He expected that he could improve his party’s standing in Parliament, perhaps even winning a majority. Thus, for the third time in less than four years, Canada had elections. And for the third time, no party won a majority.
In the 2008 election, the CPC remained the largest party but was again short of a majority. In response to the outcome, the other parties began to negotiate over an alternative government that might replace Harper’s government when the newly elected Parliament convened. The LPC, NDP, and BQ signed an agreement in which there would have been a coalition government consisting of ministers from both the LPC and NDP. The BQ would have agreed to support it, without having any ministers in the government.b The various parties shared a belief that Harper’s failure to win a majority of seats for his party indicated he had lost the legitimacy to continue his minority government, and they argued that his economic policies did not offer sufficient stimulus for the slumping economy (at the outset of the world economic recession). The BQ, which runs and wins seats only in Quebec (as discussed in chapter 4), argued that Harper was shortchanging the province in terms of federal transfers of funds.
Table 8.6. Canadian House of Commons election, 2006
Table 8.7. Canadian House of Commons election, 2008
Had this coalition been formed, it would have been a first for Canada since the Second World War. It would have been a three-party transaction establishing a two-party minority coalition, with the third party (the BQ) offering “outside” support—that is, support on confidence and supply votes in the House despite having no cabinet representation. However, Harper responded immediately by requesting (and obtaining) a temporary closure of Parliament and by denouncing the proposal as a coalition of “socialists and separatists.”c By the time Parliament reconvened, the LPC had changed its leader, due to internal disagreements about the wisdom of the coalition (particularly its reliance on BQ support). The LPC, under its new leadership, made a new transaction with Harper and the CPC, by which the latter would agree to present regular fiscal reports to Parliament.d This minor agreement between the largest party (CPC) and the largest opposition party was enough to allow Harper to continue as prime minister, still heading a single-party minority government.
Later the LPC announced it was no longer satisfied with the minimal concessions it had obtained from Harper. Without the assurance of LPC support against a no-confidence vote, and with a supply vote due, Harper turned to the NDP and transacted with it by offering an increase in Employment Insurance spending.e This was an important issue to the labor-affiliated NDP, due to the high degree of unemployment during the recession. It may seem like an implausible transaction, inasmuch as it involved a Conservative government and its most left-wing opponent in Parliament. However, it was mutually beneficial because for Harper and the CPC, it ensured their minority government would remain in power, while for the NDP it gave them a policy concession that they could claim to have delivered to their constituents.
Finally, in May 2011, following a no-confidence vote by all the opposition parties against Harper, another early election was called—the fourth election in just under seven years. This one produced a majority for the CPC, restoring—at least for now—a single-party majority government.
Table 8.8. Canadian House of Commons election, 2011
NEW ZEALAND
A second example of minority governance comes from New Zealand. As noted in chapter 5, the change of the electoral system from plurality before 1996 to Mixed-Member Proportional since then has resulted in regular coalition or minority governments. The minority governments in New Zealand have differed in a key respect from those we saw in Canada: instead of shifting transactions with different parties in Parliament, depending on the policy issue, those in New Zealand have tended to have written “confidence and supply” agreements with one or more parties in Parliament. This creates a more formalized and ongoing transactional relationship between the minority government and its support party or parties. A good example comes from the government led by Helen Clark, leader of the Labour Party, following the November 1999 election.
In this election, Labour and its prospective coalition partner, a left-wing party known as the Alliance, came up just short of a majority of seats. In order to ensure support in Parliament, Clark signed an agreement under which the Green Party (seven seats) agreed to support the Labour-Alliance coalition cabinet on confidence and supply measures, in exchange for policy consultations.
There was one especially contentious policy issue that arose during the life of this government. A Royal Commission was established by the government to study the issue of genetically modified (GM) food crops and to propose whether the government should lift a moratorium on the importation of such plant materials. The commission concluded in 2001 that GM foods should be allowed. The government agreed and announced a plan to lift the moratorium in 2003. The Green Party was implacably opposed to this proposed policy change, and as it was a veto player, due to the government’s dependence on its votes, its stance raised the possibility that the government would not survive if it pushed ahead with its plans. However, an election was due before the 2003 planned lifting of the moratorium, and Clark decided to move it up a few months.
In the July 2002 election, the Greens made the GM issue a priority and said they would not support any government that would lift the moratorium.f Labour, meanwhile emphasized other issues but also made it clear that it intended to go ahead with the Royal Commission recommendations. The election could be said to have produced a win for both the Greens and Labour. Both saw their seat total increase.g However, clearly Labour and the Greens were no longer compatible, as the Greens had made it apparent that they would not support a Labour minority government unless it dropped its plans to lift the moratorium. Clark ended up forming a minority cabinet again, only this time with the support of a center-right party, United Future, which demanded very few policy concessions in exchange for supporting the government on confidence and supply.h The Greens, meanwhile, announced they would go into opposition rather than support a government committed to a policy that they and their voters strongly disagreed with.
The example of minority governments in New Zealand shows multiple parties transacting with one another in order to jointly achieve the majority needed both to avert potential no-confidence votes against the government and to allow the veto gate of Parliament to be opened for passing policy. Unlike in the Canadian case, in New Zealand minority governments have tended to sign “confidence and supply” agreements with one or more other parties in Parliament. These agreements allow the government to be assured of the support party’s votes on confidence votes and the budget, and allow the smaller party to obtain some policy concessions. They do not commit the smaller party to vote for the government’s policies in areas not covered in their agreement, if the smaller party disagrees with the government policy, as we saw with the Green Party in GM policy. In this way, if we look back to figure 8.3, we can see that parties seek to balance maintaining accountability to the block of voters who delegated to them their votes at the last election, on the one hand, while transacting and compromising to pass legislation through the veto gate, on the other hand.
Table 8.9. New Zealand Parliamentary election, 1999
Table 8.10. New Zealand Parliamentary election, 2002
NOTES
a. The government also needed to negotiate with one independent—nonpartisan—member of Parliament in order to ensure a majority. Exactly what the independent might have obtained in the transactions was not clear, but the vote on the NDP-supported budget amendment was dramatic, and its outcome was uncertain until one of the last of the independents stood up to cast his vote for the amendment, resulting in a tie vote in the Parliament. The speaker of the House cast the tiebreaker in favor of the government, which thereby survived. A timeline of the 2004–2006 Parliament is available at http://www.cbc.ca/news/background/parliament38/index.html.
b. The agreement called for eighteen Liberal and six NDP ministers of cabinet, with the BQ committed to supporting the cabinet for at least eighteen months, http://www.cbc.ca/news/canada/story/2008/12/01/coalition-talks.html.
c. “Harper Says Dion in Bed with Socialists and Separatists,” Western Star, December 1, 2008.
d. “Ignatieff Puts Tories ‘on Probation’ with Budget Demand,” CBC, January 1, 2009, http://www.cbc.ca/canada/story/2009/01/28/ignatieff-decision.html.
e. “NDP to Prop up Tories to Pass EI Changes,” CBC, September 16, 2009, http://www.cbc.ca/canada/story/2009/09/16/ndp-election-tories.html.
f. See “Greens Given GM Warning,” New Zealand Herald, March 6, 2002.
g. The Alliance ceased to be relevant, and its ex-leader, Jim Anderton, won only two seats heading his new Progressive Coalition, which committed to remaining in coalition government with Labour.
h. The main concessions that UF received were a Commission on the Family, which would vet—but not necessarily block—government policies for their compliance with supporting families, and a commitment to consider building a new highway to bypass a congested part of UF leader Peter Dunne’s district. As of 2012, the Commission on the Family had been reduced to a single member with little policy influence, and the road had not yet built, despite UF having been a support party to every government during this time period.
. . .
How, then, are executives selected? As we have noted earlier, if the system is parliamentary, then the selection process is done within the legislature as an internal action of either the majority party or the majority coalition. If, however, we are dealing with a presidential or semi-presidential system, the executive needs to be chosen by the electorate.
We can start with the question of whether the election of the president is to be a direct or indirect process. An indirect election is one in which there are some type of institutional intermediaries between the electorate and those who select the president. The US Electoral College is such a mechanism. Voters in the United States do not actually vote for the presidential candidate of their choice (even if the ballot upon which they vote makes it appear that they are so doing). Instead, they vote for a slate of electors who are pledged to vote for the candidate who wins the plurality of votes in the given state.16 Those electors then assemble in their state capitals in December to cast their electoral votes, which are, in turn, not counted officially until the new Congress convenes in January; as such, the president of the United States is formally elected by the electors, not the citizens of the United States. While the process is indirect, the electors do not also serve as legislators, preserving the separation of origin of the executive and legislature.17
In direct elections, the voters vote for the candidate of their choice directly (rather than voting for intermediaries). Direct elections can be single events, where the candidate who receives the most votes wins (that is, plurality elections, as discussed in chapter 5), or can be processes that require a specific percentage to win and may require voters to return for a second round if no candidate receives the requisite percentage of the vote in the first round. Most two-round systems require an absolute majority (that is, 50 percent plus one), but some systems use a qualified plurality, wherein a candidate can win in the first round sans gaining an absolute majority, but only by surpassing a specified threshold (for example, 40 percent) or margin requirement. The only such case among our set of democracies is Argentina, where victory requires a plurality of at least 45 percent or else at least 40 percent if the runner-up trails by at least 10 percentage points. Otherwise there is a runoff between the top two. Rules such as these forestall electing presidents with small pluralities but acquiesce to the mathematical likelihood that a second round is likely not necessary if a candidate can achieve a large plurality, but not quite an absolute majority, in the first round.
THE US EXECUTIVE BRANCH IN COMPARATIVE PERSPECTIVE
In this section we detail nine ways in which the United States is different (if not unique) in comparison to our thirty other democracies. Of the nine differences, the first three deal with the structure of the office in regards to term, reelection, and emergency replacement. The next four deal with the nomination and election process. The last two deal with the question of the transactional relations between the executive and the legislature.
In presidential systems, as we have noted, the popularly elected head of government must transact with a separately elected legislature, which, like the presidency, has a fixed term of office. Thus a further area of variation among such systems is in the relative bargaining weight that the president has vis-à-vis the legislature. This bargaining weight is principally affected by two factors, the president’s constitutional powers and his partisan powers. Constitutional powers means the formal authority over legislation granted the president in the Constitution. Partisan powers refers to the degree of support that the president has in Congress, both from the majority or opposition status of the party and from the likelihood that he will obtain legislative support from his own party. Thus, while all presidents in a given country have the same constitutional powers (barring changes in the Constitution itself), a given president’s partisan powers depends on how well his party performs in congressional elections, as well as how much influence he has in his own party.
Table 8.11. Presidential terms of office in twelve democracies with popularly elected presidents
1. The four-year presidential term of office. As discussed in chapter 6, one of the most striking characteristics of the House of Representatives in comparative perspective is its very short term of office—only two years. Likewise, the four-year presidential term of office is short in comparison with that of other popularly elected presidents, but neither unique nor a great deal shorter. The majority of our cases with popularly elected presidents (the presidential and semi-presidential systems) have terms longer than four years (seven of twelve have terms of either five or six years). Table 8.11 shows that Argentina, Brazil, Chile, and Colombia have four-year terms, too. This increase in company that the United States has with other four-year terms is recent, as Argentina (1995), Brazil (1994), and Chile (2006) only recently changed to shorter terms.18 There was one prominent case of even longer terms: French presidents were elected to extraordinarily long terms—seven years—until a constitutional reform cut it to a more typical five years starting in 2002.
2. Term limits. In the United States, while the term may be short, presidents are eligible for immediate reelection. Thus a given president may serve for eight years. As table 8.12 shows, some presidential systems place stricter limits on how long a president can serve, by either limiting the president to a single term or allowing a second term but only after the passage of an interim term. Colombia has a limit of two four-year terms as well, but prior to the constitutional reform of 2003, presidents had been limited to one four-year term.19 France is the only country among our set in which the president can be reelected without any limits. When we combine the information on terms of office with that on term limits, it is clear that the actual consecutive length of time that a US president can serve—up to ten years,20 though more realistically, eight—is actually longer than in most of the other countries. For instance, Korea and Mexico have five-year and six-year terms of office respectively, but these terms also represent lifetime limits.
Table 8.12. Presidential term limits in twelve democracies with popularly elected presidents
Source: Based on data supplied by John M. Carey.
These variations in maximum tenure reveal that there is a trade-off between the length of a given term and the presence of term limits. With the notable and really striking exception of France—where the term length used to be seven years and reelection is unrestricted—presidents who can be reelected to consecutive terms usually are those whose terms are shorter. In fact, the recent changes to shorten the terms of presidents in Argentina and Brazil were both taken as compromises between the opposition and supporters of presidents who sought the right to run for reelection. Thus the trade-off we are referring to was embodied in constitutional compromise: potentially longer total tenure for any given president, but more frequent presidential elections. Similarly, over the years various proposals have surfaced in the United States for lengthening the presidential term—usually to six years—and these have almost always been linked to a proposal to ban reelection.21
3. Vice presidency versus early election to fill a vacancy in the presidency. In the United States, the position of vice president exists but has a limited role except in the event that the president dies, resigns, or is removed from office.22 In that case, the vice president takes over as president, serving the remainder of the term for which the departed president was elected. Most other countries with elected presidents do not have a vice presidency. In fact, only Argentina, Brazil, and Colombia have the position of vice president. In other cases, if the presidency is vacated, an interim president is appointed (typically from congress), either to fill out the term, or until early elections are held. In these cases, unlike in the United States, even though the presidency has a fixed term, presidential elections could come early in the unlikely event that a president leaves office prematurely. In most cases, if there is an early election, the president elected simply fills out the remainder of the original term, thereby returning elections to their regular schedule. However, in France, early presidential elections reset the clock, giving the new president a full term, as has happened twice: when President Charles de Gaulle resigned in 1969 after less than four years of his (then) seven-year term, and when President Georges Pompidou died in 1974, less than five years into his seven-year term.
4. Indirect election. The United States is now unique among countries with elected presidents in having those elections be indirect. Presidential elections in the United States are popular, in that voters select a presidential candidate after a public campaign by the candidates, but they are indirect, in that the ultimate selection takes place via the Electoral College. In direct elections, the candidate is elected who obtains the most votes, either in a single-shot election or in a runoff. Of the twelve countries with popular presidential elections—both presidential and semi-presidential systems—only in the United States is it possible for the candidate with the most votes in the final round of popular voting to lose out in favor of a candidate with fewer popular votes. Of course, this is what happened in 2000, when Al Gore obtained more votes than George W. Bush, but the latter candidate—after a lengthy series of recounts in the state of Florida and a ruling by the US Supreme Court—was determined to have obtained more electoral votes. In the Electoral College, each state has a voting weight equivalent to its total number of representatives and senators. States are free to choose their method for allocating their own electors—they do not even have to hold popular voting, although all do. Forty-eight of the fifty states, plus the District of Columbia, award all of their electoral votes to the candidate with the plurality (that is, most votes) in the state.23
We see in table 8.13 that the United States has not always been alone, as two other countries among our sample used electoral colleges in the recent past; both changed to direct election in the early 1990s. Argentina’s, like that of the United States, was “federalist,” in that it weighted the representation of states (provinces in Argentina) in the electoral college such that less populous states were overrepresented relative to their share of the national population. The candidate who obtained a majority of the total number of votes cast by electors—who assemble only in their states and not as a national body, providing a further federalist element—is elected. The electoral college of Finland, on the other hand, was “partisan” in that electors were elected much like legislators. Candidates for elector campaigned as individuals as well as representatives of a party promoting a specific presidential candidate, and the electoral college assembled as a single body, deliberated, and was empowered to take multiple ballots, if necessary, to produce a winner.
Table 8.13. Elected president: Method of election
Note
a The qualified-plurality rule in Argentina stipulates that the leading candidate is elected on the first ballot if he or she wins either (1) at least 45 percent of the total vote, or (2) at least 40 percent of the total vote if the second-ranked candidate is at least 10 percentage points behind; if the leading candidate does not meet either of those conditions, there is a runoff between the top two candidates.
The other eleven democracies with popularly elected presidents all use direct elections. As table 8.13 shows, most—eight out of twelve—use the majority-runoff method: If no candidate wins a majority of votes in the first round, a second election is held between the top two candidates. Korea and Mexico use the plurality method (as did Colombia until 1991), in which the candidate with the most votes wins even if he or she does not have a majority of the votes. Argentina uses a method that may be termed “qualified plurality,” which can be seen as intermediate between plurality and majority-runoff. (The note in table 8.13 explains the details.) The use of runoff methods—whether majority or qualified-plurality—is a means of preventing the possibility that a president is elected against the wishes of the majority of voters. Often under plurality rules, the candidate with the most votes is well short of 50 percent. In such cases, it is not necessarily the case that the majority is unhappy with the result, but the risk of a majority-disapproved candidate is greater to the extent that the race is very close, or the largest party is relatively more extreme ideologically than the others. For instance, in Korea all presidential elections since the return to democracy in 1987 have been won with less than 50 percent of the vote, and those of 1998 and 2002 were decided by less than 2 percentage points. The Mexican presidential election of 2006 was won by a candidate, Felipe Calderón of the National Action Party, with 35.89 percent of the votes and a runner-up who trailed the winner by just more than half a percentage point (35.33 percent).24 It is possible that some of these elections could have turned out differently if a runoff between the top two had been required. In the United States, the use of plurality rule, in indirect elections, is even more extreme, as we see next.
5. Extreme multimember district plurality and barriers to small-party participation. What makes the American system of electing presidents even more unusual is that the Electoral College is elected by plurality in districts (the fifty states and the District of Columbia), many of which have very large magnitudes; as we will recall from chapter 5, the magnitude of a district denotes the number of representatives (electors in the case of the Electoral College) to be elected in the district. District magnitude has a strong impact on election results both in proportional representation (PR) systems and in plurality systems (as well as other majoritarian systems like majority-runoff and the Alternative Vote), but in opposite directions: Increasing the district magnitude in plurality systems entails greater disproportionality and greater advantages for large parties. This is in stark contrast to the effect under PR, where increasing district magnitude results in greater proportionality and more favorable conditions for small parties (see box 5.1).
A similar hypothetical example for plurality is that in a particular area the election contest is between Democrats and Republicans and that the Democrats are slightly stronger. If this area is a three-member district, the Democrats are likely to win all three seats. However, if the area is divided into three single-member districts, a Republican candidate may well be able to win in one of the districts and hence win one of the three seats. When the district magnitude is increased further, disproportionality also increases. Imagine if the US House were elected state by state, using plurality, and, as is the case with the Electoral College, voters had to vote for an entire ticket rather than having the option to pick and choose candidates of different parties. If this were the case, then a party could beat the runner-up by a very thin margin yet would get all of the states’ seats—for instance, all fifty-five in the case of California! This is precisely the system used in the Electoral College, and it is what can make a single close state contest so contentious as was the case in the struggle to determine which party’s candidate had won Florida’s twenty-five electors in 2000. It is also one critical reason why parties other than Republicans and Democrats rarely are significant in presidential elections, as we will discuss further.25
The combination of indirect election, high district magnitude, and plurality rule ensures that the number of candidates competing in presidential elections is sharply lower in the United States than in most other countries that hold presidential elections. Table 8.14 shows that effective number of candidates and the actual number of candidates (ignoring those with less than 2 percent of the vote), averaged over all the elections held in each country between about 1990 and 2010. We see that the effective number of candidates over this time period has averaged 2.28 in the United States, which is the lowest of any of our cases save Austria’s 2.20.26 The actual number of candidates is lower than anywhere else. Compare the majority runoff elections, where the effective number averages 3.39 and the actual number winning at least 2 percent averages 5.00 (in the initial round). These systems encourage additional candidates to run, and voters to support them, because (1) it may be unclear at the outset which two will be the top two, implying competition among three or more parties for runoff slots, and (2) smaller parties can demonstrate their support in the first round and then form alliances with one of the remaining candidates for the runoff. The second round also makes it possible for a candidate to have won the most votes in the first round but lose the runoff, when voters for other, eliminated, candidates prefer the candidate who initially placed second. In other words, the plurality winner is not necessarily majority-preferred, and having the second round ensures that a majority-disapproved candidate is not elected.
Table 8.14. Effective and actual numbers of candidates in presidential elections, 1990–2010
As table 8.14 shows, direct plurality elections tend to have less fragmented competition than the majority-runoff elections, given that the plurality rule by definition means victory to the largest: The effective number of candidates averages 2.98, and the actual candidates winning more than 2 percent of the vote averages 3.83. What is particularly interesting, however, is that even direct-plurality election tends to produce participation by a larger number of significant parties than we see in US presidential elections. This reinforces the point we made previously about the impact of multiseat plurality rules being even more disproportional than single-seat plurality. When presidential elections are by direct plurality, we have a nationwide election for a single seat, and hence parties other than the top two can collect votes wherever in the country they have support, and potentially even win with less than 50 percent of the vote—or even less than 40 percent, as in the Mexican case we referred to previously. By contrast, the multiseat plurality election of electors in the United States means that a third party would have to win the largest share of votes in several states to have any realistic chance of winning. Thus the campaign for selecting the president is basically restricted to participation by large parties in the United States more than in the other countries. As a result, there is less injection of the viewpoints of smaller parties into the most important national political debate.
The disproportionality of the US Electoral college is clear from table 8.14, where we see that the effective number of candidates, when measured by their shares of the electors is only 1.85, which is even lower than when measured by votes (2.28). This contrasts with the lesser reductions in Argentina, and especially Finland, where electors were allocated via PR, and thus smaller parties could form coalitions to participate in the choice of a president.27 The 1992 US presidential election, which featured an unusually strong third candidate, demonstrates the difficulty a candidate who is not from one of the two established parties has, on account of the use of state-by-state plurality in the Electoral College. Democrat Bill Clinton, Republican George H. W. Bush, and independent H. Ross Perot split the popular vote 43 percent–37 percent–19 percent. But the electoral vote split 370–168–0.28 That is, Clinton won more than two-thirds of the electors, despite much less than half the votes. It is easy to see why the Electoral College, as employed in the United States, generally discourages many small parties from even entering—if they cannot win electoral votes, they can have no direct influence29 on the final selection of the president.
6. Regionalized competition for president and over-representation of small states. A further consequence of both the federalist nature of the Electoral College and the use of plurality rule in large multimember districts is the regionalization of presidential electoral competition. For instance, media commentary about presidential campaigns routinely refers to “red states” and “blue states,” meaning those which are likely to give their entire slate of electoral votes to the Republican or Democrat, respectively, regardless of the margin of victory.30 A small number of states is likely to be decisive in any given election year, as relatively small swings of the popular vote in a few states can shift the outcome in the Electoral College and hence determine the presidency. Under a direct-election procedure, attention would be unlikely to focus on swing states (and, within them, crucial voting blocs), because the goal would be to win a national plurality (or majority or qualified plurality, depending on the rules), rather than to amass a majority of electoral votes. Note again that it is not the Electoral College alone that is responsible for this regionalization, but the fact that each state gives its entire electoral-vote contingent to one candidate, with the winner determined by plurality.31 For instance, if electors were awarded proportionally, small vote shifts in specific states would be much less decisive in the final outcome.
In the United States the electoral-vote winner may not be the popular-vote winner, because the percentage of electoral votes in each state is only roughly proportional to the population of the state. No state has less than three electoral votes, or about 0.56 percent of the total, because each state has a number of electoral votes equal to its representation in the House and Senate combined. Some of the states that have just three electoral votes have far less than 0.56 percent of the national population, with the greatest discrepancy being that in Wyoming (0.18 percent of the population).
We can calculate an advantage ratio for each state by simply dividing its share of the total electoral vote by its share of the national population. These ratios range from about .85 for California and Texas to 3.18 for Wyoming. In other words, California’s weight in electing a president is only 85 percent of its contribution to the national population, while Wyoming’s is more than three times as great as its population. Figure 8.4 plots each state’s advantage ratio against its population (based on population estimates from 2002 and the Electoral College apportionment, which were effective in 2004 and 2008). It shows very clearly how the smallest states are significantly over-represented. Only states with around 5 million population have an Electoral College weight about equal to their share of the population. There are thirteen states with an advantage ratio greater than 1.5. Together, these states hold 4.7 percent of the population, but cast 9.1 percent of the electoral votes.
Fig. 8.4. Relationship of state population to over- or underrepresentation in the US Electoral College
Because no other country currently elects its president via an electoral college, no other country has electoral rules that regionalize the contest.32 Regionalized competition prevails in Korea because the parties have distinct regional bastions; however, regions are not weighted in the final outcome (other than by their turnout), because the election is direct.
7. Primary elections. A primary is an election between two or more candidates of the same party, in advance of the general election, and it is used as a means of nominating that party’s candidate for office. The United States uses a series of state-level primaries and caucuses33 to nominate its major party candidates, who then compete for the presidency. Rules vary by state as to whether only voters who identified an affiliation with a given party when registering to vote may vote in that party’s primary. As we noted in chapter 6, the United States widely uses primary elections to nominate candidates for most elected offices. The widespread use of primaries to nominate candidates by the major parties is a feature unique to the United States, although some countries have used primaries in a limited fashion. Table 8.15 shows that some Latin American parties have used primaries on a limited basis during our period of study.
Not only is the United States the only one of our democracies that has used the primary process consistently for its major party nomination for decades, it is also unique in the sense that it uses a state-by-state and indirect process. The US presidential nomination process is indirect because primaries select convention delegates—a partial analogue to the electors in the general-election phase.34 Additionally, there remains no national vote in the primary process, but rather a sequence of state-level contests (often and increasingly with groups of states voting on a common day) that inevitably gives greater weight in the ultimate selection of a nominee to voters in some states than in others. By contrast, other countries that have held presidential primaries have generally done so in a direct national vote. A partial exception to direct national primaries was found in Mexico’s then-ruling Institutional Revolutionary Party in 1999. In that case, the election was held on the same day throughout the country, but the decision rule was that the candidate who won pluralities in the most congressional districts would be the party’s nominee, even if that candidate did not have the plurality of the national popular vote.35
Table 8.15. Countries and parties that have held presidential primaries, 1990–2010
Note:
a In 2005 a primary was planned, but the withdrawal of one of the two candidates led to its cancellation. In 2009 a regional contest was held, and the substantial victory of the eventual nominee was large enough to preclude any further voting.
Sources: Siavelis and Morgenstern 2008; additional detail provided by Mark P. Jones and Peter M. Siavelis (personal communications).
The closest case to the United States in regards to using regional primaries was Chile in 2009, when one regional contest was used by the Concertación to nominate its presidential candidate.36
A concluding note on US primaries in comparative perspective, is that despite the lack of proportional representation (PR) mechanisms in US national elections, proportional representation rules are commonly used to allocate delegates in statewide contests. In fact, PR (albeit with very high thresholds) is required in all states by the rules of the Democratic Party, usually partly within congressional districts and partly statewide. In the Republican Party, some states use PR and others use plurality, either in congressional districts or statewide. Thus presidential primaries, especially for Democrats, are the one arena in which many Americans vote under PR rules.
8. Relatively weak constitutional powers over legislation. In the United States, the presidency is provided one constitutional power over legislation: the right to veto a bill passed by Congress, which then requires two-thirds of each house to override. It is this veto—combined with the fixed terms—that guarantees a transactional relationship between the executive and the legislature, because neither branch can accomplish a legislative agenda without the consent of the other. Assuming differences in preferred policy outcomes, either the two branches come to a negotiated agreement over the shape of a bill, or one side backs down and accepts the other’s version, or else no legislation is passed. Contrast this with what happens in a parliamentary system, where the head of government—the prime minister—can be removed via a vote of no confidence if the legislative majority disagrees over policy priorities with the executive. Thus the US president is in a strong position to say no to the legislature’s majority but at the same time in a weak position in that he has few levers to exercise in order to push the legislature into action in support of his agenda.
Contrast the US presidency’s formal powers to those of other elected presidents, shown in table 8.16. As can be seen at a glance, almost all presidents in presidential systems have at least the veto, but among our seven cases, five have more than the veto. Four of the presidencies in our set of countries are constitutionally given unilateral powers, by which we mean the right to enact decrees with the force of law. For example, in Brazil, the president may legislate on a wide range of policy matters by emitting a “provisional measure,” which is effective for sixty days. After that time, it lapses, although presidents often reissue lapsed measures. Congress may reject a provisional measure by passing countervailing legislation, and here is where the veto power comes in. Notice from table 8.16 that the Brazilian president’s veto has a lower threshold to override than that of the United States: an absolute majority, rather than two-thirds. However, this is actually not as weak as it may appear, as the Brazilian veto is based on a majority of all members, whereas the United States is based on two-thirds of members present and voting. What this means is that a determined congressional majority in Brazil could prevent a provisional measure from remaining in effect, but given that the Brazilian congress is often fragmented and has a high absentee rate, in fact the congress is not well equipped to block the president’s use of unilateral measures. Generally, then, the unilateral powers permit the president of Brazil to play the leading role in promoting his agenda to a degree that a US president can only dream of. Nonetheless, really significant policy changes usually require statutes or even constitutional amendments, so congress is still a check on even the unilateral powers of the Brazilian president.
Another case of unilateral powers is Argentina, where presidents may issue what are known as “decrees of necessity and urgency,” which take effect immediately. Given the president’s much stronger veto, these are even harder to overcome than Brazilian provisional measures. They also do not expire after a set period.
Table 8.16. Powers of popularly elected presidents
Notes:
a First chamber alone (Sejm) need to override.
p “partial” (or item) veto allowed.
Sources: Shugart and Carey 1992; Tsebelis and Alemán 2005; and respective national constitutions.
It should be noted, also, that not all vetoes are equal. Not only are there different thresholds by which legislatures can override a given veto (which affects a given president’s potential for success), but three of our cases (Argentina, Brazil and Colombia) allow for a partial (or item) veto. As Shugart and Carey note, partial vetoes “increase presidential power dramatically” because “congress cannot present large legislative packages to the president. Which effectively force the latter to accept pork projects or other objectionable amendments . . . in order to pass desired legislation” (1992: 134). However, with the partial veto, a president can target specific elements of a given legislative package, while allowing the rest to pass into law.
Table 8.16 also shows that there are three cases of presidents with “integrative” powers. We borrow this term from Cox and Morgenstern (2001), who use it to refer to powers that allow the president to control the agenda of congress. These are different from unilateral powers, which allow the president to make policy without even consenting with congress (though legislators may subsequently reverse a unilateral act, with varying degrees of difficulty). Integrative powers allow the president, for instance, to establish a maximum spending level, which congress can cut but not increase. Other integrative powers establish that in specific policy areas, the congress may not initiate legislative change but must wait for the president to do so. These powers thus give the president a powerful role in shaping policy change, especially when combined with veto power (Chile and Korea) or with both veto and unilateral powers (Brazil). With only the veto power, then, the US presidency is considerably weaker in constitutional powers than at least five of the other seven presidencies in presidential systems. The powers of the Mexican presidency are the most comparable to that of the United States. Strikingly, none of these constitutional powers over legislation apply in semi-presidential systems, except for the veto wielded by the Polish president.
9. Moderate partisan powers due to the prevalence of “divided government” and modest discipline. A president can be said to enjoy high partisan powers if his party commands a majority of seats in congress and if he himself commands the party, such that its legislators regularly vote with the president. If a president has very high partisan powers, he does not need constitutional powers to dominate the legislative agenda. At the extreme of high partisan powers, the congressional majority would become subordinated in a hierarchical relationship with the president. Such a situation existed in Mexico before the long-ruling PRI lost its majority in 1997 and then lost the presidency three years later.
In the case of the United States, presidents frequently do not have a majority. As we noted in chapter 6, parties in the United States are nonhierarchical, which means that not only do their own legislative leaders not command their parties but neither does the president, typically. Thus presidents must transact with congressional leaders, even of their own party, or with individual members. The term divided government is commonly used to describe the situation in the United States in which the Congress (or at least one house) is controlled by a majority party distinct from the party of the president. Divided government is actually the norm in the United States. In our primary period of analysis for this book, 1990–2010, both houses have been controlled by the president’s party only for eight and one-third years out of twenty-one. Specifically, the Democrats have enjoyed unified government from 1993 to 1995 and from 2009 to 2011, and the Republicans for a portion of 2001 and then from 2003 to 2007.37 Under these conditions, neither party can be said to be able to “form a government,” because while the president is free to appoint a cabinet of his choosing, he has no control over the agenda of the legislature. All he has is the veto with which he can prevent his opponents in the legislature from passing laws he dislikes (unless they can muster the two-thirds needed to override him).
Given the effective supermajority requirements in the Senate for passing major legislation (except the budget) that we discussed in chapter 8, even unified government is not all that it is cracked up to be in the US system. Only a president with both a unified government and a sixty-vote margin for his party in the Senate could be said to have a truly unified government. This is an unusual situation and has only happened once in the period under analysis, and even then it was for only a portion of 2009. The Democratic Party’s sixty-vote majority in the Senate was lost when Senator Ted Kennedy of Massachusetts died. Kennedy was replaced by a Republican (Scott Brown) in the subsequent special election.
A single-party majority opposing the president is rare in our other presidential systems. However, it is also not common in other presidential systems for the president to have a copartisan majority. The more common situation is for the president’s party to be the largest in Congress but short of a majority, necessitating the formation of coalitions to pass legislation, and the use of whatever constitutional powers the president has at his disposal in order to put his stamp on the legislative output.
It is worth comparing what happens in the absence of a pro-presidential majority in the legislative assembly in presidential and semi-presidential systems. As we noted, opposition majorities—so-called divided government—are typical in the United States. Our other country with significant experience with opposition majorities (though usually coalitions, not a single party) is France, a semi-presidential system. As discussed earlier in this chapter, France has had periods of “cohabitation,” when the parliamentary majority, and hence the premier and cabinet, were controlled by a party or alliance opposed to the president. Because, as we saw in table 8.16, the French president has no veto, decree, or integrative powers; in a situation of cohabitation, he is actually much weaker than a US president under divided government. A French president lacks the ability to shape the policy output of the government, except to the extent that he can shape the political complexion of the government itself. Obviously, he cannot shape it when it depends on a parliamentary majority opposed to him. In Poland, on the other hand, where we have also had brief periods of cohabitation—for instance from 1993 to 1995, when Lech Wałẹsa was president but an opposition alliance of the SLD and other parties controlled the parliamentary majority—the president retained policy influence through his veto. The Polish system thus can be said to have both “cohabitation” (sometimes a premier and cabinet opposed to the president) and “divided government” (understood as the ability of the president to exercise a veto against the legislative majority). Finally, it is also worth considering whether divided government is akin to minority government in parliamentary systems, which we saw represents just more than a fifth of governments in those countries (table 8.4). Superficially, they are similar: an executive of a party that has only a minority of seats in the legislature (or at least one chamber, if bicameral). However, here is where the similarity ends. US presidents under divided government face a situation in which a party other than their own has the majority; a prime minister, by contrast, would never be in such a position, because when the majority shifts from one party to another, the prime minister also shifts to the new majority. It is the separate election and fixed term of the US executive institution that allows presidents, along with the cabinets they choose, to remain in office when they are opposed by a majority in the House, Senate, or both.
CONCLUSION
We have seen that the United States is unusual, though not unique, in having a presidential form of government. Among the thirty-one democracies included in this comparative study, a large minority of them have an elected president, and of those, seven are fully presidential systems. In terms of the structure of the office, the US executive has a shorter term of office, allows for immediate reelection (but only once), and has an elected emergency replacement (the vice president). All of these are features that place the US presidency in the minority of our presidential and semi-presidential democracies, but none of these are especially significant in terms of the way the office operates.
Pure presidentialism is significant for our understanding of executive power in the United States, insofar as it makes the executive a veto gate controlled by one player (the president), and this means that for a president to successfully see a legislative agenda passed, it is necessary to transact with the other veto gates, that is the two chambers of Congress (or, specifically, with the veto players within the chambers who determine when to open the congressional gates). This is to be contrasted with prime ministers who usually have all the veto players with which they need to transact represented within their own cabinet: Either the party that the prime minister heads is the only veto player (in the case of a single-party majority government) or the veto players are the parties in coalition with the prime minister’s party.38 Presidentialism, therefore, is a key component for understanding policy making and presents a very different set of institutional parameters than is the case in the parliamentary systems (or even semi-presidential systems, depending on the powers of a given president in those cases) in our study.
In terms of truly unique features of the US executive, we can look to the method of election. No other country systematically uses primaries to nominate the major party candidates, nor does any country use an electoral college or other indirect method to elect a president.39 These features, along with the plurality system to elect electors from multiseat districts, diminish the incentives for third-party candidate to participate in the process, which further cements the two-party system discussed in chapters 6 and 7.
Notes
1. The flipside of this is that the prime minister rarely proposes a measure that he knows his own party is divided on, thus further decreasing the risk of an observed defeat, which would be politically embarrassing for the PM and for the party as a whole.
2. The head of state continued to be a president, elected by parliament, as had always been the case.
3. This type of executive-legislative arrangement can be further broken down into premier-presidential and presidential-parliamentary. In the former subtype, the prime minister and cabinet are constitutionally responsible only to the legislative majority, whereas in the latter, the president has a constitutionally granted right to dismiss a prime minister and cabinet (Elgie 2011; Shugart and Carey 1992, Samuels and Shugart 2010). All of the cases of semi-presidentialism included in this work are premier-presidential, except for Austria. However, specialists agree that Austria functions as if it were premier-presidential, (see Müller 1999; Samuels and Shugart 2010: 88–90) or even parliamentary (Lijphart 1994: 95). Other examples of presidential-parliamentary systems, with far more active presidents, outside of our set of democracies include Peru, the Russian Federation, and Taiwan.
4. In Korea, there is a position of prime minister; however, the constitutional system belongs to the presidential category, because the prime minister is not responsible to the legislative majority but rather serves at the pleasure of the president (Shugart 2005: 327; Elgie 2011: 26).
5. The Finnish president’s powers have been reduced in recent years, both formally and informally. Formally, a constitutional reform in 1999 took away the president’s power to dissolve parliament. Informally, the end of the Cold War reduced the salience of foreign affairs in Finnish politics. A head of state—especially a popularly elected one—can be expected to play a significant role in relations with other states, and given Finland’s long border with the Soviet Union, foreign affairs was the main domain in which Finnish presidents played a prominent role.
6. The year 1965 was the first in which direct presidential elections were held under the French Fifth Republic constitution of 1958. The year 1986 was the first in which the party or alliance of parties that had elected the president did not also hold a majority of parliament (and, therefore, the premiership).
7. If the legislature is bicameral, and the second chamber has significant powers (see ch. 7), then there are also two veto gates. In such cases, coalition formation may take account of the balance of parties in the second chamber as well as the first. This is the case, e.g., in Italy.
8. By not taking cabinet positions, for instance, party D is not responsible for actions taken by the government that party D’s voters dislike. However, it also has less influence over what those policies will be.
9. On the other hand, a minority coalition more closely resembles any other coalition, due to the presence of two or more parties in the cabinet, and the necessity to transact also with a party or parties that are needed to win majority support in parliament.
10. Presidential systems are excluded from these two latter categories. First, they are excluded from consideration of minority governments because this concept is less meaningful in pure presidential systems. Unlike a coalition, which results from transactions with other parties that are akin to those that take place in a parliamentary system (Cheibub 2007: 73–86)—other than the fact that the president’s position is not dependent on such transactions—minority governments may either result from situations similar to those in parliamentary systems or may be cases of divided government (discussed later). Second, presidential systems are excluded from consideration of the number of parties in cabinets because of difficulties obtaining the necessary data on all cases.
11. As mentioned in ch. 4, these governments in Spain typically rely on one or more parties from one of Spain’s subnational units (e.g., the Basque Country or Catalonia) for their parliamentary support.
12. The only other time in the entire post-WWII period when a single party did not obtain a majority of seats was in February 1974. At an election in October of that year, a majority again resulted.
13. Table 8.3 includes only two such governments under “New Zealand 1”; however, all governments of the post-WWII era were likewise single-party majority.
14. A good example of the dominance of the president, and her own party, comes from Brazil. The cabinet of Dilma Rousseff after her election in 2010 had twenty-seven members, thirteen of whom were from her Workers Party (PT), while just five were from the Party of the Brazilian Democratic Movement (PMDB). This imbalance of cabinet positions did not reflect the relative strengths of the parties in the legislature, as is normally the case in a parliamentary coalition. The two parties had almost equal shares of legislative seats, but the cabinet reflected the control of the PT over the most crucial seat: the presidency.
15. Since 2002, French governments have become much more subordinate to the president. This was the first year in which the president was elected immediately before the legislature, and both for five-year terms. This change would tend to greatly increase the chances of a pro-presidential majority and make the president more dominant than before over all the parties in a coalition (Samuels and Shugart 2010: 175–79).
16. Save in Maine and Nebraska, as we will explain later.
17. This contrasts with the electoral colleges in Germany (known as the Federal Assembly) and India, in which sitting legislators comprise a large bloc of the electors.
18. The dates refer to the year in which the first president with the shortened term was elected. Previously, Argentina’s and Chile’s terms were six years, and Brazil’s was five.
19. Under the rules of the 1886 constitution, nonconsecutive reelection was permissible. The 1991 constitution forbade all reelection until the 2003 reform was adopted.
20. A US president who has served two years or less of a term to which another person was elected (e.g., because the elected president dies or resigns) is eligible for two more terms, meaning one person may theoretically serve for up to ten years.
21. For an overview of this topic, see Neale 2009. See also Buchanan 1988 and Cutler 1980.
22. In fact, the vice president’s only constitutional duty is to preside over the Senate and cast a vote only in the event that the votes of senators are tied. Vice presidents rarely attend the Senate, unless there is a high probability of a very close vote on a matter of great importance to the president or his party. Any other administrative role performed by a vice president is purely at the discretion of the president.
23. In Maine and Nebraska, two electoral votes are awarded to the candidate who wins the statewide plurality, and the others are awarded to the winner of the plurality in each congressional district.
24. And in 2012 a president elected with under 40% proved not to be a one-time event, as the PRI candidate, Enrique Peña Nieto, was elected with about 38%.
25. It is sometimes said that Maine and Nebraska award their electors proportionally, but this is incorrect. In Maine in 1992, e.g., Bill Clinton won all four electoral votes despite winning well under 50% of the votes, because he won the plurality in each congressional district as well as statewide. With PR, Clinton would have won two electors, and George H. W. Bush and H. Ross Perot would have won one each.
26. It is worth noting that though Austria’s president on paper appears quite powerful, it is in reality mostly a ceremonial position. The 2010 elections were also especially noncompetitive, with the incumbent (Heinz Fischer) winning almost 80% of the vote in the first round. Even given the nature of the position and its less-than-competitive recent contest, Austria still averages more actual candidates (3.3) than does the US (2.6).
27. The Argentine case implies that some of the deterrence effect is not plurality allocation of electors but the need for a majority of electors to elect a president. In both Argentina (pre-1995) and the United States, if there is not a majority of the electors for one candidate, then the final selection devolves to Congress. Thus there is little incentive for parties to seek electoral votes if they have no chance at a majority. This was so even in Argentina, where the electors were allocated to candidates by proportional representation. Nonetheless, the effective number of candidates in the United States would be much lower than for Argentina if we took into account elections from before 1990, given that doing so would diminish the impact of the two elections in which H. Ross Perot ran unusually strong for an independent or third-party candidate.
In Finland, on the other hand, where their former electoral college was elected via proportional representation and could take multiple ballots among the electors themselves, participation by smaller parties was high.
28. The last time a candidate in a US presidential election outside of the two major parties won electoral votes was in 1968, when George Wallace won forty-six electoral votes by winning Alabama, Arkansas, Georgia, Louisiana, and Mississippi. Before that, in 1948, J. Strom Thurmond, running under the State’s Rights label, won thirty-nine electoral votes by winning Alabama, Louisiana, Mississippi, and South Carolina, and in 1960, there were fifteen unpledged electors selected. In recent elections we have seen “unfaithful electors” (i.e., electors who vote contrary to the plurality of voters in their states) cast protest votes in the Electoral College. Specifically, in 1988 an elector (from West Virginia) cast one electoral vote for Lloyd Bentsen (the Democratic nominee for vice president) and likewise a 2004 elector (from Minnesota) cast a vote for John Edwards. In 2000, an elector from the District of Columbia abstained rather than vote for Al Gore, as a symbolic protest against the Supreme Court’s ruling in Bush v. Gore.
29. Only indirect influence, in the sense that their presence in the race can tip the statewide plurality in some states to a majority-disapproved candidate.
30. This practice started in the 2000 presidential elections and has become a staple of the US political vernacular since—going even beyond presidential politics.
31. It is possible for a second candidate to obtain one of the four electors in Maine or Nebraska because these states do not use statewide plurality; this happened for the first time in 2008, when Democrat Barack Obama won the most votes in one congressional district in Nebraska and hence one of the state’s electors, while Republican John McCain won the statewide plurality and hence the other four electors.
32. As noted earlier, Argentina, although emulating the federalist nature of the US Electoral College, used proportional representation to allocate each province’s electors among the candidates. This proportionality somewhat mitigated the very high malapportionment of their electoral college itself. While the smallest provinces were more over-represented in Argentina than are the smallest states in the United States, no province could deliver its entire voting weight to one candidate.
33. Some hold party caucuses, where voters gather, deliberate, and select delegates pledged to specific candidates who obtain a minimum share of the caucus-goers support. Typically, caucuses receive much lower participation rates than primary elections. The most famous example of caucuses are those of Iowa, which are traditionally held before any of the other states hold their primary elections.
34. One difference is that the party conventions allocate states’ voting power more proportionally than does the Electoral College (proportionate to votes for the respective party’s presidential candidate in the state in past elections). The convention delegates further differ from presidential electors in that they are deliberative, at least in principle.
35. Because congressional districts are drawn to equalize population but not necessarily PRI voters across districts, this process effectively gave greater weight to localities where the party was weaker relative to other parties. Given that the party was going to face a tough general-election battle, a process that encouraged candidates to reach out to voters outside the party’s strongest congressional districts may have assisted the nomination of a more-appealing candidate over the candidate of the traditional party machinery. As it turned out, the winner of the most congressional districts also had the most votes nationwide—and then lost the general election.
36. On April 5, 2009, there was a regional primary in the Maule and O’Higgins regions of Chile between Eduardo Frei and José Antonio Gómez. Because Frei’s margin of victory was more than 20%, no additional contests were held and Frei won the nomination. This was the first time this type of procedure was used (personal correspondence with Peter M. Siavelis). If Chile had engaged in a region-by-region process, this would have been the closest analogue to the US system among our cases.
37. Note that years of control start and end in the January after an election, hence 1993–1995 covers a twenty-four-month period starting in early January of 1993 and ending in early January of 1995. In 2001 the Senate was a fifty–fifty tie, with the fact that Vice President Richard Cheney was the Republican leading to a tiebreak vote to allow the Republicans control. However, Senator Jim Jeffords of Vermont switched his party affiliation from Republican to Independent (who caucuses with the Democrats) effective in May of 2001, which shifted control of the Senate to the Democrats and ended a brief period of unified government.
38. The exception would be in a minority government, in which the prime minister may also have to transact with one or more parties that are not in the cabinet, but that potentially could join with other noncabinet parties in a no-confidence vote.
39. Not counting the mostly ceremonial presidents in some parliamentary democracies, such as Germany and Israel.
