FEDERALISM AND DIVISION OF POWER
As noted in chapter 2, one of the most profound choices made at the Philadelphia Convention was that concerning the relationship between the states and the central government. The political exigencies of the day were such that not only was a complex choice on the table, but it was such that they were forced to invent a new type of division of power that we now know as federalism. So fundamental was this issue that the proponents of the Constitution identified themselves as the Federalists as they advocated that the states ratify the proposed document. This innovation was one that would eventually be embraced by many countries, including thirteen of the democracies in our study (see table 4.1).
THE INSTITUTIONAL OPTIONS
What options were available to the Philadelphia conventioneers and, by extension, what options exist in theoretical space for the arrangement of relationships between localities and the central government? In terms of basic institutional design, there are four options open to a group of political engineers faced with the task of taking a set of units and establishing a basic governing system over them. These options are, in decreasing order of centralized authority: a unitary state, a federal state, a confederal state, and no state at all (that is, dissolution—the units all go their own way). These arrangements have to do with how much authority the center has vis-à-vis the subunits. In a unitary state, policy-making authority is concentrated in the center, and whatever autonomy the subunits have is as a result of delegation from that center: It is a system wherein power flows down from the center to the subunits. All democratic unitary states have at least a local (municipal) level of elected government, and several of them also contain an intermediate level that is elected, usually known as provinces or departments. However, in unitary states, the subnational governments are not guaranteed the right to make final decisions over any policy area.
Table 4.1. Population (2011) and area of unitary and federal countries
Source: CIA World Factbook.
A federal system may be defined most simply as one with multiple levels of sovereignty. To be sovereign is to be legally empowered to make binding decisions. So in a federal political system—which we can also term a federation—not only is the national government sovereign, but there is also a level of sovereign intermediate government in territorial entities typically known as states or provinces. In a federal system, ultimate sovereign power resides in the central government, but there exists also a range of policy questions upon which the intermediate-level governments hold sovereignty, implying that in these areas, the central government is constitutionally barred from overturning the decisions taken at the intermediate level.
Another way to look at these concepts in the light of the framework discussed in chapter 1 and outlined in figure 1.1 is to think of how the three power arrangements (unitary, federal, and confederal) work in terms of principal-agent relationships. In a democratic unitary system, the people as principals delegate their power to their agents in the central government, which then has a hierarchical relationship with subunit governments. In such an arrangement, the subunits do not independently make policy, but simply administer the policies created and handed down by the central government. In a federal system the citizens as principals delegate to both the central and subunit governments, whose relationship is defined in a nonhierarchical fashion. In confederal systems the central government is an agent of the subunits, and there is no direct connection between the citizens and that government. Instead, they delegate to the subunits, which then delegate to the central government.
The lines of authority in a federation are outlined schematically in figure 4.1. Using the same symbolic notation as we introduced in chapter 1, we see solid lines with arrows at their endpoints indicating the direction in which authority flows, and thick bold lines with arrows on both ends referring to transactional relationships—that is, bargaining between constitutionally independent actors. Starting at the top, we find (as in figure 1.1) that if the country in question is a federal democracy, the electorate is the source of authority. Through the act of voting, and mediated through the rules of the electoral system (chapter 5), voters delegate authority to both the national politicians and politicians in the subunit (state, province) in which they reside. That is, all voters are simultaneously members of the national electorate and of the electorate of their subunit. In the context of political systems with a presidential structure, such as the United States, this means in practice that every voter is able to vote for the national executive (president), national legislators (representatives and senators), as well as the state executive (governor) and state legislators.1 Both the national and subnational governments have their own respective bureaucracies, which implement the policies enacted by elected politicians.
Fig. 4.1. Basic principal-agent relationship under federalism
Although it is conventional to speak of the national (central) and intermediate (subnational, that is, state) governments as being different “levels”—and we will adopt this phrasing for convenience as well—they are depicted in figure 4.1 as being side by side to make an important point that is critical for understanding how federal and unitary systems differ. In any system that meets the definition of being federal, there are policy areas in which the subnational governments make binding decisions, which are not subject to being overturned by the national government. Thus in these policy areas, the national government is unable to impose its will on the units, which results in a transactional relationship between the national and unit governments (hence the horizontal depiction of relations between them). A significant amount of policy making in federal systems consists of national legislation that offers incentives to the unit governments to adapt their policies to be consistent with national preferences. Such a relationship is transactional and stands in contrast to what we see in unitary systems, where the national government has the legal authority to simply impose its preferences through legislation that is binding on the provincial or departmental governments. In addition, as also depicted in figure 4.1, much policy implementation in federal systems requires the cooperation of the national and unit bureaucracies. Because these bureaucracies are agents of different principals, again this sort of policy making is transactional: The national government is unable to dictate how state bureaucrats do their job. Indeed, in some federal systems, subunit governments have to cooperate with central government policies in terms of implementation. This can manifest in a variety of ways, which include implementation, design, and budget. For example, central governments often utilize state-level bureaucracies as the point of distribution for a policy, such as using county health departments to manage and distribute the Supplemental Nutrition Assistance Program (SNAP) in the United States (or the fact that interstate highways are constructed and maintained by state departments of transportation). In some policies subunits may be afforded leeway in how a program is put into practice (although within set parameters), such as with the Temporary Assistance for Needy Families (TANF) program in the United States, a poverty assistance program sometimes colloquially referred to as “welfare.” In that situation, US states have leeway to determine factors such as eligibility and specific benefits. Subunits can also influence policies via fiscal inputs. In the previously noted programs in the United States, states provide some of the funding and therefore influence the policies in that fashion as well. More dramatically, a study of federal countries by Stepan (2004b: 330) notes that in Argentina and Brazil, fiscal powers over central government policy are so strong at the subunit level that they amount to a veto power.
These transactional relationships depend on the very existence of distinct policy areas in which the unit and central governments enjoy sovereignty. Naturally, the boundaries between these sovereignties are subject to dispute, and they may evolve over time. Thus there must be some means of identifying which decisions rest with which level of government, and a means for sorting out and resolving disputes. The remainder of this chapter will discuss the various constitutional features that exist for protecting the units from encroachments by the center, and vice versa. Some of this sorting out is left to the political institutions themselves, for instance, bargaining between the first and second chambers of the legislature or among and within political parties. Often, however, conflicts over jurisdictional matters are left to the judicial branch, as we will discuss in our chapter on courts.
So far we have contrasted federations to unitary systems, noting that the latter lack the transactional relations that typify federal systems, whereby the central government and intermediate governments have their own areas of sovereignty. So what of confederal systems? In these, sovereign power flows “upward” from the subunits to the central government, which is typically limited in scope. Confederacies are arguably not viable models for actual states, especially in the modern era. Indeed, Madison noted multiple historical examples in Federalist 18, 19, and 20 of the problems associated with this form of government. The closest example we have at the moment is the European Union, wherein the member states retain ultimate sovereignty, although that discussion is beyond the scope of our study.2 Finally, if subunits cannot find a way of coexisting in some capacity, dissolution of the federation is an option, as in the case of the peaceful breakup of Czechoslovakia into the Czech Republic (one of our thirty-one democracies) and Slovakia in 1993 (not included in the study).3
The US founders were well acquainted with three of these options, two of which they rejected from the beginning: dissolution of the union, and what the authors of The Federalist Papers called “consolidation,” that is, what we today would call the creation of a unitary state. A third option, that of confederation, also lacked appeal, as it was the failure of governance under the Articles of Confederation, the first constitution of the United States, which brought the founders to Philadelphia in the first place. As noted in chapter 2, there was a formal proposal made to reform the confederal arrangement under the Articles (that is, the New Jersey Plan), but it was ultimately rejected. As such, they needed innovation to create a fourth option, which they did in the form of federalism. The government under the Articles was one in which each of the states was coequal, that is, each had one vote in the Congress (even if the different states sent different-size delegations) regardless of the population of the state. Further, normal business under that system required a supermajority of votes, nine of thirteen, and unanimity to alter the Articles. As such, the system was one in which the feeble powers of the central government were impossible to deploy effectively because it was quite easy for a minority to thwart the process (as discussed in chapter 2 and as outlined by Madison; see box 2.1). Clearly, then, the national government under the Articles of Confederation lacked sovereignty in a meaningful sense. It lacked national-level institutions that were agents of a national electorate, which is, as figure 4.1 indicates, required for a federation. Instead, the national institutions were constitutionally subordinate to the political institutions elected within each of the separate states.
It is important to understand that while a major purpose of federalism was the maintenance of the states as political entities with an important degree of political autonomy within the union as a whole, the basic goal of the move from confederalism to federalism was the strengthening of the central government of the United States, by endowing it with a source of authority separate from that of the sates. As such, while contemporary US political debates about federalism are often about states’ rights in the system, the purpose of creating what Madison called an “extended republic” was to create a central government strong enough to govern the union of states as a whole and to subordinate the states to that central government in policy areas reserved by the new Constitution to the national government.
WHY FEDERALISM?
At this point, we have established the options available in terms of the relationship between the central government and its subunits, and we have sketched out how authority patterns vary among unitary, federal, and confederal systems. Now we turn to a consideration of why designers of a constitution for a given country might, as a general proposition, choose federalism. Therefore, we shall turn to an examination of the specific institutional characteristics of federations. A classic reason—in the sense of having been the justification for the US federal design itself—is that separate political entities may already exist and “come together,” as Stepan (2004a) put it. That is, if existing sovereign units decide to build a common government to subsume some of their separate authorities, they may establish a federal balance of power to preserve some of their existing sovereignty even as they create a larger polity. While this was certainly the case in the United States, it would be only partly or not at all accurate as a description of most of the other federal systems. Some federal systems—notably Belgium, India, South Africa, and Spain—were unitary states at one time and later transformed themselves into federations as a result of political decisions taken in the national legislature or a constitutional assembly. Stepan refers to this pattern as “holding together,” to signify that a political decision to guarantee sovereign powers to the subnational level was made as a way to provide distinct subnational groups, such as geographically concentrated ethnic groups, incentives to remain within the larger country.
In fact, size and ethnic diversity are very strongly linked to federalism. By size, we mean here both in population and area. In fact, figure 4.2 shows that all countries that are exceptionally large in one or both dimensions are federal. The figure shows a country’s population on the horizontal dimension, and its area on the vertical dimension. Federal countries are shown with one symbol, the solid square, while unitary systems are shown with two other symbols (as discussed in more detail in the next section). The data on which the graph is based are provided in table 4.1, and show that the median population of our federal systems is approximately 47 million, while the median size of our unitary states is much smaller, approximately 14 million. Measured by territorial extent, the median federal system is almost ten times the size of the median unitary state (1,964,375 square kilometers versus 255,660).
In figure 4.2, the more a country is located toward the upper right, the more it is large in both dimensions, while the more it is located toward the lower left, the smaller it is in both dimensions. The dashed curve in the graph separates the set of countries that are very large in at least one dimension, all of which are federal. A case like Japan, which is very large in population, is nonetheless much smaller in area that any of the countries on the other side of the dashed curve. (Japan is also an unusually homogeneous society.) The dotted curve then roughly differentiates those smaller countries that are mostly unicameral as well as unitary, from those unitary systems that have bicameral legislatures yet unitary divisions of power. We will discuss the distinction between unicameral and bicameral systems further in the next section of this chapter, and then more fully in chapter 7. For now, the important observation is that bicameralism tends to be associated not only with federalism—indeed, all the federal systems are bicameral—but also with those unitary systems that are large in one dimension—such as Japan (large population, middle-size area) and Chile (middle-size in area, but small in population).
Fig. 4.2. Relationship between size (population and area) and division of powers
Key: USA, United States; AR, Argentina; AUS, Australia; AUT, Austria; BE, Belgium; BR, Brazil; CA, Canada; CHL, Chile; CO, Colombia; CZ, Czech Republic; DK, Denmark; FI, Finland; FR, France; GER, Germany; GR, Greece; HU, Honduras; IND, India; ISR, Israel; IT, Italy; JP, Japan; KR, South Korea; MX, Mexico; NL, Netherlands; NZ, New Zealand; PL, Poland; PT, Portugal; SA, South Africa; SP, Spain; SWE, Sweden; SWI, Switzerland; UK, United Kingdom.
While most federal systems are large in both area and population, a glance at figure 4.2 shows that there are some federal systems mixed in among the mostly unitary smaller countries (Austria, Belgium, and Switzerland). On the other hand, none of the unitary states are as large, territorially speaking, as the median federal state. The one that comes closest is Colombia (1,138,914 square kilometers) and it is worth noting that it had a phase of experimentation with federalism in the nineteenth century. Six unitary states have populations close to or greater than the median for federal states (Colombia, France, Italy, Japan, Korea, and the United Kingdom),4 but each of these is below the median federal system in area.5
Two of the federal systems that are small in both population and area—Belgium and Switzerland—are excellent examples of the use of federalism as a means to grant local autonomy to regionally concentrated ethnic or linguistic minorities. While these countries may be small, they have clearly distinct ethnolinguistic regions that are able to constitute local majorities in specific subnational units. Much the same can also be said about several of the large federations that likewise have subnational units where national minorities constitute either local majorities or at least much larger minorities than in the federation as a whole, including Canada, India and South Africa.
Although US state boundaries were never set with ethnic representation in mind, some national minorities constitute far greater shares of the population of specific states than they do of the US population as a whole. For instance, African Americans comprise just more than 12 percent of the US population, but more than 25 percent of the population of six states.6 Native Americans comprise less than 1 percent of the US population, but more than 4.5 percent of seven states.7 Hispanic/Latino Americans comprise about 12.5 percent of the US population, but more than 25 percent in four states.8 Federalism permits minorities potentially to enjoy greater political influence over a state’s politics than they might be able to exercise at the national level.
INSTITUTIONAL CHARACTERISTICS OF FEDERALISM
We have already indicated one necessary defining characteristic of federalism that is institutional in nature: the presence of a constitution that establishes dual levels of sovereign governments. Various institutional properties then follow as a means to put the principle of federalism into effect. Those institutional properties are:
1. independently elected subnational and national governmental institutions;
2. a constitution that guarantees sovereignty to the subnational governments and provides some distribution of responsibilities between national and subnational; and
3. institutions to represent subnational preferences within national political institutions.
The first two criteria were already discussed earlier, and shown in figure 4.1, where we indicated that in federal systems, voters elect both national and subunit governments, which exist in a “transactional” relationship with one another, rather than having the national government hold all sovereignty. Many unitary systems meet the first criterion (and some meet the third as well), but by definition they do not meet the second. If a given political system does not meet all three of these conditions, it is not federal.
In unitary states, then, only the national government is sovereign, and any policy responsibilities exercised by subnational governments are contingent on national law and, as a result, may legally be taken away by a change in national law. Similarly, in confederations, only the units are meaningfully sovereign, and the government of the confederation itself enjoys no direct democratic connection to voters in the broader confederation (whether directly or through being responsible to a democratically elected central legislature). One of the reasons unitary systems allow, at least in principle, the national government to take away local or provincial governments’ policy-making responsibilities is not only that there is no constitutional guarantee of a division of power, but also that there is usually no mechanism to represent the preferences of subnational electorates, or the regional governments themselves, in the central policy-making institutions.
We now turn to the principal means by which such interests are represented in federal systems—through a second chamber. We already noted, with respect to figure 4.2, that bicameralism tends to be associated with countries that are large and especially those that are federal. We discuss bicameralism in greater length in chapter 7. Now we turn to a more detailed discussion of why federalism goes with bicameralism, and of variations among federal systems in the ways in which they structure their second chambers.
As noted in chapter 2, bicameralism with equal representation of the states in the second chamber was one of the great compromises of the US Constitution-drafting process. Whereas the smaller states favored the so-called New Jersey Plan, which would provide for equal representation of the states in a unicameral federal congress, the larger states favored the Virginia Plan, which called for representation based on each state’s population. This division of preferences over institutional design is quite natural and rational, in that the delegates to the Constitutional Convention from smaller states feared that their states would be outvoted by a legislature dominated by representatives of larger states. On the other hand, larger states’ delegates objected to equal representation, as it would dilute the representation of the big states. The so-called Great Compromise, crafted at the Constitutional Convention in 1787, was to have a bicameral legislature with one chamber based on each of these principles. Thus the United States has a House of Representatives (first chamber, also called the lower chamber) where each state has a number of representatives based on its population, and a Senate (second chamber, also called the upper chamber) where each state has an equal number of senators regardless of its population. In other words, representation for each state in the first chamber is approximately proportionate to population, while the second chamber over-represents the smaller states. Similar compromises are common in other federal systems, as we shall see.
All thirteen of our federal democracies are bicameral. While figure 4.2 shows that many unitary systems are also bicameral, as we shall see in more detail in chapter 7, there is a notable difference between bicameralism in federal and large unitary systems. The constituencies of second-chamber legislators tend to be drawn very differently in federal and unitary systems. The role of the second chamber in a federal system is to represent the subnational units, whether or not equally. None of our unitary bicameral systems has a second chamber with equal representation for provinces (or other units),9 and many do not even have their subnational units and second-chamber constituencies coincide in any way. The equal representation of the states in the second chamber is, of course, the classic “federalist” compromise that we see in the United States. However, as we shall see shortly, not even all federal systems share this feature.
FEDERALISM AND FISCAL DECENTRALIZATION
One of the typical justifications for federalism is that it allows regional specialization in policy. Separate regional governments may decide, through their own internal democratic processes, to provide different mixes of policies than do others. They may even compete with one another to attract business investment and citizens themselves, who may be drawn by a regional unit’s specific mix of taxes, public services, and other policy characteristics. This kind of specialization and competition is less likely to be feasible in unitary states because the local governments have far less autonomy and most of their policy choices are significantly constrained by the national government.
A more immediate requirement for regional policy specialization than whether the system is federal or unitary is the availability of revenue. Almost any policy requires revenues to carry out, and countries vary widely on the share of revenue raised by the subnational levels, compared to the central government. In table 4.2 we see the various levels of government for twenty-six cases and the revenues they raise both as a function of the percent of GDP and what shares each level receives as a percentage. It is immediately evident that the degree of revenue centralization is closely related to the unitary-federal distinction. The median share of revenue raised by the central government is 93.22 percent in the seventeen unitary systems, and 78.74 percent in nine federal systems.10 The shares in unitary systems range from 68.04 percent (Sweden) to 99.09 percent in the most centralized (Greece). Among the federal systems, the United States, at 66.32 percent of revenues raised by the central government, is below the median for federal systems but higher than the two most decentralized federal systems on the list: Canada (53.29 percent) and Switzerland (59.15 percent). Two of the federal systems in which the share of revenue raised by the central government is at or above the median, and thus more akin to the unitary systems, are also relatively new to the game of democratic federalism. Belgium and Spain were once unitary states. Belgium began a process of federalization only in the 1970s, and Spain somewhat later than that. In general, the strength of federalism is linked very closely to the ability of subunits to collect revenue as well as their share of said revenue relative to the other portions of the country.
Table 4.2. Tax revenue in twenty-six democracies (1990−2010 averages)
Note: Data were unavailable for the five excluded cases.
Source: OECD Fiscal Federalism Network, http://www.oecd.org/tax/federalism.
US FEDERALISM IN COMPARATIVE PERSPECTIVE
Even if the United States was the first federal system, its institutions and political practices are atypical in some respects. We now turn our attention to a comparison of US and other variants of federalism.
1. Equal representation of states with widely divergent populations. The principle of equal representation in the Senate is enshrined in the US Constitution. In fact, it is the only feature of American government that the Constitution’s provision on amendments (Article V) explicitly says cannot be subject to amendment. If the units of a federation are not too disparate in size, equal representation may not be politically significant. However, if the largest state is very much bigger than the smallest, and the population is concentrated in one or a few states, then equal representation of states implies highly unequal representation of voters. Consider that the ratio of population size between the largest American state, California, and the smallest, Wyoming, was about 66 to 1 at the 2010 census. Yet in the Senate, their representation is, of course, 1 to 1, given that each has two senators. It is worth reiterating that this is precisely the intent of the federal bargain institutionalized in the Great Compromise for the original thirteen states—to have one chamber of Congress representing the states and another representing the population. Of course, in the House of Representatives (lower house), California, with fifty-three seats as of the 2010 census, and Wyoming, with one seat, are represented far more closely to their shares of the national population.11 This disparity of popular representation is worth putting in both historical and cross-national perspective. Historically, the ratio of largest to smallest states in the United States has increased greatly. It was only 12.7 to 1 when the Great Compromise was reached, or about 18 percent as disparate as it is in the early twenty-first century.
A good way to compare federations to one another and also to compare the United States of the early twenty-first century with that of the late eighteenth century is to construct an “effective number” of federal units by both population and Senate representation. We then can compare this effective number to the actual number of units. The effective number gives us a weighted index of how concentrated some quantity is. It weights the states’ contribution to the index by their own shares, which is to say each share is squared. The squares are then summed, and the inverse (reciprocal) of this sum yields the “effective” number. The effective number, first introduced by Laakso and Taagepera (1979), is now a standard measure used by political scientists in analysis of party vote and seat shares, and we will use it in this way in chapter 5. To illustrate how the effective number works, consider a hypothetical federation consisting of three states with equal populations; for this federation, the effective number of states based on population is 3.0. If, instead, one of the states had two-thirds of the nation’s population, another had 23 percent, and the smallest of the three had the remaining 10 percent, then the effective number of states would be only 1.97. The smaller effective number compared to the actual number indicates that the federation’s population is concentrated, rather than equally distributed across the states.
We can perform the same calculation on states’ shares of seats in a legislative chamber. Comparing the effective number of states by population with the actual number of states gives us a sense of how concentrated the population is across the units of the federation. Comparing the effective number of states in the Senate to the effective number by population gives us a clear idea of how closely Senate representation mirrors the distribution of the population across states. Throughout this discussion, we will refer to the calculation based on population as Nfed, signifying “effective number of units in the federation,” and the calculation based on Senate representation as Nsen, or “effective number of units in the Senate.”
Table 4.3. Actual and effective number of units in federal systems
Notes:
a Indicated as number of states, provinces, or other principal units, plus number of other territories with any legislative representation (if greater than zero), but not including the capital territory, if any.
b Separate capital territory (all with legislative representation, except District of Columbia in the United States).
In table 4.3, we report the effective numbers of units in both the federation and its second chamber for each of the federal systems in this book. We can see immediately that the US population is quite concentrated, such that the effective number, Nfed (22.75) is less than half the actual number of states (50). The effective number of states in the Senate, Nsen, is, of course, 50, because every state is represented equally in the second chamber. The data from the other federal systems show that some federations have even more concentrated populations than the United States, notably Argentina, where the effective number of provinces (5.72) is less than a quarter the actual number (23). Like the United States, Argentina represents its states equally in its senate, as can be seen by a figure for Nsen that is equivalent to the actual number of units (provinces plus a special capital district). Brazil’s concentration of population is similar to that of the United States, and likewise its states (and capital district) are equally represented in their senate. Argentina and Brazil are the only federations in the set of countries considered in this book that have second-chamber representation of the population more unequal than the United States has.12 Only Switzerland comes close among the industrialized states, though in Switzerland some of the smallest cantons are actually considered “half-cantons” and thus have less representation in the upper chamber than the full cantons.
On the other hand, several other federations have an effective number of units in their second chambers that are much closer to the effective number of units in the federation. What this means is that, contrary to the concept the founders of the US Constitution agreed to in the Great Compromise, these federations provide fewer seats in their second chambers to the smallest states than to the largest. Consider the example of India, where the effective number of states in the federation is less than half the actual number, similar to the United States. The ratio of largest to smallest states in India is, however, vastly greater than in the United States, at more than 300 to 1. Rather than provide equal representation to these highly unequal states, Indian second-chamber representation is somewhat proportional to state population, ranging from twelve to eighty-six members per state, resulting in a value for Nsen that is actually closer to Nfed than it is to the actual number—the opposite pattern from that seen in the United States. Several other federations provide second-chamber representation that is even closer to the units’ shares of the population than India does. Austria and Belgium have almost perfect proportionality of unit representation in the second chamber.
Two federations deviate somewhat from the principle of representing their units in the second chamber. In Mexico, the senate has three senators for each of the thirty-one states and the capital district, thus resembling the US (and Argentinean and Brazilian) models of equal representation. However, the Mexican senate also has another thirty-two national senators, or one-fourth of the total.13 This provision leads to a value of Nsen (30.58) that is slightly less than the number of units (31).14 In Spain, some senators are directly elected within the provinces, but these are not the primary units of the federation.15 The primary units are the Autonomous Communities, most of which consist of more than one province and have additional senators, appointed by the parliament of the Community.16
The federal systems can be compared at a glance in figure 4.3, which graphs the ratio of actual units to Nfed on the horizontal axis, while the vertical axis shows the ratio Nsen to Nfed. By graphing our data in this way, we are able to see at a glance how the United States compares to other federations in two of the core features of such systems: how equal the units are to one another in population, and how equal they are in representation. The closer a country comes to the upper right of the space, the more it offers equal representation despite highly unequal populations of its units. The more it is in the lower left, the more the federation approaches equal-sized units. Any case that falls on the diagonal line has adhered to the institutional principle of the Great Compromise: unit equality in the second chamber, regardless of population. (Those that are above the diagonal are those that have capital districts that are also represented, which we return to later.) The federations that are below the diagonal are those in which the units are not equally represented.
Fig. 4.3. Relationship between population concentration and second-chamber representation of units in federal systems
Key: The horizontal axis graphs the ratio of actual units to Nfed; the vertical axis shows the ratio Nsen to Nfed.
At a glance, we can see that Argentina employs equal representation in its senate despite having a population that is highly concentrated in just a few of its provinces, and we can also immediately identify a set of federations that have not adhered to the equal-representation principle of the Great Compromise—including Canada and India and the others that fall well below the diagonal line. Only Brazil and Switzerland join the United States and Argentina as federations with a ratio of actual to effective number of states greater than 1.75 in which the ratio, Nsen to Nfed, is likewise greater than 1.75. In other words, these four federations combine unequal-sized states with equal representation of those states in the second chamber. The other federations either have much less skewed population distributions or more proportionality of unit representation in the second chamber, or both.
Fig. 4.4. Actual and effective number of US states over time
As we noted, this equal representation of unequal states is one of the immutable features of the US Constitution, which raises the question of whether the framers who compromised on this fixed equality of states in the Senate possibly could have imagined how much the divergence of state populations would grow over time. As we can see in figure 4.4, the gap between the actual and effective number of states has indeed grown significantly over time. At the first federal census, the effective number of states in the union (7.9) was only about 40 percent less than the actual number (13). Since 1960, the effective number has been no greater than half the actual number. Since 1980 the effective number has actually fallen somewhat, as the largest states have grown much faster than the smallest ones. It is interesting that equal representation of widely divergent state populations has not become a political issue in the United States. The principles of American federalism, including equal representation in the Senate, are deeply rooted in American politics, but Senate representation could emerge as an issue in the future as the disparity of populations continues to expand, as it almost certainly will do.17 The original balance at the founding, when the idea of equal representation of the states was set into the Constitution, becomes more unrepresentative over time. As we have seen here, some other federations—notably Canada and India—trade off to some degree the “representation of units” principle in order to approach more closely the “one person, one vote” principle of democracy.
2. Absence of representation of the capital and other territories. In some federations, the national capital is located within its own territory, rather than within one of the states or provinces. For instance, the US capital, Washington, is contained within the District of Columbia, rather than within either Virginia or Maryland, the two states that ceded territory to create the capital district. Five of our other federal systems also have special capital districts: Argentina, Australia, Brazil, India, and Mexico. However, the United States is unique in not granting the residents of its capital territory voting representation in the national legislature. The residents of Washington, DC, have a nonvoting delegate in the House of Representatives and neither voice nor vote in the Senate. Thus DC residents are deprived of representation despite being more numerous than the residents of Wyoming, and being close in number to those of Alaska and North Dakota.18 The matter of the federal capital’s nonrepresentation in Congress is sporadically an issue in the United States and is typically framed as a question of whether the District of Columbia should be granted statehood.19 Among all of our federations, however, the US capital has the smallest percentage of its population in the national capital, by far. Only 0.2 percent of the US population lives in Washington, DC. The next lowest shares for federations that have special districts for their capitals are 1.2 percent for Brasília (Distrito Federal), Brazil, and 1.6 percent for Canberra, Australian Capital Territory.20 The smallness of Washington, DC, relative to the nation’s population as a whole may explain why the issue of representation for its residents does not loom larger in the national debate.
Similarly, many federations have other territories that are not granted the status of a state or province, often because they are noncontiguous or otherwise remote from the population centers of the nation, sparsely populated, inhabited largely by indigenous groups, or some combination of these factors.21 Examples include the Northwest Territories, Nunavut, and Yukon in Canada; Northern Territory in Australia; and six territories in India, including the Andaman and Nicobar Islands. In many federations, some such territories have legislative representation.22 However, no US territory has either a voting representative in the House of Representatives or any delegate to the Senate or a vote in presidential elections (although some of them are included in the major parties’ presidential nomination processes). US territories include the Commonwealth of Puerto Rico, which at more than 3.7 million residents, is larger than almost half (twenty-four) of the states (and only slightly smaller than Oklahoma).23 US territories also include several smaller units: American Samoa, Guam, the Northern Mariana Islands, and the US Virgin Islands.24 The largest of these, Guam, has just shy of 160,000 residents.25
3. Concurrent national-subnational elections versus staggered calendars. Because both subnational and national elections must be held in federal systems, the question of the electoral calendar arises. If many states hold their own legislative or executive elections in between elections for the national legislative or executive elections, the state elections may serve as informal bellwethers of support for the national parties. If, on the other hand, all states hold their elections at the same time as national elections, obviously this bellwether phenomenon can’t exist; moreover, there may be less political distinctiveness of the two levels of the federation. Whether unification of the national and subnational calendars favors the nationalization of regional politics or the regionalization of national politics depends on a factor we discuss subsequently, which is the nature of the party system. First, however, let us see how the United States compares to other federal systems in the extent to which national and subnational elections occur separately.
Our federal systems vary substantially on this dimension. In the United States, of the forty-eight states that have governors with four-year terms, thirty-five elect their governors at the midterm of the federal president’s term (2006, 2010, and so on). Ten elect their governors at the same time as the president (2008, 2012, and so on). The remaining four hold their gubernatorial elections in odd-numbered years: Kansas and Louisiana in the year before a presidential election, New Jersey and Virginia in the year after. Two states, New Hampshire and Vermont, have two-year gubernatorial terms with elections both in presidential-election years and at the presidential midterm. As for legislative elections, because all states and the US Congress hold elections biennially, federal and state legislative elections always coincide, except in the four states that have elections for state offices in odd-numbered years.
The electoral calendar in the United States is thus somewhere in the intermediate range in terms of the unification of state elections, both with one another and with national elections. The highest degree of unification would be if all national and state or provincial elections occurred on the same day. Brazil is the one case with complete unification of the federal and state electoral calendar. Every four years, voters vote for federal president and congress members and their state governor and legislative assembly on the same date.26 At the opposite extreme, every subnational unit could hold its elections on its own schedule, never coinciding with either another unit or with the national elections. This pattern is typical of Canada, where election dates rarely coincide. It is approximated in Germany, where between federal elections in 1998 and 2002, fifteen of the states held state elections on fourteen different dates, while one27 held elections concurrently with the two federal elections. India and Mexico also have distinct calendars, though often with several states having elections concurrent with the federal elections or on interim dates on which several states vote. In Spain, most of the autonomous communities have their elections on a single date separate from federal elections, but a few—notably the Basque and Catalonian regions (the largest)—have their own dates. Thus there is a wide variation among our federal systems in the degree to which their electoral calendars are unified, with the United States closer to the unified end of the spectrum, but not the most unified.
4. Integration of the party system. The integration of the party system refers to the extent to which the same parties compete across the federation, in both national and subnational elections. Parties may vary in their strength across the subnational units, but if the party system is integrated, the same parties run and win representation in both national and subnational legislative elections in all states or provinces. The United States is quite unusual among federal systems in this respect.28 The Democratic and Republican Parties contest elections in all states, and in no state is one of the parties so dominant that there has not been recent alternation in its state governorship. In some of the very small states that have only one congressional representative, the same party might regularly win the seat, but the other party typically contests it. Moreover, distinct state parties are very rare in the United States. Only Alaska has a distinct party—the Alaskan Independence Party (AIP)—which has been competitive in several consecutive state elections. The AIP even elected its candidate, Walter Hickel, to the state governorship in 1990. The only other states that recently have had governors who were nominated by neither the Democratic nor the Republican Party are Connecticut (Lowell Weicker, nonpartisan, 1990) and Minnesota (Jesse “The Body” Ventura, Reform Party, later Independence Party, 1998). These are the only recent exceptions to the rule that the US party system is fully integrated at the national and state levels.
This high level of integration is not found in most other federal systems, although some, such as Australia, Germany, and Mexico, come close. In each of these countries the party system is highly integrated, in that the same major national parties contest elections in all states (though all are not necessarily competitive in all states). Australia and Germany each have at least one party that competes only in one or a few states—the Christian Social Union in Bavaria, Germany, and the National Party in some Australian states—but in these cases the party is in an essentially permanent alliance with one of the national parties in federal elections.
Spain and India represent the most unintegrated party system in our set of federal democracies. In Spain, there are two national parties (the Socialist and Popular Parties) that are competitive in most of the regions, while in India, one party (Congress) is national in scope and another major party (Bharatiya Janata) competes in most states. However, both countries have numerous parties that are specific to a single subnational unit, and often these regional parties are major parties in their own state or regional assembly. In six of the nineteen autonomous communities of Spain, at least one regional party also obtains representation in the national parliament; frequently, one of the Basque or Catalan regional parties will provide support in parliament for a minority cabinet of one of the two big national parties (see chapter 8). Many state parties are represented in India’s national parliament as well, and in both countries these parties often hold the balance of power between the two major national parties, effectively determining which one of them will be in government.29
The Canadian party system lies somewhere in between the highly integrated US and unintegrated Indian and Spanish examples. The three major national parties—Conservative, New Democratic, and Liberal—compete in all provinces, and except in Quebec, they are the only parties that regularly win seats in the federal parliament from any province. However, in addition to Quebec parties,30 there is a Saskatchewan Party that has formed the government in that province twice and a Yukon Party that is active in that territory. Very small province-specific parties exist in a few other provinces. None of these parties competes in federal elections.31 In addition, even though it was for decades only the third party nationally, the New Democratic Party has long been one of the two major parties in several provinces. Thus there is a high degree of regional party diversity in Canada, though more party-system integration than we see in India and Spain.
Whether a party system is integrated or not matters for several reasons. For example, distinct subnational parties may have a stronger incentive than subnational chapters of national parties to seek to extract resources from the center as a means to build their own political bases. In more integrated party systems, on the other hand, the national scope of the parties implies a balance of interest within the parties in seeking to please national and subnational constituencies. It is also typical in federations with integrated party systems for units governed by a given party to affect the national reputation of the party and even to serve as recruitment grounds for national candidates who may run on the claimed success of their policy implementation in a state or province. The integration of the party system is also likely related to the extent of separatist tendencies within a federation. For example, Canada’s province of Quebec has held two referenda on separation from Canada, though they were defeated. Spain has an active and violent separatist movement based in the Basque territories, as well as movements seeking greater autonomy in various other regions. India has numerous separatist movements. Of course, these countries do not have separatist movements because their party systems are not integrated; in fact, the causal path may well run in the opposite direction. Nonetheless, if distinct subnational parties gain the upper hand over national parties, the federal balance may become more difficult to sustain in the face of these centrifugal pressures.
CONCLUSIONS
Federalism provides an arena of regional political competition and policy formation that is distinct from the arena of national politics. As we saw in this chapter, just fewer than half of our thirty-one democracies are federal. The federal systems tend to have a lower share of revenue collection in the hands of the central government, which at least potentially provides an opportunity for distinct jurisdictions to engage in experimentation and competition in policy in a way that is less feasible in a unitary state. We saw that federal systems are always bicameral, with the second chamber structured to represent the units of the federation. Nonetheless, the United States model of equal representation of the subnational units in the second chamber is not followed by all federal democracies, with several providing fewer seats for the smallest units in order to compromise the federalist principle of unit representation somewhat to approximate more closely the democratic principle of one person, one vote. These are contradictory principles in any federation that has states of widely divergent populations, and all federal systems need to strike some balance in them in order to be both federal and democratic.
We saw that the degree of integration of the party system and the coordination of the electoral calendar are significant for the degree of distinctiveness of subnational politics. Those principles of federalism that emphasize the sovereignty of the separate units are most strongly manifested with separate calendars of elections and relatively unintegrated party systems, such as in Canada and India. Those principles of federalism that stress the coordination of the separate units within one national polity are strongest with concurrent elections and integrated party systems. It is perhaps somewhat surprising, then, that the United States has one of the most coordinated electoral calendars and the most integrated party systems, given the primacy placed on the states in American politics. Recall that the United States has one of the lowest shares of central-government revenue collection (though not as low as Canada), as well as equal representation of highly unequally sized states in a powerful “federalist” second chamber. How can the United States have all of these decentralizing tendencies despite electoral coordination and party-system integration? The answer probably lies in distinctive features of US politics that we will discuss in later chapters. For instance, US parties are notably weaker and more locally organized than parties in most other democracies in this book, a product partly of the presidential form of government and partly of the way legislators are elected. None of our other democracies combines a presidential executive with single-member district legislative elections and federalism, as we shall see. In the context of all of these institutional features that tend to fragment political power, perhaps what is surprising is that the party system is so integrated. In this context, it could be that without coordination of the electoral calendar there would be much less coordination of states’ distinct interests at the federal level than what we see in practice. On the other hand, the much stronger parties associated with parliamentary democracy in countries such as Canada, Germany, and India might threaten subnational sovereignty and the development of distinct regional parties were it not for the separate electoral calendars that allow state politics to remain a mostly separate arena even in the context of strong parties. This discussion thus highlights the ways in which different components of the larger political system fit together, and sets the stage for us to continue our discussion of these other components.
Notes
1. The terms used for state legislators vary across states, but typically they are members of the assembly and senators. All state legislatures are bicameral, with the exception of Nebraska.
2. The European Union has its own parliament, which would seem to meet one of the core criteria for federalism: Voters wherever they live in the EU have the right to vote both for their country’s politicians (the governments of Britain, France, Germany, etc.) and for European-level politicians. However, the EU lacks its own executive that is either directly elected by European voters or accountable to the European parliament. Instead the executive is comprised of delegates of the member-state governments. There is a distinct European bureaucracy, but the enforcement of decisions made by the European institutions remains with the member-state governments. It is this last point that makes the EU fundamentally not a federation: Binding authority remains with the member states.
3. Often the breakup of multinational states is violent, such as with dissolution of Yugoslavia in the 1990s or Sudan in 2011. Indeed, the end of the Cold War led to a dramatic example of dissolution, as the Soviet Union went from one single state to fifteen separate states. However, none of these cases is the breakup of a democratic federation, and Stepan (2004a) makes a convincing argument that no authoritarian system can be considered meaningfully federal.
4. Colombia, a unitary state, has a population that places it just below the median for federal systems.
5. Only Colombia has an area more than one-third that of the federal median. (Colombia’s area is about 58% of the median federation).
6. Mississippi (37.1%), Louisiana (31.5%), Georgia (29.7%), Maryland (28.7%), South Carolina (28.3%), and Alabama (26.2%). In addition, African Americans comprise 54.4% of the population of the District of Columbia, but as we noted below, DC residents have limited political representation. Source: US Census Bureau (2006–2008 estimates).
7. Alaska (13.4%), New Mexico (9.3%), South Dakota (8.4%), Oklahoma (6.7%), Montana (6.1%), North Dakota (5.3%), and Arizona (4.5%). Source: US Census Bureau (2006–2008 estimates).
8. New Mexico (42.1%), California (32.4%), Texas (32.0%), and Arizona (25.3%). Source: US Census Bureau (2000 census).
9. The Dominican Republic (a democracy not among our thirty-one countries) is among the rare unitary systems in which the subnational units are equally represented in a second chamber.
10. This difference of means easily meets standards of statistical significance.
11. Nonetheless, even in the House, Wyoming remains over-represented, due to every state’s guarantee of one representative no matter its population. California has 53 representatives; to give it the 70 that would be needed to provide a fully proportionate share would require a larger House. As we see in the chapter on legislatures, it was formerly the case that the US House size periodically increased; however, it has been fixed at 435 since 1910.
12. Argentina and Brazil, but especially the former, also stand out in providing their smaller states over-representation in the first chamber as well, whereas the United States comes very close to providing perfect proportionality of House seats to state populations. See Samuels and Snyder 2001 for a discussion of malapportionment in first chambers.
13. These are elected in a single national district, proportional to political parties’ shares of the nationwide votes (i.e., the sum of all votes cast for senatorial tickets in each of the states).
14. For Mexico, we calculated Nsen as follows: First, we calculated the percentage contribution of each unit to the election of national senators (each unit’s share of the national population multiplied by 31. Then we added this result for each state to the number of state-based senators (always 3) to arrive at an “effective” number of senators for each state, which ranges from 7.3 for the largest state (Estado de México) to 3.14 for the smallest (Baja California Sur). This effective number of senators for each state then serves as the basis (out of the total 128) for calculating the effective number of units in the senate.
15. The Spanish constitution does not identify Spain as a federation; however, it qualifies in every definitional aspect that we employ in the chapter. The same can be said for South Africa.
16. Provinces predate the establishment of Autonomous Communities. Therefore, whereas Spain’s federation would be classified as “holding together” in Stepan’s sense, the provinces engaged in a process of “coming together” to establish the larger Communities that form the basis of federalism in Spain.
17. The most important of the Founding Fathers with respect to the design of US institutions, James Madison, was originally an advocate of representation of the states according to their share of the national population, as we discussed in ch. 2.
18. Only since 1964 have DC residents been entitled to vote for president.
19. Another option that is also discussed is the opposite of statehood: retrocession of the land north of the Potomac River back to Maryland save for a very small district in and around the National Mall where most of the buildings of the federal government reside. Such a process would allow the currently disenfranchised citizens living in the District of Columbia to vote for House and Senate members and in presidential elections as citizens of Maryland. There is precedence for such a move, as the land originally ceded to the District by Virginia was returned to Virginia, including Alexandria, in 1847 (basically all the territory originally given to the District south of the Potomac). A variation of this option was proposed in Congress in 2004, which would have granted citizens of the District voting rights as part of Maryland without formal retrocession (H.R. 3709: District of Columbia Voting Rights Restoration Act of 2004).
20. At the other end of the scale, about 38% of Argentina’s population lives in Buenos Aires.
21. The United States is unique among our federations in having noncontiguous units (excluding islands close to the mainland) that are fully incorporated into the federation: Alaska and Hawaii.
22. In Australia, e.g., the Federal Capital and Northern Territories have half the representation of a state in the federal senate.
23. Based on the 2010 US Census.
24. Surprisingly, given a lack of representation in US federal elective institutions for these territories, the integration of the party system extends to all of them except Puerto Rico. While local parties or independents are prevalent in the other territories, Democrats and Republicans contest territorial elections.
25. Each of the US territories listed has a nonvoting delegate to the House of Representatives, as entitled under a federal law that mandates such a delegate to any territory with more than five thousand residents. Federal law also allows any territory with more than sixty thousand residents to apply for statehood. Each listed territory meets that threshold as well, but given that all, other than Puerto Rico, are much smaller than the smallest existing state, these territories would be unlikely to be considered seriously for statehood. The question of statehood for Puerto Rico, on the other hand, is very much a live issue, as is the opposite position: full independence.
26. Elections for local governments are held on a common day throughout Brazil, but separately from other elections.
27. Mecklenburg-Pomerania.
28. The party system has been less integrated in the past—especially before the Civil War, when there were distinct subsystems.
29. Both India and Spain are parliamentary systems, though in India large coalitions often contain several state parties, while in Spain it is typical for a single-party minority government to form, with regional parties offering support from outside the cabinet.
30. Parti Québécois in provincial elections and Bloc Québécois in federal elections. The latter party lost most of its seats in the 2011 election, although it remains active.
31. Additionally, in the Northwest Territories and Nunavut territory, there are no parties in provincial elections, but the national parties do contest the single seat each territory has in the federal parliament.
