1 Introduction

A classical prediction with regard to the development of federal systems of government, made by the British philosopher James Bryce, is that over the long run, “the importance of the States will decline as the majesty and authority of the National government increase” (Bryce 1888, vol. III, p. 653). Similarly, German public economist Johannes Popitz (1927), while witnessing a period of rapid centralization in Germany, speaks of a force of attraction exerted by the central budget with regard to the competencies for policy-making. Reading these classical predictions, and also looking at the empirical evidence, one is tempted to conclude that there is an inevitable tendency towards centralization in federal systems. Federal constitutions appear to be inherently unstable and sub-central governments are, in the long run, in danger of being overwhelmed by a tendency of government centralization.

There are, however, only very few cases in history where formerly federal systems have experienced a sudden collapse into a highly centralized, unitary regime. More often, centralization can be observed as a gradual process, with incremental increases of central government competencies, which often do not even occur at the expense of sub-central units. The rise of the modern welfare state has, for example, in most countries been a matter of central government politics. But often this has been the result of the central government claiming responsibility for a novel policy issue, and not of an explicit, vertical transfer of competencies. Thus, not every increase in the share of central government spending (or taxation) relative to total public spending (or tax revenue) reflects a centralization of formerly decentralized competencies.

Similarly, as we will see in greater detail below, not every loss of local autonomy is reflected in tax and expenditure shares. For example, tax coordination and revenue sharing schemes involve a significant loss of sub-central fiscal autonomy, but are not reflected in budgetary indicators. Adding to this, there is also the complication that different countries can have very diverse institutional frameworks. However, to investigate how different constitutional provisions shape processes of centralization is an interesting question in itself. And given the limited reliability of quantitative indicators, we believe that an in depth comparative case study is warranted: A detailed look at two countries over a long period of time may reveal additional information on the mechanisms of centralization that are not extracted through econometric analyses.

Our research questions in this paper are the following: (a) Does an initial constitutional framework of competitive federalism inhibit centralization, compared to an initial constitutional framework of cooperative federalism?; (b) Do the pathways of centralization differ between these two types of constitutional frameworks?; and finally (c) Which general lessons on the political economy of government centralization can be learned from comparing both types of regimes?

We choose Germany and the United States as representatives of the two distinct ideal types of federal systems: Germany is an example of cooperative federalism, and thus represents a constitutional framework that, prima facie, can be expected to be very vulnerable with regard to attempts to centralize competencies. The United States, on the other hand, represents the model of competitive federalism—at least when we base this categorization on constitutional origins. As we will see in this paper, both countries have experienced strikingly similar trends of centralization. The United States started off closer to the ideal type of competitive federalism, and certainly still maintain a higher degree of sub-central autonomy than Germany today, but significant tendencies of shifting responsibilities to the central level can be observed in both countries.

Our findings can be interpreted as supporting evidence for the ubiquity of Popitz’s Law, even under very different institutional frameworks. Nevertheless, we will also see that the actual instruments of centralization differ considerably between the two countries. Even if there may be a universal law of centralization, the actual processes of centralization come in many different shapes and colors. This finding also has some implications for constitutional policy-making: A centralization-proof constitution is difficult to construct, and the preservation of federalism probably relies to a large extent not only on formal, but also on informal institutions.

The argument will proceed as follows: In Sect. 2 we briefly sketch the long-term, historical developments in intergovernmental relations both in Germany and in the United States. Within this limited summary, we can of course only hint at a small number of important junctions and not provide an exhaustive history of federalism in both countries. In Sect. 3, we attempt to identify the main instruments of government centralization in both countries, and we will show that the process of centralization differs substantially between both regimes. We discuss the political economy of government centralization against the background of the historical evidence presented earlier. Finally, Sect. 4 offers some conclusions.

2 Federalism in the United States and in Germany

2.1 The United States: a long departure from dual federalism

From an external point of view, and certainly from a continental European perspective, federalism in the United States is still perceived as an important real-world manifestation of the idea of interstate federalism. The idea of interstate federalism rests on having at least two distinct layers of government, each layer commanding full sovereignty over a clearly defined set of competencies. Such an arrangement is also often referred to as dual federalism (Corwin 1950).Footnote 1 Dual federalism is further characterized by a division of powers between the executive, legislative and the judicial branches, which is also strictly enforced on the state level. And finally, dual federalism ideally also provides a channel of influence for the states on the federal level through a parliamentary assembly of state delegates—a role that is, however, not played by the United States Senate any more at least since 1913 when election of senators by state legislatures was formally and definitely put to an end through the seventeenth constitutional amendment.

For the United States, the tenth amendment to the constitution appears, at first sight, to guarantee the persistence of dual federalism and a clear-cut allocation of sovereignties, with its provision that “powers not delegated to the United States by the Constitution, nor prohibited by it to the States, are reserved to the States respectively, or to the people”. The realm of federal government does appear to be restricted to fields of policy that are explicitly enumerated in the constitution, thus preserving state autonomy over the remaining issues. However, matters have changed since the ratification of the tenth amendment in the year of 1791. One dispute between Federalists and Anti-Federalists focused on the problem of implied powers, i.e. on the position that justifications for federal policy-making may also be merely implied in constitutional provisions. By 1819, the Supreme Court under Chief Justice John Marshall approved of the implied powers doctrine in affirming the right of Congress to establish a central bank for the United States (see Schwartz 2003).

The pre-civil war period also saw some other extensions of the leeway of federal government, for example an increasingly far-reaching interpretation of the commerce clause. On the other hand, there was serious debate regarding the doctrine of nullification, which, if it had gained acceptance, would have allowed each state to nullify a federal law that the state considered unconstitutional. The nullification debate is one example of strong tensions between state and federal governments; it is also evidence of the enormous confidence of state governments, and of the representatives of some states in the Senate, vis-à-vis the federal government at this stage of American history (Latner 1977).

The subsequent Civil War was a clash not only between the North and the South, but also between different theories of federal government. On the one hand, the southern states saw the union as a contractual agreement between sovereign states, with the federal government acting merely as a servant of the states. On the other hand, the northern states interpreted the United States constitution as a quasi-contract agreed upon by the people of the union themselves. The latter view endows central government with a legitimacy that is granted directly by the people, and this serves as one of the normative foundations of the claim that federal law has supremacy over state law. Following the turmoil of the Civil War, the view of a federal government with a legitimacy of its own had prevailed, if only through the military victory of the northern states and not through persuasion.Footnote 2

In the aftermath of the civil war, ratification of the fourteenth and fifteenth amendments in 1868 und 1870, respectively, further enlarged the competencies of the federal government to enforce individual rights throughout the Union, and the jurisdiction that has been developed subsequently has established “doctrines of federal intervention and even preemption that were later transferred to other spheres.” (Elazar 1981: 6). Nevertheless, the evolution of American federalism did not unambiguously follow a path towards greater centralization in the decades after the Civil War. An example is the approving mentioning the concept of a dual citizenship, which distinguished between state and federal citizenship, in the Supreme Court’s decision on the Slaughterhouse Cases in 1873.

The presidential campaign of 1912 can be argued to have marked an important step in the long process of transition from dual to cooperative federalism (see Elazar 1962 for a detailed description of the concept of cooperative federalism), with the two leading candidates both having, albeit for different reasons, proposed a stronger role for federal government. Accordingly, the following years saw the founding of the Federal Trade Commission and the ratification of the sixteenth amendment, which formally endows Congress with the power to levy a federal income tax. Having secured this source of revenue, the central government became increasingly interested in taking part in state and local spending decisions. It did so, however, in an entirely cooperative manner; federal government more or less continued to act as a servant to the states, funding projects (e.g. in infrastructure) that the states themselves demanded (Elazar 1981), and that served an excess demand of citizens for public goods that could not be supplied by the states alone (Kincaid 1990).

During the Great Depression and World War II, this type of cooperative federalism was increasingly perceived as an efficient means of governance. Subsequently in 1953, Congress appointed the predecessor of the Advisory Commission on Intergovernmental Relations, which was established in 1959 in order to develop and monitor instruments for cooperation between federal, state and local governments. Intergovernmental cooperation became increasingly sophisticated and differentiated in the following decades, and more importantly, the central government more and more acted as an agenda setter, rather than a servant of the states. Walker (2000) gives an overview over the development of federal grant programs, which did increase from 132 total programs in 1960 to more than 660 programs in 1998.Footnote 3

From the Nixon era onwards, politicians and commentators in the United States often used strong rhetoric in favor of decentralization and increased state autonomy. It is, however, important to note that the practical results deviated from this rhetoric. Elazar (1981) points out how under revenue sharing, Congress still reserved the right to control state spending of shared revenue in many respects, and how the federal level displayed a tendency to delegate particularly difficult and unpopular issues to the state level. Accordingly, this era is characterized as one of decentralization as opposed to state autonomy, i.e. an arrangement where the central level decides on the issues left to the states, which is obviously entirely different from an institutional framework in which the federal level is perceived as a servant to the states. Kincaid (1990), observing a rising number of federal preemptions overriding state and local legislation in the 1970s and 1980s, even diagnoses a switch from cooperative to coercive federalism. Thus, the pure numbers, which show a sharp increase of own revenue in the funding of state activities during the 1980s (see Walker 2000: 153), may be misleading.

2.2 Germany: negotiating power sharing through federalism

In 19th century political debate in Germany, the United States’ system of federalism has, to some degree, been discussed as a prototype (Fraenkel 1960: 100; Boldt 1991). However, the German nation state of 1871 was not a democratic project. While the Federalist Papers argued for federalism as a system of representation of diverse interests, and as a mechanism to control political power, the proposal of a decentralized system in Germany has in particular served as a tool of pragmatic power politics. Regional power holders could not simply be coerced into forming an unified nation state. A federal system allowed preservation of a domain of sub-central sovereignty, and thus the political cost of consenting to a nation state was reduced for regional power holders (Laufer and Münch 2006). The constitution of 1871 thus also followed the principle of explicitly enumerating the competencies of the federal level, and of leaving all other competencies to the sub-central level.

During this time period, enthusiasm for, finally, building the German nation state was the widespread sentiment within the population, and it was much more salient than concerns for checks and balances or civil liberties. Federalism in general and the preservation of sub-central political power in particular was thus essentially an elite problem in Germany, or as Elazar (1987: 147) states, reasoning on federalism remains a “most abstractly theoretical inquiry, fully in the German tradition”, or even “an abstract matter with which constitutional law theorists are concerned” (Reuter 1991: 11, own translation). Accordingly, regional power holders and not the general public have resented further centralization measures in the subsequent decades. An example is the struggle over the introduction of federal taxes during the Kaiserreich, which have been met with fierce opposition by the sub-central levels.Footnote 4

In the parliamentary assembly of 1948 and 1949, preferences for different types of federalism were rather heterogeneous along party lines, with the conservative parties by and large being in favor of competitive federalism. The eventual outcome must be understood as the result of legislative bargaining in a highly multi-dimensional policy space. On the important issue of competencies of taxation, the parliamentary assembly did even fail to reach a definitive bargaining equilibrium. Article 107, which assigns revenue from a number of important taxes to the different federal levels, was merely a placeholder in the original version of the constitutional law, the Grundgesetz. Since a compromise was not found during the deliberations in the constitutional assembly, nor between the assembly and the western allies, the only content of the original Article 107 was a deadline—December 31, 1952—for the eventual finding of a compromise between the state and the federal level (Blankart 1999).

German federalism is indeed an archetype of cooperative federalism,Footnote 5 and the need for cooperation between states and the federal government has been there since the framing of the Grundgesetz. The states have very limited exclusive competencies, which are mostly in the area of cultural and education policies, but also in law enforcement. The states are to a very large extent endowed with the task of enforcing federal law; the generic competencies of law enforcement at the central level are very limited. On the other hand, a majority of the Bundesrat, the upper chamber of parliament composed of Länder delegates, must approve of any federal legislation that in some way affects the Länder budgets. The formal constitutional framework itself therefore requires close vertical cooperation within the federal system.

On the revenue side of the budget, tax revenue of the states almost exclusively stems from taxes that are legislated for at the central level. Even for those taxes whose revenue is exclusively appropriated by the states, such as the tax on bequests or on the ownership of automobiles, legislation is made at the central level. The Länder did thus far have a competency to autonomously raise debts in order to finance their expenses. In 2009, however, a constitutional reform was enacted which in general prevents the Länder from having budget deficits, without compensating for them through extended sub-central control over any other fiscal instruments. Essentially, the most recent reform measures therefore constitute a further step towards government centralization.

In the preceding decades, there have been two milestones following 1949 where formal revisions of constitutional law have legitimized increased centralization in Germany. The Finanzverfassungsgesetz of 1955 eventually introduced a compromise on Articles 106 and 107, arranging the apportionment of tax revenue between the federal levels, and assigning the legislative competencies on taxation. The result was the cooperative system described above; the calls of the western allies for competitive federalism had apparently been forgotten by 1955. In addition, Article 107 also contained a mandate for the federal level to devise a horizontal fiscal equalization scheme between the states and, if necessary, to amend this scheme with vertical transfers from federal funds to particularly needy states.

In 1969, another step towards centralization was taken with the Finanzreformgesetz. Again, this is a change of constitutional law, and as such it relied on the consent of qualified majorities in both the federal parliament (Bundestag) and the Bundesrat. The reform has, in Articles 91a and 91b, formally defined policy areas (e.g. regional structural policy, agricultural policy) that are to be conjointly administered by the federal government and the states. The new Article 104a has legitimized vertical grants-in-aid to help state governments in financing “significant” investment projects. And Articles 106–108 have been changed such that the grand scheme of revenue sharing was formally established that, in essence, is still in effect today.

Germany, therefore, also experienced a long-term trend of centralization of competencies in the post-war period, and unlike the United States, it is also characterized by very comprehensive mechanisms of horizontal coordination between sub-central jurisdictions. A very modest deviation from the general trend towards centralization only occurred in 2006, when a reform was approved by both chambers of parliament, which formally (measured in affected articles) represents the most far-reaching single revision of constitutional law in the history of the country. Materially, the results have however been far less spectacular.Footnote 6 There are very few, and not very significant, policy areas where the Länder have regained formal autonomy. And even where this is the case, as in the example of legislation on smoking bans (see below), they tend not to make full use of their autonomy, but engage in horizontal policy-coordination instead.

3 Some political economics of government centralization

As the discussion in Sect. 2 has shown, we can observe a quite similar general trend towards centralization over much of the reviewed period in history, both in Germany and in the United States. Without any doubt, intergovernmental relations in the United States are still characterized by a degree of state and local autonomy that is much larger relative to that in Germany. This is hardly a surprise, given that the constitutional framework of the former country has, from the beginning, been guided by the concept of dual federalism, while the latter has been an archetype of cooperative federalism ever since the Grundgesetz was approved by the constitutional assembly. What is of interest to us, however, is the development of centralization that can be observed through time, and we have seen that in both countries, a loss of sub-central autonomy in policy-making has taken place.

From the perspective of political economics, which for good reason is convinced that institutions matter this is a striking development, because two completely different constitutional frameworks in the two countries have allowed for a substantial increase in government centralization. Prima facie, one could have expected the United States to be much less prone to centralization, given that its constitution provides relatively few incentives for intergovernmental cooperation. In the following subsections, we will attempt to identify the main political instruments of centralization that have been the most important drivers of the processes in both countries, and that might help us explain our observations made in the section above. From there, we will turn to a more general discussion of the theoretical implications of our case studies in the following section.

3.1 Constitutional change and constitutional reform

Much of the centralization process in the United States has been the result of a re-interpretation, rather than a re-phrasing, of the Constitution. As Robertson (1994) emphasizes, the Supreme Court has shown a tendency to grant central government ever more competencies, especially under the general welfare clause, the commerce clause and the necessary and proper clause at least throughout the first two thirds of the twentieth century. Only in some distinct periods can this be explained with political pressure. An important example is Roosevelt’s threat to fill the Supreme Court with judges who concur with his policy preferences. This may have led to Supreme Court decisions being more accommodating to central government demands for centralization during the 1930s. Such distinct political influences cannot, however, account for the persisting willingness of the Court to promote measures of centralization over many decades.

Kincaid (1990) argues that even in the 1980s, the Supreme Court engaged actively in re-interpreting the constitution such that the tenth amendment was now argued not to protect states’ rights to the previously believed extent. With a more conservative Supreme Court, this tendency has been reversed since then. For example, in its New York v. United States decision of 1992, the Court declared federal law unconstitutional that was to obligate the states to provide sites for the disposal of radioactive waste, and in the United States v. Lopez decision of 1995, the Court narrowed its earlier broad interpretation of the commerce clause.

The Supreme Court has a sufficient scope of competencies to strongly influence the balance of power between the federal and the sub-central levels, because the Constitution is deliberately vague in specifying what the exact vertical allocation of competencies should be (see e.g. Stotsky and Sunley 1997). As such, it follows a common-law tradition of refraining from detailed formal regulation of society. Restricting the constitutional law to very general provisions, subject to interpretation, opens the possibility of changing the understanding of the text at relatively low cost. It would, however, be premature to believe that the Supreme Court, with its changing ideological majorities, has been the only driving force of centralization and devolution, respectively.

Palmer and Laverty (1996) argue that conflicts between Congress and the Supreme Court need to be taken into account, because Congress can respond to judicial decisions through the legislative process. In the case of the Gun-Free School Zones Act, which was repealed by the Court in the United States v. Lopez decision, Congress did indeed adopt a modified proposal of the same provision in 1996, with the technical issues incriminated by the Supreme Court corrected and the material effects of the Act upheld. This example shows that the actual influence of the Supreme Court on accelerating or decelerating government centralization in the United States is more ambiguous than it appears prima facie. This is even more the case given the fact that, in the long run, the ideological positions of a majority of Supreme Court Justices are highly endogenous to the ideological positions of Presidents and majorities in Congress. A Supreme Court steering the federal system in one way or the other may therefore be merely a symptom for underlying policy preferences in the other branches of government, and ultimately of voters.

There is another, more subtle detail in the selection process of Supreme Court Justices that merits attention. In the United States, only the federal government is involved in the appointing of Supreme Court Justices, since members of the Senate are elected directly by voters and not delegates of the states. Even if the screening process is imperfect and multi-dimensional, it nevertheless gives federal policy-makers an opportunity to select Justices who are sympathetic towards their demand for increased government centralization. This is different in Germany, where half of the judges of the Verfassungsgericht are elected in the upper chamber of parliament, the assembly of state delegates. Interestingly, Thomas Jefferson would have preferred the German model of electing Supreme Court Justices.Footnote 7

There is another important institutional difference. The constitutional court plays a much smaller role in federal relations in GermanyFootnote 8 due to the fact that the vertical allocation of competencies is to a larger degree explicitly defined in the written constitution, the Grundgesetz, leaving relatively little room for interpretation. Consequently, the decisive measures of centralization such as the introduction of the Finanzreformgesetz have in Germany been associated with formal changes of written law. Given the institutional structure of federalism in Germany, such changes can only come into effect with a majority not only in the Bundestag (the elected parliament), but also in the Bundesrat (whose members are appointed by the Länder). Centralization in Germany can thus not be achieved unilaterally by the central level, or through legal re-interpretation of vaguely phrased constitutional law. The consent of a sufficiently large number of all Länder is necessary.

3.2 Partisanship and issue salience

Prima facie, there appears to be a strong relationship between partisan politics and the demand for centralization. In the earlier history of the United States and moving into the second half of the twentieth century, the existence of an unfortunate relationship between racist sentiments and strong support for state rights can hardly be denied. Decentralization was perceived as a prerequisite for upholding segregationist policies at least in some parts of the nation. Similarly, skepticism against large-scale redistributive policies is often associated with a preference for devolution of competencies to the state and local level. On the other hand, federal government in the United States has assumed the role of the progressive force in promoting civil rights, and left-wing political forces also tend to favor centralized redistribution, fearing a “race to the bottom”, in which competing states offering different welfare programs might become engaged (e.g. Weissert and Schram 1996).

Weaver (1996) reports that even governors have been split along partisan lines in the devolution debate of the 1990s. Republican governors tended to lobby for decentralization, for example of competencies in welfare policies. They were even willing to accept reduced amounts of federal grants if, in return, these grants were transformed into block grants, with little leverage for federal micro-management. Democrat governors by and large took a different stance and opposed a decentralization of competencies. The general ideological position of a politician does therefore appear to be a good predictor for their position on (de-)centralizing government, at least at sub-central level. But one can ask if this is also more generally the case.

Some observations regarding the use of mandates and preemptions in the United States, made, for example, by Posner (2007), point in a different direction. He shows that the phenomenon of coercive federalism, diagnosed earlier by Kincaid (1990), has prevailed under both parties controlling the White House and Congress. Posner observes that, using instruments that do not fall within the narrow scope of the Unfunded Mandates Act, the Bush era saw a significant centralization of competencies in the areas of education, welfare, homeland security, election administration and taxation. He does also observe that the Bush administration has used mandates in a number of areas to promote policies where it had clear-cut ideological preferences, such as tax cuts and welfare reform. The presumption that the Bush administration has no large stake in the issue of state rights can be corroborated by the fact that when Bush addressed the Republican national convention in 2000 and in 2004, no mention of state rights was made (Zimmerman 2007).

Indeed, it appears that centralization or decentralization as such have recently not been very salient policy issues when it comes to actual partisan policy-making in the United States, although the political rhetoric sometimes suggests the opposite. Rather, parties often use adjustments of the vertical allocation of competencies to promote policies that are of greater relevance judged by their ideological core values: The conservative party centralizes to cut taxes, or to promote homeland security, and it does so to ensure that its own values and ideological positions determine policy universally, in the entire country, with little or no leeway for single states to deviate from these positions. The same is true for the liberalFootnote 9 party—centralization is a means of pursuing nation-wide enforcement of policies according to one’s own preferences on issues that are perceived as more important than the centralization issue itself.

In fact, Milkis et al. (2007) argue that in contrast to Bush the elder and Reagan, the younger Bush was, when running for president and in his first term, the first Republican president for a long time who could reasonably anticipate governing with a Republican Congress and Senate. The expectation of being able to pursue an uncompromised Republican agenda at federal level might have tempted him into putting less weight on issues such as states’ rights from the beginning. Planning to reach, and expecting to get, full partisan control of federal government is associated with no incentive to campaign for states’ rights, regardless of one’s own ideological position.

In Germany, the process of government centralization has also been the result of bi-partisan efforts, involving the two largest parties, the center-right Christian Democrats and the left-wing Social Democrats. As we have seen above, crucial centralization measures such as the Finanzreformgesetz of 1969 have occured as formal constitutional changes, which require a two-thirds majority in both houses of parliament. In these cases, bi-partisan support for centralization has been an institutionalized prerequisite. But similar patterns of support can be found where informal cooperation has led to a de facto reduction of sub-central autonomy. There is no clear preference for decentralized policy-making associated with any party orientation. Rather, a tendency can be observed more recently that relatively affluent Länder are more reluctant to support further fiscal centralization, compared to those with relatively weak own sources of tax revenue. This feature has, however, until very recently not been closely correlated with the partisanship of the Länder governments. The reason for the relatively greater support of poor sub-central units for centralization is clear: Proposed measures of fiscal decentralization often involve a reduction of horizontal fiscal equalization schemes, and measures of centralization in turn are often accompanied by vertical compensation payments.Footnote 10

Similarly, in recent discussions on further measures of reforming German federalism (the Föderalismusreform II), the relatively affluent Länder have proposed the introduction of effective tax autonomy, but were unable to organize a majority for their proposal. At first sight, one could have argued that the division was along partisan lines with the conservative party tending to support tax autonomy and the Social Democrats opposing it. However, it is important to note that in recent years, the more affluent Länder tend to be governed by the conservative party, and that conservative-led governments in less affluent Länder have been sceptical towards proposals of tax autonomy.

On close inspection, the partisanship argument therefore turns out to have only very little explanatory power. Rather, issue salience matters. Despite frequent lip service that suggests otherwise, the vertical allocation of competencies in federalism appears to be used by political parties as a means, not an end in itself. Reluctance to decentralize (or willingness to centralize) corresponds to the perceived ability of a party to enforce preferred policies on other, more salient issues than federalism on a nationwide scale. A notable difference between the United States and Germany in this respect is that German federalism, as discussed above, offers formal mechanisms that facilitate collusion between state governments. Centralization is thus not only alluring for parties that expect to be pivotal in national politics. It is also alluring for state governments who, independent of party affiliation, can use collusive mechanisms to implement preferred policies.

3.3 Collusive behavior

We have seen above that, on the United States’ gubernatorial level and over recent decades, liberal politicians have tended to be somewhat more appreciative of centralizing measures than their conservative colleagues. However, history also shows that by and large, governors of all parties can be bribed into consenting to centralizing competencies. Posner (2007) gives education policy as an example. Here, governors have tended to lobby for a centralized definition of standards of education, hoping that such standards would be associated with increased federal grants. Resistance against centralizing education policy became reasonably strong only after it became clear that the No Child Left Behind program, established by the Bush administration, defined standards without providing for sufficient additional funding to implement them. At least in some instances, the motive of maximizing budgetary resources appears to be stronger at the sub-central level, than the motive to retain (or even expand) formal competencies.

In Germany, the Länder have at the same time been victims and offenders in the centralization process (Sturm 1997: 337), which is very much in line with the cartelization hypothesis of Blankart (2000). There does indeed appear to be a strong tendency at the sub-central level in Germany to avoid the pressure of fiscal competition through coordination and centralization. Thus, the state governments in the Bundesrat agreed to the more centralized structures that restricted their decentralized exercise of power. Unlike in the United States, where during the centralization process of the federal system Washington’s collective power—both executive and legislative—grew at the expense of the member states, all executive powers in the Federal Republic of Germany grew at the expense of all parliaments. There has been an increase in discretionary leeway for members of the executive branch at both the federal and the state level. The cost of this is a decrease in political control through the parliaments in general. But in federal terms, the cost has also been a disempowerment of the German states in their role as a counterweight to the federal government.

It is also important to note again that Germany’s constitutional law of 1949 already laid the foundations for Germany as a highly centralized country. There was not, as in the United States, a strong tradition of dual federalism that needed to be overcome in order to centralize competencies. On the contrary, the fact that the constitution allowed for a high degree of interdependence between the levels of government from the beginning facilitated the evolution of modes of cooperation between them. Among these, there are even formal instruments that, although they were not provided for in constitutional law, were developed over time. There are regular meetings of Länder ministers to coordinate their policies, and in some policy areas, such as education, their meetings are typically expected to result in a far-reaching horizontal harmonization of policies. It is important to note that this differs from modes of interstate cooperation in the United States, which typically does not aim at horizontal harmonization, but at representing common state interests vis-á-vis the federal government. In order to also facilitate a vertical coordination of policies, the Finanzplanungsrat was founded during Germany’s first post-war recession in 1968, and it still exists today. It assembles, among others, the federal and state finance ministers and serves to plan and to coordinate medium term budgetary policies. Furthermore, it also has a formal right to give recommendations on the federal budget, after the budget’s first draft is drawn up by the federal minister of finance.

There is another fundamental difference in the centralization processes of the two countries. In the United States, as seen above, the relative strengthening of the central government had a lot to do with abundant income tax revenue at central level, and a need to invest these funds in state and local level. This is different in Germany, where revenue from the quantitatively most important taxes (personal and corporate income tax, value added tax) has always been divided between the federal levels. Rather, it can be observed here that government centralization has triggered a general strengthening of the executive branch of state and local governments, relative to the influence of the legislative branch. Centralization of issues, in the cooperative manner described above, opens the opportunity for state government officials to negotiate these issues with their colleagues from other states, or from the federal level, in effect taking powers away from the state parliaments (Sturm 1997; Döring 2000).

It is also striking how the development of tax and expenditure shares of federal, state and local governments fail to tell the whole story of German federalism. Stegarescu (2005) presents a long time series on these shares from 1881 to 2001. These shares do, however, exclude the parafisci (i.e., public and semi-public social security systems which are administered by the central level), and thus tend to underestimate the true weight of the central level in particular after the Second World War, where social security systems have been expanded dramatically. As one might have expected, the two world wars are responsible for the most significant early increases in central spending (and reductions in sub-central spending). However, in the immediate post-war period, the data also show a sharp centralization of spending, and rather stable shares in total spending of the central government (or even a mild decline of central spending) thereafter. The picture on the revenue side is very similar. However, the discussion above has shown that the allocation of legislative powers has changed in a rather different fashion, with either outright centralization (on the revenue side) or instruments of horizontal and vertical cooperation (on the expenditure side) cutting into the autonomy of the Länder. And if the numbers on social security are included, then both federal shares are increasing also over the post-war period.

We can therefore observe two interrelated trends in fiscal federal relations in Germany: On the one hand, we see that the weight of the central government in terms of tax and expenditure shares is largely due to a growing weight of social security policies in overall government activity. On the other hand, we also see that the loss of sub-central decision-making autonomy, which is not necessarily reflected in shares of expenditures and revenues, is a result of entirely voluntary engagement of federal and sub-central decision makers in negotiations amongst themselves. At no point in post-war history have sub-central decision-makers actually been forced to surrender their autonomy.

This vertical collusion between elected officials is one facet of a more general problem of collusive activities; the other one is horizontal collusion. Clearly, competition between sub-central entities is a problem for many decision-makers in state and local politics. It restricts their scope for redistributive politics by providing those who are net tax payers with an exit option (Oates 1999), and it can also provide the state and local electorates with the information necessary to evaluate the relative ability of their elected officials through yardstick competition with neighboring states (Besley and Case 1995). Governors who are interested both in their own job security, and in securing real decision-making power for themselves, therefore face a dilemma: Collusion may reduce competitive pressures between states, and it may even increase tax revenue, but it is on the other hand associated with a substantial reduction of real, autonomous decision-making powers.

Weingast (1995) has argued in a seminal paper that interstate federalism, as a formal institutional framework, may be instrumental in preserving individual liberties. The exit option transforms the unitary government’s monopoly of coercive power into an, at least somewhat, competitive order, an argument that has already been made by Brennan and Buchanan (1980). Moreover, if resistance against a government trespassing on individual liberties is a public good, the coordination problem that exists in a more heterogeneous society between different groups may be mitigated in a federal system. The reason being that migration may involve a Tiebout-like sorting process, creating more homogeneous populations in sufficiently small sub-central jurisdictions. Given these effects of decentralized autonomy, one can arrive in a straightforward manner at the hypothesis that local governments competing for tax bases, like businesses competing for market demand, find it expedient to form cartels (Blankart 2000). Tax sharing schemes are an obvious choice to alleviate the pressures of fiscal competition. But even with tax prices negotiated in collusive agreements, sub-central units may still compete for tax bases by supplying heterogeneous public goods. In this sense, attempts to use horizontal instruments of policy coordination, such as those we have discussed above for Germany, are an additional piece of supporting evidence for the hypothesis that cartelization is an important motive for the surrender of sub-central autonomy.

As we have seen in Sect. 2, the state-level governments in the United States have also shown a certain willingness to exchange autonomy for fiscal security, i.e. to increasingly finance their spending through federal grants. Interestingly, however, horizontal policy coordination appears to be much less an issue in the United States. A plausible explanation for this is the sheer size of the federation. The much larger number of states in the United States is likely to be associated with higher transaction costs for negotiating coordinated policies. In addition, states in the US are, with regard to economic and socio-demographic characteristics, more heterogeneous compared to German Länder. The larger number and heterogeneity of sub-central jurisdictions does not only increase the costs of negotiation, but it also increases the likelihood of ex post defection. Apart from these problems, the United States also lacks a formal institutional framework, akin to that in Germany, which facilitates policy coordination. There is no equivalent to the Bundesrat with its accompanying bureaucracy, which provides a permanent forum to discuss and align the interests of the sub-central units.

The choice of formal institutions should ideally also be explained endogenously, and the simplest explanation is that it represents political preferences and beliefs of the relevant decision-makers at the time these formal institutions were set up. Constitutional framers who envision competitive federalism and aim at a clear vertical separation of competencies will not find it necessary to establish a Bundesrat in the first place. In this sense, differences in formal institutions may reflect the underlying differences in informal political institutions, which we will discuss in the next subsection.

In any event, the collusion argument proposed here also implies that the problem of stabilizing a federal order does not only consist of finding institutions designed to prevent central governments overawing of the sub-central units in order to appropriate rents (de Figueiredo and Weingast 2005). Rather, we have argued here that centralization is often not so much a matter of exploitation of a federation by the central government, but that centralization often is a cooperative, even a bottom-up process. Preserving sub-central autonomy is to a large degree an issue of avoiding voluntary collusion between governments, both horizontally and vertically.

3.4 Informal institutions

As political economists, we like to focus on the impact of formal institutions on political outcomes, but informal institutions—preferences, perceptions or conjectures that are widespread in a population, and thus have an impact on political decision-making—may also play a role. Hesse (1962) already coined the phrase referring to Germany as an “unitarian federal state”, alluding to a tendency to override a de jure federal framework by de facto harmonization of policies. Lehmbruch (1998) speaks of a cultural norm that drives Germany towards increased government centralization. In the collective political memory of Germany, state and local autonomy still appears to be associated with the problems that resulted from the co-existence of a large number of, in some cases very small, autonomous territories in the historical period before the founding of the nation state in 1871. The unitary nation state, in other words, is perceived as the institutional framework that resolves the problems that result from decentralized autonomy, be it transaction costs or political conflicts. From such a perspective, extensive decentral legislative competencies would also be associated with severe problems in normative reasoning: The question can be posed if a socially harmful (or even morally wrong) act should not be made illegal in the entire nation.

As recently as 2007, Germany saw a debate on legal bans on smoking in public places. Technically, the Länder have the competence to pass laws on this issue. There was, however, strong public opinion pressure to avoid a patchwork of differing legislations across the country. Consequently, ex ante coordination was pursued, and negotiations led to a relatively tight framework, such that eventually very similar rules on smoking bans have been enforced by all sub-central Länder.

Some clues for the relevance of informal institutions can also be found in the United States. Weingast (1995) points out that well into the nineteenth century there was a consensus that the role of federal government should be limited. He explains the emergence of this consensus with the fundamentally different political interests of northern and southern states, and the fear of both factions that the respective opposing faction might gain control over a federal government that is strong enough to enforce its preferred policy on the entire nation. The way this was to be achieved was, obviously, to implement institutional barriers to federal authority. This consensus, however, was not stable. Weingast argues that the founding and the success of the Republican party reflect its demise, due to the Republican demand to end slavery in the entire nation. One can, however, see this as only a first step in a long process which led Americans to become more willing to endow central government with additional competencies.

As already noted above, the apparent success of central measures, and of cooperation between the central and the sub-central levels, in ending the Great Depression and in steering the war economy was instrumental in increasing widespread acceptance for the strong federal involvement in state and local politics, which was to be established thereafter. Given that the unfortunate overlap between groups interested in states’ rights, and groups interested in preserving segregationist politics, existed well into the second half of the twentieth century, the proponents of centralization were also able to associate their positions on federalism with an aura of social progressivity, and of moral soundness. Subsequently, the federal level began using instruments such as crosscutting requirements and crossover sanctions, bundling otherwise completely unrelated issues. For example, the allotment of grants-in-aid for state infrastructure could be made to depend on the states meeting conditions on environmental or civil rights issues (see ACIR 1993).

A third federal country, which has experienced far less centralization than the two countries discussed in detail above, is Switzerland. Blankart (2000) argues that the citizens’ veto power through the instrument of a referendum, which is a Swiss peculiarity, is responsible for this difference in development. Indeed, Feld et al. (2008) present econometric evidence for Swiss cantons that supports this conjecture: Those cantons with more direct-democratic participation in the budgetary process have experienced less centralization from the municipal to the cantonal level. Schnellenbach et al. (2010) argue that citizens use the referendum instrument to veto centralization attempts in order to curb representatives’ rent extraction. However, this is most likely only true if the voters themselves do not have a strong preference for homogeneous policies across the entire federation—which is exactly what we have found for Germany in the above sections.

Eventually, the extent to which centralization takes place does therefore hinge upon the distribution of political preferences in the population. If there is a widespread aversion against fiscal competition, then even the introduction of direct-democratic instruments following the Swiss example will not significantly stabilize fiscal decentralization. But how can such differences in cultural norms regarding trust in centralized government or the need for decentralization be explained? A detailed exploration of this issue lies beyond the scope of this paper, but the threat of military conflicts might be a plausible starting point. In the United States, the lack of an external threat due to its geographical location has long led to sense of security, and hence to a low demand for security supplied by the central government. In contrast to the United States, German territories have for centuries been involved in armed conflicts with each other and with external foes, and the ideas of pacification and of a strong central government have long been closely related. In Switzerland, also located in the middle of Europe but protected by natural barriers and too small to be worth attacking, on the other hand, external threats were not as big a concern as in Germany, and armed conflicts between cantons have been very few and mostly of low intensity. As a result, the Swiss (like the United States) do not feel so threatened in a historical perspective.

It is important to note that, once such an informal institution such as a strong, widespread trust in central government is established, it is likely to remain stable even if the actual reason for its origination has long disappeared. This is obviously the case with regard to external threats: Germany is surrounded by friendly neighbor countries, and the United States are hardly subjected to military threats. But informal institutions or cultural norms tend to be self-stabilizing, e.g. through communication between individuals (see North 2005). Even norms that are not in any sense useful with regard to a practical purpose any more can be perpetuated in an institutional equilibrium.

4 Conclusions and outlook

Looking at the development of the vertical allocation of competencies over time in both the United States and in Germany, a striking pattern of gradual centralization emerges as a somewhat natural, evolutionary process. Decentralization measures do, in contrast, appear as occasional deviations from this long-term evolutionary trend. For example, with regard to the United States, Kingdon (1995) has noted that the devolution movement of the 1990s was the result of a somewhat unique conjunction of circumstances: heavy pressure to reduce a federal deficit that was threatening to become unsustainable, a widespread sentiment of mistrust in federal government, and a conservative congressional majority in combination with popular and influential governors demanding political decentralization. Milkis et al. (2007) also argue for the importance of perceived limits of the political capacity of federal government as an important rationale for devolution.

In this sense, it does indeed appear to be the case, that in federal systems, a persistent tendency towards centralization is at work. State and local governments do not enjoy the pressures of fiscal competition. As a rule, they are willing to give up autonomy if they can gain security, for example through a far-reaching fiscal equalization scheme, through other grants-in-aid, or through tax sharing. We have also seen that, contingent on the institutional framework of the economy under observation, the pathways towards centralization differ: In the United States, the issue posed by de Figueiredo and Weingast (2005) of sub-central governments being overawed by central authority is indeed a problem, given the use of coercive instruments such as federal preemptions. In Germany, centralization resembles more of a bottom-up process.

One might conclude that formal institutions by themselves are neutral with regard to the broad trend of government centralization. The process of centralization stops when it reaches exogenous limits, for example in the form of budgetary resources at central level. Federalism is likely to be preserved in democracies if informal political institutions support political decentralization, for which Switzerland is an example, and in this case, the extent of voters’ participation granted by formal institutions also matters. But beyond this, it remains difficult to see how a federal constitution can generally be made self-enforcing in the sense that voluntary, bottom-up centralization and coercive centralization by the central level are both safely avoided.

A point that merits further research effort is the fact that centralization appears to depend on the existence of an overarching, central level of government. This is in particular the case for Germany: Prior to the end of the Holy Roman Empire in 1803, there was hardly any activity of small, autonomous jurisdictions merging into larger units, or even cooperating extensively in single policy arenas. If centralization could be explained on grounds of economic efficiency such as the exploitation of economies of scale, one would, however, expect exactly such a bottom-up process: Cooperation between small, autonomous jurisdictions that may or may not eventually lead to the establishment of formal central-level institutions. Our evidence points into a different direction: The process of centralization commences once a constitution providing for strong central-level institutions is there.