"Fiscal Federations: Lessons for the Design of a European Fiscal Union" By Carlo Cottarelli

December 13, 2012

Presentation for ECB-IMF Conference in Frankfurt, December 13, 2012
By Carlo Cottarelli

The purpose of this presentation today is to discuss how issues that are critical for the design of a fiscal union in Europe, particularly over the longer run, are addressed in existing federal states.

In doing this, the first question we should ask is: how relevant is the experience of federal states for a European fiscal union? I think it is quite relevant, as some of the economic factors that should shape the design of a European fiscal union apply to federal states as well. At the same time, it is important to acknowledge that there are differences. For example, labor mobility is not as high as in many existing federal states. And, of course, the euro area is not a political union, so some of the tools available in political union would not be available in a fiscal union in Europe. So, any findings regarding federal states will have to be interpreted with a grain of salt. Moreover, the presentation describes prevailing practices, but does not assess performance under those practices, so its findings should not be interpreted as normative statements.

This presentation is based on a study that the Fiscal Affairs Department has undertaken on how federations work. The study covers thirteen federations (including some relatively decentralized ones like the United States, Switzerland and Canada, whose experience could perhaps be more relevant for Europe), essentially most of the large ones monitored by the Forum of Federations, the global network of federations established in Canada.

This study, which should be completed in early 2013, is focusing on seven dimensions that are relevant for the design of a European fiscal union in the long run.

  • The role of centralized fiscal policies, and, hence, the role and size of a federal budget
  • The role of risk sharing through transfers from the center
  • The constraints existing on macrofiscal policies at the subnational level
  • The harmonization of public financial management processes
  • The harmonization of noncentralized fiscal policies
  • The modalities of member state borrowing in a federation
  • The role of crisis management arrangements

These dimensions will also be the focus of this presentation. Before proceeding, one terminological clarification: unless otherwise indicated, the term “local” or “state” will be used to refer to the highest level of subnational government, regardless of the actual denomination that this level takes in federations (although, where appropriate, data referring to all subnational government entities will also be reported).


So let’s start with the first dimension: the role of centralized fiscal policies, and, consequently, the role and size of the federal budget. Public finance has long discussed the benefits of decentralized spending and revenue decisions: allocating fiscal decisions to a level of government that is closer to the taxpayers allows a closer tailoring of fiscal policy to local preferences. However, in spite of the potential gains from decentralization, there is a minimum size of the federal government necessary for the effective delivery of its core functions. Even in the most decentralized federations in our sample (Canada, United States, and Switzerland) the federal government revenue and own expenditure represent about half of general government final spending.

What is usually centralized? Three key functions stand out:

  • The first one is the most traditional one: fiscal federations invariably conferred on the federal government defense and foreign relations, and other traditional public goods such as the countrywide justice and law enforcement, communications (e.g., postal service) and key transportation systems.
  • Second, since the 1880s introduction of a national social insurance system in Germany, the center has also increasingly assumed social insurance and redistributive functions.
  • Third, and more recently, the central government is responsible for carrying out macroeconomic countercyclical policies, providing risk-sharing against economic shocks affecting all or some regions, and responding to other aggregate shocks (such as natural disasters).

Stabilization and social protection were a crucial factor in the growth of central governments through the 20th century. For example, in 1929 the U.S. federal government budget was about 2½ percent of GDP, about one-third of the states’ budgets, while in 1939 after the New Deal the respective figures were 10 percent for the federal and 9 percent of GDP for the state budgets.

How about revenues? Fulfilling the tasks assigned to the center requires resources. In principle, these could be transferred from member states. In practice, federal governments always finance themselves through their own taxes (except sometimes early in the life of federations when transfers are more common—for example in Germany during the 1870s and 1880s, or the U.S. before the reforms introduced by Alexander Hamilton). There are some regularities in the revenues that are centralized:

  • All federations in the sample have chosen to have a central corporate income tax. This matches the macroeconomic functions of the federal government since this tax is highly cyclical: centralizing highly cyclical revenues implies a centralization of the automatic stabilizers (which may not be allowed to operate as freely at the state level if states face more binding financing constraints than the federal government). Moreover, a central corporate income tax contributes to market integration and provides a level field for competition. Finally, the corporate income tax has a very elastic base across states, and may cause economic inefficiency if location decisions are made for tax purposes. A central corporate income tax minimizes these distortions.
  • All fiscal federations have also a central personal income tax but state personal income taxes are also widespread, reflecting dissimilar preferences or revenue needs across regions. A central personal income tax is an important instrument in those federations for the countercyclical, and inter-regional equalization roles of the central government.
  • In principle, consumption taxation meets most criteria for a good local tax (such as local governments should tax the less mobile basis like consumption or real estate). However, VAT—the most common and efficient form of consumption taxation—is not easily decentralized (as it is challenging to preserve the integrity of the VAT credit-debit chain and associated administrative advantages, without imposing internal border controls of a kind federations inherently seek to avoid). Indeed, decentralized consumption taxes are typically retail sale taxes (for example, almost all U.S. states and some Canadian provinces), with some efficiency loss.

The different degree of centralization of spending and revenue power leads to the emergence of the so-called vertical imbalances—the need for transfers from the center to members states. The extent of these transfers, however, widely differs across federations. Vertical transfers are as high as 60 percent of state revenues in Belgium, but play a relatively small role in the most decentralized countries (e.g., in Canada, United States, and Switzerland, where they account on average for 16 percent of state revenues).

The size of the EU budget remains minimal—accounting for 2 percent of general government spending in the EU. The EU budget was clearly not designed with the objectives I just described, not only those that would arise only in case of political union (like common defense and external policy) but also more economic roles, namely macroeconomic stabilization, risk sharing and a common provision of public services.


This brings me to the second item in my list: the role of transfers from the center in offsetting macroeconomic shocks, and the related issue of risk-sharing. Of course, transfers in federation serve other purposes too, not just macroeconomic purposes, but here the focus is on the latter because that has been the focus of policy debate in the euro area. Let me first clarify two points:

  • first, in addressing this issue, we need to look not only at what I called vertical transfers (unrequited payments made by the center to local budgets) but at any kind of financial transaction made between the central government and the entities resident in a certain member state (including, in addition to vertical transfers, transfers from the central government to private sector residents in the subnational jurisdiction, less central taxes paid by residents in the subnational jurisdiction), what we can call net transfers. This is relevant because, for example, a reduction of payments from the states to the center can also offset macroeconomic shocks.
  • Second, the focus should be on how these net transfers respond to output shocks (both common and idiosyncratic) and not so much on their level (which could be high because of large vertical imbalances unresponsive to cyclical fluctuations).

Econometric evidence based on the experience of United States, Canada and Australia leads to the following conclusions:

  • Changes in vertical transfers in response to cyclical shocks are not large nor common, although there are examples (such as the transfers from the U.S. federal budget to the states as part of the 2009-10 fiscal stimulus package).
  • Net transfers are relatively sensitive to shocks (particularly common shocks). They primarily reflect the operation of central automatic stabilizers, as a result of the centralization of output-sensitive spending (in particular unemployment benefits) and, especially, revenue items (such as corporate and personal income taxes, but also consumption taxes). These mechanisms offset around 15-20 percent of common shocks for the three federations considered in our study (although the econometric estimates would suggest a lower offset for idiosyncratic shocks). This result is intuitive because federal revenues represent about 18 percent of GDP in the average of the above federations, and this is a good proxy for the size of the automatic stabilizers operating on the revenue side (which are the bulk of the automatic stabilizers).

What are the implications of these findings for the role a central budget can play for risk sharing of macroeconomic shocks? It could be argued that the fact that most of the action related to net transfers reflects the centralization of automatic stabilizers implies that members of federations are not really better off in handling macroeconomic shocks: they would benefit from the central automatic stabilizers, but the state automatic stabilizers are correspondingly lower. This view, however, does not take into account that local budget may not be in a position to respond flexibly to shocks (i.e., to let the automatic stabilizers operate, particularly in the presence of large shocks) because of nominal fiscal rules or financing constraints.1

Altogether, movements in net transfers can help in offsetting shocks and involve a degree of risk sharing. While their magnitude should not be exaggerated, these effects are not trivial in federations but they only play a very marginal role in the euro area at present. Yet, in principle, the need for risk sharing is higher in the euro area because of lower labor mobility and lower synchronization of economic cycles. Of course, there are several ways of setting risk sharing mechanisms, one of which is through a central budget. Other options, less ambitious, exist such as a stabilization or “rainy day” fund—set up by pooling together resources through transfers from member states which are then distributed when states are hit by a shock. However, the centralization of policies would have additional advantages in the longer run, as it would foster macroeconomic convergence. 2


The argument that risk sharing mechanism should be stronger in the euro area than in existing federations is also reinforced by the fact that the constraint set in the euro area on member states’ macrofiscal policies are, in many respects, quite tight, which brings me to the third dimension of our analysis—the constraints on the macrofiscal policies of subnational governments.

By macrofiscal policies I mean the setting of the budget balance, the aggregate spending level and the amount of public debt. Four approaches are essentially followed in federations to constrain macrofiscal policies:

  • First, you have direct (administrative) controls by the central government. For instance, central government may set and revise every year borrowing constraints.
  • Second, fiscal rules. Although rules impose stringent constraints on fiscal discretion, they are less binding than direct controls, because rules preclude the central government from micromanaging subnational fiscal policy, and because subnational governments have generally some margins to comply with fiscal rules. Rules themselves can be either imposed by the center or self-imposed, although the distinction is perhaps less relevant from a practical than from a legal standpoint.
  • Third, you have cooperative approaches: unlike fiscal rules, they allow subnational governments to negotiate their fiscal targets on a regular basis.
  • Finally, federations can just rely on market discipline.

Our analysis of federations reached four conclusions.

First, fiscal rules—particularly in the form of balanced budget rules—are by far the most common form of constraint. In about half of the cases these rules are self-imposed, as for example, in Australia, Canada, Switzerland and the United States. This said, one could argue that, in the absence of self-imposed rules, rules would have been imposed by the center.

Second, cooperative arrangements are rare and are mainly found in European economies. For example, in Austria, annual fiscal targets are negotiated by federal, regional, and local governments via the Austrian Stability Program. A similar negotiation process occurs in Belgium through cooperation between the federal and regional levels and the High Finance Council.

Third, direct controls from the central government are also very rare in federations (contrary to what happens in unitary states), except in case of breach of fiscal targets or financial support in crisis situations (see below).

Fourth, purely market-based discipline remains atypical.

How does this compare with the current situation in Europe? There are more similarities here than in other dimensions:

  • The EU relies on fiscal rules, rather than on direct controls and cooperative arrangements: there are at present four main supranational rules—the 3 percent deficit rule, the 60 percent debt rule, an expenditure benchmark, and medium-term budgetary objectives defined in structural terms. The fiscal compact also requires countries to enshrine a structural balance rule in national legislation.
  • There is also a yearly cycle of economic policy coordination (the European Semester). But this exercise cannot be described as a full-fledged cooperative approach, as budgets are examined rather than negotiated between Brussels and Member States.
  • The EU framework does not resort to direct controls from the center, although there have been proposals in that direction for countries in breach of the rules.

All this is similar to what we found in federations. There are three important differences. First, the EU rules apply to the general government, with countries being responsible to distribute the target internally among government units. By contrast, in federations, central constraints generally apply separately to different government levels, and states are not responsible for the achievement of lower-level targets. Second, most federations tend to impose a smaller set of constraints, except Brazil and Spain which adopted extensive fiscal rule frameworks in the aftermath of severe fiscal crises. Third, sanctions are relatively mild in the euro area: they usually consist in opportunity costs from financial deposits. The conditions to convert these deposits into outright fines are strict, and have, so far, never been applied. In addition, the EU framework does not provide for administrative sanctions, which exist in several federations (for example in the province of British Columbia in Canada, ministerial salaries can be withheld; and in Brazil, officials who violate the rules may be subject to criminal penalties and fines).


The monitoring of fiscal rules is greatly facilitated by the adoption of common public financial management tools, the fourth dimension of our analysis. We have looked at the extent to which federations have standard or harmonized guidelines for member states on budget formulation, reporting and audit. We found that there is a higher degree of harmonization regarding the standardization of chart of accounts, the definition of budget and accounting classification, and the provision of ex post information, for example, end of the year financial statements. But there are some gaps. Annual financial statements are often available only after a significant time lag – anywhere up to a year. For example, Brazil takes 8 months to produce a general government financial report. In many other economies, including advanced economies, it takes a long time to consolidate the government finance statistics – Germany for example takes nearly one year to release detailed general government financial statistics.

Harmonization of procedures related to ex ante fiscal projections and budget preparation is even more limited. This is mainly because measures affecting these ex ante step are considered more intrusive. However, some federations do have more provisions for coordination at the budget preparation level, particularly in emerging markets (such as South Africa, Brazil and India).

At the euro area level, the new Budget Frameworks Directive rightly emphasizes more regular and comprehensive publication of member state data. However, while the Directive calls for independent audit of the quality of government accounts, it does not prescribe the accounting standard to be used by the auditor in determining whether such accounts represent a true and fair view of the government’s financial position.


The fifth dimension of our study covers the harmonization of fiscal policies that have not been centralized.

On the spending side, our analysis shows that, where spending is decentralized for health care, education, social protection, and infrastructure, governments have effectively implemented harmonization policies—ensuring minimum provision levels while allowing local governments wide discretion to allocate spending. For example, for education, while provision and funding is decentralized in many countries, policies are partially harmonized through national education standards. For infrastructure, central transfers reduce regional disparities or encourage local contribution to national priority projects—while retaining centrally projects with important externalities or of national scope.

On the revenue side, measures of tax harmonization are common in federations, because of the spillover arising from decentralized tax policy decisions and of economies of scale through information sharing and other collaboration among tax administrations. A common form of harmonization is when the tax base is harmonized and member states can set the tax rate either freely or within a centrally defined range. In some cases, the tax base is largely unified but member states can provide a small number of simple tax allowances or tax credits.

Here, again the EU differs significantly from federal states. There has been little harmonization of spending policies. And on the revenue side, degrees of harmonization and mandatory minimum taxation apply only to VAT and some excise taxes.


Let us now move below the line and consider the modalities of member state borrowing in federations. This issue has attracted enormous attention in the euro area during the last two years, with a heated debate about euro bonds and bills as a way of financing member states. Let’s first describe the arrangements prevailing in federations in ordinary times (the sixth dimension of our study), and then look at emergency financing during crisis situations, the seventh and last dimension.

The bulk of borrowing of member states in federations is from markets through security issues and loans. Central governments do not usually lend to sub-national governments directly, although there are exceptions, primarily related to special programs and development purposes, particularly in emerging economies. For example, central governments in Australia, Canada, Germany and the U.S. provide small specific-purpose loans (for housing and infrastructure or, in the U.S., unemployment compensation) to state governments. In India, over the last decade the central government has been phasing out its large direct loans to states. Austria is a notable exception. The Austrian federal government’s debt management agency is tasked to raise debt and on-lend to states through direct loans. While these loans are not large (about 2½ percent of GDP in 2011), they cover a significant part of states’ financing needs (in 2011 the federal government held about 32 percent of states’ debt).

In some federations, central governments also indirectly finance sub-national governments by channeling loans through centrally owned financial entities, particularly in emerging economies, where federally owned development banks are present in South Africa, Mexico, and Brazil. In emerging economies, central governments also provide guarantees to subnational governments. This is not common in federations in advanced economies, although in Germany, the federal government provides guarantees on a very limited scale to Laender for specific projects.

Altogether, the kind of common borrowing that has been proposed for the euro area is unusual in federations. This does not mean that Eurobond proposals have no merit. Historically, at the early stage of a fiscal union, the center has sometimes taken over sub-entities’ debt on a one-off basis. In the United States, for example, the federal government took over states’ debts in 1790. But operations of this kind should, if at all, only be considered as a bridge toward more long-term instruments, namely common debt backed by common revenues, as in the case of federal debt in existing federations.


Finally, let’s consider the role of crisis management arrangements in federations, or, in other words, what happens when a member state is in financial difficulties. When facing a financing crisis, subnational governments are rarely allowed to default on their private creditors. In fact, when debt restructurings have occurred, they mainly involved smaller political entities or cases where the federal government debt itself was restructured. Pre-set crisis resolution frameworks such as bankruptcy procedures are indeed absent for sub national governments, with only a few exceptions. Instead, the federal government usually provides financial support in case of crisis, although this typically occurs through a range of ad hoc mechanisms involving guarantees, loans and transfers, not through standing facilities:

  • Examples of guarantees include the introduction by the Australian government in early 2009 of a temporary guarantee scheme for state borrowing as a response to rising financing costs for subnational governments.
  • Examples of loans include the provision by the U.S. federal government of loans to the District of Columbia in 1996. Another example is New York City in 1975, which was supported by a federal credit line. There were also several examples in Brazil, when the debt of states was taken over by the central government in 1989, 1993 and 1997. A similar case is Argentina in the early 2000s.
  • Argentina, also in the early 2000s, and Mexico in the mid-1990s provide examples of ad hoc transfers. Among advanced economies, in Germany, following a decision of the constitutional court, ad hoc transfers were provided in the 1990s, and, again, in 2011, to Bremen and Saarland.

In the case of loans and transfers conditionality has been typically imposed on deficits, and in a few cases on structural reforms, including privatization.


As indicated at the beginning, the goal of this presentation was to describe common practices in federations and not to draw implications for the euro area. But it is clear that at present the mechanisms of fiscal coordination in place in Europe are much weaker than those prevailing in federations, perhaps with the exception of the constraints on the macrofiscal policies of member states (and, even in this case, with weaker sanctions in case of noncompliance). Thus, the road towards a full fiscal union is a long one. It could perhaps be walked fast but probably it can be fully walked only in the presence of steps towards political union. But, there are some priority steps that could be taken in the short run that, while consistent with longer term fiscal union objectives, would also help address the crisis that the euro area is currently facing, as discussed in a separate contribution to this conference provided by Ms. Allard.

1 Note also that, even if the econometric evidence shows that net transfers do not respond much to idiosyncratic shocks, central net transfers in response of common shocks still imply a degree of risk sharing because not all states may be able to access financial markets at sustainable rates when hit by a common shock.

2 See Carlo Cottarelli, "European Fiscal Union: A Vision for the Long Run


Public Affairs    Media Relations
E-mail: publicaffairs@imf.org E-mail: media@imf.org
Fax: 202-623-6220 Phone: 202-623-7100