Fiscal Federalism and European Health System Decentralization: A Perspective

1. Introduction

Health care systems generally refer to packages of essential services, delivered either by the state or the market. In Europe, health insurance schemes are financed by general taxes or hypothecated social insurance contributions, and private insurance plays either a complementary or supplementary role. Such public insurance schemes have developed for reasons related to both efficiency and equity, including: a) to take advantage of pooling and single payer welfare gains; b) to allow coverage for unexpected risks, and especially c) to reduce problems of information asymmetry, which make ‘accountable governance’ a challenging endeavour. It is not surprising then, that in the majority of European countries, health care is a publicly financed package, even though they increasingly exhibiting a process of political and fiscal decentralisation (e.g., Italy, Spain, France, UK and Belgium). That said, there are too some experiences of recentralisation in smaller states such as Norway, or in recently liberalised ones such as Poland ( see Costa-Font and Greer, 2012 for a review).

Given the heterogeneity in risk exposure, especially in large European states, health care, (followed by social care and education), is the most common responsibility that has been decentralised to subnational governments,. Hence, it seems reasonable that fiscal federalism scholars choose to employ health care as an area of government activity to study the impact of decentralisation in public services.

Decentralisation has a variety of meanings, depending upon the context. In this paper we essentially refer to it as the ‘devolution of fiscal competences to subnational governments’, and hence it should be regarded as a synonym for “regional autonomy” in designing health care programs.

Reasons for health system decentralisation are country specific but generally can be divided in two groups, namely:

(i) direct improvements of static (allocative) and dynamic (innovation) “efficiency” effects of the health system and

(ii) expanding participation though the existence of different political cycles and the emergence of additional sources of political competition, which indirectly result in efficiency improvements if rent seeking by interested stakeholders declines (e.g., pharmaceutical industry, doctors and pharmacists among others) due to larger scrutiny, and if political competition reduces the chances of central and subcentral government capture.

Both effects are possible because 75% of total health care expenditures are publicly financed (OECD, 2009). Health system decentralisation thus encompasses moulding health systems to allocate health care responsibilities in such a way that health policies meet the demands of the regionally heterogeneous median voter. This is made possible because of thecompetitive nature of intergovernmental decisions, as explained below.The classical decentralisation theorem (Oates, 1972) based on the preference–policy match[1], is rooted in an implicit assumption of welfare improving mobility (Tiebout, 1956). However, this assumption might not hold in a relatively immobile European population and hence, mobility cannot be the maindriving mechanism. Instead, the most powerful and crucial incentives lie in the degree of political and fiscal accountability, and - more specifically - the extent to which the design resembles a yardstick competition mechanism[2].

Furthermore, unlike other areas of public sector responsibility, health care outcomes are subject to more information asymmetries from an individual’s standpoint. Health care services, unlike educational services, are intermediated by agents (e.g., doctors), and therefore objective quality dimensions of health services are filtered by such agents. Additionally, such agents might have interests besides that of patient wellbeing. This implies that patients only judge health systems performance according to observable criteria, including the length of waiting lists and waiting times, bureaucracy and more other process related outcomes that may not necessarily be related to the adequacy of treatment or other dimensions of quality of care. Finally, another important feature of health care as services is that although health care is regarded as a “merit good”, it can also be provided by the market. Hence, if the state fails to satisfy the heterogeneous demands of all social groups, a market for private health insurance, outpatient and inpatient care can develop.

The special nature of health care and information asymmetries make it a distinct area of policy responsibility. As part of this,specific constitutional and fiscal design becomes crucial. For example, for fiscal decentralisation to have a full impact, fiscal responsibilities in the form of taxes, and to a lesser extent, subsidies, should be allocated to subcentral governments alongside other policy responsibilities.[3] However, the degree of fiscal decentralisation is not an obvious feature to identify and measure.[4]. Furthermore, some areas of health provision are global public goods, such as immunisation, and policies in one region will have spillover effects to other regions. Therefore, the decentralising the regulation of such conditions are unlikely to result in an improvement on overall welfare. For instance, Baicker et al (2010) document that most of the devolution of public policy responsibilities in the US takes place in the area of health care, and this implies a complex operationalization and design of federal grants to encourage state actions towards efficiency and innovation. Finally, health packages are one of the most costly welfare services governments provide, so it seems reasonable to expect that devolving health care responsibilities under stringent budget constraints is likely to improve government efficiency.

This paper aims to review the contending issues that arise from the decentralisation of health system responsibilities to subcentral governments from a fiscal federalism standpoint. We examine how a more balanced spread of both political and financial authority to different levels of government (thus reshaping the vertical structure of the health system) affects processes and health care outcomes. We argue that unlike the old mobility argument, health care decentralisation is likely to influence tighter political agency on the performance and dynamics of the health system. To do so, we rely mainly upon the theory of fiscal federalism and recent developments in political economy to explore the economic effects of devolution. Unlike the previous literature on welfare state federalism we incorporate the influence of policy innovation and different forms of inter-jurisdictional competition in exerting an influence on the development of regional inequalities.

The next section focuses on the reasons for fiscal decentralisation in health care. Section 3 will discuss the different sources of vertical imbalances in the allocation of political and fiscal responsibilities, whilst section 4 will provide a succinct analysis of resource allocation mechanisms. Section 5 will then discuss challenges and section 6 concludes.

2. What does Health System Decentralisation stand for?

2.1 Defining Fiscal Decentralisation

Decentralisation, as defined here, proxies for autonomy of subcentral governments (Oates, 1985), or the strength of subnational power in the form of employment control as well as devolved regulatory and taxation powers. More recently, economists have begun to see decentralisation as a way of tightening the political agency between constituents and incumbents to enhance the mechanisms of the so-called ‘political agency’ (Besley, 2006). It is different from formal (or legal) federalism, in that the former is a constitutional decision whilst the latter is the result of the political bargaining that takes place both before and after the constitution of a country is determined. But it can produce comparable effects insofar as it gives rise to inter-jurisdictional interactions, even though it is only in federal states that state owned powers operate in a similar fashion to the property rights market , and hence central governments cannot legally invade decentralised responsibilities. As we will discuss later, in unitary states the central state exerts an active role in invading state powers and in issuing framework laws that can act as an indirect way of limiting health care responsibilities, or the degree of autonomy that subcentral governments have.

Both expenditure decentralisation and tax revenue decentralisation are imperfect measures of autonomy because not all funds are expended by subcentral governments, and even if they are if financed through transfers or block grants as in the UK and Spain, decision-making on such resources has taken place at a central level. In other words, expenditure mandates on behalf of the central government should not qualify as decentralised expenditures.[5] Furthermore, the extent of legislative activities to restrict subnational government and the existence of policy coordination (e.g., as it is the case in Italy with Framework laws, or in Spain with Inter-territorial Councils of regional health ministers) restricts the autonomy of regional states. Fiscal autonomy does not necessarily contradict the existence of equalisation and redistribution (Oates, 1999), as most decentralised states explicitly or implicitly take these into account through unconditional grants.

Together with fiscal responsibilities, limits to pre-empt policy action include central government regulation. Indeed, Piperno (2000) reports that in Italy, national parliaments still invade decentralised responsibilities, and the central government frequently vetoes regional laws which lead to conflicts of competence. Most often, conflicts are solved to the favour of central government, which directly or indirectly (through parliaments) elects members of the Constitutional Court[6].

Today, there is consensus among scholars that the key to the success of decentralisation boils down to its institutional design. More specifically, the extent to which decentralisation manages to align political credit and fiscal blame for each policy within the health system. If the central government does not decentralise the “blame” of public policy action (taxation) and only decentralises mechanisms of credit claiming (expenditure), it is likely that decentralisation will bring an expansion of government expenditure with limited effects on efficiency (Costa-Font, 2010). The latter is commonly known in the literature as soft budget constraints. Similarly, insufficient subnational own resources (vertical fiscal imbalances) as a cost containment strategy (Lopez-Casanovas, et al, 2005) can hamper the degree of diversity in the system, and hence the extent of fiscal autonomy. In such cases, one might not observe generalised efficiency outcomes from government decentralisation.

The means by which decentralisation influences health systems efficiency include spotting sources of red-tape and mismanagement, and, when incentivised, producing cost–saving experimentation (Costa-Font and Moscone, 2008). Furthermore, from a political agency perspective, junior governments tend to be more accountable and seek support for new policies that make them more efficient relative to other levels of movement, and thus more likely to be rewarded electorally (Seabright, 1996). The latter mechanism would imply that some forms of competition between levels of government exist, and as we argue in this chapter, depend largely on how the health system is designed. Hence, not all decentralisation processes will result in better health system outcomes per se. The limits to the design of fiscal and politically accountable systems of governance are the main incentive mechanisms that determine whether decentralisation attains its intended aims.

2.2 Health Systems under Decentralisation

Institutional factors such as political,social, legal and historical constraints play a role in restricting the efficiency of fiscal decentralisation. A central question that remains unanswered in the literature is whether federalism is an institutional device to control expenditure, or instead an institutional structure that heightens the level of activity - by conveying different demands that do not reach consensus at the national level - to legitimize autonomy at a subcentral level, and hence health care expenditure. This chapter will provide an answer to this question by discussing different incentives that exist in different sources of funding. Figure 1 plots patterns of relative public health expenditure of health systems organised under the umbrella of a federal state, against expenditure of countries that remain unitary states and either have or have not decentralised the provision of health care to subcentral governments. Importantly, evidence on unadjusted relative health expenditure suggests that decentralised health systems do not exhibit significantly different levels of relative expenditure, but countries that are organised as federal states have traditionally exhibit about 1% more of their GDP towards expenditures than the rest, though patterns seem to reveal a similar pattern over time, which suggested that the difference might well carry a historical weight.

Figure 1. Relative Public Health Expenditure by health care constitutional form

Source: OECD, 2011.

Health systems traditionally have been designed to attain both insurance and service delivery or accessibility goals. Whilst one function includes global public goods (e.g., management of epidemics, drug patenting, etc.), other components of the health system are primarily local public goods and can be efficiently managed at the local or regional level. Even in the US, there has been an expansion of federal health care programs and expenditure together with an increase in state funds. The so called “insurance component” of health care provision means that even if individuals are not using health care presently, they might still be willing to pay for the development of new health care programs insofar as they expect to benefit in the future. However, as we explain later, the federal and unitary distinction does not explain the autonomy of subnational governments, and instead it is how governments are financed and whether they have health care responsibilities or not which makes all the difference.

The existence of externalities implies that the productivity effects of health programs might spill over to other jurisdictions, especially if individuals are later employed in other jurisdictions. Spillovers are typically internalised through costly coordination mechanisms, or when transaction costs are small enough, through cooperation and contracts (Breton and Scott, 1978).

3. Reasons for health system decentralisation

3.1 Preference Heterogeneity and Mobility

Decentralisation is naturally an institutional embrace of heterogeneity. Yet, whether decentralisation is desirable or not depends on whether the gains from addressing regional heterogeneity are greater than lower scale economies and higher transaction costs that a centralised health system would entail. Many health care services, with the exception of public health attention of vaccines and epidemics, qualify as regional public goods as information specific needs and preferences tends to be scattered over large territories. More generally, health needs tend to be far from homogenous; hence the identification of potential (often unobserved) marginal benefits and true marginal costs tends to be more efficient at a subnational level. In other words, under or overprovision of public services under centralised allocation of regional public goods come with a cost to taxpayers that would not exist under a decentralised government, unless expectations of bailout exist (Bordignon and Turati, 2010).

If, as argued, preferences and willingness to pay for different health programs differ over the territory, then some level of sorting could take place if individuals are able to vote with their feet and either choose health care outsidetheir state or reside in the area where their preferences for health care match the existing supply. Examples demonstrate why there is evidence of internal patient mobility based on the existence of observable quality differences across the territory (Levaggi and Zanola, 2007).

Nonetheless, sceptics might point out that there are limitations for the average citizen, even if well advised, to identify a region or an area where quality of care is higher. The potential benefits of mobility for many procedures might not be high enough to compensate for costs of mobility within a given territory, unless heavily specialised. Another criticism that would more generally challenge the benefit of decentralisation against a uniform central state is when scale economies are lost and especially when decentralisation brings complexity to the system; the transition to decentralisation can lead to duplicities and potential sunk costs, and can politically give rise to veto points to central level legislation. However, the question is whether complexities are a one off event or endure over time, and eventually lead to cost savings as some studies identify (Costa-Font and Moscone, 2008). Similarly, the question of veto points - although it is treated more specifically in the next chapter on political decentralization - is double-edged in that some studies reveal that decentralisation can actually help to dissolve the blame and give rise to reform (Costa-Font, 2010). Decentralisation thus might provide an opportunity to overcome a central level veto or policy neglect. This is the case for mental health care in Spain, where decentralisation has allowed experimentation and reform at a regional level and overcome lack of sufficient consensus (Costa-Font et al, 2011).

3.2 Accountability and Scale

Possibly one of the most important benefits of government decentralisation lies in the associated effect of competition between constituent governments, and the tightening of the so-called political agency. The letter results from either actual or potential mobility, or through the mechanisms of political and fiscal accountability. The close links between politics and action increases the probability that the welfare expansion of a region influences its chances of re-election. Decentralised forms of taxation and governance lead to diversity in services and prices for such services (taxes) and citizens’ capacity to associate action with taxes (and form Wicksellian connections). However, decentralisation implies the introduction of another level of government in the provision of health care, which, unless responsibilities are fully transferred to region states, can blur the lines of accountability, especially when region states lack a parliament where incumbents are held accountable for their policies.

One of the main reasons to keep some health services centralised lies in the existence of optimal scale for global public goods. This is the case for the management of epidemics. In addition to scale benefits of centralisation, it is important to mention externalities or inter-jurisdictional spillovers, both positive and negative, that can in turn lead to an under or overprovision of public goods. However, some theoretical research contends that if spillovers are high enough, decentralised expenditures are welfare enhancing (Koethenbuerger, 2008).

Another potential source of scepticism lies in the difficulty of citizens forming quality perceptions which can limit the benefits from competition, and instead stakeholders might take advantage of their informational position to form cartels or agreements to impede the effects of competition. Similarly, when incumbents in different regions belong to the same political party, they might have fewer incentives to compete, which can mitigate the effect of decentralization on the health system. The capacity to cooperate is influenced by the size of the regional health systems; generally if states are similar in size they will tend to innovate and cooperate less. In contrast, if states are dissimilar enough, this allows experimentation at a lower scale and - as some scholars argue - enhances the credibility of the fiscal contract due to the lower cost of letting a small state go bust (Wildasin, 1997).

3.3 Costs

There are some costs todecentralising a health system as well. One might argue that there are a variety of sunk costs in designing a federal structure. For instance, one would expect a certain level of duplicity in the early stages of decentralisation. This is true in the case of Spain, where the Spanish Ministry of Health remains intact and does not merge with other social policy areas and most health policy responsibilities having been decentralised; similarly, this is the case in Italy. One argument for this feature is that there are still global public goods (e.g., international health, epidemic management, information provision, etc.), which should not be decentralised. Similarly, one can expect some level of externalities or spill over effects of certain conditions in a territory that requires some specific inter-territorial coordination to be facilitated by the central state.

Another argument lies in the need for health system coordination, when cooperation is expected to be hard achieve. Issues on fiscal equalisation and guaranteeing some level of regional cohesion or equity in the provision health care are. Finally, some instance of conflict resolution might be exercised by the central state.

The operationalization of federalism may or may not encompass competition. Indeed, whilst competitive federalist systems like the US might give rise to some form of territorial competition, countries following cooperative federalist structures might engender inertia. Inertia is typically resolved with some level of negotiation between different government tiers, and when conflict emerges legal activity can create some administrative or transaction costs that otherwise would not exist in a centrally run health system.

Finally, one of the most noted potential costs of a decentralised polity lies in the capacity of the central government to enhance fiscal responsibility and to eliminate bailout expectations (Turati and Bourginnon, 2009, Crivelli eta al, 2010). The expectation of subnational governments to receive additional funds in the event of financial need weakens the budget constraint of subnational governments that instead behave strategically. The guarantee of no rescues is paramount, otherwise incentives of subcentral governments would not be aligned with that of the whole state and moral hazard effects would emerge. That is, under decentralization, states will have incentives to incur deficits with the expectation of being bailed out.

4. Vertical imbalances

Possibly one of the most striking problems of decentralised governments lies in the design of incentives to attain diversity and competition. In designing incentives there are set of features that should be taken into account, including the following:

First, budget constraints should not be perceived as being soft, as is the case in some European countries. That is, in designing the decentralisation of health systems, given the existence of common pool and moral hazard effects, it should not be expected that bail outs will be given. Else, perverse incentives to expand expenditure will exist and efficiency will be overlooked. Fiscal federalism theory predicts that allocative efficiency improvements follow from self-financing states, and thus own-taxes should be the primary revenue source. Social insurance systems reveal that states can veto tax increases, but cannot veto social security expansions, which might actually lead to expenditure expansion under federalism, as in the case of Germany. The latter may be one reason underpinning expenditure patterns displayed before. Soft budget constraints in health care are specifically problematic, as the central government cannot credibly allow subnational governments to bankrupt itself in proving highly visible services such as health care.

Second, subnational governments must have adequate resources to pursue their activities, and include a certain level of own resources. If revenues of subcentral governments do not equal or exceed their expenditures, then fiscal vertical imbalances arise. Fiscal imbalances are common in all countries as both in unitary states and in federations, fiscal revenues (as a proportion of GDP) do not equal fiscal expenditures. This imbalance is corrected though the use of transfers, which can be discretionary – and hence politically manipulated- or based on an allocation formula adjusted by differences in needs and risk across subnational governments. However, countries differ in whether health care receives a specific allocation formula, or instead is part of the general funds that are allocated to subnational governments. Overall, the more transparent and general the financing of subnational governments, the more financial planning and efficiency is encouraged.

Third, together with vertical imbalances, one can identify the effects of externalities or spillovers between regional governments, or that respond to a phenomenon that exceeds the jurisdictional domain of the regional government. Another parallel effect is that of the existence of significant disparities in the size and capacity of regional governments; the latter require either adjustments for population or risk in the allocation, and are generally known as horizontal imbalances.

In all countries that have decentralised their health system, transfers represent a large proportion of sub-national government’s revenues (OECD, 2009). The share of own taxe revenue with respect to transfers differs from country to country, as well as the specific transfer design. Intergovernmental transfers are viewed as a supplementary means of finance to address the existence of externalities, and to deal with vertical and horizontal imbalances. Transfers act as a form of redistribution as well as a source of insurance against region specific shocks (e.g., epidemics). Transfers promote innovation when there are limits in the capacity of region states to invest in innovation, and are also more generally employed to use the central state economies of scale in tax collection. As we referred to before, the obvious downside is that unless transfers adjust for fiscal effort cto incentivise efficiency, they can lead to soft budget constraints and more generally moral hazard problems.

One of the most well documented empirical regularities in the fiscal federalism literature is the so-called flypaper effect (Hines and Thaler, 1995; Gamkhar and Shah, 2007). This effect refers to the observed greater stimulatory effect of unconditional grants on local government spending than on increases in community income, whihcn is a form of . moral hazard. Hence, the design of a decentralised health system must take into account the undesirable consequences of a lack of clarity in who bears the fiscal and political blame. That is, if there are fiscal vertical and horizontal imbalances, the incentives of region states are to not keep fiscal discipline because doing so does not reward them. Similarly, if the allocation of political and fiscal responsibilities is poorly defined, then it will be difficult to trace the political credit for health policy decisions, and therefore one might expect region states to invest only on credit claiming projects and not on improving welfare more generally.

Another feature that can trigger poor financial management is the expectation of a bail out either directly, but especially indirectly, though fiscal equalisation mechanisms. Fiscal equalisation schemes exist in almost all decentralised countries and range from 3% of the GDP in Switzerland, Finland and Spain, to 1% in Greece and 2% in Germany (OECD, 2009).

5. Horizontal Imbalances

Together with vertical imbalances, the design of federal health systems considers the emergence of horizontal imbalances, which are differences in health outputs between jurisdictions at the same level of government. Such imbalances can emerge primarily as a result of differences in regional capacity to fund public services, needs, as well as due to other reasons, such as regional choices and preferences. Generally, federal inspired systems do consider the design of equalisation grants and different funds to subsidise equality. Furthermore, to deal with differences in needs, most federal systems take some risk adjustment criteria when designing block transfers, or alternatively equalisation subsidies are used. Finally, horizontal imbalances might result too from differences in preferences and values, which although challenges a certain notion of uniformity-equity, also allows for choice and low cost experimentation.

Several studies suggest that health expenditure per capita (a measure of unadjusted output) appears to decline (or not to increase) with decentralisation. Baicker et al (2010), in their examination of fiscal federalism in the US, consistently find that programs that have been devolved to the states - including education, public welfare and a share of health care - exhibit lower regional inequalities in 2002, ascompared to 1957 before devolution took place. Similarly, in Spain, some studies find that regional inequalities in health, education and social care have declined. Figure 2 below compares regional inequalities in Spain, UK and England. Importantly, regional inequalities in Spain, where devolution is managed regionally, have decreased by 50%, whilst in the UK we see a more modest decline, but in England, a highly centralised health services exhibits high regional inequalities, with rates of more than double that of Spain, which in turn appear stable over time.

What can explain such a phenomenon? One explanation lies in the effects of equalisation mechanisms and a certain failure in England to deal with regional specific needs and preferences. Whilst this is true, it does not fully explain why we do not observe the same downward trend in Spain or in other countries in the UK. A second explanation links policy diffusion as a mechanism to externalise the innovations, whereby traditionally lagging regions import the innovations of front-runner regions, a phenomenon previously documented in Spain (Costa-Font and Rico, 2006a). These mechanisms would not exist in centralised health systems. Therefore, although decentralisation does indeed give rise to diversity, in the longer run diversity might well decline if the mechanisms for policy diffusion, and more generally credit claiming by innovative governments, become fully operative.

Figure 2. Regional Inequalities (unadjusted health care output)

6. Challenges

6.1 Race to the bottom?

Decentralisation can be seen as a means to reduce the size of the state. This is the hypothesis put forward by Brenan and Buchanan (1980). According to this hypothesis, decentralisation stands out as a pro-competitive mechanism to tame the Leviathan as follows:

“Total government intrusion into the economy should be smaller, ceteris paribus the greater the extent to which taxes and expenditure are decentralised.”

Hence, as government intervention would be expected to decline with decentralisation, one would expect a waning of unnecessary expenditures and red tape . Alternatively, Oates (1986) suggests a counterbalancing argument, namely that while decentralisation is more efficient by tailoring programs to heterogeneous preferences, it implies a loss of some scale economies that alone can be large enough to trigger expenditure to increase. In the case of health care, empirical evidence is suggestive of the second effect. As Figure 1 displays, the argument does not receive empirical confirmation in the case of health care.

Different explanations have been put forward to explain why public expenditure increases after decentralisation:

a) Short term scale loss vs. long-term efficiency gains. Health expenditure might increase but the total welfare expenditure in the long run would not increase due to the longer term savings that come from allocative efficiency gains from decentralisation in administrations withmore experience in managing budgets in comparison to centrally managed models. This is the evidence Costa-Font and Moscone (2008) find in the Spanish system of regional health care services. Their findings suggest that experience in managing health care responsibilities is associated with lower per capita expenditure.

b) Collusion (Brenan and Buchanan, 1980) due to horizontal cooperation or vertical coordination that typically takes place when there are fiscal imbalances resulting from expenditures being decentralised, but a higher level of government collects taxes and assigns them through block transfers to the states. This is the case in the UK with the Barlett formula and in Spain for ordinary regions subject to common financing . Alternatively, one can imagine the influence of the central state through laws that set out framework packages. Examples from Italy show that regulation has managed to reduce the extent of diversity, which might explain a moderate interregional competition.

c) Vertical competition and policy innovation can explain to an extent why a standard race to the bottom does not take place. Vertical competition, as we explain below, refers to competition for underfunded policy responsibilities when there are opportunities for credit claiming. Costa-Font and Rico (2006a) reveal that the rationale of vertical competition in health care is to expand rather that reduce health care expenditure. An important consequence of vertical competition is the development of policy innovation at the subcentral level in order to differentiate themselves from other region states, and to avoid competition. Evidence of this effect on pharmaceuticals regulation explains significant policy innovation that when successful tends to be diffused (Costa-Font and Puig Junoy, 2007). Oates and Wallis (1988) use an alternative explanation for expenditure increases based on the existence of government differentiation, which is consistent with findings which suggest policy innovation is boosted to keep the cannibalisation effects of competition under control.

d) Political markets. Decentralisation brings power closer to the citizenry, and hence enhances political incentives for incumbents to influence policy action to guarantee re-election. If the incentives of regional incumbents are not driven by mobility but exclusively through the political system, then governments will attempt to accommodate the preferences of the median voter. If the median voter favours widening health care coverage, as is the case in many European countries, one would expect inter-jurisdictional competition to lead to an expansion of the size of the health system, and more generally to health care reform.

e) Finally, an alternative explanation for the absence of a race to the bottom in health care lies in the fact that decentralisation fiscal designs rely too much on central level funding, such as block transfers, and in turn exhibits a high degree of borrowing autonomy which engender fiscal deficits, as is the case of Italy and Spain (Crivelli et al, 2010).

6.2 Mobility and voting with one’s feet

A potential source of (horizontal) government competition is mobility-creating welfare or quality-driven migration. Patient mobility is less of an issue in Europe compared to the US. In the United States, 40% of the population live in a different state than that of birth, and the percentage increases to 50% if we look at college graduates (Baicker et al (2010). Similarly, 2.5% of US residents change state every year. Mobility is far more limited in Europe for a variety of reasons, including the fact that individuals build significant regional attachments and networks, as well as other social barriers like language that pertains even within countries such as Switzerland and Spain. In addition to the latter constraints, given that decentralisation empowers regional incumbents to improve the quality of their regional health systems, patient mobility becomes a residual in sorting out short term health care needs rather than a competitive instrument, as Tiebout models would predict.

6.3 Political Agency and Accountability

For decentralisation mechanisms to work, the mechanisms of the political agency need to be in place. That is, decision makers should be responsive to the demands of their constituents. The most obvious way for this to take place is through regional or statewide electoral processes so that officials in subnational governments align their own preferences for improving lives with that of their constituents. Elections should be based on region or state specific affairs and not intertwined with other countrywide matters, as is the case in many of Spanish regions (Costa-Font, 2009), and in many developing countriessuch as China, where officials are not elected. One way to evaluate how well this model performs is by examining users’ perceptions. Figure 3 displays data on the best performing health systems based on such a criteria, and strikingly the three countries topping the rank -namely Spain, Belgium and the UK -,are decentralised health systems, and countries at the bottom are either centralised health systems or federal system that have shifted to more central control of their health systems. Of course, these data are insufficient to perform a full evaluation, but it is suggestive of some trends.

Figure 3. Health System Improvement Perceptions

Question: “Compared with five years ago, would you say things have improved, gotten worse or stayed about the same when it comes to…Healthcare provision in (our country)?”

Source: Eurobarometer, 2010.

6.4 Experimentation

The link between decentralisation and experimentation is been an old argument that dates back to Hayek’s (1939) argument that decentralisation, by increasing experimentation, produces more information on how to run a government. Health care is one of the most clear-cut examples of a natural public policy experiment.. The US shows how federal health care reform shares significant knowledge from health care reform in Massachusetts. Evidence from different countries reveals that experimentation takes place after devolution for a variety of reasons. First, junior governments tend to legitimise themselves by introducing innovation in the way they run the health system (e,g., free long term care and no prescription charges in Scotland and Wales respectively). Second, decentralisation can provide voice to the opposition party or regional minorities which would be lacking under a decentralised system. This gives rise toa degree of vertical competition with the central government that can provide additional political incentives for innovation (Costa-Font and Rico, 2006a). Finally, if soft budget constraints are corrected, decentralisation can provide fiscal incentives for innovation, especially if innovation produces costs savings - some evidence of this is found in European countries (Costa-Font and Moscone, 2008).

6.5 Political Competition

The fiscal federalism literature (Breton 1996) contends that governments compete. However, in understanding the wide range of competitive relationships one must distinguish vertical from horizontal forms of competition, such that interactions between differing levels of governments are differentiated. The most obvious means of competition comes out of tournaments theory, whereby citizens of one jurisdiction evaluate the performance of their own constituency against other jurisdictions (Salmon, 1987). The main downside of such a mechanism is that performance is not easily observable, especially the quality dimensions which motivate citizens to either move or use political agency to punish or reward the incumbent party ruling the health system. Not even the World Health Organisation is able to fully evaluate the performance of different health systems. Nonetheless, even if citizens can evaluate the performance of the health system, there is no way to vote on the health system in isolation, as general elections do not tend to serve that purpose. Hence, if health care is one of the key areas of policy that has been decentralised, devolution can help citizens to express approval or discontent with health policy specifically. Finally, even when regional voting occurs, regional elections must be sufficiently differentiated from other electoral contexts in order for it to convey the preferences of the regional median voter, which does not always take place.

6.6 Local vs. country level capture

One of the common concerns about the decentralisation of health policy is that of capture, leading to policy failure. Decentralisation can bring local producers and regulators closer together, which might reduce information asymmetries - but if mechanisms of public sector purchasing are not transparent enough, this may lead to the risk of local capture (Laffont, 2000). On the other hand, it is well documented that decentralisation increases the transaction costs of capture at the country level. Hence, whether decentralisation gives rise to or serves as an incentive to contain the effects of regulatory capture of European health systems depends generally on the effects it has on transparency and corruption in general, and/or whether the welfare loss from regional capture exceeds that of lesser captures resulting from higher transaction costs in a decentralised health system.

7. Conclusion

This paper has attempted to bring together a broad set of questions on the decentralisation of health systems. It argues that decentralisation is a proxy for subnational autonomy, and its success in tackling heterogeneity in preferences and needs depends on its design. Particularly important design features include addressing fiscal imbalances, promoting competition, policy innovation and making sure that the mechanisms of the political agency align individuals’ preferences and needs with that of their incumbent’s priorities. However, there are several limits to the success of decentralised health systems, including the alignment of fiscal and political accountability, the design of resource allocation mechanisms that bypass soft budget constraints, and more generally the development of incentives to policy diffusion that, if successful, can keep long-term inequalities in health output down. More importantly, decentralisation can help to enhance the political accountability of a health system by giving rise to a parallel political cycle where citizens can evaluate specifically policies that have been devolved.