THE ROLE OF PROBLEM PRESSURE AND IDEAS IN HEALTHCARE SYSTEM CHANGE

Transcription

1 Paper to be presented at the 6th Annual ESPAnet Conference Cross-border Influences in Social Policy in Helsinki (Finland), September 18 20, 2008, Stream 13: Changing ideas and consequences of reforms in healthcare THE ROLE OF PROBLEM PRESSURE AND IDEAS IN HEALTHCARE SYSTEM CHANGE --Preliminary version, please do not cite without permission-- Achim Schmid, Ralf Götze, Mirella Cacace and Heinz Rothgang Achim Schmid, University of Bremen, Collaborative Research Center 597 Transformations of the State Linzer Str. 9a, Bremen (Germany), +49 (421) , Ralf Götze, University of Bremen, Collaborative Research Center 597 Transformations of the State Linzer Str. 9a, Bremen (Germany), +49 (421) , Mirella Cacace, University of Bremen, Collaborative Research Center 597 Transformations of the State Linzer Str. 9a, Bremen (Germany), +49 (421) , Heinz Rothgang, University of Bremen, Collaborative Research Center 597 Transformations of the State Linzer Str. 9a, Bremen (Germany), +49 (421) , Abstract The paper attempts to explain the changing role of the state in OECD healthcare systems. The analysis of 23 OECD countries has revealed broad trends with respect to financing, service provision and regulatory structures. We identified increasing similarities between the three system types National Health Service, social health insurance and private health insurance. The public financing share tends to converge over time and the funding mix has grown increasingly similar. At the same time service delivery shows a common privatization trend. With respect to regulatory structures, we observe convergence as healthcare systems have integrated non-system-specific elements of regulation. Our broad explanatory model starts with the identification of driving forces in the form of globalization, demographic change and medical-technological progress which cumulatively create problem pressure on healthcare systems as they raise healthcare expenditure while economic resources are limited. While the direction and the timing of system transformation may be influenced by various intervening variables, such as actors, their interests and perceptions, we argue that the healthcare system type is an essential factor to explain the increasing similarities seen amongst systems over time. As healthcare systems vary, so do their adaptive responses to problem pressure. Healthcare systems tend to develop specific deficiencies which can not be solved by routine mechanism. As a consequence non-system-specific elements are implemented which lead to the emergence of mixed types. However, this model based upon problem pressure and system type fails to address the mechanisms of change and convergence. Therefore we aim to combine the rather functional model with interpretative approaches which include policy learning and diffusion, both broadly captured under the heading of ideas. We do this by describing the diffusion of regulatory mechanisms (DRGs and Reference Pricing) across healthcare system types. 1

2 The Role of Problem Pressure and Ideas in Healthcare System Change 1. Introduction Analysing the development of 23 OECD healthcare systems 1 since the oil crisis of the 1970s which are generally considered to mark the end of unbounded welfare state expansion reveals broad trends with respect to financing, service provision and regulatory structures. The public financing share tends to converge over time and the funding mix has grown more similar. At the same time, in healthcare service delivery privatization trends can be observed as a common pattern. With respect to regulatory structures, we observe increasing similarities between the three healthcare system types National Health Service (NHS), social health insurance (SHI) and the private health insurance (PHI), as exemplified by England, Germany and the US as healthcare systems have integrated non-system-specific or innovative elements of regulation (Rothgang et al. 2005). As a consequence of these developments we observe the emergence of mixed type healthcare systems. Our main research interest in this paper is to investigate explanatory factors concerning the above mentioned changing role of the state in OECD healthcare systems. In doing so we can make use of different branches of theory: functionalist approaches referring to system s needs, actor-centred approaches focusing on interests and related political or social power resources, institutional explanations highlighting the role of political institutions and path dependency and interpretative approaches pointing towards ideas, i.e. cognitive factors, values and belief systems as well as social and policy learning (Hacker 1998; Nikolentzos and Mays 2008; Béland 2005; Béland and Hacker 2004). Of these approaches the functionalist perspective, in the sense of regarding policy change as a response to problem pressure, seems to serve best as a starting point to explain trends of convergence. As the golden age of welfare state expansion ended in the 1970s, healthcare systems increasingly began to experience problem pressure, which raised the need to contain healthcare costs and at the same time limit the amount of additional resources going into the healthcare system thus causing the need to find more efficient ways to organize healthcare systems. Due to functional requirements which differ according to the healthcare type, distinct reactions have been required leading to convergence of healthcare types. Interpretative arguments and social learning in particular may then provide the mechanism for transforming functional requirements into political action. Conversely, institutional and path-dependence explanations and also power resource approaches rather highlight differences across countries or systems which are structured by the country-specific configuration of institutions or power resources, respectively. Respective explanatory elements should thus be added if differences in the timing and speed of developments within one system type are to be explained. Our broad explanatory model starts with the identification of driving forces in the form of demographic change and medical-technological progress on the one and globalization on the other hand, which cumulatively create problem pressure on healthcare systems as they raise the demand for healthcare funds while at the same time limiting economic resources available to the system. While the direction and the timing of system transformation may be influenced by various intervening variables, such as actors, their interests and perceptions, we argue that 1 Our sample includes all countries (except Turkey) that joined the OECD before the first oil crisis starting in autumn 1973, namely Australia, Austria, Belgium, Canada, Denmark, Finland France, Germany, Greece, Iceland, Ireland, Italy, Japan, Luxemburg, the Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, the United Kingdom and the United States. 2

3 the healthcare system type 2 itself is an essential factor to explain the increasing similarities amongst systems. This seemingly paradoxical statement can be understood as follows: As healthcare systems vary, so do their strengths and weaknesses and their adaptive responses to problem pressure. Healthcare systems tend to develop specific deficiencies which can not be solved by routine mechanisms. As a consequence, in the search for more efficient solutions, non-system-specific and/or innovative elements are implemented which lead to the emergence of mixed types and increasing similarities. Of course, adaptive responses to system deficiencies do not develop automatically. Within a competitive environment efficient answers to problem pressure may spread via evolution. For healthcare systems which hardly compete with each other on a given territory an additional mechanism for structural innovation and diffusion is needed. Adaptive responses may require decisions of policy-makers and subsequent implementation through local actors. Adaptation processes may also occur in the absence of policy reforms through actors empowered under the existing regime (Hacker 2004). Searching for more efficient solutions, decision makers can fall upon experience with existing national policies as well as identifying a range of regulatory alternatives in other countries. The monitoring and evaluation of reform alternatives gives the opportunity to estimate reform effects before implementation. Thus, besides problem pressure mediated by healthcare systems, we have to consider cognitive factors. The aim of the paper thus is to combine the rather functional approach based on healthcare system types with approaches that include policy learning and the spread of policy ideas across healthcare systems. We proceed with a short account of the changing role of the state in healthcare systems (section 2) and a presentation of our broad explanatory model (section 3). In the following we describe health-system-specific deficiencies and steps of adaptation taken according to healthcare system types (section 4) thus outlining the functional core of the explanation. Next we use DRGs and reference pricing as two examples of policy instruments which reflect new ideas to regulate healthcare sectors (section 5). We describe problem pressures which triggered the implementation of these regulatory mechanisms and describe their diffusion across healthcare system types. While problem pressure may evoke adjustment processes, the role of policy ideas and their diffusion will have to be considered as a transformation mechanism. The conclusion offers a summary and hypotheses about features which facilitate and hamper the diffusion of policy ideas (section 6). 2. A short account of the changing role of the state in healthcare systems Our analysis of healthcare system change in the OECD world begins with an account of the developments since the oil crises of the 1970s which are generally considered to mark the end of unbounded welfare state expansion. In order to give a systematic overview, we distinguish between three dimensions of the healthcare system: the financing dimension, the service provision dimension and the regulatory dimension (Rothgang et al. 2005). While many authors examine health expenditure levels and growth trends, a most interesting finding with respect to the role of the state in the financing dimension is the convergence of the public financing share (Cacace et al. 2008a). The public financing share subsumes tax financing and social insurance financing in percentage of total health spending (cf. OECD 2004). Convergence occurred as the dispersion of the public share of total healthcare financ- 2 See also Wendt et al. (2008) for a conceptual framework for defining healthcare system types. 3

4 ing decreases (sigma convergence). This is mainly due to countries with a low public financing share in the beginning of the observation period tended to increase the public financing share, while those countries which already displayed high values of public financing in 1970 rather turned to private financing (beta-convergence). Countries like the US, Greece and Switzerland, for example, show above average increases of the public financing share from low initial levels. In contrast to this, we observe an increase of private financing in several countries like Canada, Germany, New Zealand, Ireland, Italy and the Scandinavian countries, which all had above-average public financing shares in 1970 (Rothgang et al. 2008). In the service provision dimension, we find that public provision decreases in nearly all of the 15 OECD countries for which appropriate data is available (Rothgang et al. 2008) 3. Over almost all healthcare systems, two major factors contribute to the common privatization trend. First, due to efficiency pressure and the requirement to provide care in the most suitable setting, but also medical progress enhancing outpatient care, the locus of care is shifted away from the inpatient to the outpatient sector (Getzen 2004; Tuohy et al. 2004). Since the state traditionally plays a greater role in the provision of hospital care, while private providers prevail in the outpatient sector, this means an implicit shift to private provision (implicit privatization). Second, the state as a direct service provider is on the retreat, since governments in many healthcare systems divest their facilities (explicit privatization). Mainly for-profit organizations buy up public hospitals in order to turn the industry into a profitable market segment. Besides privatization which refers to a growing share of both, for-profit and not-forprofit providers, therefore, there is also evidence for a profitization in the delivery of services (Schmid et al. forthcoming). Healthcare regulation refers to the fundamental relationships between the three main actors in healthcare systems, i.e. the service providers, the financing bodies and the potential beneficiaries. For our analysis, we focus on the structural features of regulation by examining the actors (thereby addressing the question, who regulates the healthcare system?) and the mode of interaction describing the mechanism by which the system is co-ordinated (Tuohy 1999; Rothgang 2006). As concerns actors, we refer to three specific kinds that are relevant in the regulation of healthcare systems, namely the state, corporate actors (as e.g. the associations of sickness funds or providers) and private market participants. With regard to the specific interactions taking place between actors, we identify three basic modes of coordination: (1) the exertion of hierarchical control (2) the engagement in bargaining processes where contract partners interact on equal footing and (3) the condition of competition which implies rivalry between individuals or groups. In order to analyse the regulation dimension, we chose the cases of Britain, Germany and the United States as these countries most closely resembled the respective ideal types of a stateled NHS, a corporatist SHI and a market driven PHI in the beginning of our observation period in the 1970s (Giaimo 2002; Giaimo and Manow 1999; Boscheck 2005; Rothgang 2006). Beginning with the case of the British NHS, we find an example of a state-led healthcare sys- 3 In order to measure public service provision in healthcare systems we have constructed a Public Provision Index (PPI). The PPI is constructed as a weighted mean of all sectors, i.e. the inpatient, outpatient, dental and pharmaceutical sector. As a formula for assessing the role of the state across all sectors, we examine changes in the public/private-mix within healthcare sectors and changes in the size of healthcare sectors (Schmid et al. forthcoming). In the inpatient sector, the public share is determined by the number of beds in public ownership. In the outpatient sectors (outpatient care, dental care and pharmaceuticals), the employment status of health personnel is used to delineate public from private provision. We define self-employed outpatient care providers as private (for-profit) entities, while healthcare personnel employed by the government are public. The PPI has been constructed for 15 OECD countries: Australia, Austria, Denmark, England, Finland, France, Germany, Iceland, Italy, Japan, the Netherlands, New Zealand, Norway, Switzerland and the US. 4

5 tem that has introduced market elements to improve efficiency. The prevalence of market mechanisms in the British healthcare system has been accompanied by a stronger state involvement as indicated for example by the increasing role of the state in the regulation of service providers (Klein 1996; Giaimo and Manow 1999; Hacker 2004; Grimmeisen forthcoming). While there is a continued reliance on corporatist self-regulation in the German healthcare system (Giaimo and Manow 1999; Wendt et al. 2005). Germany now makes room for public (state) and private (market) actors, thereby squeezing the traditional system of selfregulation in between (Rothgang et al. forthcoming). Thus, the introduction of competition between sickness funds in 1996 has also triggered a lot of subsequent reforms which in sum profoundly changed the modes of regulation in the financing system. Meanwhile, the introduction of Diagnosis Related Groups (DRG)-based remuneration in 2003 has afforded a greater role to the state in the regulation of healthcare policy and, at the same time, to the market. Finally, with regard to the private healthcare system of the US, hierarchical state regulation rose significantly as a result of the creation of the public programmes Medicare and Medicaid. Yet within the realm of private insurance, hierarchical state regulation remained weak, paving the way for considerable policy conversion (Stone 2000; Hacker 2004). Thus at the same time, private insurance experienced the emergence of managed care proliferating quickly from the 1980s on, which, although purely private by nature, has given way to a series of hierarchical arrangements (Cacace forthcoming). As a consequence of the backlash against managed care, hierarchical regulation of providers and patients was amended by bargaining elements. Due to the spill-over of managed care to the public programs and as private insurers adopt remuneration methods from the Medicare program, the boundaries between the public and the private blurred (Cacace 2007). Thus in all three healthcare systems, the regulatory structure has altered profoundly. Our main finding is that these healthcare systems incorporate elements that are non system-specific, which led to a considerable hybridization of healthcare systems (Rothgang et al. forthcoming). As healthcare systems blur by incorporating elements from one another, mixed types emerge. The convergence hypothesis in the regulation dimension therefore suggests that healthcare system became more similar as they adopt instruments and structures from each other. Within this process, the diffusion of instruments and policy learning have to be considered as mechanisms, which support increasing similarities in response to common or systemtype specific problem pressure (Wilsford 1995; Gilardi et al. 2005). 3. Explaining Healthcare System Change: The General Explanatory Model In order to explain the phenomena of convergence and common trends, it is hypothesized that healthcare systems share a common source of problem pressure which forces them to induce change. According to the standard functionalist argument, we should witness a uniform reform path in all industrialized countries (Hacker 1998; Giaimo and Manow 1999; Tuohy 1999). In particular the assumption is that reform paths lead to a privatization of social risks, to the introduction of market competition and to a retreat of the state across OECD healthcare systems. Yet by analyzing the empirical data, we find that common privatization trends are solely observed in the service provision dimension. Privatization trends alone, therefore, tell only part of the story. Regarding their financing structure, by contrast, we find that the healthcare systems have approached each other with respect to the role of the state. 4 Moreover, con- 4 Here we observe common privatization trends mainly during the 1980s and the 1990s and a tendency towards higher public spending levels for the rest of our observation period. We also show evidence for convergence when comparing public expenditure growth with its baseline in 1970, as countries with low public financing caught-up with the leaders (Cacace et al. 2008a). 5

6 vergence in the regulation dimension implies a blurring of systems i.e., public elements grow in the private US healthcare system, while market competition is considerably enhanced in the state-led UK system and in the Germany s social insurance scheme as well (Rothgang et al. forthcoming; Cacace et al. 2008a). Thus we find that the chosen trajectories vary considerably among the distinct types of healthcare systems. A crucial explanatory variable, therefore, is the healthcare system itself, its deficiencies and functional requirements as reflected in the specific system type. The direction of change is ultimately subject to the particularities of individual system types yielding a modified problem pressure hypothesis. To structure our argument, we propose a general model of healthcare system change, which serves as a heuristic device (see Figure 1). The explanatory model addresses the interplay of specific driving forces, which cumulatively exert problem pressure on healthcare systems. This system stressor can then be seen to interact with intervening variables, which structure change by veering systems toward a specific direction and by setting the pace; thus accelerating or retarding and even temporarily arresting transformation. Figure 1: General Explanatory Model Driving Forces Intervening Variables Transformation (dependent variable) Political-economic System MATERIAL FORCES: Globalization Medical Technological Progress Demographic Change INSTITUTIONAL FORCES: Europeanization NAFTA, WTO, WHO IDEATIONAL FORCES: Individualization Problem Pressure Health care System Type Actors Institutional Structures Value System Ideas Perceptions Convergence Common Privatization Trends Blurring of Healthcare Systems Source: own depiction To start with the driving forces, we distinguish between material (e.g. globalization, medical/technological progress, demographic change), institutional (e.g. Europeanization) and ideational forces (individualization 5 ). Without going too much into detail about the mechanisms of how these forces act in an isolated manner, it can be said that especially demographic change, medical progress and individualization tend to increase demand for scarce resources, whereas globalization 6 and partially also Europeanization rather set limits on their 5 Individualization trends have two major consequences. On the one hand, changing life patterns require new forms of risk protection. On the other hand, increasingly informed and self-conscious patients successfully demand the responsiveness of health care services and thus ask for more choice. 6 Conversely, the compensation thesis (see: Rodrik 1996; Swank 2002; Rieger and Leibfried 2003) assumes that globalization has rather an expansive effect on welfare spending. However, it can be argued that compensation policies will rather include active and passive labor market policies to protect those groups which are most vulnerable to globalization. Healthcare spending, by contrast, is less likely to be used to compensate the marginal- 6

7 availability. For the time being, however, we will argue that their cumulative effects exert problem pressure on healthcare systems (Giaimo and Manow 1999; Tuohy 1999). As a consequence, efficiency has been the catchword in the political debate of the 1990s, requiring healthcare systems to implement cost-containment strategies while simultaneously achieving (or preserving) state-of-the art treatment, responsiveness and choice (Wilsford 1995; Hacker 2004). At the same time, institutional forces require the harmonization of national healthcare policies and the compliance with international law. Regarding the intervening variables, the most crucial factors are the healthcare systems themselves: their functional requirements, their respective strength and deficiencies, their value systems, the associated veto points and organized interests, which of course are also part of the broader political-economic system. In line with the modified problem pressure hypothesis, it is both the nature and degree of problem pressure, subsequently mediated by the component factors of the healthcare systems themselves, which can be said to define the direction of healthcare system transformation observed (dependent variable, see Figure 1). This hypothesis suggests that in search of solutions to common problem pressure, systems develop distinct policy responses, thereby quitting traditional paths and taking up new elements that are nonsystem specific. As system types vary so too do their adaptive responses. By borrowing from each other healthcare systems transform into mixed-types (Rothgang 2006). The modified problem pressure hypothesis takes into account that different lines of adaptation emerge depending upon the specific healthcare system type the observed countries resemble. 4. Healthcare Systems and their Response to System-specific Deficiencies According to our argument, the direction of change is indicated depending on the specific features and deficiencies of the healthcare system type. In a somewhat stylized manner, we line out the general characteristics as well as the system-specific deficiencies by examining the system logic of (1) state-led NHS systems, (2) social insurance schemes and (3) private healthcare systems, which serve as examples of most pure-type healthcare systems (Rothgang 2006; Giaimo and Manow 1999; Tuohy 1999). We thereby draw on the general system logic, as well as on some empirical examples, which may serve as first evidence. 4.1 National Health Services Healthcare in National Health Services (NHS) is characterized by universal coverage based on citizenship. Ideally, a full range of health services is provided free at the point of delivery. Consequently, services are typically financed through taxes. Health service delivery is characterized by the dominance of public provision i.e., state-owned hospitals as the dominant providers in the inpatient sector and public employees in the outpatient sectors. Moreover, the relations between financing institutions, service providers and (potential) beneficiaries/ patients are mainly regulated through state hierarchy. Within our OECD sample, the Nordic countries (Denmark, Finland, Iceland, Norway and Sweden), Australia, Ireland, New Zealand and the UK, as well as Italy, Spain and Portugal can be counted as NHS systems. However, this large group of NHS countries is heterogeneous and some countries significantly deviate from the ideal type description as the outpatient, dental and/or pharmaceutical sectors are characterized by private providers (Schmid et al. forthcoming). ized workforce (Burgoon 2001). Therefore we assume that concerning healthcare globalization rather reinforces cost containment pressure. 7

8 NHS systems are successful in cost-containment, but tend to develop specific malfunctions such as long waiting lists for certain treatments, insufficient investment in healthcare facilities, poor responsiveness and low productivity or low motivation of providers. These problems are related to budget cuts which entail forms of rationing, but can also be more generally related to the idea of state failure. In short, the notion of state failure assumes that stateorganized healthcare may guarantee equal access to services but fails to provide services efficiently (Scott 2001). Therefore, in NHS systems problem pressure mainly translates into remedies to state failure. Driven by the need to produce healthcare services more efficiently (cost-effective and technically efficient), mainly market-based alternatives come under consideration. The developments in state-led systems with respect to the role of the state in healthcare can be crudely summarized as follows. The role of private healthcare financing has increased on average and service provision has been privatized in various ways. Moreover, NHS systems have implemented market elements as modes of regulation. However, the implementation of market-oriented reforms does not necessarily mean a loss of state authority, since markets need to be regulated. Therefore market-oriented reforms tend to go hand in hand with intensified state hierarchy ( seesaw effect ). The increasing salience of private financing in state-led systems may be simply a side-effect of cost-containment in the public system. It has to be considered, however, that this form of privatization may involve high political costs. In the UK, public financing has increased massively from 2003 on, after years of continuous decline of the public financing share (OECD 2008). Private financing can also reveal dissatisfaction with the public system, e.g. undue waiting times or low standards of treatment provide incentives to choose private alternatives (Klein 2005). As a further option, the rise of private financing may also reflect attempts to avoid excessive demand. Cost-sharing can be implemented to provide an incentive for using more primary care, like in Italy and some Scandinavian countries (Saltman and Figueras 1997; Figueras et al. 2002; Freeman 2000). In the service provision dimension we observe diverse forms of privatization. Functional privatization (Schneider and Tenbücken 2004) has often been a first step, particularly the outsourcing of non-clinical services to private corporations. Less often, health-related functions and hospital management have been outsourced (Maarse 2004). The trend to abolish command-and-control management and to grant more autonomy to hospitals goes hand in hand with formal privatization (Busse et al. 2002). Formal privatization means that hospitals adopt private legal forms and private management structures while remaining in public hands. The sale of hospitals to private owners, i.e. material privatization, however, appears to be rather marginal in most NHS countries. Instead, private providers have often been strengthened implicitly as resources have been shifted towards the outpatient sector, where private providers play an important role (Schmid et al. forthcoming). These diverse forms of privatization can be understood as attempts to increase incentives to provide services more efficiently and to steer resources away from cost-intensive inpatient care. State hierarchy as the dominant mode of regulation in NHS systems has supported costcontainment policies which can easily be implemented through global budgets (Grignon 2006). At the same time, hierarchical modes of regulation often in the form of centralized planning tended to produce allocative inefficiency which e.g. became manifest in long waiting lists. As a result of system-specific problems, NHS systems have supplemented state hierarchy with market-style modes of regulation (Freeman 2000; Saltman and Figueras 1997). Several NHS systems such as England, Finland and Italy have established a purchaser- 8

9 provider split, though in some cases limited to certain areas, like in Sweden and Spain (Figueras et al. 2002). Accordingly, providers have to compete for purchaser contracts. Similar developments have taken place in New Zealand in However, according to policy analysts, the competition of public and private providers for public service contracts did not produce efficiency gains. By 2000 the new Labour government redirected the system towards local planning (Ashton 2005). Also in Italy, managerial behaviour has generally been encouraged (Freeman 2000). Probably the most comprehensive reform steps have been taken by the English NHS through the creation of internal markets (Grimmeisen forthcoming). The introduction of market mechanisms, however, provoked new state interventions. For example, during the implementation of the purchaser-provider-split in the UK hospitals ran into difficulties because of massive problems regarding calculating costs and contract models with purchasers. By state interference, cost calculation was standardized through patient classification systems (Health Care Resource Groups, see section 5) (Grimmeisen forthcoming). Thus, hospital remuneration exemplifies a seesaw effect as the state introduces competition by means of hierarchical intervention and is forced to apply further regulatory measures in order to support the functioning of markets. 4.2 Social Health Insurance Systems Social health insurance (SHI) systems account for the second largest group in our OECD sample, consisting of Austria, Belgium, France, Germany, Japan, Luxembourg, the Netherlands and since 1996 Switzerland as well. The system logic of SHI is characterized by corporatist self-regulation based on collective bargaining between sickness funds and provider associations (Rothgang et al. 2005). On the financing side, wage-related contributions play a dominant role. With regard to the health service delivery, inpatient care mainly relies on public or private non-profit hospitals, whereas the other healthcare sectors (i.e. the outpatient, dental or pharmaceutical sector) are dominated by private for-profit providers (Eeckloo et al. 2007). SHI schemes usually cover a large part of but not the entire population. Although insurance is generally mandatory, the self-employed, civil servants and high-income employees often have an exit-option (Carrin and James 2004). The eroding financial basis of SHI schemes can be identified as a major system-specific deficiency. First and foremost, the share of waged on overall economies income decreases since the late 1970s, while capital gains (which are not subject to contributions) become more important in all OECD countries (IMF 2007). The decline of standard fulltime employment also decreases the sickness funds revenue. Furthermore, demographic change negatively affects the ratio between contributors and recipients (Kanavos and Yfantopoulus 1999). At the same time, global competition forecloses the raising of SHI contributions as they represent a (visible) part of labor costs (Ferrera et al. 2001). Thus we observe attempts to promote alternative funding sources, like taxes or private payments. In Germany and the Netherlands for example, private out-of-pocket expenditures increased since the mid-1990s (OECD 2007). In addition, social insurance contributions were transformed from wage-related to per-capita flat-rate contributions. This occurred recently in the Netherlands, where almost half of the total healthcare expenditure was financed through income-independent contributions (Greß et al. 2007). In service provision, trends towards privatization and profitization, i.e. the growth of private for-profit providers at the cost of private non-profit providers, are mainly driven by permanent fiscal austerity which limits the state s support for public or private non-profit hospitals. Regarding public hospitals, governments often react with formal privatization to raise incentives 9

10 for efficient behavior by making them responsible for profits and losses (Busse et al. 2002). In addition, private non-profit hospitals have to cope with reduced support due to secularization trends. Public and private non-profit hospitals with unfavorable cost structures are vulnerable to acquisition by private for-profit organizations. The advantages of for-profit organizations include more flexible personnel structures and easy access to the capital market. Since forprofit operators serve investors interests, they are forced even more strictly to provide services efficiently. Concerning the regulation of SHI systems, we observe a decreasing role of corporatist selfgovernment and an increase of state hierarchy and market competition at the same time. We conclude that this most striking change is a reaction to self-inflicted steering failures and the tendency towards deadlock. These phenomena which are mainly caused by rent-seeking behavior of corporatist actors are commonly labeled institutional sclerosis. In a system of mandatory assignation to sickness funds or full retrospective reimbursement, the insurer has no incentive to contain costs while the collective arrangements rather force him to be responsive to providers claims. As a consequence, since the early 1990s Belgium, Germany, Switzerland and the Netherlands enhanced the legal framework and incentives to allow for competition between sickness funds (Laske-Aldershof et al. 2004). As soon as the state intervenes with the abolition of fund assignation, competition restrains the funds freedom to increase contribution rates (Sappington 1991). Yet a considerable side-effect of competition on the financing side is the incentive for risk selection (Cacace et al. 2008b). In all SHI systems with free choice of sickness funds, the state reacted to this adverse result of market competition with the introduction of risk-compensation schemes (van de Ven et al. 2007). As the sickness funds needed parameters to compete with, the government started to cut back the collective contracting of providers by introducing opportunities for selective contracting (Cacace et al. 2008b). Germany and the Netherlands are examples for this spill-over effect of competition from insurer to provider markets (Maarse and Bartholomée 2007). Furthermore, market-oriented changes in the reimbursement structure were implemented to improve costeffectiveness of the healthcare system and to reduce the rent-seeking of providers. In the inpatient sector this is strongly tied to the switch from reimbursement schemes according to incurred cost to lump-sum payment systems, calculated by using Diagnosis Related Groups (Cacace et al. 2008b). Generally, the pharmaceutical sector became more affected by hierarchical price control. Nearly all social insurance systems have introduced fixed or reference prices in their pharmaceutical markets (Mrazek 2002; Ess et al. 2003). As the definition of the benefit package is a major challenge of a system regulated by collective bargaining procedures, this aspect is especially prone to state intervention. In SHI systems, the failure of corporate actors to define an appropriate scope of medical goods and services to be collectively financed therefore also leads to greater state intervention. Institutions for health technology assessment have been founded in nearly all SHI systems, like in Austria, Belgium, Germany, France and the Netherlands (Banta and Oortwijn 2000). Sometimes, hierarchical regulation is covered by integrating self-regulatory structures, but all these measures are implemented and often controlled by the shadow of hierarchy. In conclusion, poor cost-containment capacities and the lack of cost-effectiveness in collective bargaining systems provoked the state to intervene hierarchically as well as to introduce competitive market mechanisms which in turn led to even more hierarchical re-regulation. This seesaw effect squeezed corporatist self-regulation in between and is the key to understanding the blurring of the social insurance systems. 10

11 4.3 Private Health Insurance Systems The logic of private health insurance (PHI) systems is characterized by extensive reliance on consumer demand and market competition as a mode of coordination (Hsiao 1992). In our sample of 23 OECD countries, Switzerland (until 1996) and especially the United States come close to the competitive market model. In private systems, the purchase of insurance is not mandatory, nor are the insurers obliged to accept all applicants. Insurance companies sell a variety of contracts that are contingent upon health status and also reflect consumer s willingness and ability to pay. Service delivery is mainly performed by private providers. This also means that healthcare is an industry, where private entrepreneurs are responding to the high mobility of capital (Tuohy 1999; White 2007). In the outpatient sector, physicians are generally self-employed and therefore mainly for-profit providers. While private healthcare systems in general score relatively high with regards to responsiveness and choice, their ability to achieve social objectives is limited (Nichols et al. 2004). Private health insurance markets suffer from market failure caused by adverse selection 7 and moral hazard 8 due to asymmetric information (Cutler and Zeckhauser 2000). If insurers on the other hand are able to classify risk thus avoiding the problems of adverse selection by riskrating premiums and by imposing (high) co-payments, this method renders health insurance unaffordable particularly for those population groups who most urgently need it, i.e. sick and poor individuals (Cacace et al. 2008a). Furthermore, private insurers intend to reduce uncertainty by excluding pre-existing medical conditions from the benefit package or by completely denying coverage for an applicant (Glied 2001). PHI systems thus cannot achieve full coverage of the population with a comprehensive benefit package. The result is an increase in the number of uninsured and underinsured persons, as experienced drastically in the US healthcare system. In addition, as long as financing agencies are just payers and do not actively manage service provision they are particularly prone to moral hazard behaviour. Private systems therefore tend to have high healthcare expenditures (Colombo and Tapay 2004). In fact, the US and Switzerland have the most expensive healthcare systems in the world, consuming in 2005 over 15 percent of the GDP and 11.5 percent, respectively (OECD 2008). As a consequence of efficiency pressure, hierarchical elements of regulation have increased in PHI systems over time. Switzerland even moved away from a private to a social insurance scheme by establishing mandatory insurance for all citizens and thus tackling the problem of non-insurance and underinsurance (Moser 2005). The system change was enabled by state interference in 1996 through the introduction of the New Law of Health Insurance (Krankenversicherungsgesetz, KVG). The main objectives of this law were to contain costs, strengthen solidarity and to provide access to high quality health services for every individual (Gerlinger 2003). Also in the US, public elements have been growing considerably. On the side of the government, public programs for the aged, the disabled and the indigent were enacted at a time when many elderly persons faced challenges finding affordable coverage within private 7 Adverse selection occurs when an insurer is not able to rate the health risk of the applicants properly. In this case he will calculate a premium based on average risk expectation. This premium will attract mostly individuals with relatively unfavorable risk structure. As a consequence, the insurer will be forced to increase the premium in the next period and more favorable risk groups will be further deterred. This process can repeat itself with the consequence of a break-down of the insurance market. At best, good risks accept co-payments and co-insurance thereby signalling that they are good risks. As a result good risks only get partial insurance. 8 Moral hazard refers to a change of behaviour once insurance exists. It includes reduced prevention (ex ante moral hazard) as well as over-utilisation (ex post moral hazard) and may happen in private as well as in social insurance systems or NHS types. If insurance (on part of the insured) and underwriting (on part of insurances) are not mandatory, however, it may even lead to the ruin of the market. 11

12 markets (Colombo and Tapay 2004; Marmor and McKissick 2000). With US Medicare, even a social health insurance program was implemented for the aged and the disabled. The considerable increase in public funding also entailed a substantial rise in hierarchical state regulation (Marmor 2000; Patel and Rushefsky 1999). Yet, in the realm of private insurance, government regulation remained weak especially vis-à-vis service providers (Stone 2000). In private insurance, vertically integrated private managed care organizations gained a foothold. Due to their hierarchical structure, they were able to impose instruments for steering patients and providers. Private managed care arrangements, therefore, operated as a functional equivalent to government regulation. Health Maintenance Organizations (HMOs) achieved considerable savings, especially in the first decade of their spread. As a consequence of the backlash against HMOs, virtually integrated provider networks emerged, giving some leeway for bargaining procedures in the interaction between insurers and service providers (Cacace forthcoming; Hsiao 2002). Managed care was introduced in Switzerland as well, yet in this case, it was a consequence of state legislation (KVG 1996) and therefore not initiated by private actors. Nevertheless, the managed care instrument was adopted in order to bring more hierarchy into the delivery system. On the side of service providers, the system logic of private, market-based healthcare systems would lead to the expectation that most providers are private for-profit. We find this assumption affirmed only for the outpatient sector, where most physicians are self-employed. In inpatient care by contrast, non-profit providers (public or private) prevail, as they are either part of the safety net (US) and/or because they are subsidized by the state (Switzerland) (National Center for Health Statistics 2007; Farsi and Filippini 2006). In the US, public inpatient care provision decreased as a direct consequence of the growth of managed care. Here, the government sold its municipal hospitals in light of increasing market competition (Getzen 2004). Functional privatization through outsourcing activities is advanced, especially in the US. For example, hospitals commission the service industry not only with catering and laundry services but also with management functions, such as human resources management and payroll accounting. Most of these highly specialized suppliers are private for-profit businesses. In coping with system deficiencies, hierarchical elements have increased considerably in private healthcare systems, emanating from the side of public and private actors as well. More collective financing too is observed, either through the introduction of social health insurance or by an increase in public funding. At the same time, profitization tendencies in the health service sector are clearly to be discerned, but are limited in their extent as the state subsidizes the non-profit industry. 5. Problem pressure, policy ideas and convergence In the previous section we have tried to show how efficiency pressure and the healthcare system as an intervening variable may contribute to explain increasing similarities across OECD healthcare systems. We have done this by reviewing healthcare systems in a stylized way, describing the respective dominant system logic, its deficiencies and responses to these deficiencies. Our explanatory model suggests that adaptive responses to system specific problems shape healthcare system change. Of course, adaptive responses to system deficiencies do not develop automatically. They may require decisions of policy-makers and subsequent implementation through local actors. Adaptation processes may also occur in the absence of policy reforms through actors empowered under the existing regime (Hacker 2004). 12

13 In the present form the explanatory model focuses on structural forces (i.e. problems and health systems), but does not specify different mechanisms of change and convergence. Searching for more efficient solutions, decision makers can recur on experience with existing national policies. Independent problem-solving (Holzinger and Knill 2005) which does not take into account developments in other countries can still lead to convergence if common problems evoke similar answers or as shown above if distinct answers to problem pressure increase the similarities between systems. However, since international mutual monitoring gives the opportunity to identify a range of regulatory alternatives and to estimate reform effects before implementation, forms of transnational communication will also have to be considered. The latter include (rational) policy learning from other countries or the common, transnational development of solutions as well as emulation or mimicking of policies of pioneer countries (Holzinger and Knill 2005; Powell and DiMaggio 1991). 9 Our explanatory model tries to include these mechanisms under the heading of ideas. Policy ideas can be understood as policy paradigms, i.e. a set of principles and causal beliefs which shapes the perception of problems and offers an acceptable set of potential solutions. Thus policy ideas serve as a route map and help to structure political decision making and policy learning (Béland 2005). In this sense ideas may support path dependent developments, since belief systems are rather stable. However, as paradigm shifts occur new ideas can induce change. Alternatively, policy ideas refer to the framing of problems in the political agenda setting process. Here, ideas are used as an instrument to generate public support (Béland 2005). Moreover, policies itself can be seen as ideas to cope with relevant societal problems. In this sense policy innovation in other countries and experience with policies offers a reservoir of policy options and a fundament of policy learning. 10 The following sections therefore aim to include policy ideas into the explanatory framework which focuses on problem pressure. Therefore we chose two regulatory instruments which have been successively implemented in many modern healthcare systems and for which some evidence of policy learning and diffusion can be found in the literature: Diagnosis Related Groups/Health Resource Groups and reference pricing of pharmaceutical products. Both instruments are also related to the role of the state in healthcare systems. Diagnosis Related Groups can be seen as a hierarchical instrument which supports the effectiveness of market- 9 Further mechanisms of policy convergence have been identified in the literature (Holzinger and Knill 2005): imposition, international harmonization and regulatory competition. Considering healthcare system change in our sample of 23 OECD countries the application of imposition, which refers to coercive policy transfer through supranational institutions or foreign countries and competition as mechanisms seem less plausible. International harmonization means that national regulations will have to comply with international or supranational law. Harmonization has some impact in the EU. Healthcare belongs to the responsibility of EU member states and by now there is no direct harmonization concerning the core of health systems, but spillovers from regulations in other policy fields will have to be considered (Greer 2006). Various regulations affect the regulatory dimension of health systems. Some examples are non-discrimination laws which have forced private health insurance companies to offer unisex rates. Further, the Third Directive of Non-life-insurance constrains government s influence on the design of PHI plans, e.g. rules which implement an obligation to contract for private insurance companies. The working time directive had major influence on hospital doctors and forced hospital to organize their workforce more efficiently. Small hospitals with low staff numbers and therefore less flexibility to organize working time may run into severe difficulties. Harmonization efforts through the open method of coordination, i.e. commitments of EU members to achieve common health targets, have not been fully implemented yet. The EU also influences national healthcare systems via negative integration, i.e. by ruling out national regulations as they are regarded as incompatible with European law, particularly free market obligations. So, in a number of subsequent decisions the European Court of Justice has encouraged cross-border purchase of medical goods and services (Obermaier 2008). 10 Apart from values and as a heavy influence on belief systems knowledge is a crucial factor. Due to changes in informational technologies nowadays knowledge about health technology and its assessment is known everywhere immediately. Of course a increasingly more common knowledge may lead to converging belief systems. 13

14 like mechanisms. Reference pricing also affects the role of the state, corporative actors and markets. Reference prices set constraints to market mechanisms either through state regulation or corporative actors. At the same time they involve less coercive power than fixed prices. 5.1 Diagnosis Related Groups The spread of Diagnosis Related Groups (DRGs) and comparable inpatient grouping systems across countries can be used as an example for policy diffusion and policy learning in healthcare. While in the mid-1980s DRGs were only used in the US Medicare system and partially in France and Canada, nowadays DRGs are implemented in most countries of our OECD 23 sample (France 2003; Fischer 2007). In this sense we may also consider the spread of DRGs as a convergent trend towards the use of inpatient classification systems though their use and implementation differs considerably from country to country (Lüngen and Lauterbach 2000). DRG-systems aim to classify hospital patients according to their case-mix (as measured by diagnoses and patient characteristics) and according to homogeneous treatment costs (Fischer 2001). Thereby DRGs define the hospital product and can serve as a basis for hospital financing schemes, hospital management, planning and utilization review (Rodrigues 1993). The observation that the costs of seemingly comparable hospital services differed remarkably triggered the development of DRGs in the late 1960s at Yale University (Pouvourville 2004). The first implementation occurred in the USA in New Jersey 1977 and 1983 in the US- Medicare program together with a prospective financing scheme. The main motivation for the implementation of the new financing scheme was to contain hospital expenditure (Coffey and Louis 2001). Up to 1983 hospital expenditure increased much faster than total healthcare spending and real growth of hospital costs since 1970 ranged between six and nine percent (OECD 2008). Hospitals were paid on a retrospective cost-reimbursement basis which involves negative incentives for cost containment (Cacace 2007). By contrast, prospectively defined payments by average cost per case force more expensive hospitals to cut down costs. Indeed, the introduction of DRGs and prospective financing schemes caused hospital reorganizations, made hospital managers more aware of economic thinking and a more efficient way to produce hospital services. Consequently, increasing specialization, outsourcing and reduced length of stay could be observed, while as a side-effect quality control schemes were developed to impede volume increases and quality reduction (Coffey and Louis 2001). In the following DRGs were implemented in France and Canada in the 1980s. In the early 1990s Belgium followed and several NHS countries such as Britain, Ireland, Portugal, Australia, Finland and Sweden. By 2000 Norway and Denmark had joined the other Scandinavian countries using a commonly developed DRG system. More recently in 2003 Germany implemented DRGs while Japan and the Netherlands as well as Austria (1995) developed comparable systems based on disease and treatment procedure rather than diagnoses alone (cf. Figure 2). The list of non-adopters among our OECD 23 sample includes only Greece, Luxembourg and New Zealand which stopped the program in 2000 (Forgione et al. 2005). DRG-systems are increasingly used to reimburse hospitals on the base of prospective financing (Lüngen and Lauterbach 2000; France 2003). Other priorities are observed in France where the definition of global budgets is based upon DRGs. Similarly Norwegian DRGs determine regional budgets whereas the mechanism of allocating funds to hospitals is left to the regions. Decentralized healthcare systems regularly use DRGs to account for cross-border 14

15 provision of care, e.g. in the Scandinavian countries and Italy. Belgium and Ireland aim particularly at incentives to the reduction of length of stay. Mostly, DRGs affect only part of the hospital budgets (e.g. Denmark, Portugal), a limited number of hospitals (e.g. Switzerland, Finland) or are not fully implemented in all regions (e.g. Sweden, Spain) (Lüngen and Lauterbach 2000). Figure 2: DRG-systems across OECD countries USA Yale-DRG 1977 USA HCFA 1983 Canada CMG 1983 France GHM 1986 France GHM 1997 France GHM 2008 Belgium AP-DRG 1990 Switzerland AP-DRG 2002 in some Cantons New Zealand AN-DRG 1998 USA AP-DRG 1988 USA APR-DRG 1991 Australia AN-DRG 1992 Belgium APR-DRG 2002 Australia AR-DRG1999 Germany G-DRG 2003 Switzerland Swiss DRG 2012 USA CMS 2001 USA R-DRG 1989 USA MS-DRG 2007 Italy HCFA 1995 Portugal HCFA 2001 Japan DPC** 2003 Ireland HCFA 1990 Portugal HCFA 1992 N-DRG 1996 Sweden 1997 Norway 1997 Finland 1998 Denmark 2000 Iceland 1998 Netherlands DBC*** 2003 Finland HCFA 1987; FinDRG 1995 Sweden HCFA/ AP-DRG 1992 Denmark Dk-DRG 2002 Great Britain HRG Great Britain 1992 HRG Austria LDF* 1995 Canada CMG/Plx 1997 Canada CMG Source: Adapted from Fischer (2007), Lüngen/Lauterbach (2000) and Erlandsen (2007) GHM: Group Homogène des Malades; N-DRG: NordDRG; Dk-DRG: Danish DRG; DBC: Diagnose Behandelings Combinaties; LDF: Leistungsbezogene Diagnose-Fallgruppen; HRG: Health Resource Groups; AP: All Patient DRG; HCFA Health Care Financing Administration; G-DRG: German DRG; AN: Australian National; AR: Australian National Refined; CMG Case Mix Groups; CMG/Plx Case Mix Groups with Complexity Overlay and Age Adjustement; DPC Diagnosis Procedure Combinations * based on treatment procedure ** Diagnosis and Disease-based groups *** Diagnosis and treatment-based groups While the main motivation for DRGs and prospective financing in the US was the desire to control costs, this aim was seen less relevant in NHS and SHI countries (Donaldson and Magnussen 1992). Although cost-reimbursement schemes for hospitals involved incentives to expand hospital services in NHS and SHI countries these systems more often managed to implement fixed budgets in order to impede excessive growth. Budgeting however involves no incentives to efficient service delivery and may cause rationing. In particular NHS had to suffer from waiting lists. Here, the motivation for introducing DRGs is to give hospitals incentives to act efficiently by encouraging them to (a) increase their activity, (b) increase their efficiency while (c) holding costs constant (Donaldson and Magnussen 1992: 60f.). Since hospitals receive case-based lump sums, there are incentives to deliver services more effi- 15

16 ciently, while budget control impedes excessive volume increases. The developments across systems may be interpreted as common efficiency pressure which translates differently into healthcare systems due to their institutional conditions to control costs through hierarchical regulation. While independent problem solving could have led several countries to develop case-based financing strategies, there is obvious incidence for diffusion and policy learning. A clear-cut incidence for policy diffusion and policy learning is simply given by the different DRG-types healthcare systems have implemented. The flow chart in Figure 2 tries to illustrate the links between countries and DRG types which are chronologically ordered. The dotted lines link systems which have been influenced by the related older system while the solid lines link further developments of DRG types within a country. Hence, the early French GHM are inspired by the US-HCFA system which has been influential in early adopting countries. Clear-cut links can also be drawn between the German G-DRG system and the Australian system, which had been chosen as a model (Lüngen and Lapsley 2003). Since the early 1980s there is also evidence for considerable transnational communication on DRGs and their use, such as conferences, common research groups, the cross-national development of common datasets etc. (Rodrigues 1993) 11. Further, the Scandinavian NordDRGs document a process of common problem solving and have been implemented successively in the Nordic countries. Conversely, Japan, the Netherlands and Austria have developed own DRG systems, which are rather based on disease and treatment-based classifications. Moreover, these systems aim to cover outpatient and inpatient services in order to maintain or foster the integration of health sectors. Why did DRGs diffuse? Gilardi et al. (2008) show that the probability to implement DRG based hospital financing increases with the ineffectiveness of the existing financing scheme and the relative success of DRG adopters. Moreover, some policy characteristics promote diffusion. DRGs offer a solution to problems across different healthcare systems and are flexible enough to be adjusted to different systems and different purposes. Since their development and implementation involves considerable costs common problem solving strategies as observed in the Scandinavian countries are attractive. Further, the implementation of this rather technical device is less prone to ideological concerns and party struggles. Indeed, only in New Zealand the DRG scheme and the purchaser provider split have been stopped subsequent to a change in government. The availability of promising policy ideas can also help to overcome national blockades. In Germany, corporate actors were commissioned to expand a case-based financing scheme for surgeries to further hospital services, but the responsible actors of the corporate health system failed to agree upon a treatment based grouping system. Consequently the government stipulated the introduction of a scheme which follows the methodology of AR-DRGs. The recurrence on international experience is therefore also related to the failure of the corporate health system regulation (Lüngen and Lapsley 2003). 5.2 Reference Pricing Since the 1980s OECD countries were facing the problem of fast growing public expenditures for pharmaceuticals. Between 1980 and 1990 public expenditures on drugs grew much faster than the GDP (OECD 2008, own calculation). As a consequence several OECD countries developed instruments for pharmaceutical price control (see Mrazek and Mossialos 2004). Germany introduced in 1989 a new policy tool called reference pricing (RP) which aims at 11 In 1985 the Council of Europe started a research program on DRGs, while the WHO supported the transformation of ideas through conferences on DRGs. The European Union supported DRG and case-mix related research and organized workshops aiming at the standardization of DRGs at the European level which however failed due to distinct national conditions (Rodrigues 1993). 16

17 containing costs and improving efficiency. RP schemes set fixed reimbursement limits for pharmaceuticals assigned to the same group. If the producer charges a price above the reimbursement limit, the consumer has to bear the rest. Thus, reference pricing is an indirect method of price control (Ess et al. 2003). There are several ways to implement a RP scheme in a specific healthcare system. First of all, schemes can be internal or external. Internal schemes compare product prices within a country, while external ones take the prices of other countries into account (Danzon and Ketcham 2004). Second, RP can be generic or therapeutic. While generic schemes cluster only identical off-patent drugs with the same active ingredients in groups, therapeutic clusters can also include off- and in-patent drugs which have similar active ingredients or a comparable effect (Puig-Junoy 2007). Thirdly, fixed reimbursement levels are mostly set at the median, the average of several lowest or the lowest price of a pharmaceutical group (Brekke et al. 2007). Since its introduction in 1989, Germany uses an internal RP scheme. In the beginning the scheme was mainly generic and 1996 the inclusion of in-patent drugs was even stopped. Within groups with the same active ingredient sickness funds maximally reimbursed the price at the first tercile (Schneeweiss et al. 1998). Just two years after RP implementation in Germany the logic spilled over to the neighboring Netherlands which also introduced an internal scheme. In contrast to Germany, the Dutch version used therapeutic groups including in- and off-patent pharmaceuticals. The reference price was the median of all drugs belonging to a specific cluster (Danzon and Ketcham 2004). While RP concentrated first on two central European SHI systems, in 1993 it spread over to the Nordic NHS countries. During one calendar year, Denmark (Vallgårda et al. 2001), Sweden (Gross et al. 1994), and Norway (Haga and Sverre 2002) introduced internal RP for generics and off-patent brand drugs. Two years later, Iceland followed with a quite similar scheme (Martikainen and Rajaniemi 2002). Even the logic for setting the reference price within the groups was quite similar: Denmark took the average of the two lowest priced drugs, the Swedish calculation includes the lowest available price plus ten percent, Iceland allowed only a surcharge of five percent, and Norway, which had to cope with sky-rocketing public expenditures on pharmaceuticals, reimbursed just the lowest price. In 1993 New Zealand also implemented an internal RP scheme as one part of their major NHS remodeling. The advisory board was inspired by health policy reforms in other Anglo- Saxon countries and the Netherlands (Jacobs and Barnett 2000). Following the Dutch experience New Zealand s policymakers chose therapeutic clusters, but instead of the median used the lowest price within a group as the reimbursement price (Danzon and Ketcham 2004). The Australian RP scheme, which started in 1998, shares many configurations with its New Zealand pendant (Lopez-Casasnovas and Puig-Junoy 2000). In the mid-1990s the idea of RP spilled also over to the North-American OECD member states. Although neither Canada nor the US introduced a comprehensive national RP scheme, it came into force in limited regions and programs. As Canada was facing the highest growth rates of public pharmaceutical expenditure in the OECD world, the provinces were searching for solutions to cope with this situation. In 1995 British Columbia (BC), the third largest Canadian province, introduced an internal RP scheme with therapeutic clusters and reimbursement limits usually fixed at lowest level. (Morgan 2008). Even some US healthcare programs introduced RP-like schemes. Today, nearly half of the states (including British Columbia s US neighbors, Washington, Idaho and Montana) make use of this instrument to control cost in their Medicaid program. While some of these states use a less rigorous calculation with the 17

18 lowest price plus 50 percent others tend toward stricter limits (Abramson et al. 2004). Even some private managed care plans only reimburse multi-sourced pharmaceuticals up to groupspecific ceilings (Brekke et al. 2007). While the idea of internal RP spread over the world, the Netherlands started in 1996 an additional external RP scheme. It allowed the Dutch Minister of Health, Welfare and Sport to set maximum selling prices on the basis of the average wholesaler prices for identical pharmaceuticals in Belgium, France, Germany and the UK. As the Dutch policymakers did not update the internal RP scheme, the external one became in fact the most important steering tool for in- and off-patent drug prices (Danzon and Ketcham 2004). Just one year later after the Dutch implementation, Ireland introduced an external RP scheme which reimburses either the UK wholesaler price or the average of Denmark, France, Germany, the Netherlands and the UK, depending on which of the two is the least expensive (Martikainen and Rajaniemi 2002). Figure 3: Implementation of reference pricing schemes in selected OECD countries Status Year of Internal scheme External scheme change (Cluster method and pricing) (Countries and pricing) Pilot 1989 Generic: Price at the first tercile None Germany Therapeutic: Weighted average Recent 2006 Therapeutic: Price at the first tercile None Netherlands Pilot 1991 Therapeutic: Median None Recent 1996 Therapeutic: Median Average of BEL, FRA, GER and UK Pilot 1993 Generic: Average of the two lowest None Denmark priced drugs Recent 2005 None Lowest price within the EU Sweden Pilot 1993 Generic: Lowest price +10% None Recent 2002 None None Norway Pilot 1993 Generic: Lowest price None Recent 2003 None Average of the three lowest prices of AUT, BEL, DEN, FIN, GER, IRE; NLD, SWE and UK New Zealand Recent 1993 Therapeutic: Lowest price None Iceland Recent 1995 Generic: Lowest price +5% None Canada (British Recent 1995 Therapeutic: Usually the lowest price None Columbia) Australia Recent 1998 Therapeutic: Lowest price None United States Pilot Mid Generic (in half of the States): Lowest None (Medicare) 1990s price +50% (or less) Ireland Recent 1997 None UK price or average of DEN, FRA, GER, NLD and UK Pilot 1998 None Lowest Price of EU-15 Greece Recent 2006 None Average of two lowest EU-15 and the lowest new EU member price Pilot 2000 Generic: Weighted average of the None lowest-priced drugs which account Spain 20% of market share Recent 2004 Generic: Average of the three lowest None priced drug Italy Pilot 2001 Generic: Lowest Price None Recent 2004 None None Portugal Pilot 2001 None Lowest price of ESP, FRA and ITA Recent 2007 Generic: Lowest price Average of ESP, FRA, GRC and ITA Belgium Recent 2001 Generic: Original brand price -26% None Austria Recent 2004 None Average of EU-15 for new and special approvable pharmaceutical Own depiction, italics refer to practically irrelevant instruments In the most recent expansion phase starting in the late 1990s, internal as well as external schemes have been used to implement RP in Southern Europe, in particular Mediteranean 18

19 NHS systems. While the bigger countries like Spain or Italy introduced internal RP schemes with generic clusters, the smaller like Greece and Portugal tended to external ones. The Greek external RP came into force in 1998 and was set at lowest price within the European Union (Yfantopoulos 2008). Portugal adopted the idea of external RP in 2001 and picked like Greece also the least pharmaceutical price but from a smaller country sample consisting of France, Italy and Spain (Stargardt and Schreyogg 2006). At the same time Spain and Italy finished the implementation of their internal RP scheme. Whereas the Italian NHS reimbursed only the lowest price among the off-patent groups (Folino-Gallo et al. 2008), Spain chose with the weighted average of the lowest-priced drugs that account 20 percent of market share a rather complex method (Puig-Junoy 2007). While Spain and Italy just started an internal generic RP scheme in the early 2000s, the Scandinavian countries abolished it as it did not fulfill the expectations. Norway and Denmark replaced it with an external RP scheme. Since 2003 Norway has calculated the average of nine EU member states (Dalen et al. 2006) and Denmark simply takes the lowest price within the EU (Strandberg-Larsen et al. 2007). Since 2004 Austria uses also EU-15 average prices for a special segment of the pharmaceutical benefit basket. Summing up, we see that over a period of nearly 20 years clusters of neighboring or similar countries (i.e. Nordic, Oceanic and Mediterranean NHS) synchronously introduced and changed RP mechanisms (see Figure 3) which is a hint for policy learning. According to our problem pressure argument, the necessity for reforms is mostly explained by fast growing (public) expenditures for pharmaceuticals especially due to medical technical progress. Pure internal RP schemes which are limited to generic clusters were mostly replaced or supplemented over time with internal RP including in- and off-patent drugs or external RP schemes. The latter even increased the role of bigger European pharmaceutical markets like Germany, France and the UK as their price changes directly influence other countries (Stargardt and Schreyogg 2006). Although we find similar developments within groups of countries, further research is needed to draw concrete links of policy transfer between the countries. Especially important would be research that could shed light on what happens in the moments in which the idea of RP spills over to a new country or group of neighboring or similar countries. Such research can shed fruitful information regarding how ideas spread and transform into policy learning leading to healthcare system convergence. 6. Conclusion In this paper, we have shown that convergence and the common trends we observed in the healthcare systems of 23 OECD countries can be explained by the reaction of distinct system types to problem pressure. In our examination of the changes for National Health Services, social health insurance systems and private health insurance systems, we demonstrate that it is this diversity, which finally ends in convergence amongst systems. Whereas our underlying modified problem pressure hypothesis explains why healthcare systems change in certain directions mechanisms of change and convergence are not specified. Therefore, we made a first attempt to explain how these changes take place and to include mechanisms of change into the explanatory model. One mechanism leading to convergence might be policy learning through mutual monitoring, as it offers a big advantage for policymakers: positive and negative effects of policy instruments can observed in other coun- 19

20 tries. As examples for policy learning and diffusion, we examined the spread of new developed reimbursement instruments: Diagnosis Related Groups (DRGs) for inpatient services and Reference Pricing (RP) for pharmaceuticals. After the first implementation in pioneer countries like the Untied States (DRGs) and Germany (RP) the basic idea of these policy instruments flew over the OECD world including all three healthcare system types, as policymakers expect that both help to contain costs and to improve efficiency. While some countries may just emulate influential pioneer countries, we find also signs for rational policy adaptation and transnational problem solving. Convergence occurs in this case as a common trend of integrating innovative reimbursement methods leading to similar regulatory principles regarding the interaction of insurers and providers. The spread of DRGs and RP is likely be fostered by program characteristics since both seem to be very technical instruments which are difficult to understand for the public and which do not directly collide with dominating norms and values. Therefore, the potential of ideological conflicts constraining the adoption process is rather low. Further, the development and implementation of both instruments involves considerable costs which can be reduced through common problem solving or by recurring on the experience of pioneer countries. Considering policies with a high potential of ideological conflict, policy learning and diffusion is likely to be more dependent on institutional constraints, the configuration of relevant actors and policy ideas in the sense of core beliefs and values. Concerning instruments such as gate keeping, priorization lists, flat rate contribution schemes, competition and mandatory insurance which affect core beliefs of choice and justice, policy makers will have less leeway to learn from others. 20