Fixing Federal Faults. Complementary Member State Policies in Swiss Health Care Policy

This article aims to understand what makes member states complement federal healthcare policy beyond the instruments planned by federal policy. We employ a Multiple Streams approach to study how Swiss member states use their discretion in order to complement federal healthcare regulation with the aim of decreasing outpatient healthcare expenditures at the cantonal level. Based on a written survey in the Swiss cantons, we perform a Fuzzy Set Qualitative Comparative Analysis (fsQCA), which places a keen emphasis on complex patterns. The method identifies what combinations of determinants make it particularly likely that a canton opts for complementary policy activity. Several configurations prove to foster such activity. While this is important, it is also important to pay attention to the constellations that precisely do not foster complementary policy activity. Our analysis of the cantonal choices on governing outpatient healthcare reveals that party politics in the executive and/or the public administration play a major role in this task, whereas neither organized interests within the medical profession nor individual policy entrepreneurs are crucial. Federalist systems offer opportunities for policy innovations the federal level ultimately may benefit from.


Introduction
This article aims to understand what makes member states complement federal policy beyond the instruments planned by the federal policy. The common narrative in member state implementation research is deficit-oriented. Compliance research in the EU and other multi-level systems asks whether member states' policy execution is in concordance with the central state's regulations (Thomann & Sager, 2017a, b). The basic implication of this narrative is that member states by default deviate from the central state directive they are supposed to implement. In this contribution, we want to complement this one-sided view of member state implementation by taking the Swiss case to show how member state discretion can have a positive effect on public service delivery by allowing member states to complement federal regulation. The cantonal policy reaction to federal policy, thus, is not part of actual member state implementation of federal policy but is an additional and deliberative political decision beyond over-compliance or gold-plating as observed in compliance research (Thomann, 2015(Thomann, , 2019. The specific example we selected is that of outpatient healthcare, as stronger regulation in this largely privately organized sector is seen as a means of controlling healthcare costs. Outpatient care, or, synonymously, ambulatory care is medical care that does not require a stay of more than 24 hours in the care delivering facility. As the Organization for Economic Cooperation and Development (OECD) and the World Health Organization (WHO) (2011, p.43) put it: "The ambulatory care sector provides general medical advice, diagnostic services, obstetrics care, perinatal care, care for children, family planning, minor surgery, rehabilitation, dental care and home-based care". Ambulatory care is largely provided by independent physicians and so-called outpatient departments of hospitals.
Since the early 1980s, there has been growing pressure to reform the Swiss healthcare system. The main problem was, and still is, the continuous increase in healthcare costs. However, due to institutional veto points, health policy reform with regard to governance of the outpatient healthcare sector is largely deadlocked at the federal level. We argue that the way out of this deadlock is the cantons' discretion in policy making. In the Swiss healthcare governance framework, the cantons' powers to intervene in the domain of outpatient healthcare are very limited. Nevertheless, there is room for manoeuvre because, as is often the case in federal settings, the extent to which the Swiss cantons make use of their discretion and the strategies they apply differ.
In this paper, we study how the Swiss Confederation's member states, the cantons, use their discretion in policy making in order to tackle the problem of rising healthcare costs within the limits of national regulation. Given how difficult it is for Switzerland to further regulate outpatient healthcare at the federal level, it is of particular interest to examine what determinants foster the development of complementary healthcare policies at the subnational level. Therefore, the article answers the following research question: Which factors drive cantonal governments to govern outpatient healthcare beyond the instruments planned by federal policy?
More concretely, we seek to understand what factors explain the extent of cantonal activity in regulating a sector in which cantons can complement federal policy. We employ the Multiple Streams Framework (MSF) as our overall theoretical approach to answer the given research question. The MSF provides a realistic approach to policy decision-making as it assumes the coupling of different streams rather than a monocausal explanation for policy decisions. Recent research has shown the MSF to be well suited to understand policy implementation deci-sions (Sager & Thomann, 2017), which is why we consider it apt to help us understand complementary policy activity beyond actual implementation.
The research design consists of a comparative analysis of the Swiss cantons as fully fledged political systems. Based on a written survey in the Swiss cantons, we perform a Fuzzy Set Qualitative Comparative Analysis (fsQCA), which strongly emphasizes complex configurational patterns. The method allows us to focus on what combination of determinants make it particularly likely that a canton opts for complementary policy activity in the regulation of healthcare. Several configurations prove to foster complementary activity. While this is important, it is also necessary to pay attention to the constellations that precisely do not foster complementary activity.
The paper is structured as follows: First, we provide an overview of both Swiss federalism and the Swiss healthcare system. Second, we describe the phenomenon we aim to explain -the complementary cantonal policy activity in the field of outpatient healthcare. Third, the conditions favoring such governance are derived from MSF informed theory and presented. Fourth, we explain the approach and the particularities of the method employed. Finally, we present and discuss the results of the systematic comparison, as well as the lessons that may be learned from the Swiss case.

Swiss Federalism, the Swiss healthcare system and the case of outpatient healthcare
In Switzerland, federalism is one of the most distinctive features of the political system. Cantons are fully-fledged political systems in their own right and primary political units that enjoy policy making responsibilities in important domains such as health, education, justice, police and social assistance. Most federal programs are implemented by the cantons, as they are not only executing, but also programming authorities (Sager & Zollinger, 2011, p.31;Sager et al., 2017, p.35;Sager & Thomann, 2017).
Thus, in Switzerland, there are principally 26 cantonal healthcare (sub)systems linked by a national health insurance system financing most health care services (Okma et al., 2010, p. 93;De Pietro et al., 2015;Rüefli, 2020;Trein 2018Trein , 2019. However, the cantons' autonomous governance powers in health care concern almost exclusively the domains of hospitals and nursing homes (so-called inpatient care). In outpatient care, cantons lack financing competences. Their regulation competences in this domain are basically limited to the authorization and supervision of health service providers. Furthermore, they can provide financial incentives to attract service providers and "soft measures" to influence service provision (Vatter & Rüefli, 2003;Wyss & Lorenz, 2000;. Service provision in outpatient care is -within the framework of the Federal Health Insurance Law (HIL) -governed almost exclusively in a corporatist setting by non-state actors (De Pietro, 2015, 164ff.;Rüefli, 2020). Private health insurance providers are obliged to conclude tariff agreements with all service providers who fulfil the requirements to practice within the framework set by the HIL. Thus, de facto, there is a contracting obligation (Paris et al., 2010, p.15). Principally, all insured persons have unlimited access to all physicians working in private practice and have freedom of choice. Due to these circumstances, in Switzerland, physicians in private practice have not only a great deal of autonomy, but also strong political influence (Immergut, 1992;Groenewegen et al., 2002, p.209).
Since the early 1980s, there has been growing pressure to reform the Swiss healthcare system. The main problem was and still is the continuous increase in healthcare costs. In the period observed in this article, the domain of outpatient healthcare grew particularly fast. Between 1996 and 2010, it was the cost item with the highest average annual growth rate (4.5%) in the healthcare system (Federal Statistics Office). Another issue that has been the subject of intense debate is the development of primary care. While there is an oversupply of doctors -particularly of specialists -in urban regions, rural regions face a growing undersupply of general practitioners (Jaccard Ruedin et al., 2007;OECD & WHO, 2011, 88ff.). This undesirable tendency is another motive for more governance in outpatient healthcare, with the objective of promoting general medical practice.
For many years already, public authorities at the Federal level have made various attempts to intervene in the outpatient sector in order to control cost (see Sager et al., 2010, 13f.). However, most projects aimed at reforms in the outpatient sector have either been declined by Parliament or in popular votes. Between 1996 and 2014, the only intervention at Federal level affecting the outpatient sector that passed Parliament was a restriction in the admission of outpatient service providers (see OECD & WHO, 2011, p.91;Rüefli, 2004;Rüefli et al., 2016). As a consequence of this political deadlock at the national level, interest in the cantons' potential to regulate outpatient healthcare increases. In the following section, we discuss the opportunities for cantons to govern outpatient healthcare.

The phenomenon to be explained: Complementary policy activity of outpatient care by the cantons
As is customary in federalist states where the member states are sovereign political units with a strong autonomy (Sager, 2007(Sager, , 2009, the degree to which the cantons make use of their discretion varies. We focus on three specific types of policy instruments the cantons can apply within the Swiss health care system's governance structure in order to intervene in the outpatient healthcare sector should they choose to do so (Sager et al., 2010, 29ff.;Saltman & Busse, 2002;Simoens & Hurst, 2006). We employ a blend of different taxonomies to categorize the instrument type mixes  chosen by the cantons. First, we employ Howlett's (2005) distinction between substantive and procedural instruments to separate content from process. Our first two instrument types regard substantive instruments while the third assembles procedural instruments. Second, we employ Vedung's (2011) typology of carrots, sticks and sermons to differentiate the substantive types. We separate binding regulation, i.e. Vedung's (2011) sticks, from financial incentives, i.e. Vedung's (2011) carrots. Sermons, i.e. mere provision of information and persuasion, do not play a measurable steering role in the regulation of outpatient healthcare in the cantons. The distinction also resonates with Hoods' (1983) NATO typology, concretely the A and the T: binding regulation regards authority-based (A) instruments and financial incentives regard treasure-based (T) instruments.

Binding Regulation:
The cantons are responsible for regulating and supervising the conditions of admission to professional practice and to the health insurance system, and have the power to set requirements concerning qualifications or equipment, quality standards and the like. One important regulatory instrument with regulatory discretion is the competence to restrict the admission of medical practitioners to the health insurance system. This instrument was introduced by the Federal government in 2003. This licensing restriction gave the cantons an instrument to control the number of physicians in private practice and other types of outpatient care providers, enabling them to counter both over-and undersupply in certain regions or in certain medical disciplines. Within the rules defined at the federal level, the cantons are free to adjust this instrument to their specific situations and needs (Rüefli, 2004;Rüefli et al., 2016).

Financial instruments:
Cantons can also intervene in the outpatient sector within the boundaries of their financing powers. They do this directly in the domain of home nursing and indirectly in the hospital sector (outpatient units in public hospitals). Furthermore, the cantons are free to create financial incentives. They can, for example, contribute financially to medical provider networks and to innovative projects in quality assurance, or promote the opening of medical practices in undersupplied regions via loans or by providing practice rooms.
Procedural instruments: Cantons can use "soft measures" in order to attempt to influence the behavior of individuals or organizations. In this approach, public officials act as facilitators, mediators, or initiators who set stimuli, bringing together private actors and motivating and/or empowering them to act in a certain way. They might for instance initiate and facilitate processes that lead to new models of service provision developed and implemented by the service providers, or they might form working groups with hospitals and private practitioners to establish new collaborative organizational models for emergency care. An interesting feature of these instruments is that most of them do not require an explicit legal foundation and are thus not necessarily subject to political decision processes, but can rather be applied by the administration itself.
In the research design section, we propose an index in order to measure cantonal complementary policy activity in outpatient health care and describe to what extent the cantons make use of the different instruments. We measure complementary policy activity by the number of policy instruments employed in each canton.

Theory: Conditions for complementary policy activity in the outpatient sector
In the following, we treat the question of how the differences between the cantons in complementary policy activity in the outpatient healthcare sector can be explained. We employ the Multiple Streams Framework as our overall approach. Sager & Thomann (2017) show that the MSF is well suited to understand implementation decisions in federalist systems. Zahariadis & Exadaktylos (2016, p.62) view "policy outputs as constituting implementation windows". Accordingly, the federal policy decision can be considered an implementation window launching additional political decision-making processes at the level of the constituent states (Sager & Thomann, 2017). In our case, we do not study implementation of the HIL but complementary policy activity. We consider reform failure at the federal level rather than a federal policy decision or window of opportunity for complementary policy activity at the cantonal level. As stated above, all reform projects at the federal level between 1996 and 2014 failed except for one. We take this political deadlock at the national level as a political opportunity for the cantons to become active in outpatient healthcare policy. We therefore consider MSF an adequate approach to answer our research question.
We take advantage of Herweg et al.'s (2015) extension of Kingdon's (1984) original model as applied to policy learning by Kuenzler (2018). Herweg et al. (2015, 444ff.) propose splitting the stages of agenda-setting from decision-making. Kingdon's (1984) original analytical interest was agenda-setting. However, policy studies mainly use MSF to explain decision-making (Zahariadis, 2014). Herweg et al. (2015) state that the second coupling follows from the first, i.e. when "a policy proposal has reached agenda-status (…) the window for the second coupling opens automatically. Politicians, interest groups and other relevant actors do not want to miss the opportunity of influencing a choice that is up for decision and start negotiating immediately" (Kuenzler, 2018, p. 271-2). The two decisive streams are the political stream and the entrepreneur (Herweg et al., 2015, p.445). Zohlnhöfer et al., 2016(also Zohlnhöfer, 2016 illustrate their approach in single cases. To our knowledge, however, Kuenzler (2018) provides the only comparative test of the approach. We follow this example with our focus on the decision-phase. As stated we consider the policy window to be open due to the political deadlock at the federal level. We consider the policy stream to be constant because it is determined by the implementation obligation of the HIL as a constant policy context. We made this choice for empirical reasons in order to restrict the number of conditions. However, we argue this choice is justified with respect to the limited scope in outpatient care regulation provided by the HIL as shown above in the presentation of the Swiss healthcare system. Accordingly, we test the combined effects of the following five conditions: the politics stream captured by party majority in government (first condition) and the presence of organized interests (second condition); the institutional setting of the coupling process as proposed by Sager & Rielle (2013) measured by size of the cantonal bureaucracy (third condition) and its degree of centralization (fourth condition); and the presence of policy entrepreneurs (fifth condition). (2013), we focus not on mono-causal explanations, but on the settings of conditions to explain complementary policy activity. We therefore use the theory to identify relevant conditions and formulate directional expectations rather than mono-causal hypotheses (Schneider & Wagemann, 2012). What we are interested in is how these conditions interact in specific settings.

Politics Stream:
Our first condition regards partisan politics. The "partisan theory" (Hibbs, 1977) attributes particular importance to the party-political composition of parliament and government when explaining differing political results. The instruments identified in the cantons are mainly in the competence of the executive rather than the legislative. We therefore consider the government's composition. Swiss cantons have coalition governments with parties from different ends of the ideological spectrum sharing executive power (Vatter & Milic 2019). Comparative welfare state research emphasizes the importance of a strong left-wing for state intervention preferences, which also has an effect on health policy. With their support for a strong (social) state, the political Left also takes a more positive stance with regard to active health policy than market-liberal forces do. We therefore expect a high proportion of left-wing parties in a cantonal government to be a favorable condition for complementary policy activity.
Theories of power resources explain the degree of state activity with the strength of vested interests. Because of their sovereignty in wage agreements and their high degree of political organization, doctors are able to block health policy decision-making processes aimed at stronger regulation of their activity, which limits the scope for regulation in this sector (Immergut, 1992;Groenewegen et al., 2002, p.209). We therefore expect the strongly organized interests of the medical profession to be an unfavourable condition for complementary policy activity.
Institutional Setting: Sager & Rielle (2013) have both argued and empirically shown that the inclusion of institutional and organizational factors are crucial to understand the actual coupling of the streams. As Lodge (2003, p.163) states, "[o]ne of the basic tenets of the statist and historical institutional literature has been its emphasis on the internal organization of the state, particularly the political-administrative nexus." One such dimension is size. The number of roles assumed by an organization is indicative of the capacity of that organization to initiate policies, develop alternatives and ultimately to implement decisions (Egeberg, 2003), i.e. ca-pacity must be considered an important condition: A larger administration can become active because it has the capacity to do so. Thus, it can be expected that larger cantonal administrations are more likely to intervene actively in the outpatient sector than smaller ones.
Another important dimension relates to the administration's horizontal specialization. Those areas dealt with by the same organizational unit can be expected to be more likely to be coordinated than those dealt with by different units (Egeberg, 2003). Cantonal administrative structures differ to a large extent with regard to hospital planning and outpatient healthcare regulation in that in some cantons the two tasks are centralized in the same unit, while in others they are in the competence of the same department, but in different units or located at different levels. While there is much more regulation experience in hospital planning than in outpatient healthcare, simplified access to information and experience regarding cantonal hospital planning are likely to have a positive effect on regulation in the outpatient sector. We therefore take the responsibility for outpatient care and hospital planning in the same unit as our fourth condition favouring complementary policy activity.

Policy entrepreneurs:
Entrepreneurs take an important place in Kingdon's framework. Kingdon (1984, p.188) describes policy entrepreneurs as "advocates who are willing to invest their resources -time, energy, reputation, money -to promote a position in return for anticipated future gain in the form of material, purposive, or solidary benefit." With the current distribution of power between the parties in mind, the entrepreneur tries to forge a coalition for his or her pet proposal (Herweg et al., 2015, 445f.). If his or her efforts are successful, the policy is formally adopted. Policy entrepreneurs can be found within as well as outside the administration. Both ways, their commitment can be expected to make a difference in favor of complementary policy activity. The presence of a policy entrepreneur in the canton who strongly supports regulation in the outpatient sector is our fifth condition. In the next section, we present our empirical approach.

Research design, method and operationalization
The research approach taken in this study is a standardized comparison of cantonal health policies. We systematically compare the 26 cantons, whereby the role of single conditions is explored not as to their isolated, but their combined effects. Accordingly, the selected method does not identify mono-causal explanations using the ceteris paribus principle, but rather takes account of the complexities of the political reality and, correspondingly, does not control interaction between the various variables, but instead explicitly takes account of them in the form of various combinations of factors with the same effect. To this end, in the empirical analysis the fuzzy-set QCA (fsQCA), a technique stemming from the method family of QCA (Qualitative Comparative Analysis) is used (Ragin, 2009). The original QCA method, developed by Ragin (1987), is based on Boolean algebra, i.e. the logic of the binary system with the two logical operators 'or' (+) and 'and' (*). The basic idea is that if an outcome D is found in a case displaying A, B and C as well as in another case displaying A and C, but not B, it obviously does not make a difference for the occurrence of D whether B is present or not. The respective formula is: A*B*C+A*b*C→D, which accordingly can be minimized to A*C→D. The notion of multiple conjunctural causation is crucial in this respect, meaning that different paths can lead to the same outcome. Complexity includes also the notion of asymmetric relationships, i.e. if the presence of a feature can explain an outcome, its absence does not necessarily imply the opposite outcome. Accordingly, we will analyze both occurrence and non-occurrence of complementary policy activity.
We employ the fuzzy set version of QCA. Fuzzy-set QCA works with so-called 'set memberships'. In addition to complete membership in a set (Fuzzy-value 1) and complete non-membership in a set (Fuzzy-value 0), various degrees of partial membership are possible. The crucial distinction is the crossover point at 0.5, which establishes the qualitative difference in kind between degrees of non-membership and degrees of membership in the set. A Fuzzy-value of 0.8 for instance shows that a case is located more within than outside a set, without fully belonging to it (Ragin, 2009). "In the process of set calibration (…), it is particularly crucial to specify qualitative anchors (the set membership scores of 0 and 1 in csQCA and the scores 0, 0.5, and 1 in fsQCA)" (Schneider & Wagemann, 2010, p.404).
The phenomenon to be explained is the degree of the canton's complementary policy activity in outpatient healthcare. Because up to 2009 a systematic overview of the canton's governance activities in the domain of outpatient health care was lacking, in a first step it was necessary to gather data to describe the cantonal regulation regimes. To this end, we conducted a survey (Sager et al., 2010) in which in a standardized questionnaire the appropriate decision-making and implementation bodies in each Swiss canton were asked to describe a number of features concerning their governance activities in outpatient health care. We collected qualitative information on the governance instruments in use (Which of the instruments mentioned above were applied to what extent? Which categories of health care providers were affected? Are there governance activities at local level?) as well as on administrative features and the existence of policy entrepreneurs describing the organization of cantonal governance activities. Gaps in survey data were filled with additional research via public records and the cantonal administrations' websites.
In a second step, in order to operationalize complementary policy activity, we transformed the qualitative data obtained into a standardized index, expressing to what extent a given canton makes use of the various policy instruments described above. We define activity as the degree of policy intervention in outpatient healthcare. We assume higher activity the more instruments we find in a canton. This means we do not differentiate between instruments but sum up the number of policy instruments employed in each canton. The information regarding policy activities concerning the different categories of service providers was transformed into points according to the coding rules presented in Table 2. The maximum value, representing the strongest degree of activity, that can be obtained in this coding scheme amounts to 8 points. To calculate the complementary policy activity value, the points are then divided by this maximum. The result is a value between 0 (no activity) and 1 (maximum activity).  Table 3 summarizes the conditions and the cantonal policy activity of outpatient care and shows the data sources used to measure these values. As for the conditions 'shared responsibility' (RESP) and 'policy entrepreneur' (ENT), we collected the data by way of the survey. We directly asked whether outpatient care regulation was located in the same organizational unit as the one responsible for hospital planning to measure shared responsibility, and we asked whether there were outstanding individuals engaging themselves for outpatient healthcare regulation to measure the existence of policy entrepreneurs. The cantonal raw data can be found in Table 7 and the calibrated data in Table 8 (see Appendix). The calibration of the conditions is presented in the supplementary online appendix.  Ragin (2009, p.111). Based on the theoretical considerations in Table 1, we make directional expectations about the outcome which would be produced by those combinations of conditions which are not empirically observed (logical remainders). The truth tables and the complex and parsimonious solutions are listed in tables 10 and 11 in the Appendix. As criteria for setting the raw consistency threshold, we looked for "gaps" in the raw consistency values, for low proportional reduction in consistency (PRI) values, and checked whether the latter indicated the presence of a contradictory case with qualitatively different membership scores in the configuration and the outcome set (Schneider & Wagemann, 2012, p.123).

Results: Complementary policy activity in the cantons
The results of the survey show that numerous cantons display initiative and use their room for manoeuvre within the existing federal regulation setting in combination with the HIL to develop innovative solutions. The variety of cantonal regulation arrangements is complex. 1 Table  4 illustrates the heterogeneity with respect to governance, but also with respect to population and the situation in outpatient health care. The highest value found for complementary policy activity is 0.69, the lowest 0.13. The mean value is 0.34, standard deviation 0.15. The average level of complementary policy activity is thus rather low, with only two cantons reaching a value surpassing the theoretical average of 0.5. The main regulation instrument employed by the cantons is the license restriction for new service providers. While all the cantons made use of it at the time of the survey, its specific design and the modalities of application varied greatly among the cantons. All cantons restricted system access for physicians, while only a minority of cantons applied the instrument to regulate access for other types of service providers as well. Financial instruments were applied by a small number of cantons in order to influence outpatient hospital departments or in home nursing. Only five cantons supported municipalities' regulation efforts, most of them only in the weak form of consulting. Four cantons actively promoted managed care with different measures. In three other cantons, activities were in preparation. We found more activities aimed at fostering cross-sectoral integrated care, in most cases consisting in the reorganization of emergency care-structures. Other reported activities concerned the organization of multidisciplinary integrated care centres or networks.
We could not identify specific patterns of diffusion or coordination among cantons, nor did we find any systematic correlation between certain patterns of regulation and the degree of application of other types of instruments. There is, in conclusion, a broad variety of governance activities and complementary policy activity, which needs further analysis and explanation. We do so in the next section.

Explaining complementary policy activity: Findings of the QCA
No necessary conditions for neither high nor low complementary policy activity were detected (cf. Table 9 in the Appendix). The sufficient solution for a high degree of cantonal outpatient policy activity that cannot be further minimized by the applied algorithm is displayed in Table  5. The single case coverage refers to the cantons that have a membership > 0.5 in the respective path. The raw coverage expresses the share of the outcome that is covered by a single path, while the unique coverage indicates how much a path uniquely covers (Schneider & Wagemann, 2012, p.139). The arrow "�" indicates that the conditions are sufficient for the outcome, i.e. these combinations always lead to the outcome.
Accordingly, we find four paths that typically lead to high complementary policy activity. In a first path, a low proportion of left-wing parties combines with weakly organized interests. This is the case in nine cantons. In a second path, covering five cantons, the political left is weak, there is no policy entrepreneur, and the responsibility is shared in one administrative unit. A policy entrepreneur is also lacking in the third path, which combines with a high proportion of left-wing parties and a large administration in four cantons. Finally, in only one canton the left is strong, interests are strongly organized, responsibility is shared in one administrative unit, and a policy entrepreneur exists. Contradictory case (membership in path > 0.5, membership in outcome < 0.5). The raw consistency threshold was set at 0.84. The next highest consistency score is 0.80 and the row contains only one case, with membership of less than 0.5 in the outcome (TI 0.53, 0.34). Prime implicant: ADM*ent. This produces the more parsimonious intermediate solution.

Source: the Authors
The first path covers nine cantons in which a left minority in the government and weakly organized physicians lead to a strong regulation of outpatient healthcare. The second path covers five cantons in which again a weak left in government, this time together with centralized responsibility in the administrative organization and the absence of policy entrepreneurs, lead to high complementary activity in the regulation of outpatient healthcare. Both of these paths find an explanation in the responsiveness of middle and right parties in government in spite of these parties' programmatic preference for competition and market mechanisms. The will to accept factual constraints of political reality may be more explicit in the executive than in the legislative. While in the first path organized opposition from the medical profession is lacking, in the second path, complementary policy activity is arguably pushed by a centralized administration that makes up for the lack of a policy entrepreneur.
The third path covers four cantons where a strong left in the government together with a strong administration and absence of an entrepreneur lead to high complementary policy activity. This path fully complies with the expected effects for the three factors that are corroborated in the combination observed in the three cantons. The fourth path only covers the canton of Fribourg, where we find again a strong left together with a strong organization of the medical profession, a centralized administrative structure and the presence of a policy entrepreneur. The case is interesting as it is the only case where a policy entrepreneur plays an active role. Whereas the combination of a strong left in government and the organizational factor of centralized responsibility conforms to the effect found in path three, we see the effect of the policy entrepreneur in this case as counterbalance to the political influence of the vested interest. We argue that in Fribourg it was the policy entrepreneur in the person of the responsible minister who promoted and ultimately enabled the complementary policy activity in the realm of outpatient health care in the course of a difficult hospital reform. Figure 1 illustrates the good consistency and excellent coverage of these results. In fact, only one case of high complementary policy activity could not be explained, specifically, the canton of Glarus. Furthermore, Aargau is the only contradictory case. Here, a weak leftist government with weakly organized interests has led to low complementary policy activity. In this canton, the left was too weak in government to take advantage of the weak organization of the medical profession. We now turn to the two paths that are sufficient for low complementary policy activity, as illustrated in Table 6. Following the Enhanced Standard Procedure (ESA) (Schneider & Wagemann, 2012, 200ff), we avoided untenable assumptions during minimization: specifically, logical remainders that contain one of the paths of the intermediate solution for STEER were excluded from the analysis of steer. The raw consistency threshold was set at 0.885. The next highest consistency score is 0.851 -this and all other rows with raw consistency > 0.75 cover only cases with membership of less than 0.5 in the outcome (exception: GL (0.53, 0.69) in row 8).

Source: the Authors
In a first path covering two cantons, the political left is weak, organized interests are strong, and a policy entrepreneur exists. The constellation indicates that in the two cantons of Ticino and Grisons with a weak left in government, well organized physicians' interest groups offset policy entrepreneurship. The second path covers two cantons where the left is strong, organized interests are weak, the administration is small but responsibility is shared in one unit, and no policy entrepreneur exists. The main reason for low complementary policy activity in this constellation lies with the weak administration that does not take up potential inputs from the left in government and from the absence of a policy entrepreneur who could provide such inputs. The weak opposition in form modestly organized vested interests do not make a difference in this constellation nor does bureaucratic centralization of responsibility.
We can see in Figure 2 that contrary to the solution for high complementary policy activity, these results can only explain four out of eleven cases with low complementary policy activity. The low coverage of these results indicates that while the theorized factors have an excellent ability to explain high complementary policy activity, other explanations must exist for low complementary policy activity that were not considered in our assessment.  Table 1.
Cases situated above the diagonal are consistent with the statement of sufficiency. Cases in the upper left quadrant are deviant cases for coverage. Cases in the lower right quadrant are deviant cases consistency in kind. Cases in the lower left quadrant are not directly relevant (Schneider & Wagemann, 2012, 67ff, p.308).
In the next section, we discuss our empirical results in the light of theory.

Discussion
The most eye-catching result is that the political condition "strength of the Left in the government" proved to be a decisive explanatory factor. However, the direction of its influence depends on how it is combined with other factors: both strong and weak leftist parties can lead to both high and low complementary policy activity, depending on the circumstances. This can be interpreted such that while the decision in favour or against regulation activities in the outpatient sector is partly taken at the meta-level of 'high politics' in the cantonal governments, a closer look at the meso-level of the administration and its working environment is necessary to understand the role of party politics. The strength of the target group of the regulation, the physicians, also plays an inconsistent role for high complementary policy activity. In the majority of the cantons with high complementary policy activity, weak organized interests fail to counter public budget sensitivity of the right majority in government (path 1). In only one case, the canton of Fribourg, do strong organized physicians face strong outpatient healthcare regulation because they were offset by a policy entrepreneur in the person of the responsible minister, who was a strong proponent of outpatient health care in the context of a large hos-pital reform in the canton. In the first path for low activity, we find the inverse constellation when, in the absence of a strong left in government, the organized interests trump the policy entrepreneur. While the effects of organized interests and policy entrepreneurs interact, the strength and organization of public administration sets the stage for and impacts complementary policy activity both as moderator and policymaker. In two paths for high complementary policy activity, centralized responsibility in the administrative structure enables the government to develop and implement outpatient healthcare regulation (paths 2 and 4); in another path (path 3), a strong administration, together with a strong left in government, establishes complementary outpatient healthcare regulation. In the second path of low complementary policy activity, the two conditions interact in that organizational centralization does not make up for a weak administration.
Based on our findings, we state firstly that politics is key. The role of party strength in the executive is ambiguous; both left and right in government decide for complementary policy activity depending on the setting. Second, organized interests and policy entrepreneurs interact and, again, it is the circumstances that are decisive as to whether one or the other wins. Finally, the size and capacity of a cantonal administration have shown to be key factors for stronger policy activity in the outpatient sector. While centralization of responsibility sets the stage for enabling regulation, a strong administration acts as policymaker in negotiation with interest groups and entrepreneurs. The last two findings underline the importance of meso-level politics in the decision-making and implementing processes regarding regulation in the studied outpatient sector.

Conclusion
In this study, the example of outpatient healthcare was chosen to illustrate how cantons use their discretion to adapt federal law and, more importantly in the present case, to complement it with additional policy activity. The configurative analysis with fuzzy-set QCA leads to the key finding that the decision for or against policy activities in the outpatient sector is taken at the level of cantonal party politics in the executive, but can only be understood taking into account specific constellations of factors primarily at the administrative level, but also with regard to organized interests and political entrepreneurs. Thus, a strong administration with adequate organizational structures is best equipped to identify problems, develop solutions and negotiate these with the relevant organized interest groups.
Due to their veto power, the actors concerned -namely physicians -are able to politically block reform attempts at both the federal and the cantonal level. There are examples of cantons making use of their discretion to influence outpatient healthcare provision, setting up governance structures which actively include stakeholders. An example at the Federal level is the master plan for promoting primary care (Federal Office of Public Health [FOPH], 2013), which was developed in cooperation between the Federal level, the cantons and medical associations, and relies mainly on procedural policy instruments that promote specific professional training and development, research, or the development of innovative models of service provision.
There are various implications of these findings for the case of reform in federal governance settings. First, as Okma et al. (2010, p.104) put it, "reforming healthcare systems is not a one-shot effort, but rather a process of permanent change, requiring a certain level of "after reform maintenance". Such maintenance, however, is difficult, when governance structures are as highly fragmented as they are in Switzerland (Braun & Uhlmann, 2009). This leads to a second conclusion: in order to be effective, reform needs to take into account system complexity, institutional constraints, and political and cultural values to find a balance between ambition and feasibility. In Switzerland, this means relying on decentralized strategies, integrating the relevant stakeholders into governance arrangements. Thus, public administrations willing to actively intervene in the domain of outpatient healthcare can best do so by initiating and facilitating collaborative processes (Ansell & Gash, 2008;Emerson et al., 2011), ideally triggered by needs and interests articulated by the actors affected by the reform -in this case, general practitioners struggling to improve their role within the healthcare system.
Our study shows that Swiss member states do in fact use their discretion to complement federal healthcare regulation with the aim of decreasing outpatient healthcare expenditures at the cantonal level. Our findings are a case of multilevel governance in both directions, top-down and bottom-up. In accordance with research on member state implementation, we show the importance of formal politics in the member state response to federal input, which not only applies to implementation but also to complementing federal policy. The finding that public administration plays a major role both in terms of capacity and as enabler and initiator implies that administrative capacity at the member state level provides a possible way out of federal policy deadlock. Federalist systems thus offer opportunities for policy innovations the federal level ultimately may benefit from.        Parsimonious solution: left*org* ENT + LEFT*ORG*adm*RESP*ent � steer (solution consistency 0.89, solution coverage 0.29).

Outcome: Complementary policy activity of the outpatient sector (STEER)
As for the outcome, the threshold for full membership in the set of cantons with pronounced regulation in the outpatient sector was set at 0.75. Thus, no canton achieved full membership in the set of strongly regulating cantons, which appears justified given the fact that none of the cantons came even close to reaching the maximum number of points. The point of indifference is set in the large gap between the regulation value 0.38 and 0.25 at 0.32. The value for absolute non-membership in the set of strongly regulating cantons should lie at a regulation value of 0.

Condition 1: Substantial share of the political Left within the government (LEFT)
As for the first condition, for each of the 26 cantonal governments the number of seats on the cantonal government held by the Left for each year from 2002 to 2008 is recorded. For the relevant years, the average number of seats held by the Left varied between 0 and 53%. In view of the fact that there were only three cantons that on average had a Left proportion of seats of just over 50% at all between 2002 and 2008, the threshold for full membership is set at 50.1% average proportion of seats. The lower qualitative anchor (Fuzzy-Value 0.5) for absolute non-membership of the set of the Left-wing governments lies at 5% or less seats. The point of indifference, which is where it cannot be said whether a canton is within or outside the set of the cantons with a substantial proportion of Left in government, is set at 24%.