Introduction
This research aims to analyze the conditions under which governments show high levels of formal accountability for public policies. We focus on Spain, seeking to explain differences across its 17 autonomous communities (ACs) in formal accountability for health policy implementation. The quasi-federal design of health policy in Spain and the differences in formal rules for implementing health policy across regions allow analyzing differences in their levels of accountability. The research questions are as follows: Why is health policy more accountable in some regions than in others? Which conditions make regional health policies more accountable? To answer these questions, we first compare accountability for health policy across Spain’s ACs, analyzing empirically the links between the presence/absence of five causal conditions and higher formal accountability for health policy. As the analysis covers Spain’s 17 ACs, we use fuzzy-set Qualitative Comparative Analysis (fsQCA), a technique well suited for analyzing a medium to small number of cases. The analysis allows examining which (combinations of) conditions are necessary and/or sufficient for achieving higher levels of accountability for health policy.
While accountability is ‘an ever expanding concept’ (Mulgan, Reference Mulgan2000) few empirical studies investigate which variables affect it. The most important focus is on the institutional factors influencing accountability in independent regulatory agencies (Koop, Reference Koop2011, Reference Koop2014; Maggetti, Reference Maggetti2012; Busuioc, Reference Busuioc2013), in the European Union, and in other decentralized polities with multiple administrative and decision-making levels (Bovens, Reference Bovens2007; Fisher and Hobolt Reference Fisher and Hobolt2010; Papadopoulos, Reference Papadopoulos2010; León, Reference León2011; Brandsma and Schillemans, Reference Brandsma and Schillemans2013). We draw on this last approach, which focuses on the conditions which endow public policies with higher accountability levels, to compare accountability for health policies across Spain’s 17 autonomous regions. We consider most of the explanatory factors used in these general studies, like the policy’s salience and government’s ideological orientation, but also others like private vs. public management of health services and government’s minority vs. majority status in multiparty systems, which have been less explored in the past. Additionally, we pay attention to Spain’s peculiar process of decentralization by considering the parliamentary strength of non-statewide parties (NSWPs) in Spain’s 17 ACs.
The structure of the paper is as follows. In the next section, we present the theoretical approach used to frame the concept of accountability for public policies and explain how we applied it to measuring health policies in Spain’s ACs. Next, we review the scholarly literature to identify some relevant factors that are thought to lead to variations in levels of accountability, and elaborate appropriate expectations about their validity. We transform these expectations into specific hypotheses about regional differences in accountability for health policies in Spain, and define and operationalize the conditions used in the analyses to test them with the help of fsQCA. The results from applying this method are reported next. In the last section, we draw the conclusions.
Subject of study: accountability for public policies
Accountability involves the capacity of account-giving and holding-to-account between at least two parties. This social relationship implies (i) one actor A (or more) who has the obligation to inform and justify its decisions; (ii) an issue about which to be accountable (M); (iii) one actor B (or more) who is the recipient of said accountability and has the right to ask A for information and justification about its decisions and actions, and to evaluate and sanction them (Grant and Keohane, Reference Grant and Keohane2005; Bovens, Reference Bovens2007, Reference Bovens2010; Dubnick and Frederikson, Reference Dubnick and Frederickson2009; Lindberg, Reference Lindberg2013). Our approach includes both dimensions of accountability: the informative/justifying dimension, carried out by actor A, and the evaluative/sanctioning dimension carried out by actor B.
We concentrate on accountability for public policies because these are key instruments for governments to implement their political programs (Sabatier, Reference Sabatier1999). Accountability can refer to the formulation or implementation phases of policies (Pressman and Wildavsky, Reference Pressman and Wildavsky1973; Bardach, Reference Bardach1977; Howlett, Reference Howlett2011). We focus on implementation because accountability for policy formulation is often performed in restricted political ambits, like the parliament, and has been extensively studied within the process of political competition. In particular, we distinguish three aspects of the implementation of public policies frequently studied by accountability scholars: the people responsible for policy implementation (e.g. who is in charge, if they incur in work incompatibilities) (Page, Reference Page2010); budget allocation and execution (e.g. how much was spent, on what, who were the recipients, presence of budget deviations) (Rabovsky, Reference Rainey and Chun2012); and the effects of the policy (e.g. what was accomplished) (Bovens, Reference Bovens2007). In short, we analyze how much information and justification is required and which mechanisms are contemplated for evaluating and sanctioning key matters pertaining to policy’s actors, resources, and results.
We use the word ‘required’, because we only focus on ‘formal accountability’. In line with other studies (Koop, Reference Koop2011; Hanretty and Koop, Reference Hanretty and Koop2012), we aim to assess the degree to which accountability is formally regulated. We argue that formal rules, while insufficient, are necessary for accountability to occur. Formal rules provide a measure of order, homogeneity, and certainty to the implementation of accountability, allowing it not to depend on agents’ will, but on legally established obligations. Without them, accountability would be erratic and arbitrary. Formal and actual accountability need not to coincide, as agents’ accountability practices may exceed expectations (Koop, Reference Koop2014) or be fraught with problems (Busuioc, Reference Busuioc2012, Reference Busuioc2013). However, at this point, it would be difficult to obtain valid indicators of the implementation of accountability rules across Spain’s ACs, due to the short time elapsed since their promulgation. We leave this type of analysis for future research.
As we have argued in detail elsewhere, the level of formalization of accountability can be measured with the four characteristics of its regulatory framework (Pérez-Durán, Reference Pérez Durán2015). We follow Stinchcombe (Reference Stinchcombe2001) and argue that practices are more formal when they are more specific, binding, autonomous and public. Concretely, the level of formal accountability in the implementation phase of a public policy is high: (1) when it has a detailed regulatory framework, that is, when higher order rules have been developed into lower order regulations specifying what and how to inform/justify and to evaluate/sanction in concrete cases (specific character); (2) when this legal framework compels those who account to provide information and explain it when prompted, and obligates the recipients of accountability to evaluate and sanction the accounting actors when necessary (binding character); (3) when these rules provide for independent bodies to gather information and request explanations about policy decisions and evaluate and sanction policy deviations (autonomous character); and (4) when it stipulates the open nature of the information/justifications, and evaluations/sanctions occurring in the process of accountability (public character).
The case of Spain
Our analyses compare accountability for health policy across Spain’s 17 regional governments. Current health policy in Spain is the end result of a broader process of political devolution whereby regions have been increasingly (and, regarding some policies, asymmetrically) assuming political competencies typically reserved to the central government. In the case of health policy, devolution has affected all regions similarly, granting them authority to make important political decisions affecting the provision of health services. While the central government establishes some basic laws (e.g. pharmaceutical regulation), ACs are responsible for implementing them and have ample legislative powers, including on the regulation of the private or public provision of health services.
Spain’s decentralized territorial design is quite different from simpler systems where higher organizational or administrative levels delegate responsibility to lower levels, but where the latter ultimately remain accountable to the central government (Rondinelli et al., Reference Rodden1983; Vrangbæk, Reference Vis2007). It is also different from purely federal systems where all states, regions, or local authorities have the same constitutionally recognized powers. Spain’s system of decentralization is uneven, with some regions having more powers than others, but with increasing pressures toward federalization or homogenization (Moreno, Reference Moreno2002). For example, some regions (e.g. the Basque Country, Navarra, Catalonia, Galicia, and Andalusia) followed a special, faster route to autonomy, and obtained additional political jurisdiction over issues such as language (Galician, Basque, and Catalan), a civil code (Catalonia and the Basque Country), and/or a special fiscal status (the Basque Country and Navarra). This unevenness has reflected the socio-economic, cultural-linguistic, and political particularities of the so called ‘historical nationalities’ – a consequence of what Anderson (Reference Aguilar and Sánchez-Cuenca1974) once described as Spain’s historical failure at building a nation-state. This failure resulted in the incorporation to the state of the regional élites of the northeast quadrant of the country (particularly, of the Basque Country and Catalonia) on a ‘most favoured lord’ basis, that is, with local rights and privileges similar to those held by the central élites (Laitin, Reference Laitin1998). These prerogatives, and easier access to European commercial networks, contributed to the economic prosperity of these regions and to the uneven process of modernization experienced by Spain in the late 19th century (Díez Medrano, Reference Díez Medrano1995). However, it was never strong enough to counteract the military and political prowess of the central state in Madrid, eventually giving rise in the 20th century to conservative national movements in Catalonia and the Basque Country (Rothschild, Reference Rondinelli, Nellis and Cheema1981). In his celebrated study of the factors accounting for Spain’s civil war, Brenan (Reference Brenan1990) once noticed that these middle-class movements were less oriented at achieving independence from Spain than, under the threat of secession, pushing for state reforms that would help create larger national markets on which to sell their products. They sparked authoritarian reactions from the central élites and the abolition of regional rights and privileges during long spells in the history of the country, which would only resurface after the corresponding democratic restoration (Linz, Reference Linz1978; Carr, Reference Carr1982).
Such was the case of the democratic restoration of 1975, when the ‘historical nationalities’ resumed their claims to political autonomy. Unsurprisingly, the transfer of competencies over health and other policies started in these historical nationalities. However, in what some have considered as an attempt of the central state to dilute the centrifugal and nationalist aspirations of the historical nationalities (Agranoff, Reference Agranoff1993), the transfer of competencies was already comprised of other regions like Galicia, Andalusia, Valencia, or the Canary Islands, with socio-economic and cultural peculiarities but with shorter or no history of political autonomy, and which now claimed similar rights to ‘devolution’ (Catalonia obtained competencies over health in 1981, Andalusia in 1984, Basque Country and Valencia in 1987, Galicia and Navarre in 1990, Canary Islands in 1995). This opened the way for all remaining regions to claim the same rights, and for the state to counteract historical nationalities’ political aspirations by raising the political powers of all regions, in what has been termed the policy of ‘café para todos’ (coffee for all) (Noël, Reference Noël2013).
While ‘café para todos’ did not affect all policies equally (such as the fiscal autonomy of ACs), it did affect health policy. Thus, in 2002 full health policy competencies were transferred to all the remaining ACs (Aragon, Asturias, Balearic Islands, Cantabria, Castile-La Mancha, Extremadura, Castile-Leon, La Rioja, Murcia, and Madrid). More importantly, in 2002, the way in which financial resources were allocated to ACs changed radically. All regional governments increased their involvement in tax collection, and in return took responsibility over the full costs of health care (Moreno, Reference Moreno2002; Rico and Costa-Font, Reference Rico and Costa2005). As Lago and Cantarero point out, health care is, with education, ‘the foremost policy responsibility of the ACs’; together they ‘account for 60–70% of total public funds in the hands of ACs’ (Reference Lago and Cantarero2012: 21).
This quasi-federal design of Spain’s health system – which, to repeat, does not extend to all other policies – allows the studying of variations across regions in levels of formal accountability for health policies and in the conditions that explain it, controlling for many other factors that cannot be easily considered in cross-national research. However, the peculiarities and asymmetries of Spain’s overall process of power devolution pose some difficulties in the generalizability of our findings. We discuss these limitations in the conclusions.
Measuring AC’s levels of accountability for health policy
As our aim was to measure formal accountability, we examined all health policy regulatory dispositions and official records (valid as for 2011) concerning high-ranking posts’ activities, the functioning and organization of health services, and the presence of the operation of auditing bodies at state and regional levels. A total of 108 laws were analyzed (e.g. Andalusia’s Law 2/1998 on Health Provision, Extremadura’s Law 3/2005 on Health Information and Patient Autonomy, Catalonia’s Act 18/2010 on Public Audit Office).
As noted, we assessed two dimensions of accountability – information/justification and evaluation/sanctioning – and three aspects of AC’s implementation of health policies – responsible actors, resources, and policy results. In each case, accountability was deemed high if formal dispositions were highly (as against medium or lowly) specific, binding, autonomous and public. We used a simple numerical valuation to reflect these three levels: 1, 0.5, and 0. Valuations across multiple dimensions were added and transformed into 0–100 scales to build three synthetic accountability indexes for each policy aspect (responsible actors, resources, and results). Finally, the three subindices were averaged to create an overall accountability index (see Table 1).
Table 1 Accountability index in the autonomous communities (ACs)
Elsewhere, we have detailed how we applied this method to code the documentation analyzed. In general, when an AC had its own formal rules for health policy accountability, the degree of accountability was considered high. For example, regarding the informative/justifying dimension of policy results, the level of specificity was coded high when an AC had passed laws requiring hospitals to publicize patients’ waiting lists or satisfaction surveys, as against others making vague requirements affecting all policies (medium level) or applying only statewide regulations (low level). Similarly, regarding the publicity of the information and justifications provided by the actors responsible for implementing the policy, we considered highly accountable any AC that formally required high-ranking health managers’ performance to be made public (top managers include members of the bodies governing ACs’ health policies – such as secretaries and general directors – and heads of public health foundations or health consortia). If these requirements applied vaguely to all health personnel working in each region, or did not exist at all, publicity was considered medium or low, respectively. Likewise, regarding the autonomy of the evaluative/sanctioning mechanisms overseeing the use of policy resources, an AC’s health policy was considered highly accountable when it had formally established independent bodies for evaluating and sanctioning the allocation and execution of health budgets, as compared with others that relied only on regional multi-policy supervisors (medium) or statewide controls (low).
Table 1 presents the results of this exercise at classifying ACs by their levels of accountability in each policy aspect, and the overall index averaging them. For example, Madrid and Aragon have low levels of accountability because, among other reasons, their regulations about public servant’s activity records do not include high-ranking managers within semi-private health consortia, and are not made public. In contrast, Galicia has a high level of accountability because, among other reasons, it has established citizens’ rights to be informed about health policy results (service quality, extent of coverage, waiting lists, etc.). Galicia’s legislation also establishes that health data and evaluations must be made public and has established a Patients’ ombudsman or autonomous administrative body in charge of defending users’ rights.
Once we identified variations in accountability levels, the next step was to explain it. We analyzed five conditions that could lead to higher accountability in health policy implementation.
Causal conditions of accountability
As mentioned, scholars have only recently started investigating which variables – or causal conditions – might affect accountability (Koop, Reference Koop2011, Reference Koop2014; Grimmelikhuijsen and Welch, Reference Grimmelikhuijsen and Welch2012; Maggetti, Reference Maggetti2012; Brandsma and Schillemans, Reference Brandsma and Schillemans2013; Busuioc, Reference Busuioc2013). Four conditions stand up in this research. They do not exhaust the list of potential factors fostering accountability for health policy but are among the most frequently mentioned in the literature. To these conditions we add a fifth – strength of NSWPs in each region – so as to investigate the impact of Spain’s territorial tensions on health policy accountability.
Private vs. public management of public policies
Some scholars defend public service management arguing that it fosters accountability because there are more formal controls. They stress that public managers ‘often operate under greater public scrutiny than do private sector managers’ from the media or oversight authorities, and that they ‘face stronger expectations for fairness, responsiveness, honesty, openness, and public accountability’ (Rainey and Chun, Reference Ragin2005: 92–93). Minow (Reference Minow2003) argues that externalization creates opportunities for avoiding public norms and government control of private operators (Donahue and Zeckhauser, Reference Donahue and Zeckhauser2008).
If these arguments were correct, accountability should be lower when private concerns play a higher role in public policy implementation. Note that this expectation is about how accountable publicly funded policies are when implemented by private operators, not about how accountable the providers of a private good can be, which depends on the market (Hodge and Coghill, Reference Hodge and Coghill2007; Willems, Reference Willems2014).
Alternatively, some scholars argue that public management may have ‘greater diversity and intensity of external informal political influences on decisions’ (Rainey and Chun, Reference Ragin2005: 92). Interest groups may distort the policy direction and deflect accountability (Trebilcock and Iacobucci, Reference Trebilcock and Iacobucci2003).
Salience of public policy
Salience has traditionally been ‘used to designate the importance of issues, particularly for voters’ (Wlezien, Reference Wlezien2005: 555). In the United States, Ringquist et al. (Reference Ringquist, Worsham and Eisner2003) show that when an issue is salient it attracts more congressional attention and efforts. In the European Union, Egeberg and Trondal found that ‘the concerns of formally political bodies such as ministries, the European Parliament and the Council are significantly more emphasized when issues get contested’ (Reference Egeberg and Trondal2011: 880–881). Sulitzeanu-Kenan (Reference Sulitzeanu-Kenan2010) finds that issue salience determines government’s decisions about appointing commissions of inquiry. Koop shows that ‘independent agencies which operate in more salient issue areas are also subject to more extensive accountability arrangements’ (Reference Koop2011: 228). Most importantly, political salience affects organizations’ institutional design, fostering development of formal accountability rules. There is a widespread expectation that if a public policy is salient there will be more formal mechanisms to increase accountability. As the saliency of health care varies according to subpopulations’ needs and experiences with health services and with views about a health system’s performance (Soroka et al., Reference Soroka, Maioni and Martin2013), we expect the salience of health policies to differ across Spain’s ACs, and these variations to have an impact on health policy accountability.
Ideological position of governments (left-right ideology)
A distinction frequently applied to political parties regards their left or right ideologies. Regarding the economic dimension, ‘[p]arties on the economic left want government to play an active role in the economy. Parties on the economic right emphasize a reduced economic role for government: privatization, lower taxes, less regulation, less government spending, and a leaner welfare state’ (Bakker et al., Reference Bakker, de Vries, Edwards, Hooge, Jolly, Marks, Polk, Rovny, Steenbergen and Vachudova2015). Regarding their stance on social issues, right-wing parties favor ‘moral conservatism’ and stability, while leftist parties favor ‘social liberalism’ (McElroy and Benoit, Reference McElroy and Benoit2011), that is, a balance between social justice and ‘greater democratic participation’ (Bakker et al., Reference Bakker, de Vries, Edwards, Hooge, Jolly, Marks, Polk, Rovny, Steenbergen and Vachudova2015). As left-wing parties prioritize social justice and promote participatory policies in their political agendas, they could also be more interested in being held accountable.
Alternatively, as right-wing political parties prioritize market principles, they might promote higher government accountability as a means to open public information that they perceive as being biased and to control bureaucracy’s and politicians’ interests (Stiglitz, Reference Stiglitz2000).
Government’s parliamentary strength
Scholars have emphasized that minority governments carry out their programs through parliamentary agreements and/or coalitions with other parties (Strøm, Reference Strøm1990). ‘[A minority] cabinet is most likely to use policy concessions as a bargaining chip to build coalitions around specific legislation’ (Godbout and Høyland, Reference Godbout and Høyland2009: 8). Thus, one might expect that in minority governments opposition and/or coalition parties would push for accountability mechanisms to control the government.
Not all scholars agree with this expectation. According to Müller, majority parliamentary governments – those with at least half the seats plus one – ‘can not only survive in office but also enact their political program’ (Reference Müller2008: 204). Applied to accountability, one might expect majority governments to implement more effectively long-term accountability policies without finding parliamentary opposition to their approval.
Strength of NSWPs
The recent process of political modernization in Spain was characterized by two processes: democratization and power devolution/decentralization. ‘This double process has resulted in [the] configuration of a party system, with its traditional state-centered logic, and, in parallel with this, the configuration of political arenas in the autonomies’ (Pallarés et al., Reference Pallarés and Keating1997: 137), especially in the historical nationalities. Up until recently, the party system in Spain had two main statewide parties – Spanish Socialist Workers’ Party and Popular Party (PP) – and several NSWPs, which were dominant in the Basque Country and Catalonia – for example, Basque Nationalist Party and Convergencia i Uniò (Pallarés and Keating, Reference Pallarés, Montero and Llera2003; Barrio et al., Reference Barrio, Rodríguez, Baras and Barberà2010: 7).
In Spain, NSWP ‘project their independence from state-wide parties as the best guarantee of their defense of the interests of the autonomous community’ (Pallarés et al., Reference Pallarés and Keating1997). Thus, one might expect that in systems such as Spain’s in which important public policies have been decentralized and transferred to regional governments, a strong presence of NSWPs in regional parliaments will result in greater accountability for policies over which they have competencies, to differentiate themselves from the central government and, in the case of the traditionally more prosperous but politically subordinated historical nationalities, to set an example and demand central government’s democratization and modernization.
On the other hand, in a study of electoral accountability in Spain, Aguilar and Sánchez-Cuenca found that ‘nationalist voters excuse poor management of the regional government to a greater extent than non-nationalists’ (Reference Anderson2007: 62), because they prioritize claims of belonging or cultural identity over other issues. As several Hispanists have noted (Brenan, Reference Brenan1990), the regional élites of the historical nationalities found it useful in the past to mobilize their regional constituencies along the nationalist divide to pursue their own interests with fewer controls. In both cases, one might expect a strong presence of NSWPs in a region to be associated with lower accountability for public policies like health.
Other conditions
Two other conditions are often mentioned in the literature on the factors that foster accountability: degree of decentralization and government’s financial capacity.
Some scholars argue that decentralization ‘makes politicians less remote, more visible and more accountable’ (Pollitt, Reference Pollitt2005: 381). In contrast, others argue that multiple levels of decision-making – such as those accompanying administrative and political decentralization – blur governments’ responsibility (León, Reference León2011, Reference León2012). As noted by Rodden, ‘when decentralization amounts to adding layers of government and expanding areas of shared responsibility, it might facilitate blame shifting or credit claiming, thus reducing accountability’ (Reference Rothschild2004: 494).
While hypotheses on decentralization can be best tested by comparing centralized and decentralized systems in cross-national research, they can also be approximated by considering regional differences in the degree of power devolution brought about by Spain’s uneven process of decentralization. This can be done, for example, by adding a variable measuring the timing of transfers of basic powers to each region (generally earlier for the so called ‘historical nationalities’, which have higher competences), and analyzing its relationship with accountability. We do this in additional sensitivity analyses performed to check the robustness of the main results (these do not change substantively when the timing of the transfer power is added, see below). There are two reasons for not including this variable in the primary analyses. First, devolution timing did not affect much the assumption of health competencies in each region, for, as noted above, the complete transfer occurred simultaneously for all ACs when they gained full financial responsibilities in 2002. Second, the timing of devolution correlates very highly with NSWPs’ strength, which is higher in the historical nationalities that obtained political autonomy earlier. Thus, any hypotheses about the effects of regions’ different competencies on health accountability can be incorporated into the hypotheses about the effects of NSWPs’ strength.
As for government’s financial capacity, Grimmelikhuijsen and Welch (Reference Grimmelikhuijsen and Welch2012), focusing on environmental policy transparency, argue that transparency ‘of policy outcomes’ is associated with government’s higher financial capacity. Like for the timing of political devolution, we only use this variable in additional sensitivity analyses (results do not change substantively when added). This is because regional government’s financial capacity, as measured by how large health budgets are relative to AC’s total public expenditures, depends on the number of policies transferred to each region, and thus on the uneven decentralization experienced by Spain, which we have already considered. While we pondered on using a different indicator – regional government’s health spending as a percentage of regions’ GDP – we could not use it because Spain’s central state makes extra payments to the poorest regional governments to guarantee the same basic assistance to all citizens across regions (regions are still responsible for the use of these funds). This distorts assessment of regional governments’ financial commitment to health services, which appears to be stronger in poorer regions.
Method
As the number of Spain’s ACs is relatively small (17), in assessing the expectations presented above we chose a method that could work well with this small number of cases – fsQCA (Ragin, Reference Rabovsky1987, Reference Ragin2000). fsQCA assesses if the presence/absence of a specific condition or combination of conditions is necessary and/or sufficient for an outcome of interest to occur. In Annex 1 provided in supplementary materials, we provide a brief technical description of the fsQCA method and of the criteria used for considering a condition as necessary or sufficient.
Our choice of fsQCA as method of analysis required reformulating our expectations – previously stated in terms of associations – in the form of hypotheses about whether any of the conditions presented above was necessary (alone or, more likely, in some sufficient combination of conditions) for a regional government to display high levels of accountability in health policy implementation. This reformulation led us to hypothesize that private management of health services (Hypothesis 1), salient health policies (Hypothesis 2), left ideological position of regional government (Hypothesis 3), minority position of regional government (Hypothesis 4), and strong presence of NSWPs in regional parliaments (Hypothesis 5), are necessary to observe high levels of health policy accountability in an AC, as against the null hypotheses that they are not.
In order to test these hypotheses, ACs must first be measured in absolute terms using appropriate scales for the outcome and the conditions. These measurements must be followed in fsQCA by a more qualitative relative assessment of the degree of membership of each case (AC) in the logical sets defining the outcome and the conditions. In both cases, it is necessary to set up clear criteria of classification. Annex 2 describes in detail these criteria and how they were operationalized in fsQCA. The results are displayed in Tables 2 and 3. Table 2 shows the absolute values assigned to each AC in the outcome and causal conditions. Table 3 provides ACs’ membership scores in each set given the qualitative definitions or ‘anchors’ chosen for each in this research. These are the scores used by fsQCA in the analyses of necessity and sufficiency, the results of which are reported next.
Table 2 Operationalization for the outcome and causal conditions
NSWP=non-statewide parties.
Table 3 Membership for the outcome and causal conditions
NSWP=non-statewide parties.
Cases that have membership in a specific condition (>0.5) are shown in bold.
Results of fsQCA analyses
Necessary conditions for accountable health policies
In this section, we identify conditions that need to be present or absent to observe accountable health policies in an AC. As can be seen in Table 4, no condition can be considered necessary (none reaches the recommended consistency score of 0.9, see Annex 1). However, three conditions approximate the necessity threshold: absence of private management, high policy salience, and low presence of NSWPs. We consider them to be ‘quasi-necessary’ to produce accountable health policies in Spain’s regional governments.
Table 4 Analysis of necessary conditions for accountable health policies in autonomous communities
NSWP=non-statewide parties.
Logical No (~) refers to the absence of a condition.
The results suggest that the absence of private management in implementing public health policy is ‘quasi-necessary’ for policy accountability. This result gives little credence to Hypothesis 1, according to which greater intervention of the private sector in publicly funded health policies entails more control over its implementation. It gives more credence to the null hypothesis that public management facilitates accountability. Our analysis of regional laws confirms that top managers in privately run health centers are not subject to the same formal controls as high-ranking posts in the public sector – there are no formal mechanisms making private managers accountable for the public budget they administer.Footnote 1
The results similarly suggest that salient health policies are ‘quasi-necessary’ for observing accountable policies. This finding gives credence to Hypothesis 2 and is in agreement with Koop’s (Reference Koop2011) finding that an agency’s accountability is greater when the issues they deal with are perceived as more salient. In seven of the eight ACs with higher levels of accountability for health policy, citizens consider this policy as highly important (Andalusia, Cantabria, Castile-La Mancha, Castile-Leon, Balearic Islands, Extremadura, Galicia, and Navarra). It is only in the Balearic Islands where health salience is low and health policy accountability is high.
Finally, the results suggest that ACs with strong presence of NSWPs in regional parliaments (Catalonia, Basque Country, and Canary Islands) have fewer accountability mechanisms for health policy implementation. This result comes close to refuting Hypothesis 5, which expected a strong presence of NSWPs in regional parliaments to lead to greater accountability for policies over which they have competencies, so as to differentiate themselves from the central government and/or set a modernizing example for the state. Instead, it gives more credence to the alternative hypothesis that cultural identity issues and blame-shifting deflect demands for higher accountability in regions where NSWPs are strong.
Sufficient conditions for accountability for health policy
As noted, it is often the case that no condition is necessary to produce an outcome but nevertheless forms part of a combination of conditions that is sufficient to do it. As shown in Table 5, fsQCA analysis reveals three paths or ‘recipes’ – in Ragin’s terms – that lead to higher accountability for health policy.Footnote 2
Table 5 Sufficient combinations conditions for accountable health policies in autonomous communities
NSWP=non-statewide parties; *=logical AND (intersection of sets); ~=logical NO (absence of a condition).
Consistency: ‘how closely a perfect subset relation is approximated’ (Ragin, Reference Ragin2008a: 44). Solution coverage: proportion of membership explained by all paths identified. Raw coverage: proportion of memberships in the outcome explained by a single path. Unique coverage: ‘proportion of memberships in the outcome explained solely by each individual solution term’ (Ragin, Reference Ragin2008b: 86).
First, the results show that policy salience, previously identified as ‘quasi-necessary’, is present in the three paths leading to accountability. Moreover, in two of the three paths private management and strong NSWPs must be absent for an AC to have accountable health policies, which reinforces our previous analyses of necessity. Just two of the eight ACs that were deemed to be accountable in the analyses do not meet these three conditions: Navarra and the Balearic Islands. In Navarra, there is a strong presence of NSWPs but considerable private management of health services. In the Balearic Islands, health salience is low.
Second, two further combinations of conditions appear in the three routes to accountability: left-wing and majority governments (first path) and right-wing and minority governments (second and third paths). Andalusia, Castile-La Mancha, and Extremadura exemplify the first conjunction; Navarra and Cantabria, the second. The results suggest that left-wing parties are more likely than right-wing parties to pursue accountability for health policy but succeed only or primarily when they enjoy parliamentary majorities. In contrast, right-wing governments in a minority position are more likely to introduce accountability mechanisms, perhaps because they face stronger controls from the leftist opposition.
Fourth, the first path/combination has higher explanatory power than the other two.Footnote 3 This combination explains the accountability results obtained by three ACs led by left-wing parties: Andalusia, Extremadura, and Castile-La Mancha. The second and third paths have lower explanatory power, because each covers only one AC led by right-wing governments: Cantabria and Navarre, respectively. These paths cannot explain the occurrence of accountability for health policies in other traditionally conservative regions like Galicia, the Balearic Islands, and Castile-Leon. This suggests the presence of some other conditions that might explain the results obtained for these ACs. For example, the Balearic Islands have been ruled since 1983 mainly by the conservative PP but in two of the three legislatures analyzed they were led by a coalition headed by the Socialist Party. Similarly, Galicia has been ruled almost uninterruptedly by the PP since 1989 except in one of the legislatures analyzed, when it was led by a coalition headed by the Socialist Party. More in-depth analyses are required to explain additional conditions applying to these regional governments.
Sensitivity analyses
Following good practice advice (Schneider and Wagemann, Reference Schneider and Wagemann2010, Reference Schneider and Wagemann2012), we performed an fsQCA analysis for the absence of accountability. Only one condition approached the threshold we set for conditions to be considered necessary:Footnote 4 absence of private management. We identified two combinations of conditions associated with absence of accountability. They are mostly a mirror of those obtained before for its presence, thus giving more credence to our previous results.Footnote 5
Finally, two further variables were added to the analysis: timing of devolution (Early) and government’s financial capacity for health (Capty). As explained before, we had some reservations about the use of these two variables, for both were highly correlated with NSWPs’ strength. We use them only to check the robustness of our main results. The addition of these variables does not change the results substantially, as neither of these two conditions is necessary to produce accountability. Their inclusion in the sufficiency analyses mostly adds inconsequential details to the three paths identified above. It separates the three AC’s that followed the first path to accountability (~Priv*Sal*Min*Left*~NSWP) into two subgroups, identifying those that initiated the process of decentralization of health policy at an earlier stage (Andalusia) and later (Extremadura and Castile-La Mancha). It qualifies the other two paths corresponding to Cantabria and Navarre, respectively, by showing that these ACs were cases of late decentralization. And it identifies a unique path followed by the Balearic Islands (~Capty*~Early*~Priv*~Sal*Min*Left*~NSWP).
Discussion and conclusions
Social phenomena are often characterized by their causal complexity – the same outcome can be achieved through different routes (Ragin, Reference Ragin2000, 2008Reference Ragina). In this study, we used fsQCA to assess the necessity and/or sufficiency of five conditions (and of their combinations) for fostering accountability for health policy in Spain’s 17 ACs. Different combinations did succeed, confirming the causal complexity of accountability.
Our analysis of necessity revealed that three of the five causal conditions – policy salience, absence of private management and absence of strong NSWPs – were ‘quasi-necessary’ to produce accountable health policies in a region.
We found that policy salience (understood as the perception that citizens have of the importance of health policy) was present in the three sufficient combinations of conditions leading to accountability. Although the salience of a policy had already been highlighted in other studies as fostering accountability, its quasi-necessary status regarding health policy accountability in Spain’s ACs suggests that in representative democracies accountability is not just a mechanism of hierarchical control between different administrative levels, but also a tool for users/beneficiaries of social policies to influence governments’ political agendas (Wlezien, Reference Wlezien1995). The result was not trivial, for there was enough variation in the salience of health policy across regions to assess its impact on accountability.
We also found that absence of private management in the implementation of health policies is ‘quasi-necessary’ for health policy accountability. This condition also appeared in two of the three sufficient combinations leading to accountability. Health policy accountability is more common when there is less intervention by private agents in the implementation of a public policy, suggesting that any efficiency gains derived from the private provision of publicly funded health services may come at the price of lower accountability. We want to stress that we did not test whether private management is more efficient than public administration in providing health services, but rather which form of health care provision is more transparent and subject to greater mechanisms of control and sanction. Our results suggest that the mixed forms of private/public mixed management of health services that have spread across many advanced societies since the 1990’s, which Hood (Reference Hood1995) defined as ‘New Public Management’, have not been accompanied by higher accountability in the Spanish regions that introduced them more extensively. Future (preferably cross-national) research should assess if low accountability characterizes any system – decentralized or not – in which private operation of health services is widespread, perhaps because private operators rely on market mechanisms rather than on formal provisions to account for their performance – what Klenk and Pieper (Reference Klenk and Pieper2013: 349) defined as ‘accountability for outputs’ vs. ‘organizational accountability’. Alternatively, it could be that the association observed for Spain between private provision of health services and low accountability might ensue from the absence of a strong central authority that could counteract the centrifugal forces generated by the proliferation of private health care providers at the regional level. According to this explanation, low accountability would be characteristic of any highly decentralized health system like Spain’s, where health care is increasingly provided by private agents under the scrutiny of multiple territorial authorities (e.g. the ‘provinces and territories’ in Canada, Australia’s ‘states and territories’, the ‘länder’ in Germany, the ‘regioni’ in Italy, or the ‘landsting’ in Sweden) (López-Casasnovas and Saez, Reference López-Casasnovas and Saez2007; Klenk, 2011; Klenk and Pieper, Reference Klenk and Pieper2013; Duckett, Reference Duckett2015). A final possibility regarding the aforementioned association between privatization and low accountability is that it may be peculiar to Spain’s historical model of power devolution. In this model, some regions – the richest – have obtained higher prerogatives from the state than others. This would have allowed the economic élites of these regions to pursue their private interests with fewer controls, also regarding the provision of health services for which all regions have the same legislative and executive powers.
This last interpretation is consistent with another finding from our analysis of necessity, whereby the absence of NSWPs in regional parliaments was found to be quasi-necessary for accountability to occur. This condition also appeared in two of the three sufficient paths leading to this outcome. The results suggest that a strong presence of NSWPs in regional parliaments hinders accountability for health policy. As noted, some studies on the determinants of voting in autonomous governments in Spain (Aguilar and Sánchez-Cuenca, 2007) have found that in ACs with a strong presence of nationalist parties, voters judge their governments moreover issues related with their representative capacity (i.e. for the representation or defense of what they consider the interests of their community), than over their management (i.e. for the results of government administration). Our fsQCA analysis supported this idea that a strong presence of NSWPs in regional parliaments leads regional governments to prioritize claims of cultural identity over accountability for health policies, diverting attention over policy implementation. Future research should assess the validity of this interpretation by examining its applicability to other similarly decentralized systems where power devolution to regional authorities has been as ‘asymmetric’ as in Spain, like in Italy, where ‘some regions enjoy greater powers than others’ (Bevan et al., Reference Bevan, Connolly and Mays2015: 88), or in Canada, where the decentralization of health policy ‘has been exacerbated by provincial politics, and especially Quebec nationalism’ (Fierlbeck and Palley, Reference Fierlbeck and Palley2015: 110). This explanation shall be contrasted with an alternative one stressing the blurring of government responsibility, which characterizes any decentralized system – asymmetric or not – with multiple levels of decision-making, due to increased blame shifting and red-tape (Rodden, Reference Rothschild2004; León, Reference León2011, Reference León2012). A final possibility, worth being explored in the future, is that the association between regional nationalist aspirations and low accountability may be the consequence, not just of asymmetric decentralization, but also of the peculiar form this has taken in Spain, favoring the richest regions. Spain’s territorial tensions have been traditionally associated with an unsuccessful process of national unification, which resulted in economically powerful regional elites remaining politically subordinated to a central state dominated by the élites of less developed regions (as against other more typical forms of separatism directed against a center that was both economically and politically powerful). This eventually gave rise in Spain to middle-class nationalist movements in the most prosperous regions which sought, on the one hand, to influence the state’s activity and modernize its structure by threatening it with secession, and on the other, to maximize their economic interests by subsuming social tensions under the territorial cleavage. According to this Spanish-centric explanation, in those culturally distinct regions of Spain where the élites were traditionally powerful in socio-economic terms (and by imitation, also in other regions where newly powerful élites rose along with Spain’s recent modernization), claims to autonomy and independence would have helped the regional élites to pursue their interests with fewer controls and less accountability.
This explanation is consistent with other results from the sufficiency analyses presented in this paper and, in particular, with our finding that health policy accountability is low in regions where NSWPs are strong and privatization is high (provided that health is salient and left-wing parties have not been majoritarian in the regional administration). It appears that in regions where the élites have successfully mobilized their middle classes along nationalist or particularist lines, they have also succeeded in pushing a privatization agenda for the regional health services and at being less accountable for it. More research is necessary to assess if our results apply also to accountability for other policies (e.g. education), and if they will be different in other countries where claims to regional autonomy have been monopolized by left-wing rather than right-wing parties.
Future research should also assess the generalizability of other interesting findings from our sufficiency analyses. We showed that the combination of left-wing majority governments or of right-wing minority governments, appears to be sufficient to produce accountability for health policy. This suggests that while left-wing governments require strong parliamentary support to implement their preferences for more accountable health policies, right-wing governments are compelled to make this choice only when facing strong parliamentary opposition. More research is necessary to assess if, as suggested by our results, left-wing parties are more accountable than right-wing parties for most other policies, perhaps because as argued above, right-wing parties defend particularistic interests and accountability by results, while left-wing parties support common interests that need more formal mechanisms of control.
One possibility is that the apparent higher accountability of left-wing vs. right-wing governments found in our results can be explained by our decision to focus only on formal accountability, without considering compliance with accountability norms, that is, with de facto accountability. In the future, once formal provisions have had a chance to be fully implemented, we shall be able to test if formality is necessary or must be combined with other conditions to get de facto accountable policies, and if the same paths that lead to formal accountability also lead to real accountability.
Despite the obvious limitations of this research, its results have provided a rich picture of the conditions that favor accountability for health policies across Spain’s ACs and of the alternatively complex routes that lead to it. The results set up a baseline against which future research on accountability may be contrasted, contributing to the development of a line of research of increasing interest for academics and policy-makers.
Acknowledgments
The authors would like to thank the four anonymous reviewers for their useful comments and helpful advice. The authors would also like to thank Mark Bovens, Gijs Jan Brandsma, Nuria Font, Jacint Jordana, Christel Koop, Albert Meijer, and Thomas Schillemans for their valuable feedback on previous versions of this article.
Supplementary Material
For supplementary material/s referred to in this article, please visit http://dx.doi.org/doi:10.1017/S1755773915000405
