The benefits and risks of decentralising the active welfare state

The question of why central governments transfer power to regional levels has been answered in many ways by the vast literature on comparative decentralisation, devolution, and federalism (e.g., Treisman, Reference Treisman2007; Bednar, Reference Bednar2011). Overall, decentralisation entails trade-offs, because, while it can improve policy implementation, it can also produce negative effects. Below we elaborate on these points and apply them to the case of ALMP.

Some have emphasised the comparative advantages of lower levels of government in providing goods and services. For instance, scholars who highlight the notion of efficiency (e.g., Peterson, Reference Peterson1995; Hooghe and Marks, Reference Hooghe and Marks2009) have argued that downward shifts of power can provide for more efficient policy-making and effective provision of goods and services by bringing policy decisions closely in line with citizens’ preferences. In the same vein, the Public Management literature has pointed out that decentralisation can generate a more efficient handling of resources and better results through flexibility (e.g., Trigilia, Reference Trigilia1991). Decentralisation could be also a means of upgrading governments’ decision-making capabilities, given that lower levels of government can provide informed and pragmatic solutions to local and individual problems. These benefits are especially noteworthy when addressing contexts with a high degree of diversity across levels of government (Seabright, Reference Seabright1996).

These propositions are relevant in the case of the decentralisation of ALMP, as highlighted by the literature on the governance of activation (e.g., Borghi and van Berkel, Reference Borghi and van Berkel2007; Eichhorst and Konle-Seidl, Reference Eichhorst and Konle-Seidl2008). More specifically, a decentralised approach fits scenarios in which there is a need for a personalised and local approach to (re-)inserting welfare clients and for experimenting with policy solutions – an underlying idea behind the notion of activation and the modernisation of Public Employment Services (PESs).

Nonetheless, the benefits of decentralisation have been disputed. Using the principal-agent framework, Giguère (Reference Giguère2003: 14) notes that, ‘The main cost of decentralisation is the loss of control over agent's actions. In a decentralised framework, the agent uses the greater flexibility granted to pursue his own interests, which may differ from that of the principal.’ Other authors have argued that transparency and accountability can be weakened through decentralisation because the division of responsibilities among the participating levels of government becomes blurred (Stoker, Reference Stoker, King and Stoker1996: 3). In these systems, the dangers of inconsistencies through the unequal application of law are especially relevant.

When it comes to social policies, given increasing administrative, political and/or fiscal regional or local autonomy, decentralisation might be linked to salient variations across the territory. As a result, welfare clients might experience different levels of benefits and burdens across different regions (e.g., opportunities, services, conditions, obligations). This flexibility might become challenging when regional inconsistencies and inequalities start to violate the principles of equal access to benefits and services, therefore fragmenting social rights (van Berkel and Borghi, Reference Van Berkel and Borghi2008: 396). Regional inequalities can also potentially trigger migration and negative externalities (e.g., race to the bottom, clients ‘voting with their feet’).Footnote ⁴ These issues become especially relevant when we consider that labour markets are nested spaces: they are national (even supranational, if we account for the notion of ‘free movement’ in Europe), not only local and regional spaces. In this context, decentralisation could hinder redistribution, solidarity and social citizenship across regional boundaries (see, e.g., McEwen and Moreno, Reference McEwen and Moreno2005; Rodríguez-Pose and Ezcurra, Reference Rodríguez-Pose and Ezcurra2010). These problems become more salient when central institutions or mechanisms for overseeing, enforcing and/or evaluating implementation are weak or lacking.

We argue that discussions regarding the benefits and drawbacks of decentralisation should not be viewed as ‘black and white’, as some intergovernmental arrangements might be conducive to more salient variations across the territory than others. More specifically, when discussing shifts in the governance of activation and the effectiveness of active welfare states, it is important to capture differences in the intergovernmental organisation of activation as different models of decentralisation could be linked to specific developments (Falleti, Reference Falleti2005; Greer, Reference Greer2009). Drawing attention to this ‘independent variable’ is extremely relevant given that ‘the big question is: how to design . . . institutional settings through which unemployment benefits/social assistance and activation services are delivered in a coordinated manner and according to integrated approaches’ (Karjalainen, Reference Karjalainen and Kazepov2010: 143).

Capturing intergovernmental arrangements

In all countries, regardless of whether they are federal, regionalised or unitary, the relationships between central, meso, and low levels of government are characterised by interdependence. Yet the nature and the depth of this interdependence vary along multiple dimensions, therefore providing actors below the central level with different types of powers. To capture and specify the architecture of decentralisation across political systems, this section develops a framework of intergovernmental arrangements. More specifically, we take a closer look at the nature and diversity of intergovernmental (central–regional) allocations of power to understand how these settings might be organised in a state, especially those undergoing decentralisation.

The first step in this exercise is to determine whether central and lower levels have exclusive or shared powers over a policy area. When a level of government has exclusive competencies over a policy area, a governmental entity cannot interfere with the other's reserved competencies. In instances when the central level has exclusive competence, lower levels are mainly administrative spaces that operate within the strict restrictions of the centre.

Once power is transferred downwards by providing exclusive powers to the regions or local governments, or by sharing powers, intergovernmental institutions and relations get more complex.Footnote ⁵ Given these new arrangements, in the case of regionalised countries, we could ask: to what degree are regions autonomous from the central level? In other words, to what degree is the central level superior to regions? To answer these questions, we must specify the degree to which these levels of government can act autonomously and/or override each other.Footnote ⁶ Accordingly, to categorise the nature of intergovernmental balances of power and relations, we examine whether the relationship between levels of government is organised horizontally or hierarchically. These two categories are policy area specific, which allows differentiation between policy realms in a country. In addition, it is acknowledged that there are different degrees to which central and regional levels are autonomous from each other. Yet we develop a stylised version of this argument by referring to both ends of the spectrum, which is especially helpful to understand various models of decentralisation and regionalisation, as well as different types of federations.

The architecture of ALMP in Italy and Spain

The Spanish and Italian cases, at first hand, stand out for their institutional similarities. Since the 1970s, both Southern European states have transferred significant powers to their respective regions, thus creating a complex web of intergovernmental relations. In addition, in these states the ‘regional question’ serves as one of the drivers of change (Putnam, Reference Putnam1993; Moreno, Reference Moreno2001). More recently, both countries transferred ALMP competencies to their regions. However, as we look closer, there are notable differences in the way these countries have designed and implemented the decentralisation of ALMP. The results here presented come from in-depth, semi-structured interviews with twenty-six civil servants, local policy-makers and scholars and were conducted in these countries at the central and regional levels (Spanish CAs: Catalonia, Seville, Madrid; Italian regions: Florence, Venice) in the summers of 2008, 2009 and 2011. On average, each interview lasted an hour. Interviewees were ensured confidentiality.

From hierarchy to horizontality: the central level exits ALMP

When it comes to the intervening role of the centre, how were these legal mandates implemented? The Ministry of Labour allowed ‘Italia Lavoro’ and the ‘Instituto per lo Sviluppo della Formazione Professionale dei Lavoratori’ (ISFOL, Institute for the Development of Skills for Workers) to support the legal process of ALMP devolution.Footnote ⁹ These central agencies lack the legal jurisdiction to intervene on regional matters. Rather, they serve as facilitators and providers of information to the regions. Rome, therefore, did not create central agencies with the authority to define and/or enforce a general line of what ‘Italian ALMP’ should be. Accordingly, the centre has virtually lost its ALMP powers given that regions gained exclusive administrative and political powers over ALMPs. As an interviewee put it, ‘with these changes the centre does not tell the regions what to do because each region should have some room to manoeuvre in deciding on how to organise itself and what to delegate to the provincial governments’ (interview with high rank civil servant, Rome, 2009). Another interviewee argued, ‘we expected the central level to coordinate, to establish a uniform direction. And this has been done in a very, very weak manner’ (interview with civil servant, Rome, 2009). Along these lines, other interviewees observed that this change institutionalised the exit of the centre from ALMP.

Despite calls for fiscal federalism, mainly by the Northern regions, the wide-ranging decentralisation of political and administrative ALMP powers has not increased the taxing powers of the regions (Arlotti, Reference Arlotti and Kazepov2009). In fact, the centre plays a weak role given that about 75 per cent of the funds employed to run regional PESs are obtained from the European Structural Fund (interviews with civil servants, Rome, Italy, 2009).

This new intergovernmental allocation of powers, grounded in weak central intervention (i.e., the horizontal model), has been controversial given that it is linked to regional institutional variations in the area of ALMP. The risks associated with horizontal decentralisation are very relevant for Italy given the limited institutional capacity of the regions in the Mezzogiorno and the existence of marked regional gaps. These gaps are illustrated, for instance, by its high levels of dispersion of unemployment – the highest in the European Union along with Belgium (Eurostat Regional Labour Market Disparities, 2008), another country with horizontal structures (see Kovziridze, Reference Kovziridze2002; López-Santana, Reference López-Santana2009). In addition, as illustrated in Table 1, Italy shows higher levels of regional dispersion of participation rates than Spain. Time after time, regions in Northern and Central Italy outperform Southern regions, which consistently score very low. By contrast, regional dispersions of participation rates in Spain are not as marked (with the exception of the year 2002). For example, in 2008 the score for Spain was almost half of the dispersion rate of the Italian case.

TABLE 1.

Regional dispersion of participation rates in Italy and Spain, 1999–2008 (OECD Statistics) (units are standard deviations; authors' calculations)

To sum up, even though de jure the Italian ALMP architecture is designed around the ‘shared competencies + hierarchy’ model, de facto it works as an ‘exclusive + horizontal’ model, given that the role of the central level is very weak and the regions control their administrative and political destiny in the ALMP area.

A hierarchical model of regionalisation: the Spanish case

In contrast to the Italian case, Spain has adopted a model of regionalisation in which the central level is guaranteed a dominant position over the CAs even if they have acquired new powers. In this way, in this policy area, Spain employs a hierarchical model of intergovernmental relations.

Since the adoption of the Spanish Constitution in 1978, decentralisation has been an ongoing process that has responded to nationalist pressures and large regional socio-economic disparities (Moreno, Reference Moreno2001: 5; Agranoff, Reference Agranoff2010). Title VIII of the Constitution granted a considerable degree of autonomy to the seventeen CAs, and these powers vary across sub-units (León, Reference León2010). The central government retained the faculty to formulate framework legislation (legislación básica), whereas the CAs were granted the power to design and adopt regional legislation in some policy areas. This allowed the CAs to tailor policies to the needs of each nation and region (Moreno, Reference Moreno2001: 2; Gallego et al., Reference Gallego, Gomà, Subirats, McEwen and Moreno2005: 106). Yet, the centre is guaranteed a dominant role, which it actively exercises through sharing competencies with the CAs and through intergovernmental mechanisms.

With regards to ALMP, from 1998 to 2010, Spain has moved towards a model of administrative federalism given that the centre has devolved administrative powers for ALMP to all CAs (with the exception of Ceuta and Melilla), while retaining exclusive responsibility over passive labour market policy (PLMP).Footnote ¹⁰ The decentralisation of ALMP competencies was done through a set of bilateral agreements between the centre and each of the seventeen CAs. As a high level CA civil servant put it: ‘When we talk about the transfer of competencies, it is important to say that we are referring to the capacity of the CAs to implement ALMP. In this way, many ALMP competencies are still a responsibility of the central level. Thus, devolution is a limited process’ (interview with PES civil servant, Seville, Spain, 2008). Other interviewees in Catalonia expressed a similar point:

Prior to the launch of Law 56 of Employment (2003), the National Institute for Employment (INEM), an agency of the Ministry of Labour, was exclusively responsible for employment policies across the territory and local employment offices were part of the INEM network. The aforementioned law formally abolished the central monopoly over LMPs by creating ‘The National Employment System’ (Sistema Nacional de Empleo, SNE), composed of the ‘National Public Employment Service’ (Servicio Público de Empleo Estatal, SPEE) and the seventeen ‘CAs Public Employment Services’ (Servicios Públicos de Empleos de las CAs).Footnote ¹¹ In turn, in Spain ‘there is not one unemployment service, but seventeen with their own logos, offices, staff and payrolls. They are organised differently and have their own systems of employment matching’ (interview with former Minister, Spain, 2011).

The ALMP institutional architecture assigns Spanish central institutions the responsibility for facilitating cross-regional coordination and directing and overseeing cross-regional policies. The SNE is mainly responsible for coordinating and monitoring LMPs across levels of government and for bringing cohesion between ALMP and PLMP. In addition, it defines the centre's policy line, which is in part established by the ‘National Conference on Labour Issues’ (Conferencia Sectorial de Asuntos Laborales, CSAL) – an intergovernmental arena to ensure intergovernmental coordination and political discussions between the centre and the directors of the regional PESs, and to allocate ALMP financial resources to the CAs.Footnote ¹² In contrast to Italy, where the intergovernmental Conferenza Stato-Regioni is not active in the coordination of ALMP devolution, in Spain CSAL plays a key role. Once a decree is approved by the CSAL, each CA must implement it within the boundaries of their competencies. When it comes to the fiscal dimension, the CAs have limited flexibility in spending ALMP monies given that they must comply with centrally specified directives and standards. Recently these procedures have become somewhat more flexible (interviews at the regional and local levels, Catalonia, Spain, 2011).

This description shows that the Spanish architecture of LMP institutions and competencies is grounded in a nested model in which the centre plays a dominant steering role and the CAs are given the flexibility to administrate and organise their own employment services within the boundaries of central mandates. Thus, the Spanish model of decentralisation shifted from an ‘exclusive (central level) + hierarchy’ to a ‘shared competencies + hierarchy’ model. In contrast to the Italian model, in Spain there are no major discrepancies in the implementation of decentralisation as the de facto intergovernmental arrangements and practices match the de jure institutional framework; in fact, when we capture de facto intergovernmental practices in Spain, the shadow of hierarchy becomes even more apparent.

Having described the building blocks of the new architecture of ALMP in Italy and Spain, we must ask: what are some of the manifestations of these two different types of intergovernmental models? What could these intergovernmental balances of power tell us about the overall nature of active welfare states in each country?

The manifestations of two models of ALMP decentralisation

This section shows how each model of decentralisation is linked to a distinctive set of developments regarding regional institutional diversity. When it comes to ALMP, the Italian horizontal model is characterised by noteworthy institutional variations across the territory, in comparison to the Spanish hierarchical model which is more cohesive. Given the lack of cross-regional and cross-temporal data (prior to decentralisation), it is not possible to establish the direct relationship between the overall levels of regional divergences on one side, and the adoption of an intergovernmental model on the other. Instead, we assess these issues by paying attention to a number of factors which look at institutional variations of the ALMP system in both countries. More specifically, we ask: were institutions and/or mechanism put in place to promote cohesion across the territory, or reduce regional divergences? If so, are these institutions active? How many independent ALMP systems are operative within the territory? Are there central institutions that possess ‘police patrol oversight powers’ (McCubbins and Schwartz, Reference McCubbins and Schwartz1984)?Footnote ¹³Are there institutions and/or mechanism entrusted with coordinating PLMP (an exclusive competence of the centre) and ALMP?

Between cohesion and variations: the implications of two models of active labour market policy decentralisation

Designing and managing new intergovernmental settings for implementing the activation paradigm are delicate matters, as decentralisation processes are full of tensions and paradoxes, especially when it comes to the fine line dividing the expansion of flexibility and the emergence of salient variations across the territory. On the one hand, decentralisation fits the activation paradigm as it allows for the expansion of regional and local flexibility, thus potentially bringing policies closer to welfare clients and local circumstances. On the other hand, certain shapes of decentralisation can expose welfare recipients to very different policy regimes depending on where they are located – a significant issue if we refer to the solidarity and cohesion goals of the welfare state. As Obinger et al. (Reference Obinger, Leibfried and Castles2005: 3) argue, ‘Central to the idea of social protection is the provision of nation-wide uniform social rights that supplement basic civil and social rights.’

The horizontal Italian model is grounded in substantial levels of administrative and political autonomy of regional, provincial and local actors over operational and formal aspects of ALMP (Carmel and Papadopoulos, Reference Carmel, Papadopoulos and Millar2003). In regions with sufficient administrative and fiscal capacity, mainly those located in the Central and Northern parts of Italy, decentralisation is linked to experimentation (Bifulco et al., Reference Bifulco, Bricocoli and Monteleone2008) and the development of a comprehensive activation system, thus promoting innovation, and bringing labour market policies closer to individual and local circumstances – a significant goal if we take into account that activation implies a shift from ‘people-sustaining’ to ‘people-changing’ measures (van Berkel, Reference Van Berkel2010: 29).

Nonetheless, as shown in the last section, the introduction of significant levels of flexibility through administrative and political means has also resulted in prominent institutional variations across the Italian territory, which translates into welfare clients exposed to very different types, conditions, quality and standards regarding accessibility of benefits and services based on their location. These developments might not be detrimental if central standards were enforced (e.g., universality of benefits across the territory, principles regarding conditionality), and/or regional or local governments had equivalent capabilities to provide and deliver ALMP. However, in Italy these conditions are not met. For instance, a report on the new employment structures concluded that institutional variations in the provisions of ALMP services have led to the development of an ALMP system with an overall institutional capacity that is ‘unequal’ and ‘asymmetric’ (ISFOL, 2007: 12). In particular, the report highlights the existence of noticeable differences between the Central and Northern regions and the rest of the country, which in turn ‘exposes citizens as well as firms to a long-lasting condition of unequal treatment and unequal access to the job market’ (ISFOL, 2008: 12).Footnote ¹⁵ Thus, a resident of Naples experiences a very different system of ALMP than a resident of Venice or of Puglia, for example. In the same vein, across regions, private providers have different types of opportunities and relationships with the regional public sector and welfare clients. At the individual and collective levels, these disparities are significant, as they can affect the likelihood of being active participants in the labour market. At the structural level, they could be one of the factors to help us understand cross-regional variations in unemployment, employment and participation rates in Italy (on this point see Altavilla and Caroleo, Reference Altavilla and Caroleo2006). These findings suggest that, despite significantly increasing regional autonomy, horizontal decentralisation might sacrifice the social citizenship and solidarity goals of the welfare state. In this country, these consequences become particularly relevant if we take into account that the central regulative framework was already frail, unemployment protection policies are highly fragmented (Bifulco et al., Reference Bifulco, Bricocoli and Monteleone2008) and that there are great regional disparities in administrative, institutional and financial capabilities across the territory.Footnote ¹⁶ Therefore, the results are not much different from Putnam's remarks more than fifteen years ago (1993: 61) about the future prospects of regionalisation and institutional stability: ‘the regional reform appears to be exacerbating, rather than mitigating, the historical disparities between North and South. The reform freed the more advanced regions from the stultifying grasp of Rome, while allowing the problems of the more backward regions to fester.’

By contrast, the administrative decentralisation of ALMP in Spain has not provided as much decision-making autonomy to the CAs. This means that, compared to Italy, citizens are exposed to similar conditions across the territory, including levels of conditionality. In this context, regional actors have called for the flexibilisation of central mandates which would allow the CAs to have more leeway in the implementation of ALMP. Consequently, the central level has started to, slowly, give sub-national PESs and private actors more flexibility in the use of central funds, as well as in the organisation and implementation of activation programs; yet this process is still pending (interview with trade unions and civil servants, Catalonia, Spain, 2011).

A noteworthy consequence of decentralisation from the centre to the Spanish CAs is the reduction of municipal powers over ALMP. For instance, a local civil servant noted,

Furthermore, the interviewee claimed, only European Union funds allow them to experiment with ALMP, given that there are many strings attached to central and regional monies.

The evidence provided on the Spanish case suggests that, through the development of central standards and interventions, a hierarchical intergovernmental model of decentralisation can limit the range of regional and local flexibility. However, such model attempts to maximise cohesion by reducing variances across the territory, thus, better upholding the solidaristic and equality goals of the welfare state. In this way, hierarchical systems are linked to a more homogeneous LMP system, but less regional autonomy.

As spatial re-configurations of the welfare state become more salient, these cross-national findings have important theoretical and practical implications for the design and governance of active welfare states, as well as for welfare clients. Future studies should expand upon our findings by capturing intergovernmental activation models, and applying the categories developed here to additional countries. Such studies would allow us to have a better grasp of the implications of different architectures of decentralisation on activation regimes, as well as on welfare states, including health policy, for instance.