The reform of European cohesion policy or how to couple the streams successfully

ABSTRACT Cohesion policy is one of the European Union’s central spending policies, which is characterized by path-dependency and only incremental change. However, the policy has been reformed significantly in 2013. This case study sets out to trace this ground-breaking and significant reform in a fairly path-dependent environment by using Kingdon’s multiple streams approach (MSA).The approach is adapted to the characteristics of policy-making in the European Union, the particularities of the Union’s multilevel governance with its institutionalized and well-practised policy-making and almost predictable policy windows. The modified approach helps to separate two windows of opportunity with two different policy entrepreneurs. The first agenda window with the European Commission as entrepreneur prepared the grounds for the decision window with an entrepreneurial group of Member States. The second coupling reframed and refocused the proposals of the agenda coupling and linked cohesion policy more closely with the Union’s economic policy co-ordination.


Introduction
Cohesion policy is one of European Union's central policies, however, a very complex one and maybe 'one of the most difficult EU policies to define and categorise' (Bachtler, Mendez, and Wishlade 2013, 11). The policy aims at reducing disparities in prosperity and welfare, and promoting economic and social convergence within the EU, its Member States and regions. For more than 25 years the promotion of convergence and cohesion are the almost traditional objectives of the policy.
Though, the policy has always been subject of constant debate and of pressure to adapt and reform, comprehensive, wide-ranging changes and fundamental reforming of the policy had been difficult (Piattoni and Polvari 2016;Bachtler, Mendez, and Wishlade 2013). In principle, only path-dependent modifications and incremental adaptions of the policy and its tools had been achievable (Mendez 2012). Nonetheless, a significant reform in 2011/13 established a new architecture of the policy, defined new goals and finally changed the task and the raison d'etre of the policy. Inspired by the so-called 'Lisbon-strategy' and the succeeding 'Europe 2020-strategy' this 'Lisbonization' of the policy (Mendez 2011;Baun and Marek 2014) re-focused the policy and connected it with new instruments and mechanisms of European economic policy co-ordination and thus overwrote the traditional goals and tasks of convergence and cohesion. The policy 'underwent the most significant and substantial set of regulatory changes . . .' (Bachtler et al. 2017b, 1) and developed into a 'clear investment strategy in every region' of the European Union (European Commission 2010a, 10). European Commission officials characterized this recent reform as a 'wide ranging reform' (Berkowitz et al. 2015, 2) that indicates a 'paradigm shift from a transfer/absorption-oriented policy to a growthoriented investment policy focused on results and fully embedded in the coordination of economic policy' (Deffaa 2016, 157). The policy switched the path of limited modification and became the subject of fundamental change.
The question hence is, why and how this paradigm shift and significant reform had been possible and successful in a fairly path-dependent environment. One of the bestknown approaches to analyse agenda changes in political systems is still the 'Multiple Streams Approach' (MSA) by John W. Kingdon. According to his approach, agenda change can take place whenever a policy window opens the opportunity to discuss a problem that is placed high on the political agenda by political entrepreneurs, perceived there by policy makers and coupled with a specific solution (Kingdon 1995). The opportunity for change hence is the result of coupling three relatively independent 'streams' of problems, possible solutions and political attention flowing separately through the political system. The streams must come together at the right time when a window of opportunity for change opens. Then policy entrepreneurs, looking for a problem to link their pet solution, 'apply manipulative strategies to achieve their goal of coupling the three streams' (Zahariadis 2003, 156). MSA thus combines political process with characteristics and structures of a policy in a specific situation and provides a framework for process tracing (Barzelay and Gallego 2006;Rüb 2016). While initially developed by John W. Kingdon to explain the agenda setting in the US presidential political system, Nikolaos Zahariadis extended the approach to parliamentary systems and to additional stages of policy-making including decisionmaking (Zahariadis 2003(Zahariadis , 2008. The approach has been developed further, applied in a number of case studies and developed to one of the most-cited theoretical approaches to analyse policy-making (Cairney and Jones 2016;Jones et al. 2016;Zohlnhöfer, Herweg, and Rüb 2015).
MSA emphasizes the ambiguity and contingency of agenda-setting and policymaking and is certainly an approach 'that explains how policies are made by national Governments under conditions of ambiguity' (Zahariadis 2007, 65). Therefore, the approach seems to be suitable to capture the extraordinary complexity of the EU political system and the ambiguity, which 'is an indispensable part of the EU policy process.' (Zahariadis 2008, 527; also Ackrill and Kay 2011). The EU's multilevel governance system lacks transparency and comprises distinctive complexity and ambiguity. 1 This multilevel governance system comprises public and private actors and institutions not ordered hierarchically on supranational, national and regional levels. All are engaged in bargaining over policy and bringing in a set of specific interest and goals. As interdependence and linkages between levels and interaction between actors increases, so does the complexity, ambiguity and lack of transparency of the policy process. This multi-level polity with its specific dynamics '. . .make it difficult to distinguish clearly between developments at different levels and to determine in a general sense which level or institution plays the leading role. Rather, debates take place on different levels simultaneously and mutually reinforce (or at least affect) each other' (Princen 2009, 157). However, this dynamics vary between policy areas and sectors. And cohesion policy is maybe the best example to analyse this ambiguity of European policy-making, because this policy is characterised by a specific form of European multilevel governance (Sutcliffe 2000, Hooghe andMarks 2001) and stands for complex and opaque policy-making on different territorial levels comprising a huge number of governmental and non-governmental actors (Marks 1996). Gary Marks defined cohesion policy therefore as a process of 'continuous negotiation amongst nested governments at several territorial tierssupranational, national, regional, and local -. . .' (Marks 1993, 392).
The main argument of the paper is that by using MSA a comprehensible structure of these simultaneous processes in this multilevel polity and a sober analysis and process tracing of agenda setting and decision-making is possible. This article will first unfold MSA and try to adapt the approach to the particularities of EU multilevel governance and then use its tools to trace the process and analyse the results of the reform of European cohesion policy.

Multiple streams and European multi-level governance
MSA identifies three streams flowing independently through the political systemproblems, policies, and politics. Policy entrepreneurs can 'couple' these streams by promoting a particular policy and by manipulating policy-making to advance their policy aims. In the problem stream real or constructed problems struggle for attention of political actors. Situations can become a problem and then can attract political attention when 'we come to believe that we should do something about them' (Kingdon 1995, 109). The politics stream determines the responsiveness of political actors to certain problems and their favoured policy answers. These answers are already flowing around in the 'primeval soup' of ideas and options (Kingdon 1995, 116-144) in the policy stream, where they are constantly discussed, examined and reconsidered in communities of policy experts comprising researchers, stakeholders and lobbyists, government officials and Members of Parliaments. Out of this broad spectrum of ideas and proposals, only a few will win political attention. In contrast to the deliberations in the policy stream, the actors in the politics stream will not try to convince other actors with new arguments and better ideas, but bargain about political concepts or solutions and create new majorities and coalitions. Occasions and opportunities to couple the streams are rare and coupling only possible when the time has come. Kingdon developed for this particular moment the concept of 'policy windows'. Why and when a window will open (or close) is determined by events in the politics-or in the problem-stream, which is why Kingdon distinguished between politics and problem windows (Kingdon 1995, 173). Furthermore, Kingdon separated windows that open predictably from unpredictable windows. Howlett (1998) combined these categories of predictability and origin of policy windows and developed a model of policy window types according to their institutionalization ranging from routine to random windows. Institutionalized windows open more frequently according to their regular cycles or patterns and include more routine. The more routine, the more an issue can be handled by specialists through standard operating procedures (Kingdon 1995, 186). This leads to the differentiation developed by Princen and Rinhardt, who discuss the distinction from where issues can be placed on the agendafrom below or from above of a polity. An issue is placed by political leaders from above on the agenda or by experts and specialists from below (Princen and Rinhardt 2006).
When the time has come and a policy window opens, a 'policy entrepreneur' is needed to couple the streams. Then, the origin from where an issue has come on to the agenda is important for the process of framing the issues by the political entrepreneurseither broad political outlines or specific technically sound proposals Rinhardt 2006, 1121). Entrepreneurs are 'advocates who are willing to invest their resourcestime, energy, reputation, moneyto promote a position in return for anticipates future gain in the form of material, purposive, or solidary benefits' (Kingdon 1995, 179). Their role as 'power brokers and manipulators' (Zahariadis 2007, 74) is to put an issue on the political agenda, to prepare politics and then to couple the streams. Four elements are central for their entrepreneurshipdisplaying social acuity, defining problems, building teams, and leading by example (Mintrom and Norman 2009;Crespy and Menz 2015). Zahariadis distinguishes framing, salami tactics and the use of symbols as manipulative strategies used by entrepreneurs to manipulate 'the various dimensions of policy images to activate the values within' (Zahariadis 2003, 156) and so finally to gain the support of the political decisionmakers. Framing problems by entrepreneurs is decisive to categorize an issue as specific problem, to highlight certain aspects or to tone down others, and to connect the preferred solution with gains and others with losses (Knaggård 2016). 'The way problems are defined and their match to policy-maker norms and biases influences where they end up on the agenda' (Zahariadis 2016, 7). The entrepreneur develops an appropriate frame for a problem and thus manipulate the 'perceptual interpretive element' (Kingdon 1995, 110) of the definition of the issue. Their success depends certainly on their access to decision makers, the resources as for example time and money they can use, and their strategies.
When using MSA to analyse European policy-making the approach and its elements require some adaptation to the particularities of EU's multilevel political system (Ackrill and Kay 2011). EU policy-making comprises various levels of governance with a multiplicity of actors involved in policy-and decision-making, and hence 'multiple access points' (Princen 2016, 353) for policy entrepreneurs (Peters 1994). The complex political system of the European Union with specific forms of decision-making and the variety of actors and stakeholders influencing the policy making correspondents with the multiple streams approach of policy making under constraints of ambiguity. European legislation follows decision-making procedures with predetermined roles and thus strengthens the dominance of institutions. This institutionalized and wellpractised policy-making with clear allocation of roles for actors facilitates planning of political processes and therefore the predictability of policy windows.
In addition to the supranational EU-institutions Member States' governments and administrations are still playing a central role in EU policies (Kohler-Koch 2005;Bulmer and Lequesne 2005). This means, the factors influencing the politics stream Kingdon had worked outchange of government, national moodare not accurate and sufficient to capture policy-making in a Union of 28 Member States. National elections, the change of government and hence changes in the political composition of the Council happen continuously without changing European policy-making fundamentally; a European 'national mood' or a pan-European 'Zeitgeist' is seldom and rare to find and finally only occasionally public campaigns of social movements are able to influence the European decision-making process. This problem of difference and divergence between the streams flowing on the European and the national levels is also obvious for the policy stream. The possibility that an 'emerging consensus' (Kingdon 1995, 139) may change the continuous discussions in the policy stream of the European Union and its Member States is small. Political stakeholders in the Member States may interpret the criteria, technical feasibility and normative acceptance for the selection of a specific solution flowing through the European policy stream differently.
To convince and manipulate the political actors in streams on European and national levels the discussion, the definition of problems and the framing of solutions have to converge (Princen 2009). The importance of national governance and its abilities to influence agenda setting and decision-making in the EU are dependent, however, on the considered policies (Heinelt and Knodt 2011). As cohesion policy and its main instruments, the structural funds, developed into main points of reference for the concept of European multilevel governance, this nested policy incorporates specific forms of convergence. The policy's 'vertical ambiguity' circumscribes the variety of actors from different levels of governance involved in the policy-making (Sarmiento-Mirwaldt 2015)with even more access points for political entrepreneurs.
In general, the Commission is perceived as entrepreneur for European policy-making because of its right to initiate legislative processes and policy changes (Crespy and Menz 2015;Mintrom and Norman 2009;Copeland and James 2014;Laffan 1997). The Commission is able to analyse and to converge problems as common problem and thus to define an issue, to link its analysis with a specific solution, and to generate the necessary convergence or even consensus between the Member States (Ackrill et. al. 2013, 875). As entrepreneur, the Commission will frame the problem according its pet policy solution. In accordance with the distinction between the level of governance from where an issue is placed on the agenda, from above or below, Princen describes two ways framing an issue may be done in the EUthrough 'big words' or with 'small steps' (Princen 2011). The origin of the opening of a window and the style of framing by entrepreneurs, however, can influence the width and the duration of an open window and thus can predetermine the extent of possible changes. If a window opens 'from above' by political leaders in the European Council or the issues are placed on the agenda 'from below' by experts in working groups of the Council or the European Commission (Princen and Rinhard 2006) will have different consequences for entrepreneurs and the way how to couple the streams. In case a window opens 'from above' and the issue will be framed through 'big words', political and ideological aspects play a significantly greater role and thus increase the opportunity for fundamental or sudden changes. Policy windows, which open 'from below' follow usually the technical and administrative procedures of European policy-making and allow typically for incremental modifications. This often is combined with lengthy processes of discussion and selection in the policy stream, which idea might be best, and of 'softening up the system' (Kingdon 1995, 200/201). However, the existence of an entrepreneur, successfully using a policy window to couple all three streams, 'will ever be enough to cause policy reform: we always require an account of the context' (Ackrill et.al. 2013, 879). In other words, a successful entrepreneurship will depend on the environment and the situational conditions of the process and probably also on the stream in which the entrepreneur is located (Mukherjee and Howlett 2015). Zahariadis (2008, 520-525) specified four conditions, which determine the likelihood of successful coupling: first, the effectivity of the policy entrepreneur; second, the framing of the policy proposals to match the preferred solution of political decision-makers; third, the specific arena or venue for agenda and decision coupling and, fourth, the specific policy or governance mode, which influence the degree of influence and leeway for political actors.
The maybe broadest and most far-reaching adaptation of Kingdon's theoretical approach is to extent the framework to the process of decision-making (Zohlnhöfer, Herweg, and Huß 2016). Including the stage of decision-making means that, according to John W. Kingdon, an issues moves from the 'governmental agenda', which is 'a list of subjects to which governmental officials are paying serious attention', to the 'decision agenda', which is 'a list of subjects that is moving into position for an authoritative decision, such as legislative enactment or presidential choice. . .' (Kingdon 1995, 202). Although agenda setting and decision-making are fluid and blurred phases of the same policy-making process, distinguishing both phases seems necessary (Zahariadis 2003;Princen 2009). The distinction allows to separate only 'partial couplings' from 'complete joining of all three streams' (Kingdon 1995, 202) and hence to analyse successful policy change (Herweg, Huß, and Zohlnhöfer 2015) at least in multilevel policy systems. This means to differentiate between two coupling processes including two windows -first 'agenda coupling' and then the 'decision coupling' (Herweg, Huß, and Zohlnhöfer 2015). Both couplings take place at different points of time during the policy process and will be opened in different streams. A fact, which has consequences for the windows, the coupling, the actors and entrepreneurs. The first coupling is possible when the agenda window opens, either in the problem or in the politics stream. After policy entrepreneurs used successfully this window to place their preferred solution on the political agenda, the second window opens with the start of the formal decision-making process. This decision window, however, opens usually in the policy stream as a 'worked-out proposal that gained agenda status as a result of the agenda coupling' (Herweg, Huß, and Zohlnhöfer 2015, 444). This means, that the actors and stakeholders in the political stream start to bargain and to influence the proposal immediately. 'Hence, the political stream is of prime importance for the decision coupling while the problem and the policy stream are of minor importance' (Herweg, Huß, and Zohlnhöfer 2015, 445). Entrepreneurs are needed to start agenda coupling, but not necessarily to trigger decision coupling. Entrepreneurs are no game changer. The distinction between two windows therefore can also mean to differentiate between different entrepreneurs.
To sum up, the adaptation of MSA to European policy-making means to differentiate between various actors, their specific roles during the phases of policy making on different levels of governance. Moreover, using MSA for agenda-setting and decisionmaking means to distinguish between two separate couplings with different windows of opportunities and entrepreneurs.
Assessing change in European cohesion policythe streams and their coupling Financial instruments and efforts to address economic and social convergence and cohesion at EU-level did exist since the first steps of European integration. The first qualitative turn from simple European co-financing of national policy measures according annually fixed national quotas to a genuine European policy opened 'a new era' for cohesion policy (Bachtler, Mendez, and Wishlade 2013, 45). The political context had been favourable for comprehensive change. The European Commission acted very successfully as policy entrepreneur and linked this '"big bang" reform' (Mendez 2012, 158) with two projects, its proposal for a new instrument, the financial perspective (the Delors I-Package), and its internal market programme. And the Commission used the external impact of the accession of Spain and Portugal to substantiate the need and the urgency for policy reform (to history and development of the policy see Allen 2005; Bache 2015; Bachtler, Mendez, and Wishlade 2013;Baun and Marek 2014). After the 'birth of cohesion policy' (Brunazzo 2016, 22) in 1988 the policy has changed in a continuous process of subsequent modifications, adaptation and reform, in order to improve effectiveness and efficiency of the policy and its instruments. Then, after the first introduction of new governance instruments (Medenz 2012), the policy has been fundamentally reformed in 2013. The old redistributive task of structural funds and the objective of reducing regional disparities with the help of European funds have been replaced by a new focus on growth and investment combined with enhanced efficiency and effectiveness of the policy.
Cohesion policy as an important spending policy is closely linked with European budgetary policy, especially the negotiations to settle the multiannual financial framework (MFF), which defines the ceilings for spending policies. But, although the MFF and the multiannual programmes for implementing cohesion policy in European regions are running in parallel, the legislative procedures of both are different. Whereas the MFF is decided by unanimity and with consent of the European Parliament, the structural funds regulations are decided according the ordinary legislative procedure, i.e. with qualified majority in the Council and with the European Parliament as co-legislator.
This parallelism as well as the difference in legislative procedures, however, have consequences and implications for both negotiation venues and for negotiation tactics. It allows influencing the opening of a policy window on one venue by blocking or slowing down negotiations on the other. To use the terms of MSA, the parallelism has consequences for policy and politics streams, entrepreneurs, and policy windows. The discussions in the policy stream of budgetary policy can make options discussed by policy networks in the policy stream of cohesion policy more likely or can block ideas developed in the cohesion policy stream. Timing, leeway and leverage for specific options to reform cohesion policy are framed and pre-determined by decision coupling of MFF negotiations. The politics stream of cohesion policy hence is closely linked with the MFF-politics stream. This again has consequences for policy entrepreneurs trying to couple the streams of cohesion policy. They have to know their streams and policy options when the window opens, and additionally they have to observe the parallel debates and negotiations of the MFF-streams. The agenda window of cohesion policy will open predictably and predetermined by the MFF-negotiations. It will open from below, when the Commission tables its legislation proposals, and the decision window will open from above, when the European Council could hammer out a compromise on the MFF, which sets the frame for cohesion policy reform options. The analysis of streams and couplings of cohesion policy hence has to observe also the MFFnegotiations.

The problem stream
Cohesion policy has been subject of massive and constant criticism since its establishment in the eighties. In general, these debates circled around the task, the purpose, and the significance of the policy. For some critics the policy aims at redistribution and, as kind of side-payment, at compensation of certain Member States to win their support for economic integration initiatives (Leonardi 2005). The major complaint has been that cohesion policy cannot satisfy its tasks and objectives as laid down in the European treaties (Sapir 2003, 55-67) and that the policy lacks effectiveness and cannot prove empirical evidence for convergence as a result of structural funds expenditures (Santos 2008). Although the Commission tries to prove regularly that European funds work effectively and help to achieve clear progress of convergence and cohesion not all observers share the appropriateness of the Commission's methods and results of these calculations and econometric models (Cappelen et al. 2003;Marzinotto 2012).
In the run-up of the negotiations for the period 2014-2020 the policy became a field of extensive evaluation, continuous monitoring with continuous performance measurement and evaluation (Bachtler et al. 2017a). A number of other studies recognised a positive effect of structural Funds on growth and employment, even if the effect is often small von Ehrlich 2012, 2018). Critics therefore doubt whether there is a clear causality between EU funding and the positive impact on growth and employment (Le Gallo, Dall'Erba, and Guillaint 2011;Hagen 2010, 2011) or even conclude that EU funding shows negative spatial effects (Breidenbach, Mitze, and Schmidt 2016). Criticism is connected with the accusation of disproportionate promotion and achieved effect (Heinemann 2013).
Other observers underline the allocative function of cohesion policy to enhance economic growth and raise the welfare of regions and citizens . For these observers the policy provides added value (Mairate 2006) and is 'by its very nature . . . the most elaborate expression of European solidarity in its myriad forms' (Jouen 2017, 4). The empirical studies employ different approaches, models, and analyse different time frames and thus come to differing results and different conclusions. The attempts to empirically prove the effectiveness of European structural funds and the corresponding econometric models remain, at least, controversial. Furthermore, the conclusions drawn on the basis of the predominantly positive econometric impact and evaluation studies are very different (Pienkowski and Berkowitz 2017).
These more or less general and fundamental points of critique of cohesion policy coincided with a wide-ranging mid-term review of European budget in 2009/10 covering all spending polices, including Cohesion policy. In the framework of this review the policy came under 'threat' and high legitimation pressure (Petzold 2013, 7). This fundamental challenge became even more pressing as consequence of the severe financial and economic crisis in the EU and the Eurozone in 2010/11. The European sovereign debt and economic crisis left deep marks: Public investment fell by 20 percent and in the hard hit crisis countries like Greece, Spain and Ireland by around 60 percent (European Commission 2014); unemployment increased significantly and social inequality became a major problem within the European Union. This break down of investment combined with the growing need to finance supply-side measures for economic recovery and employment increased the importance of European funds (Berkowitz et al. 2015). The dramatic fall of economic activity, investments and employment in 2010, hence, was 'a wake-up call for EU cohesion policy' (Deffaa 2016, 155).
On the one hand, the policy remained the most important sign and instrument of European solidarity that became even more necessary in times of crisis. The main recipient countries asked for more European assistance and solidarity and their expectations grew, particularly in the crisis countries. On the other hand, net-paying countries argued for a more efficient use of European funds to promote growth and jobs and denied to increase their payments to the EU budget. They argued that for many decades the policy could not achieve the objective to reduce divergence.
The crisis reduced the financial leeway of all Member States, and intensified traditional conflicts over distribution of European funds. European cohesion policy had to struggle with other policies for scarce funds, political attention and strategic steering capacities. This led to a mismatch of expectations and proposals for the future of cohesion policy, and exaggerated the debates on its legitimacy, purpose and tasks. Given the economic and social consequences of the crisis, cohesion policy came under heavy pressure and the need to justify the policy fundamentally.

The policy stream
Since its first fundamental and path-breaking reform in 1988 the policy developed an unique network of European, national, regional and societal experts, stakeholders and decision-makers. Gary Marks described this policy community as 'the leading edge of a system of multilevel governance in which supranational, national, regional and local governments are enmeshed in territorially overarching policy networks' (Marks 1993, 401).
Very early, this policy community started to discuss the next steps for further reform of cohesion policy. These deliberations in the policy stream started with a new narrative to re-define and to re-focus the policy. Carlos Mendez argued: 'More so than in any other reform, the post 2013 review was underpinned by a structured and sustained debate with stakeholders, politicians and experts . . . ' (2013, 640). In parallel, the Commission launched a academic policy review process. The Commissioner for regional policy Danuta Hübner commissioned the Italian Director-General at the Ministry of Finance and Economy Fabrizio Barca to elaborate a comprehensive report. DG Regio in the Commission organised between April and October 2008 a series of five policymaker seminars with government officials of all Member States (in a personal capacity) and representatives of European regions and several hearings on various issues. The final report vehemently argued for refocusing the policy on spatial tasks and to develop a socalled 'place-based development strategy' (Barca 2009). To improve the efficiency of the policy the report suggested a greater concentration of measures and funds on a few narrowly defined core priorities and a more result-oriented policy approach including clear objectives. Additionally, the report argued, that the reformed policy should strengthen the social dimension of cohesion policy (Fargion and Profeti 2016, 482/ 483). The Barca-report thus included a 'politically motivated subtext to protect and relegitimize in the face of growing criticism about its rationale and effectiveness' (Mendez 2013, 644). The report offered arguments to significantly reform cohesion policy, and set the frame for debates in the policy stream. Not only the report but in particular the process for its preparation prepared the background and a resonance board for the deliberations in the policy stream and impacted on thinking and positioning of the European Commission.
So, when the policy window opened the policy community had already started a broad and comprehensive debate on the fundamentals as well as on specific issues of the policy. This made it easier for the Commission as policy entrepreneur to come back to this debate and to pick up specific ideas. Following the role as policy entrepreneur, the Commission was engaged in this debate and targeted the discussion in the policy stream on its own questions and ideas, its own pet solutions.

The politics stream
For the funding period 2014-2020, the Lisbon Treaty established with Article 175 TFEU a new legal basis and context for the negotiations on the new legislation. The legislation procedure changed from the old 'assent procedure' to the new 'ordinary legislative procedure' providing for a more decisive role to the European Parliament as fullyfledged co-legislator. As usual, the European Union negotiated the legislation for the structural funds and the future development of cohesion policy in the context of the negotiations on the MFF 2014-2020. Both processes were closely interconnected and these parallel negotiations followed the choreography of almost traditional lines of conflict between net contributors and net recipients of the European budget.
The Member States had already grouped themselves into two camps, the 'Friends of Cohesion' consisting of 16 Member States 2 receiving high amounts via structural funds, and the 'Friends of Better Spending' representing the group of net paying countries. In April 2012, the German Federal Government launched this initiative to make cohesion policy a more effective policy generating higher economic growth and more jobs (Friends of Better Spending 2012). 3 Given the limited financial resources and the need to reform, the Union's budget, structural funds should be better targeted and used more efficiently. Moreover, structural funds should be directly linked to the instruments of European economic policy coordinationthe Europe 2020-Strategy and the country specific recommendations in the framework of the European Semester. The structural funds should become instruments to implement European economic policy-making, to set incentives for national structural reforms and even to sanction Member States in case of missing the common targets defined by the Europe 2020 strategy.
As an answer to this initiative of net payers, the 'Friends of Cohesion' adopted in December 2012 a joint declaration in which the Prime Ministers of these 16 Member States argued that cohesion policy must remain a powerful instrument of European solidarity and the major instrument to finance infrastructure in less developed regions (Friends of Cohesion 2012). The 'Friends of Cohesion' understood the better-spending initiative as an attempt to reduce structural funds and consequently rejected the initiative sharply.
In addition to the Member States, the European Parliament as co-legislator influenced the development of this central European policy. The Parliament's committee on regional development developed a close collaboration with the Commission's Directorate general for regional and urban policy and also with the Council (Hübner 2016). Already in November 2009, the Committee on regional Development set up an informal working group to discuss the future cohesion policy. This group held various hearings and elaborated a position paper endorsed in July 2010 to influence the Commission's proposal. In July 2011, the European Parliament adopted a first resolution at the beginning of the bargaining process (European Parliament 2011). The Parliament hence defined at an early stage its negotiation objectives and ideas to reform cohesion policy and tried subsequently to enforce its objectives during the legislation procedure (European Parliament 2014).

The European Commission as policy entrepreneur and the first agenda coupling
The policy stream with its various networks of experts and practitioners on the several levels of European multilevel governance system prepared the grounds for the entrepreneurial European Commission. The Commission was able to start, to structure, and to guide debates in the policy stream, respectively in the policy community. Already the fourth Cohesion Report tabled by the Commission on 30 May 2007, included a list of questions that should deliver incentives and guidance to structure the debate on the future of cohesion policy in these policy networks (European Commission 2007). The Commission then tried to elaborate its proposals in more detail and to define a new concept of territorial cohesion with a Green Paper (European Commission 2008) and a subsequent public consultation on this concept (European Commission 2009). Already in the run-up for the next funding period two Commissioners for Regional Policy, Danuta Hübner and her successor in that office Paweł Samecki, tabled personal reflection and position papers (Hübner 2009, Samecki 2009).
Both Commissioners used the Barca-report to frame and set the stage for the debate on the future of Cohesion policy and to put forward concrete orientations to consolidate and reform the policy. Carlos Mendez argues that the 'place-based narrative' of the report could be understood 'as a part of a legitimization strategy used by DG Regio and its proponents to rebuild an identity for cohesion policy, to reconcile long-standing ambiguity and tensions surrounding the policy's rationale, and to protect it against attacks during the EU Budget review' (Mendez 2013, 648/649). Hence, the Commission, and especially DG Regio, used their entrepreneurial tasks and opportunities to promote new regional policy ideas and intensified its contacts and deliberations with institutional partners, academics and international organisations 'to provide new perspectives on the economic rationale of Cohesion policy' (Berkowitz et al. 2015, 8/9). To organise and guide the follow-up of the Barca-process the Commission established a High-Level group of experts in the Member States's governments with the experts in the Commission services. Finally, in its fifth cohesion report in November 2010, the Commission gave some hints on the direction of reform and how the Commission intended to design the policy more 'result-oriented' (European Commission 2010b). To achieve this goal, the policy should be equipped with new and modified planning and implementing instruments. European funds should in the medium and the long run be refocused and targeted more precisely on economic growth, job creation and the promotion of investments.
In a further step and prior to the preparation of the new legislative proposals, the Commission sought the dialogue with the European Parliament, the Member States, regions, and stakeholders and launched a public consultation on the future cohesion policy from November 2010 to January 2011. The Commission collected ideas, suggestions and proposals and tried to gain a comprehensive picture of the debate. The Commission in its role as policy entrepreneur exploited this discussion for the preparation of the next funding period 2014-2020 (European Commission 2011a).
Thus, when tabling its legislative proposals for the structural funds regulations the Commission had an additional momentum in its double role as entrepreneur and agenda-setter (Princen 2009). The Commission tried at an early stage as agenda setter with diverse and innovative initiatives to identify new challenges and to pre-structure the anticipated debate and difficult negotiations on the adaptation and reform of cohesion policy after 2014. Using its role as entrepreneur, the Commission thus prepared its legislative proposals and outlined its ideas and thinking on how to reform the policy and its strategic goals (Bachtler, Mendez, and Wishlade 2013, 143-161). The Commission used its almost exclusive right for legal initiative to define the policy priorities and thus set the negotiating issues (Tömmel 2016).
The first coupling started when the Commission tabled its proposal for the MFF 2014-2020 on 29 June 2011. The MFF-proposals set the annual ceilings for spending policies and developed the framework for financial planning and also allocated the funds to individual Member States. In its MFF-package the Commission proposed a financial volume of 376 bill. Euros (in prices 2011) for cohesion policy, which means the share of the structural funds in total MFF-expenditures should increase from 35 to 36.7 percent. Though cohesion policy funding should fall by 5% in absolute terms, the policy should slightly increase in relative terms to the overall size of the budget. The bulk of the funds should again be reserved for the poorest regions in the European Union. However, as new idea the Commission proposed to introduce an intermediate category for so-called 'transition regions' with a GDP between 75 and 90 percent of the average EU-GDP, that means an additional category for regions between the less developed and the more developed regions. The Commission, however, tried with its proposal to reframe the legitimacy and the raison d'etre of the policy. 'Cohesion policy is no longer only a tool to accompany market integration and address its territorial consequences, but is a lever to promote development at national, regional and local level in line with EU-wide long term policy priorities' (European Commission 2011b, 50).
Then, on 6 October 2011, the legislative proposals for cohesion policy followed with the draft regulations for the structural funds. This comprehensive legislative package included six draft regulations and specific proposals to simplify the management of the funds and to make the policy more consistent and effective (Mendez, Bachtler, and Wishlade 2012). The Commission proposed major changes for programming and implementation, such as to establish new forms of conditionality to ensure better effectiveness and successful implementation of programmes. The introduction of a performance reserve of 5 percent of national allocations should serve as an incentive for more efficiency and better spending and should be paid out at the end of the funding period to particularly successful programs. The ex-ante conditionality means that fundamental cross-cutting objectives and strategies like anti-discrimination or gender policy should be in place at the beginning of the funding period, and thus guarantee the most efficient use of European funds. Regions should have in place all requested administrative prerequisites with regard to public tenders, statistical surveying systems and other necessary legal requirements for the successful implementation of their implementation programmes, the so-called 'Operational Programmes'. Moreover, macroeconomic conditionalities should link cohesion policy to the macroeconomic objectives of economic policy co-ordination in the EU. Structural funds should better serve the fundamental economic objectives of the EU as stipulated in the strategy Europe 2020 and as laid down in the country-specific recommendations for reform and for sound finances as part of the national stability programs. Finally, to better assist the Europe 2020 strategy the structural funds should concentrate on objectives derived from this strategy for smart, sustainable and inclusive growth, such as promoting research and innovation, energy efficiency, climate change, SME competitiveness and the fight against poverty.
The Commission hence picked up the proposals of the Barca-report and the ideas discussed in the primeval soup of the policy stream. The funds should be concentrated on less but more strategic objectives and serve as instruments of a new common investment strategy for growth and jobs. But following in general the 'Better Spending'initiative the Commission then aimed at developing the structural funds to become the funding instrument of its pet-strategy 'Europe 2020ʹ and 'launched the policy as an appropriate tool for improving the competitiveness of the member states as well as their less-developed regions' (Tömmel 2016, 110). The framing of a 'place-based'-cohesion policy had been overwritten by the 'Europe 2020 master narrative', the pet-solution of the Barrosso-Commission (Copeland and James 2014; Bache 2013).

The second couplingthe decision window with a new entrepreneur
After the Commission had submitted its legislative proposals for the structural funds the agenda window closed. The first policy entrepreneur, the European Commission, dominated agenda coupling, tabled its policy proposals and coupled the streams. However, this role as entrepreneur fades away and the Commission could only indirectly influence the decision-making process, mainly through the aggregation of data and figures to serve the negotiations in the Council. The centre of policy-making shifted from the Commission to the Council. In the run up to the second, the decision window, the importance of the problem stream decreased.
The debate on the future of European cohesion policy took place in a severe and difficult policy context. The close linkage between the MFF-negotiations and the attempts to reform cohesion policy complicated the negotiations even more. The sovereign debt and economic crisis in the Euro area made it more difficult to hammer out a compromise between the Member States on the next MFF and in that framework also on cohesion policy reform. This severe crisis hence influenced as 'focusing event' the negotiations, to use Kingdon's terms. The focus and the function of the debate on the future of cohesion policy changed, because the overall budget has to be passed unanimously, which means that a consensus has to be reached also about the question of how much money should be provided for cohesion policy. To get a deal, the Member States obviously need cohesion policy as an adjusting screw for this package deal on MFF. The Member States, their national interests and their big budget bargain could not be by-passed. Because of the importance of cohesion policy for the MFF and the close linkage between both negotiation arenasno cohesion-policy legislation without a consensus on the MFF and vice-versathe Council had to search for consensus on both venues. This means, despite the option to decide with qualified majority, the Council had to find also a compromise acceptable for all Member States.
Because of their strong position in the MFF negotiations the net payers took over the role as policy entrepreneurs for cohesion policy negotiations and opened the policy window 'from above'. They were able to use the deep economic crisis in the problem stream to couple their policy goals, i.e. enforcing economic structural reforms in crisis countries and increasing efficiency with cohesion policy funding. This entrepreneurial group in the Council could dominate the preceding MFF-negotiations and could set ceilings for the spending priorities and limit the overall size of the EU budget and for the cohesion policy heading. After a first attempt failed in November 2012, the European Council ended its negotiations on the MFF-package in February 2013 after more than 24 hours of continuous negotiations. The president of the European Council, Herman van Rompuy, had already in November took over the negotiations and bypassed the established negotiation venues in the Council, Coreper and the Friends of the Presidency group. He sought close bilateral deliberations with all heads of State and Government, but especially with the group of net paying member states. (Ludlow 2012).
The final political compromise of high politics in the European Council on the MFF set the stage and limited the reform options for the negotiations on cohesion policy. The European Council reduced the financial size of the heading for cohesion policy in the MFF significantly from EUR 339 bn. as proposed by the Commission to finally EUR 325 bn. (in prices 2011) as set in into the final MFF-compromise in February 2013. Moreover, the European Council also fixed the distribution of these funds between less developed regions, transition regions and developed regions and hence pre-allocated the funds on Member States and their regions. The scope of negotiations issues for cohesion policy reform thus was restricted to technicalities and implementation issues. For the Member States, the compromise hammered out by their Heads of State and Government in the European Council was non-negotiable. So, decision coupling began with another predictable opening of the window. However, the Council had to recognize the role and the positions of the European Parliament as co-legislator.
During the agenda coupling, the Commission had already proposed to reform cohesion policy and connected the new policy closer to the Europe 2020-strategy. The net contributors now were successful with their Better Spending approach and could reframe the Commission proposals, disconnect them from the place-based narrative and could re-focus the new policy primarily on more efficiency and better results, more monitoring and new conditionalities.
Negotiations on the new regulations for the structural funds intensified and became more concrete in 2012 under Danish Presidency. The frequency of meetings of the Council Working Group 'EU structural policy' increased significantly to one or two meetings per week. The responsible General Affairs Council under the Danish Presidency met almost every month to negotiate the cohesion policy package. The strategy of the Danish presidency followed the 'building bock'-approach and separated the cohesion package into thematic blocks. This building block-approach should, on the one hand, accelerate the negotiation progress on the new legislation and secure compromises already achieved, and, on the other hand, induce the Member States to define their national negotiating positions at an early stage. At the same time, the regions should be provided with some provisional legal fundament and thus with a degree of certainty for their work to prepare their regional Operational Programs. The Danish Presidency reached consensus on two 'partial general orientations' and under Cypriot presidency a second package of building blocks could be settled. The Council finally could close the negotiations with provisional compromises on all negotiation blocks and on all legal requirements needed to start the implementation in the Member States and their regions by end of 2012. Highly controversial was in particular the connection between the programming of structural funds with the country-specific recommendations of the European Semester and thus linking separate policies with different instruments (Bachtler and Mendez 2016).
In parallel to the negotiations in the General Affairs Council, the Council presidency launched so-called Trialogue meetings between the Council presidency, the Commission and the representatives of the European Parliament to inform parliamentarians about the progress of the negotiations and, conversely, to listen to their wishes and demands and introducing them into the deliberations in the Council. The aim of these weekly meetings was to prepare a preliminary political agreement between the legislative bodies in order to accelerate the legislature process as far as possible. This should facilitate the rapid start of the next funding period in time and with all preparatory documents needed for programming.
The Irish and the Lithuanian Presidencies in 2013 faced the biggest problems in their negotiations with the EP on the question of macroeconomic conditionality and the idea to introduce a performance reserve. Both topics had been decisive negotiation points of the 'Friends of Better Spending'-group and thus became a corner stone of the Council's compromise. Members of the European Parliament but also some European regions criticized and rejected this idea, because the possibility of withholding funding when a Member State does not comply with country specific recommendations might be contrary to the fundamental raison d'etre and the task of cohesion policy. The Regional Committee of the European Parliament argued that especially Member States and their regions in crisis and in financial troubles would strongly need investments and were therefore dependent on payments via structural funds.
In their negotiations on the legislative package, the Council and Parliament agreed on three of the five legislative proposals by September 2013. Only after additional very difficult Trialogues meetings on high political and on technical levels, a final compromise had been endorsed in November 2013 (in detail European Parliament 2014). The EP could not prevent these instruments but wanted to receive for its assent the right of codecision in case of using and implementing these new instruments. Not only Member States could make use of the parallel negotiations on MFF and cohesion policy. The Parliament held back and postponed its formal vote on the MFF-regulation until it had found a compromise with the Council on the regulations on the Union's spending policies. Only on 19 November 2013, the EP voted on the MFF and the cohesion policy regulations. With the release of the new regulatory texts in the Official Journal of the European Union on 17 December 2013 the decision window has closed.

The result of the negotiations
With this ground breaking reform cohesion policy had been transformed into one, maybe the most important instrument of the Union's economic policy. The policy should become firstly more efficient and effective and secondly develop a stronger focus on growth and competitiveness. In order to achieve these goals, cohesion policy had been equipped with a set of new instruments, that will determine the future of cohesion policy in the long run: (1) more strategic guidance with a Common Strategic Framework (CSF) and specific Partnership Agreements (PA) between Member States and the Commission; (2) spending objectives derived from the priorities of 'Europe 2020ʹstrategy for smart, sustainable and inclusive growth; (3) thematic concentration of limited European funds on eleven thematic objectives outlined in the regulation and (4) new types of conditionality to ensure better performance of the policy and the funds. Especially the new 'macro-economic conditionality' combines structural funds with instruments of economic policy coordination, especially with the country specific recommendations in the framework of the European Semester.
This result of negotiations, that started with the tabling of the Commission's legislation proposals in June 2011 and ended with the release of the new legislation in December 2013, shows that the focus of reforms and the direction of evolution of European cohesion policy changed during the negotiations. When proposing to reform the policy, to change the redistribution logic, and to develop a place-based legitimacy and thus to establish a new focus of funding the Commission had been successful in its role as policy entrepreneur when the agenda window opened. The Commission made use of the discussions and ideas developed in the policy stream and could define and legitimize the policy as common place-based European investment policy with new instruments. However, the focus of reform, the justification and the legitimization of this spending policy changed when the 'Friends of Better Spending' took over policy entrepreneurship. Their argument was no more to finance common and place-based investment policy with structural funds but to increase efficiency and effectiveness, to link the policy to the EU's economic policy coordination, the European Semester and national structural reforms, and finally to reduce the financial volume of the policy.

Conclusion
This case study set out to trace the process of recent significant reform of European cohesion policy by using the multiple streams approach.
The adapted MSA proved to be helpful 'in identifying and helping to resolve puzzles of EU policy processes and policy-making' (Ackrill et. al. 2013, 883). The approach helped to structure and analyse the complicated negotiation process on reforming European cohesion policy. The distinction between two different windows with two different policy entrepreneurs could show how the narrative and finally the result of reform changed in the course of the process.
'Cyclical windows' (Kingdon 1995, 186) of European legislation open predictably and allow for anticipated procedures and processes of policy-making. This has consequences for the various actors in the three streams and for the policy entrepreneurs. The policy community knows when ideas and proposals should be aired to attain the attention of politicians and thus to influence the politics stream. The specific strength of entrepreneurs to perceive an open window of opportunity and thus to see earlier than others when the time has come to place their pet solutions on the agenda seems not that important, because all actors involved can foresee when the window will open. What seems more decisive are the political skills, the connections and the influence of the entrepreneur on agenda setting and decision-making.
As the European Commission acted as first policy entrepreneur when the cyclical agenda window opened and presented its draft regulations, this window opened 'from below' at an anticipated point in time. Thus, the type and the timetable of this institutionalized window were predictable. This predictability together with the origin in the policy stream and the fact that the window opened from below pre-determined the width of the window.
The subsequent decision window opened also predictably, albeit from 'above', when the powerful 'Friends of Better Spending'-group of net-paying Member States assumed political entrepreneurship. This 'collective entrepreneurship' was, as predicted by some authors (Hellweg, Huß and Zohlnhöfer 2015), located inside the governmental system, in our case in the Council. Both entrepreneurs, the European Commission for the first agenda coupling and the 'Friends of Better Spending'-group in the Council came from inside the policy-making system of the European Union. Both had direct access to agenda setting and decision-making and both entrepreneurs had political skills and influence, as needed for successful entrepreneurship. This helped to shape and to finalize the process successfully.
The second window opened in the politics stream after the European Council could hammer out a compromise on the multiannual financial framework. The new entrepreneur modified the pet solution presented by the European Commission and pushed for its own pet-solution, which followed the traditional juste retour-approach of net paying Member States in European budgetary policy. The decision making process had been reframed and refocused on more efficiency and effectiveness of the new cohesion policy. The origin of the policy window from above facilitated this substantial reform and the switching of the path. The solution did not only chase a problem, the solution could choose between various problems. Notes 1. This multilevel polity is the reason why Ian Bache concludes that MSA '. . .does not acknowledge the complexity of multilevel governance in which processes (policy, politics and problem streams) and participants operate at different territorial levels simultaneously'. For Bache, the application to the EU is only of limited help (Bache 2013, 34). 2. The Friends of Cohesion-group consists of Bulgaria, Cyprus, Czech Republic, Estonia, Greece, Hungary, Latvia, Lithuania, Malta, Poland, Portugal, Romania, Slovakia, Slovenia, Spain and after its accession Croatia. 3. The non-paper of 24 April 2012 had been signed by Austria, Denmark, Finland, France, Italy, the Netherlands, and Sweden.