UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The politics of subsystems : agenda management and policy change in education King, Conrad A. 2018

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2018_may_king_conrad.pdf [ 1.99MB ]
Metadata
JSON: 24-1.0364035.json
JSON-LD: 24-1.0364035-ld.json
RDF/XML (Pretty): 24-1.0364035-rdf.xml
RDF/JSON: 24-1.0364035-rdf.json
Turtle: 24-1.0364035-turtle.txt
N-Triples: 24-1.0364035-rdf-ntriples.txt
Original Record: 24-1.0364035-source.json
Full Text
24-1.0364035-fulltext.txt
Citation
24-1.0364035.ris

Full Text

   THE POLITICS OF SUBSYSTEMS: AGENDA MANAGEMENT AND POLICY CHANGE IN EDUCATION  by Conrad A. King B.A., The University of Toronto, 2003 B.Ed., Simon Fraser University, 2005 M.A., The University of British Columbia, 2008  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF  THE REQUIREMENTS FOR THE DEGREE OF   DOCTOR OF PHILOSOPHY in  THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Political Science)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  February 2018    © Conrad A. King, 2018     ii  Abstract  Why is the policy status quo more resilient in some contexts than others?  What are the conditions by which policy instruments can be transformed, in lieu of other degrees of change? I argue that the properties of policy design institutions have played an important role in determining the degree of policy change during recent reforms of public education in Germany and France, following from ‘PISA shocks’ in each country. The OECD’s Program for International Student Assessment (PISA) is a cross-national comparison of education systems that revealed unexpectedly poor results for Germany in 2001 and France in 2007. The subsequent (and surprising) policy change outcomes can be explained by examining sector-specific policy design institutions (i.e., policy subsystems). It is from analysis of subsystem politics in the education sector that I begin to develop an agenda-management model of policy change. More specifically, I found that the existence in Germany of a stable, consensus-oriented institution for education policy design – what I call a policy oligopoly – was critical for managing the reform agenda and for facilitating a transformation of policy instruments in lieu of greater or lesser forms of change. Furthermore, the existence in France of a conflict-ridden subsystem – an unstable monopoly for policy design – resulted in a disorderly reform episode and eventually the obstruction of many of the proposed policy changes. The research involved a comparative process tracing of educational policy-making in Germany and France, involving 58 participant and expert interviews and an intensive analysis of primary documents and secondary materials. From this evidence, I describe different types of policy change involved in each case, explain the connection between subsystem politics and forms of policy change, and evaluate the plausibility of this agenda-management model when compared to more conventional explanations, such as policy internationalization or partisan-politics. The dissertation makes a number of contributions to the literature on change in public policy. It provides a novel account for how and why certain policy-making regimes facing acute external pressure   iii  can significantly change the instrumental logic of policy yet retain aspects of the political and policy status quo, while other regimes fail in this regard.      iv  Lay Summary The status quo for compulsory education is often taken-for-granted by the public and politicians. Recently, however, international comparative assessments of education have demonstrated failures in the educational status quo and catalyzed reform episodes for France and Germany. The policy responses were not dramatic, despite the evidence of problems. I argue that in these cases, forms of educational policy change were the product of problem-solving for reasons of politics rather than policy. Policy change was largely determined by the institutionalized relationships between elected government, state bureaucracy, and teachers’ associations, and the compromises they could achieve regarding policy design. Policy-making institutions deliberately restrict wider participation, yet having fewer actors in the process did not lead to significant yet stable reforms in France. Somewhat surprisingly, Germany was capable of greater and more stable change (although not change that upended the status quo) because their policy design institutions incorporated a broader range of perspectives.       v  Preface This dissertation is original, and the intellectual product of the author, Conrad King. The fieldwork reported herein was covered by UBC Ethics Certificate number H13-00949.     vi  Table of Contents  Abstract ......................................................................................................................................................... ii Lay Summary ................................................................................................................................................ iv Preface .......................................................................................................................................................... v Table of Contents ......................................................................................................................................... vi List of Tables ................................................................................................................................................. x List of Figures ............................................................................................................................................... xi List of Abbreviations ................................................................................................................................... xii Acknowledgments ....................................................................................................................................... xiv Dedication ................................................................................................................................................... xv Chapter 1: Introduction ................................................................................................................................ 1 1.1 Introduction to the Argument and Research Questions............................................................... 1 1.2 Educational Policy Change: Surveying the Field ........................................................................... 3 1.2.1 Determinants of Education Policy Change: Internationalist Explanations and Ideas ........... 4 1.2.2 Determinants of Educational Policy Change: Pluralist Explanations and Partisanship......... 8 1.2.3 Determinants of Educational Policy Change: Statist Explanations and Political Inst………..12 1.2.4 Conclusions: Where from here? ......................................................................................... 14 1.3 My Argument: An Agenda-management Model of Policy Change ............................................. 15 1.3.1 Middle-range Theorizing: Scope Conditions and Forms of Policy Change ......................... 17 1.3.2 Sectoral Institutions and Subsystem Politics ...................................................................... 20 1.3.3 Policy Oligopoly ................................................................................................................... 22 1.3.4 Summing up the Argument ................................................................................................. 23 1.4 Contributions to Scholarship on Public Policy ............................................................................ 24 1.5 Methods and Evidence................................................................................................................ 26 1.5.1 Research Design .................................................................................................................. 26 1.5.2 Case Selection ..................................................................................................................... 30 1.5.3 Evidence .............................................................................................................................. 32 1.6 Conclusion ................................................................................................................................... 35 1.6.1 Dissertation Roadmap ......................................................................................................... 36 Chapter 2: Theoretical Framework for an Agenda-management Model of Policy Change ........................ 37 2.1 Introduction: Subsystem Politics and Education Reform ........................................................... 37   vii  2.2 An Agenda-management Model within a Punctuated Equilibrium Framework ........................ 38 2.2.1 The Standard Model: Agenda-setting and Disruptive Policy Change ................................. 39 2.2.2 An Agenda-management Model of Policy Change ............................................................. 40 2.3 Patterns of Policy Change within a Punctuated Equilibrium Framework (PEF) .......................... 41 2.3.1 Forms of Policy Change ....................................................................................................... 43 2.3.2 The Logic of Public Policy Instruments ................................................................................ 44 2.3.3 Education as a Policy Area: The Importance of Evaluation Policy ...................................... 48 2.4 Subsystem Politics and a Policy Oligopoly .................................................................................. 50 2.4.1 Properties of Policy Subsystems ......................................................................................... 52 2.4.2 Defining a Policy Oligopoly.................................................................................................. 55 2.4.3 Operationalizing a Policy Oligopoly .................................................................................... 57 2.5 Causal Mechanisms Linking Sectoral Institutions to Policy Change Outcomes .......................... 59 2.5.1 An Agenda-Management Model: Subsystem Actors and Strategies of Resistance............ 60 2.5.2 An Agenda-Management Model: Institutions and Mechanisms of Stability ...................... 62 2.6 Propositions of an Agenda-management Model ........................................................................ 69 2.7 Scope and Antecedent Conditions .............................................................................................. 71 2.7.1 Scope Conditions: Decision-Making under High Pressure and High Uncertainty ............... 72 2.7.2 Antecedent Conditions: The Power and Peculiarities of Cross-National Comparison ....... 76 2.8 Alternative Explanations ............................................................................................................. 77 2.8.1 Alternative Explanation: Ideas ............................................................................................ 79 2.8.2 Alternative Explanation: Partisan Politics ........................................................................... 82 2.9 Conclusions ................................................................................................................................. 85 Chapter 3: German Education Reform and a National Policy Oligopoly .................................................... 88 3.1 Introduction ................................................................................................................................ 88 3.2 Historical Background ................................................................................................................. 91 3.2.1 Political Context: The Structure and Culture of German (Educational) Policy-making ...... 91 3.2.2 Policy Context: The Structure and Culture of German Education Systems ........................ 99 3.2.3 Historical Context: External Pressure and the Importance of Timing and Cumulation .... 103 3.2.4 The Status Quo: Quality Assurance in German Education ................................................ 106 3.2.5 Conclusions from Historical Background .......................................................................... 108 3.3 Patterns of Policy Change in Germany: Managed Reforms ...................................................... 109 3.3.1 The Catalogue for Action [Handlungskatalog].................................................................. 110 3.3.2 Technical Dimension of Instrumentation: Standard-setting ............................................. 111   viii  3.3.3 Social dimension of Instrumentation: ‘Governance-by-numbers’.................................... 113 3.3.4 Punctuation in the Policy Equilibrium: 2001-2006 ........................................................... 115 3.4 Evidence of a National Policy Oligopoly .................................................................................... 115 3.4.1 Emergence of a National Oligopoly .................................................................................. 116 3.4.2 Structure of the National Oligopoly .................................................................................. 118 3.4.3 Policy Image of the National Oligopoly ............................................................................. 124 3.4.4 Stability of the National Oligopoly .................................................................................... 129 3.4.5 Conclusion: Properties of a Policy Oligopoly .................................................................... 131 3.5 Effects of a Policy Oligopoly: Agenda-management ................................................................. 132 3.5.1 Image Control: Framing as Political Strategy .................................................................... 132 3.5.2 Image Control: Framing as Causal Mechanism ................................................................. 137 3.5.3 Venue Control as Political Strategy ................................................................................... 139 3.5.4 Venue Control as Causal Mechanism ................................................................................ 141 3.6 Alternative Explanations for the German Case ......................................................................... 142 3.6.1 International Policy Transfer and an Advocacy Coalition Framework .............................. 143 3.6.2 Partisan Politics and Competition in Macro-Political Arenas ........................................... 148 3.7 Conclusions from the German Case .......................................................................................... 150 Chapter 4: French Education and an Unstable Policy Monopoly ............................................................. 152 4.1 Introduction .............................................................................................................................. 152 4.2 Historical Background ............................................................................................................... 155 4.2.1 Political Context: The Structure and Culture of French (Educational) Policy-making ...... 156 4.2.2 Policy Context: The Structure and Culture of French Education ...................................... 165 4.2.3 Historical Context: External Pressure and the Importance of Timing and Cumulation .... 168 4.2.4 Quality Assurance in French Education ............................................................................ 170 4.2.5 Conclusions about Historical Background ......................................................................... 173 4.3 Patterns of Policy Change in France: Disorderly Punctuations and Obstructed Reform .......... 173 4.3.1 Darcos Reforms (2008): ‘Lycée à la carte’ ......................................................................... 174 4.3.2 Chatel Reforms (2010): ‘Lycée pour tous’ ......................................................................... 176 4.3.3 Technical and Social Dimensions of Change: New Techniques but Traditional Logic ...... 177 4.3.4 The Punctuation in the Policy Equilibrium (2007-2010): A Disorderly Reform Episode ... 179 4.4 Evidence of an Unstable Subsystem ......................................................................................... 179 4.4.1 Evolution of the French Policy Subsystem ........................................................................ 180 4.4.2  Structure of the French Subsystem................................................................................... 188   ix  4.4.3 Policy Images of the French Subsystem ............................................................................ 191 4.4.4 Instability of the French Subsystem .................................................................................. 194 4.4.5 Properties of a Policy Monopoly ....................................................................................... 195 4.5 Effects of an Unstable Policy Subsystem: Venue-shopping in lieu of Agenda Management ... 196 4.5.1 Framing as a Political Strategy .......................................................................................... 196 4.5.2 Framing as a Causal Mechanism ....................................................................................... 200 4.5.3 Venue Control: Venue-shopping as Political Strategy ...................................................... 201 4.5.4 Venue Control: Venue-shopping as Causal Mechanism ................................................... 204 4.6 Alternative Explanations for the French case ........................................................................... 207 4.6.1 International Policy Transfer and an Advocacy Coalition Framework .............................. 208 4.6.2 Partisan Politics and Competition in Macro-Political Arenas ........................................... 211 4.7 Conclusions from the French Case ............................................................................................ 213 Chapter 5: Conclusion ............................................................................................................................... 216 5.1 Scope conditions: The Possible Cases ....................................................................................... 219 5.1.1 Micro-level Scope Condition: Political Ambiguity ............................................................. 220 5.1.2 Macro-level Scope Condition: State Insulation ................................................................. 221 5.1.3 Macro-level Scope Condition: Acute External Pressure ................................................... 222 5.2 Cross-national Comparison: Implications of Subsystem Politics for Other Models ................. 224 5.3 Extending the Theory: Complementarity with Other Theories of Change ............................... 229 5.3.1 Relationship between Institutional Change and Policy Change ....................................... 229 5.3.2 Relationship between Subsystem Politics and Knowledge Mobilization ......................... 232 5.4 A Future Research Agenda ........................................................................................................ 234 5.5 Contributions to Comparative Politics ...................................................................................... 237 5.6 Implications for Reform of Education ....................................................................................... 239 References ................................................................................................................................................ 243 Appendix A: Indicators for Structure of a Germany National Oligopoly .................................................. 263 Appendix B: Indicators for Policy Image of a Germany National Oligopoly ............................................. 264 Appendix C: Indicators for Stability of a Germany National Oligopoly ..................................................... 265 Appendix D: Indicators for Structure of French Subsystem...................................................................... 266 Appendix E: Indicators for Image of French Subsystem ........................................................................... 268 Appendix F: Indicators for Stability of French Subsystem ........................................................................ 270 Appendix G: Interview Methods ............................................................................................................... 271 Appendix H: Primary Documents .............................................................................................................. 277   x   List of Tables Table 1: Properties of an Oligopoly: Indicators .......................................................................................... 58 Table 2: Interview Methods Table ............................................................................................................ 271      xi  List of Figures Figure 1: Institutions in the Policy Process ................................................................................................. 69 Figure 2: National Policy Oligopoly in Germany (2001) ............................................................................ 124 Figure 3: Unstable Policy Monopoly in France (2008) .............................................................................. 191     xii  List of Abbreviations International DG (EAC) Directorate-General, Education and Culture EU   European Union IEA    International Association for the Evaluation of Educational Achievement  OECD  Organisation for Economic Co-operation and Development PISA  Program for International Student Assessment TIMSS  Trends in International Mathematics and Science Study  Germany BDA Confederation of German Employers Associations [Bundesvereinigung der Deutschen Arbeitgeberverbände] BLK  Bund-Länder Commission for Education Planning and Research Funding [Bund-Länder-Kommission für Bildungsplanung und Forschungsförderung] BLLV Bavarian Teachers’ Association [Bayerischer Lehrer- und Lehrerinnenverband] BMBF Federal Ministry of Education and Research [Bundesministerium für Bildung und Forschung] CDU   Christian Democratic Party CSU   Christian Social Party  DBB  German Civil Service Foundation [Deutscher Beamtenbund und Tarifunion]   DGB  Confederation of German Trade Unions [Deutscher Gewerkschaftsbund]  DIPF German Institute for International Education Research [Deutsches Institut für Internationale Pädagogische Forschung DL  German Teachers’ Association [Deutscher Lehrerverband]  DPhV  German Philological Association [Deutscher Philologenverband]  FDP   Free Democratic Party  GEW  Education and Science Union [Gewerkschaft Erziehung und Wissenschaft]  IQB  Institute for Quality Development in Education [Institut zur Qualitätsentwicklung im Bildungswesen] KMK Conference of German Länder Ministers of Education and Culture [Kultusministerkonferenz]  MPIB  Max Planck Institute for Education Research [Max-Planck-Institut für Bildungsforschung] NRW   North Rhine-Westphalia  SPD   Social Democratic Party  VBE  Association for Education [Verband Bildung und Erziehung]    France CFDT French Democratic Confederation of Labour [Confédération Française Démocratique du Travail] CGT  General Confederation of Labour [Confédération générale du travail] CNRS  National Centre for Scientific Research [Centre National de la Recherche Scientifique] CSE   Superior Council for Education [Conseil Supérieur de l’Éducation] CSEN Syndicated Confederation of National Education [Confédération Syndicale de l'Éducation Nationale] CSP  Superior Council for Programs [Conseil Supérieur des Programmes]  DEP Department for Evaluation and Forecasting [Direction de l’Évaluation et de la Prospective]   xiii  DEPP Department of Evaluation, Forecasting and Performance [Direction de l’Évaluation, de la Prospective et de la Performance] DESCO Department of School Education [Direction de l'Enseignement Scolaire]  DGESCO  Department (General) of School Education [Direction Générale de l'Enseignement Scolaire]  EPLE  local public educational institutions (lycées) [Établissement Public Local d'Enseignement] FCPE Federation of Councils for Students’ Parents [Fédération des Conseils de Parents d'Élèves] FEN  Federation for National Education [Fédération de l'Éducation Nationale] FIDL  Independent and Democratic Federation of Secondary Students [Fédération Indépendante et Démocratique Lycéenne] FO   Open Force [Force Ouvrière] FSU  Federation of United Unions [Fédération Syndicale Unitaire] HCE  High Council for Education [Haut Conseil de l’Éducation] HCEE  High Council for School Evaluation [Haut Conseil de l'Évaluation de l'École] IFE  French Institute of Education [Institut Français de l’Éducation] IGEN  General Inspectorate [Inspection Générale de l'Éducation Nationale] INRP  National Institute for Education Research [Institut National de Recherche Pédagogique] IREDU Institute for Research in the Sociology and Economics of Education [L'Institut de Recherche sur l'Education : Sociologie et Economie de l'Education] LOLF  Organic Law relative to Financial Regulation [loi Organique relative aux Lois de Finances] MEN  National Ministry of Education [Ministère de l’Éducation Nationale] PEEP Federation of Parents of Children in Public School [fédération des Parents d'Élèves de l'Enseignement Public] SGEN   General Union for National Education [Syndicat Général de l’Éducation Nationale]  SE-UNSA  Teachers’ Union – National Autonomous Trade Union (Syndicat des Enseignants – Union Nationale des Syndicats Autonomes] SNES National Union of Secondary Schoolteachers [Syndicat National des Enseignements de Second degré]  SGEN  General Union for National Education [Syndicat Général de l’Education Nationale]  SNALC National Union for Upper and Lower Secondary Schooling [Syndicat National des Lycées et Collèges]  UNL  National Union of Secondary Students [Union Nationale Lycéenne] ZEP  Zones of Educational Priority [Zones d’Éducation Prioritaire]     xiv  Acknowledgments I offer my enduring and heartfelt gratitude to my dissertation supervisor, Professor Antje Ellermann, without whom this would not have been possible. Her consistent intellectual and moral support were instrumental at every stage of the PhD process, but especially during the most difficult phases.    I would also like to express my utmost gratitude to my committee, Professor Alan Jacobs and Professor Kurt Hübner, for their insightful suggestions and incredible patience. My thanks especially to Kurt, who has become both a mentor and a friend.    I thank the Department of Political Science and the Institute for European Studies at UBC, for cultivating an environment of collegiality and academic excellence, as well as for providing financial support. There are a number of colleagues from UBC and Capilano University that I would also like to acknowledge for their generosity and contributions to this outcome:  At Capilano University: Ed Lavalle, Cam Sylvester, Tim Schouls and Ramjee Parajulee.  At UBC: Agustín Goenaga, Jason Tockman, Serbulent Turan, and Sule Yaylaci.  Of all the wonderful friends made during my years at UBC, I most grateful for the support and friendship of Jennifer Allan.     Finally, I would like to thank friends and family for seeing me through: Kyle Blake, Joanne Horrigan, Mark Pervan, and my heroic sister, Romana King.    Above all, to my never-flagging mother: thank-you.      xv  Dedication  For John and Zena.    1  Chapter 1: Introduction 1.1 Introduction to the Argument and Research Questions What makes the status quo so powerful when creating policy? Under normal political conditions, the risks and costs associated with policy change create incentives for powerful actors to adhere to existing policy frameworks. The reasons for the power of the status quo seem abundantly clear: cultural norms, delicate balances of power, cognitive uncertainty, and the structuring effects of historical practices produce a knowable and bankable equilibrium in many policy areas. Most policy remains consistent with an existing policy framework and is adaptive to a stable external environment. Yet there are episodes when the risks, costs or problems associated with preserving the status quo exceed those of reform. Compulsory public education is a case in point: education policy is under pressure from new technologies, the involvement of international organizations, and increased demands by societal stakeholders. Elected politicians have become more involved in compulsory education, as they frame it as a panacea for the many challenges facing their societies, from social disintegration to economic stagnation. Education policy reform has been justified in the context of changing public opinion, ‘permanent austerity’, international competitiveness, current best practice established at the European or international level, ‘evidence-based policy-making’, and comparative international assessments.  Amongst the latter, the most influential has been the Program for International Student Assessment (PISA), a large-scale international assessment comparing the skills acquisition of 15 and 16-year-old students, conducted by the Organisation for Economic Co-operation and Development (OECD). PISA conveyed unexpectedly poor national rankings for Germany and France, yet sparked very different politics for education reform, and unexpected policy change outcomes. Historically, both systems of compulsory education exemplified the resilience of a powerful status quo, due to venerable national traditions, well-entrenched educational bureaucracies, and influential vested interests. Of the two,   2  Germany was more change-resistant at the national level because compulsory education is a sub- national jurisdiction. There is not one education system, but sixteen. For meaningful national policy change they must coordinate reform together, seeming to destine (or doom) Germany to lowest common denominator outcomes – if not outright policy stagnation, then at most, incremental adjustments that reinforce the status quo. Yet recent national education policies in Germany and France displayed puzzling outcomes. Despite a large number of veto players and relatively weak educational bureaucracies, Germany significantly changed its evaluative policy instruments between 2001 and 2005, coinciding with a very public ‘PISA shock’. In France, the reform opportunity came later, following poor PISA results revealed in 2007. Yet in spite of France’s early involvement in international assessments, power concentrated in the central executive, government elected on a reform mandate, and evidence of policy failure, the French government was unsuccessful in transforming education policy. The political will for reform existed in both countries, yet the status quo had greater resilience in France than in Germany, which was surprising given the antecedent conditions.  How does one account for reform in the logic of national-level policy instruments in Germany, yet the resiliency of the status quo in France? Why did Germany, despite low reform capacity and significant distance from international policy models, manage to affect significant educational policy change, while France, with centralized and concentrated decision-making processes and closer proximity to international models, could not?  In short, what are the conditions for significant reform of instrumental logics in policy areas well-insulated by the state yet facing pressures from globalization, such as compulsory public education?     3  1.2 Educational Policy Change: Surveying the Field Despite the rich literature in other disciplines, the study of education has been somewhat of a blind spot for political scientists until recently (Busemeyer and Trampusch 2011; Jakobi, Martens, and Wolf 2010).1 This is surprising considering how important compulsory education has been for the modern nation-state; it has served political functions like creating citizens or fostering societal integration, as well as economic functions like training workers or facilitating economic growth (Gellner 1983). Furthermore, education is an ‘old’ policy area: if Bismarck’s insurance schemes indicate the birth of modern public policy, then public education (in some form) preceded this by several decades in both Germany and France (Iversen and Stephens 2008, 602). Despite its functional importance and old pedigree, education policy has often been neglected by political scientists because it was considered special, different from social policy, and typically excluded from analyses of the welfare state (Wilensky 1975, 3). What is lacking is a thorough application of political science theories to the determinants of education policy change (Busemeyer and Trampusch 2011). Where political science has examined these determinants, I review the literatures below, dividing the approaches into three categories: internationalist; pluralist; and statist.2 Political science has begun to rediscover education as a policy area, yet it remains an area with much potential for generating further insights.                                                            1 Amongst non-political scientists, education policy has typically been treated as an exogenous, independent variable, with sociologists, economists and educational scientists responsible for the bulk of the research on education. Their contributions tended to focus on the effect that policy had on other outcomes, such as social stratification (Dronkers 2010), human capital formation (Goldin and Katz 2008), or educational governance (Clark 1983). “[Political science] can complement [these] efforts by analyzing how differences in political institutions and processes have contributed to creating and maintaining some distinctive policy emphases” (Heidenheimer 1997, 6). 2 Each approach emphasizes a particular independent variable and argues for the importance of a particular set of actors. Many approaches acknowledge that multiple variables are often necessary, in line with early multivariate analyses of educational expenditure which indicate that it “is an arena in which monocausal explanations are wholly inappropriate” (Castles 1989, 431; see also Castles 1982; Verner 1979). For more complex policy outcomes, perhaps even multiple theories are needed to explain the process. “This is one of the more problematic issues for scholars working in the public policy field: it is not easy to convincingly explain the ongoing process of public policy (with all its various steps) using the same theoretical framework for all the different components of the process itself” (Capano 2009, 14).    4  1.2.1 Determinants of Education Policy Change: Internationalist Explanations and Ideas International political economists have made a causal connection between the behaviour of transnational or international actors and education policy change at the national level. This is not an entirely new approach, as sociologists and educational specialists have been examining this phenomenon since at least the 1970s.3 Education internationalization is thus a well-developed research agenda, one which overlaps with broader political science literature on policy diffusion, convergence, transfer or translation (Stone 2012).4 However, the early 2000s sparked fresh interest in the internationalization of education policy, because international and supranational organizations became increasingly involved in processes of internationalization and Europeanization via comparative assessments like PISA, or agenda-setting exercises like the Lisbon Strategy (2000), the Bologna Process (1999) or the Copenhagen Process (2002).  There was something distinctive about PISA, in particular. For one, it had obvious policy effects: a number of countries seemed to respond directly to PISA rankings in the development of education                                                           3 One theoretical framework was developed by sociologists John Meyer and Brian Rowan (1977), who argued that education systems are rule-conforming institutions which adapt to shared (global) norms rather than the functional demands of their own polity or economy. This ‘world institutionalism’ sustained a major research agenda within the sociology of education, and was partly taken up by organizational sociologists for more general explanations of how institutions emulate one another transnationally (see Dimaggio and Powell 1983). However, world institutionalism in education was critiqued on normative grounds as an implicitly neo-liberal modernization theory, on theoretical grounds for failing to account for the actual mechanisms of policy diffusion, and empirically for ignoring the ways in which national education systems remained quite distinct (see Boli, Ramirez, and J.W. Meyer 1985; Dale and Robertson 2002, 2007; Dale 1989, 2000, 2005; H.-D. Meyer and Rowan 2006; J. W. Meyer, Boli, and Thomas 1987; J. W. Meyer et al. 1997; Robertson, Bonal, and Dale 2002; Robertson 2005).  4 Comparative educationists have developed a research niche known as ‘educational transfer’, with the bulk of analysis devoted to the sending side of transfers (Waldow 2012, 411. See also Halpin and Troyna 1995; Perry and Tor 2008; Phillips and Ochs 2004; Steiner-Khamsi 2012). For the less-studied ‘receiving’ side; the consensus among comparative educationists is that effects vary enormously across countries. They argue that much of the educational terminology is discursively constructed (or reconstructed) in a domestic context, so it is unlikely that there will ever be complete emulation of a foreign education system, or full convergence on an international policy model. Local governing bodies actively respond to global discourses and local contexts (respectively referred to as ‘refraction’ and ‘reflection’), and global agendas become manifest in national policy decisions only when they resonate with local constituencies (Perry and Tor 2008, 523). This suggests that the internationalization of education is indeed occurring, but perhaps not at the level of whole systems of education.     5  policy, and many national policy discourses identified PISA as a catalyst for change. Moreover, the argument that PISA was causing domestic policy change seemed compelling when one considered some of its unique features. First, unlike its predecessors, PISA was policy-oriented. It produced more than just data for consumption by education researchers and specialists; it modelled high-performance education systems for the explicit purpose of lesson-drawing by policy-makers (Figazzolo 2009; OECD 2007; Rose 1991). Second, national governments had solicited this advice. Even prior to PISA, government bureaus approached international organizations such as the OECD’s Directorate of Education, in order to learn more about educational performance from a comparative perspective.5 Finally, PISA was a new technology in a number of different respects, and therefore made an impression on national policy-makers regarding the limits to their own knowledge. PISA was a novel epistemological technology and measurement tool because it used psychometrics and large sample sizes to generate new types of data about educational performance, promoting a kind of ‘governing by numbers’ (Grek 2009). It also adopted a new political technology – international rankings – which condensed large amounts of data into oversimplified yet easily accessible information, putting political pressure on actors involved in national policy-making. These features suggested that high-level national decision-makers were responding to PISA, with policy change caused by social                                                           5 The OECD has long been an ‘ideational arbiter’: “the organization, historically, has not only been a provider of raw empirical data, it has also sought to reflect and shape the parameters of policy debates as they emerge across its members” (Harmsen and Braband 2015, 32). For compulsory education, the OECD began to develop educational indicators in the mid-1980s, driven by political interests in the USA and France. American federal bureaucrats were concerned with losing the Cold War because they were not sufficiently developing the technology skills of young people, given obstacles thrown up by a decentralized education system (the Fed wanted the OECD to help them “get a grip on the individual school systems”(Troehler 2011, 152)). France’s (leftist) government worried about the lack of educational opportunities for socio-economically disadvantaged children. Both countries wanted comparable education data, and their national experts worked closely with the OECD Secretariat to develop indicators and assessment tools. By the 1990s, other countries were soliciting the OECD for data, and the Secretariat responded with their ‘Education at a glance’ series, providing (mainly qualitative) overviews of various national systems of education. A veritable transnational epistemic community had begun to develop (Knodel, Windzio, and Martens 2014, 3; Leuze, Martens, and Rusconi 2007).    6  learning.6 What was not clear was who exactly was doing the learning, nor was it clear the degree to which they were learning from past (national) experience or novel (international) information. Nonetheless, a theory began to develop around the ‘internationalization of education policy’, which suggested that learning from international ideas like PISA was a key mechanism for national policy change.   Given assumptions about learning, political scientists advanced this theory by arguing that a global convergence of education policy was driven by international policy models and the agency of transnational organizations (Breakspear 2012; Dobbins and Martens 2012; Jakobi 2009; Martens et al. 2010; Martens, Knodel, and Windzio 2014; Martens, Rusconi, and Leuze 2007; Niemann 2009, 2010, 2014). Rather than assume that norms or cognitive theories diffuse globally via osmosis or isomorphism, these scholars looked for the mechanisms by which ideas are transmitted, and found them in the governance instruments used by transnational actors like the OECD. Thus, the primary explanatory variable for the internationalization of education policy is international ideas, and the causal mechanisms are the tools of communication or governance used by international organizations to convey these ideas. This ‘transmission’ of information is constrained by national ‘reception’ factors, with intervening variables being the national principles or values surrounding education, and veto players in domestic political institutions (such as the government executive, legislatures or judiciary). This theory is primarily interested in explaining international convergence, yet in so doing, it asserted a set of causal variables and mechanisms for policy change occurring at the domestic level. However, limited case studies show that even when domestic political actors directly reference PISA, there have been a whole range of policy reactions from comprehensive reform to seeming willful ignorance (Bieber and Martens                                                           6 Peter Hall (1993, 278) defines social learning as “the deliberate attempt to adjust the goals or techniques of policy in response to past experience and new information. Learning is indicated when policy changes as the result of such a process.”  The difficulty with this definition is that it assumes learning from the outcome – just because policy has changed, it does not mean that learning occurred.  Policy goals or techniques could be changed for other reasons, with policy-makers citing experience or new information being little more than ‘cheap talk’.    7  2011). Furthermore, national policies have changed in unexpected ways.7 In terms of empirics, there is a demonstrable need for comparative and longitudinal analyses of outlier cases, in order to tease out the mechanisms that allow a domestic policy status quo to be resilient or undermined. In other words, if domestic factors merely filter the internationalization of education, why is policy change so disparate across cases – and in unexpected directions?   Analytical choices are part of the reason that the ‘internationalization of education policy’ model has failed to explain the degree of policy change in certain national contexts. Their analysis operates at the macro level – examining global trends – and they have been primarily concerned with explaining international convergence rather than national policy processes. In generating a theory that could ostensibly incorporate many highly diverse cases, their dependent variable looks for changes that are discontinuous and teleological (i.e., measuring change vis-à-vis proximity to an international model), when other change dynamics could be occurring in specific cases, such as utilitarian, cyclical, or dialectical policy change (see Howlett and Cashore 2009). Furthermore, the treatment of national actors, institutions and ideas as intervening instead of independent variables under-specifies domestic causal mechanisms. This literature does not analyze exactly how domestic actors engage in learning about educational issues, which actors do the learning, nor how this relates to the strategic use of international ideas to formulate policies that are possible within a given national context (see May 1992; Quirk 1988; Toens 2010). Theories of education policy internationalization are important for drawing                                                           7 France and Germany are amongst these ‘outliers’. Martens et al. (2010) suggest that despite the risks and costs associated with reform, veto players in Germany were convinced not to exercise their veto. However, it is only clear what each veto player did (or did not do), it is not clear why they did it. In France, the power of international ideas and transnational governance appeared insufficient for radical policy change, despite more favourable ideational and institutional conditions. Dobbins and Martens (2012) argue that although veto opportunities in France were fewer, France did not reform in response to poor PISA results because of uneven manifestation of ‘local problem pressure’ (the exact meaning of this is not clear) and widespread aversion to international (neo-liberal) ideas.   8  attention to international discourses and transnational actors, yet in some cases, the influence of international actors on domestic policy-making might indeed be contextual rather than causal.8     1.2.2 Determinants of Educational Policy Change: Pluralist Explanations and Partisanship Domestic agency is a presupposition of a strand of literature located partway between comparative politics and historical macro-sociology, which has tried to establish if political parties matter for educational policy change. Early work in this tradition (Castles 1982; Verner 1979) revolved around the question: Do explanations of educational development and change even need political variables, or are socio-economic forces sufficient?9 Subsequent empirical analyses have suggested that parties do indeed matter for spending and school structure (Castles 1989, 1998), although we lack information about partisan influence on other policy outputs. Because reform is in part due to government attention to educational issues, the question remains whether political parties (in government) are a determinant for every aspect of education policy, and if so how.   If partisanship matters for educational policy change, then two models can explain how parties form preferences and translate them into policy: power-resource theory; and partisan theory (also                                                           8 We know that transnational actors influence domestic policy development by constructing and diffusing norms and epistemes, the ‘constitutive ideas of policy paradigms’ (Skogstad and Schmidt 2011, 18). Yet it is less clear how this ideational content gets translated, processed, mobilized or shaped by domestic policy dynamics. Small-N comparisons of education policy have assessed the influence of international organizations on national actors, institutions and norms, and these studies demonstrate the importance of domestic politics and the involvement of actors from outside the traditional realm of ‘high politics’ (Dakowska and Harmsen 2015; Dobbins and Knill 2009; Toens 2010; Trampusch 2009; Witte 2006). None have examined compulsory education, instead focusing on higher or vocational education (with each sector of education involving a different set of civil society actors).   9 This question emerged from a ‘partisan neutrality hypothesis’ (the null hypothesis adapted from median voter models), which suggested that political parties were not important because they do not determine which issues get on the policy agenda. Empirically, we should see no difference in issue-attention by governments of the left or right, because political parties do not ‘own’ particular issues in order to bring them to public attention once they are in power. Instead, three other factors drive policy change. First, when in government or in power, party leaders (or members of legislature) discount partisan policy preferences and ideology in favour of policy performance and electoral success (manifest in personal credit-claiming or blame avoidance behaviour). Second, new governments want to erase the policies of outgoing governments, and what drives issue attention is change in government, not partisanship. Third, political parties notwithstanding, governments are driven to act by external events, such as the media, public opinion, international events, or the economy (Baumgartner, Brouard, and Grossman 2009).    9  called partisan differentiation theory).10 Power resource theory argues that political parties (in or out of government) are important because they mobilize class interests (Esping-Andersen 1990; Huber and Stephens 2001). In essence, parties act as ‘transmission belts’ for the political demands of particular classes, with social democratic parties expanding education and conservative parties contracting educational spending.11 Certainly, this has helped us understand the establishment of the status quo in terms of both partisan preferences and policy outputs; it made clear that parties mattered for the historical development of education policy and for the distinct characteristics of education systems, because parties represented class divisions and formed stable coalitions with other societal interests (Ansell and Lindvall 2013). However, analyses of gross spending and early educational expansion did not make clear if parties continued to matter for well-established education systems, especially given efficiency-driven reforms that tried to redistribute funding without significantly changing overall levels of investment.    More compelling and nuanced models were needed to explain what seemed to be the ‘new politics of public investment in education’ (Busemeyer 2009a; Iversen and Stephens 2008). ‘Partisan                                                           10 There is also a third possibility that encompasses both the ‘partisan neutrality hypothesis’ and the ‘partisan hypothesis’ in a dynamic way: that there has been a decline in partisan influence over time, due to globalization or other factors (Boix 2000). This is an interesting question in light of policies investigated here.  However, for the sake of analyzing the discrete episodes in this dissertation, I am trying to ascertain if political parties were the most significant proximate causal factor.  11 Early empirical analyses confirmed generalizations about gross levels of spending. For example, across OECD countries in 1960 and 1970, there was an inverse relationship between conservative party strength and educational spending, and the strength of parties on the right was the strongest predictor of percent GDP allocated to public education (Castles 1982, 63). Until the 1980s, there also seemed to be a stable correlation between left parties and educational spending, with social democratic parties instrumental for increases to primary schooling funding, and the ‘massification’ of secondary schooling during the post-war period (expanding both access and administrative oversight). The causal logic was that left-wing parties try to maximize redistribution between classes, with public education a means to do this (assuming progressive taxation). Public education is also a means to facilitate supply-side capitalism, so left-wing parties can pursue preferences for a big state and extensive redistribution within a capitalist system using education funding. However, there are redistributive aspects within education as well: these theories predict that social democratic parties spend more overall, but also prefer to spend on primary rather than on tertiary education (Boix 1997, 1998). It should be noted that compared to other social welfare policies, education is an inefficient redistribution mechanism because the wealthy often take disproportionate advantage of publicly-funded education, especially at higher levels (Jensen 2011).  Hence the need for more nuanced explanations, such as partisan theory.      10  theory’ argued that political parties have significant impact on public policy outputs by suggesting that governments of the left and right behave differently, and pursue different policies. Parties have issue ownership, demonstrated by left governments consistently trying to spend on welfare, the right on defense, or green parties on the environment (as examples). Beyond spending habits, governing parties try to put their issues on the agenda, and attach their name to particular policies, according to partisan ideological preferences. Parties build a reputation based on past policy stances, and represent constituencies that prioritize and dedicate attention to particular issues (Baumgartner, Brouard, and Grossman 2009). Thus, political parties in power are the important actors, not as mere transmission belts for class interests, but rather as strategic actors who use policy outputs – spending especially – to attract voters or consolidate their base (Busemeyer 2009a, 107). Empirically, partisan theory has made clear that education policy outputs are more than just the aggregation of economic preferences for particular groups of voters.12 Furthermore, these theories have been adept at explaining the redistributive preferences of governing parties on the left, during an era of globalization.13 While                                                           12 As Ben Ansell (2010) has shown, even if individual preferences for educational spending derive from economic factors, actual policy outcomes are driven by political bargains in which partisan ideological preferences interface with the opportunities and constraints of political institutions. Furthermore, Busemeyer (2013) demonstrates that the distribution of voter preferences for education policy is not uniform, and many factors beyond class – like age, cultural background, and so on – shape preferences on educational spending and access. Therefore, social democratic governments can no longer be certain how well they represent the economic interests of their base, when they expand education. We know even less about how contemporary conservative governments form preferences and manage education reform. Typically associated with limited public investment and promotion of market-enhancing governance reforms, conservative governments occasionally increase public investment in education, presumably when this seems less redistributive than social transfers (Ansell 2010; Jensen 2011; Lundahl 2002). Yet there might be other conditions when conservative governments invest in education, such as when the status quo incurs little material cost, but significant political costs in the short-to-medium term, like probable electoral defeat if reforms are not undertaken. It would seem that electoral competition incentivizes parties to prioritize issues and create particular kinds of public policies, in order to attract and retain certain groups of voters and get re-elected (M. G. Schmidt 1996).  13 Part of the puzzle has been why left governments try to increase spending on upper secondary or higher education, given this was an area from which elites traditionally benefitted (due to narrow access and public subsidies). There have been a number of complementary explanations for this. One is the perception of (higher) education as vocational skills-training for the working class (Iversen and Stephens 2008). Another is that education spending is an inter-temporally redistributive opportunity structure (i.e., over time lower classes benefit disproportionately from all education investments (Ansell 2008)). Related to this, social democratic governments use tertiary spending as a means to forge cross-class alliances – to open up access to tertiary education for lower classes, their core constituency, yet also provide an electoral alternative for middle classes who wish to retain   11  instructive for an understanding of the relationship between governing parties and education spending, we still lack information on how contemporary political parties form preferences on policies other than spending or school structure, and how they leverage preferences into policy outcomes when in government.14   Theories of policy change that emphasize partisanship and pluralist interests remain underdeveloped for the field of education.15 The models above assume that party preferences are stable, but differ over whether those preferences are a function of fixed class interests, wider electoral strategies such as cross-class coalitions or intertemporal redistribution, or the pluralist interests of pressure groups or peak associations. Whatever assumptions guide preference formation, if partisan theories are correct then governing parties should ‘own’ issues – or aspects of issues – and have the capacity to enact pre-existing preferences, changing policy to the degree that their partisan preferences do not conform to the status quo. Aside of spending and school structure, we lack information about the relationship between parties and policy preferences, and about the relationship between partisan                                                           access to higher education (Ansell 2005). Finally, there is the argument that left wing governments are less motivated by direct redistributive preferences, than by risk assessments of skills redundancy in the working class, with (higher) education investments a kind of social insurance scheme (Jensen 2011).     14 Another possibility is an underdeveloped variant of partisan theory that I refer to as ‘pluralist-partisan theory’. This proposes that for some policy areas, governing parties cannot define their electoral opportunities solely vis-à-vis the interests of particular voters or cross-class alliances. Parties strategically consider blocs of voters, but when elections are not imminent, they consider things like policy performance and the degree to which policy choices affect powerful pressure groups. No party ‘owns’ education issues outright, but each party tries to monopolize (or ‘colonize’) certain aspects of education policy that are consistent with their alliance to powerful interest groups (rather than voting blocs alone). A pluralist-partisan theory therefore suggests that governing parties use policies to cater to their base, to build alliances, and to signal (or appease) particular pressure groups. There is little analysis of education policy that explicitly considers the link between pressure participants and political parties (one exception being Moe and Wiborg 2017). So one could look to other areas, such as health policy, to see how political parties are responsive to powerful interests within particular sectors (see Immergut 1992). If this theory is meaningful, then researchers must pay closer attention to the kinds of policies being changed (i.e., not just assessing policies that affect end-users, like spending and access, but also evaluation and governance policies). In this dissertation, pluralist-partisan theory is considered as an alternative explanation.   15 “From a political science perspective, issues such as the role of electoral politics, political institutions and government partisanship as important factors shaping educational institutions and education policy output remain under-researched” (Busemeyer and Trampusch 2011, 425).   12  preferences and policy outcomes. In short, do parties matter for all forms of education policy change in an age of globalization, and if so, how?   1.2.3 Determinants of Educational Policy Change: Statist Explanations and Political Institutions During the 1980s, comparative politics scholars advocated ‘bringing the state back in’ to theories of policy change, treating the state as an autonomous actor with independent policy preferences or varying capacities to affect reform.16 As education remains an under-explored policy area for political science, it is perhaps not surprising that it has been somewhat neglected by statist (and historical institutionalist) approaches as well.17 Where theories of the state and political institutions have been applied to education policy, institutions function as arenas for veto in which political representatives of interests or epistemic communities block reform. Inasmuch as institutions enable educational reform, theories of state agency and reform capacity remain underdeveloped.   In the literature on education reform, domestic political institutions have been treated as intervening variables. They are arenas where veto players block reform, with the actual reform proposals flowing from actors who can form preferences independently of institutional considerations, like international organizations or socio-economic interest groups. In other policy areas, veto player                                                           16 Initially, ‘statist’ approaches underspecified exactly who the state was, and it was seldom clear if the state was a reference to legislative structure, the bureaucracy, the high executive, or if other actors – such as the judiciary – had independent preferences or capacities. As statism developed into historical institutionalism, this approach was more adept at explaining stability rather than change, although mechanisms of institutional change have been better developed over the last decade.  17 With important exceptions in the work of Kathleen Thelen (2003, 2004) in the area of vocational training and Kathrin Toens (2010) in higher education. There is a parallel literature that has ‘brought the state back in’ to comparative political economy, and evaluated education in terms of its supply of skills to the broader production regime. This ‘Varieties of Capitalism’ approach has provided robust theoretical tools and testable hypotheses regarding institutional complementarities between education, the labour market and other welfare state policies (Estévez-Abe, Iversen, and Soskice 2001; Hall and Soskice 2001; Iversen and Stephens 2008; Streeck 2012; Thelen 2004). However, it offers limited theoretical application for understanding episodes of acute education reform or cross-national variance regarding degrees of change. This is largely due to the treatment of education systems as a functional component of wider political economies, with the relationship between production regime change and education policy change overdetermined by intervening variables. Moreover, this framework focuses more on vocational and higher education, emphasizes economic rather than political actors (firms especially), and highlights the stabilizing effects of institutional complementarity rather than mechanisms of change.     13  theories have been useful for understanding cross-national variance in degrees of policy change, because they allow one to compare the capacity for change across highly disparate political systems, and can account for political behavior beyond institutional position (Jahn 2010; Koenig and Debus 2010, 271–72; Tsebelis 2002). However, the operationalization of veto player theory has not been especially sophisticated in the literature on education policy.18 If institutions are to be useful for understanding educational policy change, especially in deviant cases, then a more nuanced approach to behavior in institutions is required, as well as a more sophisticated understanding of institutional arrangements.  For starters, we must move beyond counting formal veto players in the institutions of ‘high’ politics, and begin to analyze the interface between institutions of ‘high’ politics and ‘low’ politics (such as schooling, and the educational bureaucracy).19 The assumption has been that the latter merely                                                           18 The tendency has been to assume the preferences of veto players, then count the number of formal veto players in national institutions of ‘high’ politics. This approach neglects recent innovations in veto player theory. First, insufficient attention has been paid to ideological congruence between veto players, and how a policy choice winset can vary within a given sector depending on the type of veto player and the issue. For example, Steffen Ganghof (2003, 10) has argued that political parties formulate policy preferences based primarily on office-seeking motivations, with policy-seeking as secondary, and accounting for this can dramatically change (and expand) the decision winset. Second, the degree to which actors cohere as veto players has been relatively neglected. Tommasi, Scartascini, and Stein (2014) demonstrate that a high number of (incongruent) veto players need not necessarily result in policy stability, when one accounts for time. Under certain conditions, such as when political institutions function as repeated games with relatively stable participants, there are opportunities for strategic behaviour such as log-rolling, which can increase cohesion as a counterweight to ideological incongruence. Finally, counting formal veto players makes little distinction between types of veto players (such as institutional or partisan veto players; agenda-setters; formal or informal types of vetoes). Recognizing variations in veto-type is a theoretical innovation that was suggested by Johannes Lindvall (2010), who distinguished between veto players based on their institutional arrangement. De jure (or formal) veto players are constitutionally-enabled to tread the halls of political power – the institutional and partisan veto players who are elected or appointed to govern, create policies, and represent the interests of their constituents or society at large. De facto (informal) veto players are bureaucratic or societal stakeholders with privileged access to the governing executive, and vary by policy area. Each has the power to maintain (or change) the policy status quo, but their veto depends on institutional variations in opportunity. Even in the earliest incarnations of veto player theory, George Tsebelis (2002, 34–36) briefly discussed agenda-setters as a type of veto player that can propose ‘take it or leave it’ offers to other veto players, their significance declining as policy stability increases, and inclining as they move to the ideological center of the winset. However, agenda-setting (and agenda-management) could be much better fleshed out for veto theories, and for theories of education policy change.  19 An early assumption was that education was naturally a politicized issue area, meaning that “to effect educational change, all groups must move outside the educational field to engage in political interaction at the national level” (Archer 1979, 618). Policy change was thus generated in ‘high’ political institutions, with the status quo merely reinforced by the size and complexity of education systems, the autonomy of policy implementation, strong pressure groups and entrenched interests, and perennial shortage of funding (Ambler 1987; Hall 1983).    14  anchor the status quo, even more so if prestigious or specialized agencies are allowed to operate unquestioned within their own paradigm (Carpenter 2001). Yet a bureaucracy made accountable for policy failure can actually undermine the status quo, especially if they perceive a correlation between broad reform trends and ‘locally-generated ideas’ (Blyth 2007, 764). For example, strong federalism was thought to be non-conducive to policy change in general, because of the higher number of veto players or additional veto points. However, Margaret Archer (1984) found that decentralization could be conducive to education reform, although innovations were due to bureaucratic (and not political) decentralization, where policy innovation can take place “in the relative shelter of a bureaucracy rather than in the bright lights of public debate” (Bleich 1998, 96). Undoubtedly, institutions of high politics are an important part of the policy-making process, but we need to know more about the role of the educational bureaucracy and sector-specific policy design institutions, which are understudied in many areas of public policy (Jenson 2012). In some sense, ‘bringing the state back in’ to explanations of education policy change means evaluating the preferences and agency of the permanent bureaucracy, and examining sector-specific policy design institutions.   1.2.4 Conclusions: Where from here? Research on educational policy change remains underdeveloped in political science. Gaps in our understanding of education policy-making include where precisely internationalization pressures are felt in the domestic sphere, how political parties capture educational interests, and the relationship between institutions and policy change. When considering the determinants of education policy change, researchers should heed the following:  “Moving beyond the assumption of strategic and rational actors, on the one hand, and connecting research into reform policies and politics in the context of internationalization and Europeanization with the analysis and theories of institutional change…on the other, are, in our view, the two main challenges with which this strand of literature is confronted. In this more actor-oriented perspective, questions of timing and sequencing are of interest, as it is still unclear why we observe this growing influence of international organizations increasingly since the 1990s and not earlier” (Busemeyer and Trampusch 2011, 424).   15   Analyses of education policy-making which take these issues seriously might also illuminate areas where generalized theories of public policy change are also lacking, such as roles for bureaucracy and public-sector interest groups within the institutions of policy-making.20 In the next section, I present an argument about how changes to policy instruments can be explained by interactions between sectoral policy design institutions and decision-makers in the macro-political arena.   1.3 My Argument: An Agenda-management Model of Policy Change  In this dissertation, I propose an agenda-management model of policy change, during reform episodes when the policy status quo has been discredited, yet the political status quo remains influential. This model explains the variance between managed and obstructed reforms to education systems, under a particular confluence of policy-making conditions. I argue that sector-specific policy design institutions (‘policy subsystems’) can manage the reform agenda in macro-political arenas if they are inclusive, coherent, and stable. To substantiate this argument, I introduce a new type of subsystem: a policy oligopoly. A policy oligopoly is discernible from other types of subsystems by its inclusive structure, its coherent problem interpretation, and its mechanisms for managing conflict. This type of stable subsystem has the capacity to coherently frame policy issues and control policy development venues, and in so doing, draw the attention of high-level decision-makers (and other actors in the macro-political arena) towards technical aspects of an issue, and away from aspects which might discredit the political status quo. Executive decision-makers then adopt the design recommendations which emerge from an oligopoly, resulting in a significant change in instrumental logic in lieu of other forms of policy change.                                                           20 “One area which requires more work, for example, concerns the links between the nature of the political system and its policy ‘subsystem’…too often the features and workings of political systems have been poorly handled in public policy studies. Mainstream political science, for example, has often dealt with policies as simple, direct outputs of political dynamics…[yet] policy dynamics, and political agency and structures, are strictly intertwined and political and policy arenas very often overlap to a large extent. This requires further investigation, since it is in this intertwining and overlapping that the various different drivers of change have the chance to be productive” (Howlett and Migone 2011, 60).    16  Meanwhile, actors in unstable and more exclusionary subsystems present conflicting images of problems, with some actors resorting to the risky political strategy of venue-shopping to bring their framing of issues to the attention of influential decision-makers or the public. The result is that outcomes are less predictable – unintended consequences can lead to radical transformation or blocked reforms. Thus, an agenda-management model posits that a policy oligopoly greatly increases the likelihood of a significant change in instrumental policy logics, given conditions of state insulation of a sector and acute external pressure from globalization.  This model advances three inter-related claims about the relationship between sectoral policy design institutions and a distinct form of policy change. First, I make the assertion that we must better integrate our conceptualizations of policy change and stability, to account for changes to instrumental logics. The scope conditions and change dynamics found in these cases point to the need for descriptions of a qualitatively different forms of policy change, namely ‘change in instrumentation’.  Second, I argue that subsystem politics determine this form of policy change, meaning that our analytical focus should be on ‘meso-level’ institutions of policy design rather than variables in the macro-level political arena. Third, I introduce the concept of a ‘policy oligopoly’, a subsystem type that builds on Baumgartner and Jones’s notion of a policy monopoly, the cohesive community of experts and policy implementers who try to manage issues on the policy agenda. Policy oligopolies are sector-specific policy design institutions. They direct attention within issues and manage the policy agenda in order to affect significant change to the logic of policy instruments, in lieu of a regression to the status quo or a re-orientation of policy aims. The agenda-management model advances these claims, yet this is not a generalized theory of policy change, nor of generalized instrument change. Rather, it is a descriptive middle-range theory about changes to instrumental logic under a particular confluence of conditions.   17  1.3.1 Middle-range Theorizing: Scope Conditions and Forms of Policy Change An agenda-management model is a middle range theory that explains successful changes to instrumental logic under distinct, though not unique, conditions for reform.21 The phenomenon to be explained is the significant change of policy instruments in the field of compulsory education, given particular historical and institutional conditions. However, the arguments presented in this model can illuminate policy-making processes elsewhere, beyond these cases and beyond education policy. Therefore, this dissertation proposes a modest form of theory-building which is compatible and complementary to more general approaches to public policy, especially the Punctuated Equilibrium Framework (PEF) proposed by Frank Baumgartner and Bryan Jones (1993, 2009). Below I develop my first argument, that we must account better for more intermediate forms of policy change, with a discussion of the scope conditions for this type of reform, and why descriptions of policy change should be supplemented with a new category: changes to instrumental logic. This analysis pertains to policies reformed during periods of acute external pressure, within policy sectors that are highly insulated by state sovereignty. Modern welfare states seem besieged by globalization, yet not all public policies encounter external pressures in the same way. Some sectors must contend with persistent signals of a slow degradation of policy capacity, effectiveness or efficiency. Other sectors experience sudden and very public depictions of policy failure, which emerge from credible sources (often outside of government and even the national sphere) who pressure policy-makers into reconsidering the status quo. An agenda-management model is relevant for the latter,                                                           21 Middle range theories analyze a specific set of phenomena across a limited universe of cases, and thus are a crucial addition to the single case study and generalized theories of political or policy outcomes. Grand theories, such as the structural-functionalism of Talcott Parsons, seeks to explain general phenomena at the societal level. Narrower are general approaches to public policy, such as the Advocacy Coalition Framework or Punctuated Equilibrium Framework, which explain the policy process across many sectors and a large range of political jurisdictions. Middle range theory is comparatively ‘small-N’ in terms of both cases and policy areas.  It can inform about a finite set of cases (such as Western Europe, or the OECD), facing certain political conditions (such as the effects of globalization), and for certain policy areas (such as social policy, or labour market policy).    18  where policy actors must react relatively quickly, rather than adjust slowly. Acute pressure is a condition of reform and not itself an agent of change – when the policy status quo is discredited, it is national actors who decide what to pay attention to, and how to intervene in a policy area.  The other conditions for this model involve state insulation. This is actually two conditions, because the features of the policy-making regime and of the policy area must combine to produce a context in which the state insulates a sector from external pressures. The political status quo is resilient in public policy sectors where the state has both a national interest and high degree of sovereignty, and where provision of a public good leaves limited space for governance by market or other actors. As a condition of reform, the insulating state has certain characteristic features. One is that state activity dominates the sector: because the state has taken responsibility for provision of a public good, it devotes considerable resources to it, including human resources. Another feature is that elected government (and to a lesser degree, bureaucracy) are held responsible not only for provision of the public good, but also for shielding state employees from the pressures of marketization and globalization. These features of the policy area and policy-making regime mean that governance becomes especially salient for reforms that threaten to alter the political status quo. Political interventions that affect how the ‘implementation experts’ (teachers, doctors, nurses, local administrators, etc.) do their job or understand their role can have long-lasting effects and are therefore subject to scrutiny by various actors. Notably, state insulation exists along a continuum, not as a binary.22 An insulating state and powerful political status quo can be a strong buffer, but it is not an                                                           22 On one end of the spectrum are sectors that are very highly insulated because policy implementers are not just state employees, they are also skilled workers with relatively high levels of autonomy from government decision-makers. Typical examples (and potential objects of analysis) include health care, education, and public transport, although this will vary across welfare states. On the other end of the spectrum are sectors with strong market dynamics, yet with some smaller role for an insulating state. Relatively absent the condition of state insulation, there might be only a few elected officials and bureaucrats who act as a buffer between ‘end-users’ of policy (such as citizens, consumers, firms, and so on) and external pressures (such as markets, globalization effects or transnational actors). Under these circumstances, decision-makers are likely to be concerned about effects of policy on end-users, and little concerned about how policies affect the handful of state employees charged with implementing policy or regulating a market.      19  impermeable barrier: state insulation reinforces the policy status quo, but cannot reproduce it indefinitely.   The scope conditions of acute external pressure and state insulation (involving regime-type and policy-type) created political ambiguity during the reform episodes under examination. Key concepts in PEF (selective attention; agenda-setting; framing; venue-shopping; policy monopolies) were evident during these punctuations to policy equilibria, when problems were identified and various actors strategized to solve them and establish a new policy status quo. For many such punctuations, the problems that draw attention have a technical dimension as well as engender ideological conflict, because powerful actors often disagree over cognitive ideas and normative ideologies. What is distinct about the sub-set of policy reform episodes examined with an agenda-management model, is that the policy status quo was punctuated by an event or actor outside the apparatus of the state, and this suggested a core governance problem that also should be addressed (even if implicitly). In other words, discrediting the policy status quo implied a critique of the political status quo, because the state seems to have failed some test of their governance capability. The circumstances described here are not unique to education policy, nor to the cases in question. Where they do exist, acute external pressure and state insulation demand that we focus on particular aspects within reform proposals, such as policies that affect the rationales of governance and implementation (i.e., instrumental logics).23  Given the scope conditions above, the range of potential policy change outcomes remains large yet finite. This analysis examines one such form of policy change, change in instrumentation, and compares it across political jurisdictions. Where such change in instrumental logics is successful, new policies serve the existing political status quo – they are a means for political actors to achieve their                                                           23 Absent conditions like war, economic collapse, or technological revolution, it is rare to see radical changes that re-orient the aims of public policy and redefine the interests of stakeholders in a given sector. A more likely path of reform, even significant reform, entails changes to programs or instruments that can – though not must – lead to ‘downstream’ transformations in overarching policy goals, the politics of policy-making, or even the polity itself.     20  abstract policy aims. Yet they are also innovative in that they introduce new programmatic methods, evaluation mechanisms, or modes of governance. I argue that we must consider a significant change in instrumental logic as a dependent variable in its own right, because it is qualitatively different from incremental adjustment or paradigm shift, and determined by change dynamics which are neither wholly incremental nor altogether disruptive.  1.3.2 Sectoral Institutions and Subsystem Politics Much of the scholarship on educational policy change (and policy change more generally) has focussed on the macro-political arena and the variables that matter there: ideas and social learning; partisanship in legislatures, or veto players or points in the institutions of ‘high politics’. However, the notion that policy can be explained as ‘simple, direct outputs of [macro-]political dynamics’ is misleading because policies themselves are increasingly complex and multifaceted (Howlett and Migone 2011, 60). To understand more complex forms of policy change, one must analyse factors beyond the macro-political. Explaining change is not just about which issues draw the attention of executive or legislative decision-makers, but also how an issue is framed on the agenda, and by whom. Policy design thus becomes an important factor for decision-makers, with types of change contingent on policy designs which can solve pressing problems without shifting blame onto those with the power to block solutions. Therefore, what is required is greater analysis of the institutions responsible for developing and preparing the policy reform agenda, and of the politics which determine the structure and stability of these institutions. Sectoral-level policy design institutions – policy subsystems – are essential for explaining changes to the logic of policy instruments.    An analysis of subsystem politics entails conceptualizing subsystems as well as their politics. A policy subsystem is an entity behind the government high executive; it is “a stable, tight and continuing arrangement” involving high-level political decision-makers with those that must implement policy decisions, such as bureaucrats and representatives of public-sector interest groups (Jordan and Maloney   21  1997, 559). Subsystems allow elected politicians and high civil servants to focus attention, avoid problem-overload, and manage the complexity of issues (for which they are seldom experts). As institutions, subsystems have been treated as relatively uncontroversial aids for the ‘parallel processing’ of information (Baumgartner and Jones 1993, 44). However, subsystems are not just providers of information, they are also potential sources for conflict or cooperation. On one hand, subsystems reduce uncertainty regarding cause-and-effect and help solve technical problems, on the other, they can reduce (or produce) uncertainty regarding the competition between different ideas or partisan interests. In effect, a stable subsystem with a coherent interpretation of a problem can manipulate the agenda with which high-level decision-makers operate, while unstable subsystems will struggle to control that agenda, meaning design decisions get hashed out in the macro-political arena. This makes an enormous difference in terms of change outcomes. Stable, well-structured subsystems can manage reform agendas (set by others), because they can combine innovation with the status quo, such as a significant change in instrumental logic. Unstable subsystems with conflicted images of a policy area will have unpredictable results: they might lead to paradigmatic transformation of the policy area, or to obstructed reforms that render policies ‘stagnant’.  Subsystem politics can shape the relationships between policy actors and political decision-makers in significant and (often) predictable ways, and therefore, I argue that our analytical focus must be on policy design institutions like subsystems, rather than macro-political variables, when given scope conditions prevail.    In terms of integrating this argument into a middle-range theory of change in instrumental logics, policy subsystems are ‘meso-level’ institutions which connect the ‘micro-level’ of individual and small group behaviour to the ‘macro-level’ of national policy outputs and international trends. At the micro-level, individual actors form policy preferences, and behave in strategic ways to amend or achieve those preferences (vis-à-vis policy learning, rational utility-maximizing, and so on). Meso-level institutions can help us to understand strategic behaviour at this level, such as why an actor might   22  decide against vetoing a policy that ostensibly contradicts their ideological or partisan interests. Actors perceive their interests differently (and re-order their preferences) if an institution offers policy designs that are stable, predictable, and suitable for credit-claiming, even if those policies are ideologically sub-optimal. At the macro-level, states appear as unitary actors susceptible to the diffusion of international ideas and trends. Meso-level institutions can help us understand the divergent behaviour of states, such as why some political jurisdictions do not conform to expectations regarding policy change or the internationalization of education policy. Policy change outcomes at the national level can be more comprehensible when one looks beyond the macro-level political arena of government executives, legislatures, and international organizations, to the types of cooperation and competition occurring within meso-level institutions.24 Just as a middle-range theory can bridge a detailed analysis of a single case study to general theories applicable across a larger-N, meso-level analysis can explain the connection between the behaviours of individuals, and the activities of states.  1.3.3 Policy Oligopoly Having established the importance of sectoral institutions for changes to instrumental logics, I introduce the novel concept of a ‘policy oligopoly’, a stable subsystem-type that allows actors within it to collectively manage the reform agenda in spite of disparate educational aims. This notion is based on Baumgartner and Jones’s idea of a policy monopoly, a cohesive community of experts and implementers within a given policy sector, who try to monopolize understanding of issues and problems.25 Policy                                                           24 Building on an ‘externalization thesis’ adapted from Niklas Luhmann by Florian Waldow (2009, 2012), I suggest that transnational ideas and knowledge function as a kind of ‘external irritant’ to relatively ‘closed’ educational policy subsystems. Actors within a policy community might use knowledge generated from outside the system (as a point of reference, to legitimize preferences, and so on) but this externalization is instigated and processed from within the system. In other words, education policy is often a distinctly national (or sub-national) affair, with clear and well-defended jurisdictions, and well-established domestic policy communities. Incidentally, externalization need not be transnational, it could also be intra-national (across sub-national policy communities) or even cross-sectional (borrowing from other policy area subsystems). It just so happened to be transnational in the case of PISA.   25 Conceptually, policy oligopolies (and policy monopolies) are not to be confused with monopolies and oligopolies in economic theory. These economic concepts have influenced subsystem-type in a heuristic way, in that policy   23  oligopolies – indeed, subsystems in general – are an analytical concept rather than a formal institution in the policy arena. But if these concepts have causal effect in empirical reality, then one should see evidence of them in terms of policy outcomes (i.e., one could hypothesize forms of policy change based on subsystem type), as well as evidence independent of outcomes (i.e., there would be institutional and behavioural indications that a policy oligopoly existed).    What makes oligopolies distinct from other types of subsystems – and a novel concept for understanding policy change – is how they vary regarding the core features of a subsystem, and how this variance affects policy outcomes. Oligopolies are subsystems for which compromise is effectively ‘baked-in’ to the institutional structure and communication procedures. They have an inclusive yet bounded membership, can agree on certain aspects of problems and solutions (such as instruments) without full consensus on aims, and can manage conflict in ways that include rather than exclude diverse viewpoints. This is not to suggest that oligopolies are the product of goodwill. They are likely borne of necessity, due to pressures of institutional complementarity, or new information and ideas which challenge powerful actors’ capacity to design policy alone. Yet policy oligopolies can achieve outcomes involving compromise, such as the managed reform found in changes of instrumentation.  1.3.4 Summing up the Argument To sum up, an agenda-management model of instrumental policy change is a means to understanding a new dimension of change in formal policy: significant change in instrumental logic. In order to explain this phenomenon, we need to look at the politics of subsystems, and to conceptually develop a new kind of meso-level policy design institution: the policy oligopoly. This model, and the arguments therein, explain the forms of policy change in France and Germany following poor PISA results. It also has a                                                           oligopolies describe a small, stable group of political actors trying to dominate (‘monopolize’) a social field, given limited internal competition. This is where the similarity ends. Policy oligopolies (and monopolies) need not lead to sub-Pareto outcomes, can persist in spite of inefficiency or instability, and pertain to images of a policy area, not to the behaviour of firms in a market (Kreps 1990, 325–26).    24  number of implications for our understanding of the relationship between the status quo and policy change. It is to these theoretical contributions I turn to next.  1.4   Contributions to Scholarship on Public Policy Beyond the merits of explaining an important set of empirical phenomena and analyzing a policy area underexplored in political science, this dissertation develops a middle-range theory of policy change that makes three contributions to the study of public policy.  First, an agenda-management model contributes to new conceptualizations of the relationship between policy change and the status quo, by positing likely conditions for an intermediate form of policy change during reform episodes when other forms of policy change were anticipated. This finding builds on and refines the Punctuated Equilibrium Framework. For some punctuations to policy equilibria, there will be modes of change where existing policy aims are combined with innovative implementation logics to produce significant yet not paradigmatic policy change. Thus, not all punctuations are fully transformative. Nor are punctuations necessarily unpredictable: certain types of sectoral institution can render volatile reform circumstances more stable and calculable. Finding a new form of policy change also suggests that models within the PEF should not always select on the dependent variable. Rather than define a punctuation by outcomes that are transformative of the status quo, one might instead focus on disruptive episodes where radical policy change might be expected but not always delivered. There has been much progress in our understanding of the relationship between stability and change in institutions, but less developed is our understanding of policy change dynamics – with the meanings of incrementalism vis-à-vis transformative policy change becoming the subject of recent debate. This dissertation contributes to this discussion by proposing concrete examples of significant change that do not establish an entirely new status quo.  Second, the comparative study of meso-level (policy design) institutions begins to establish a range of cases when subsystem politics is likely to matter for policy change. Variables in the macro-  25  political arena will continue to have much salience for theories of policy change, yet this dissertation explains how certain types of insulated policy community can translate these variables into reform preferences, and given the right conditions, managed reform agendas. The close and comparative analysis of a positive and a negative case of change in instrumentation, allows us to see the ways in which subsystem politics can guide – or misguide – the strategies of high level decision-makers when a policy area is in the spotlight. An agenda-management model of policy change will prevail under a relatively narrow set of conditions and in the presence of distinct causal factors, but those conditions and factors are not unique to German education policy (post-PISA). This dissertation proffers the descriptive building blocks for a middle-range theory of (limited) policy change, and begins to establish the institutional conditions and subsystemic causal factors that exist in cases of changed instrumentation.     Third, with the novel concept of a ‘policy oligopoly’, this dissertation introduces an identifiable and causally-significant form of policy design institution which exists (conceptually) at the sectoral level of the policy development process. Existing analytical frameworks provide an elemental description of the relationship between subsystems and specific policy content during periods of relative stability. In short, insulated communities use content to frame issues and set the policy agenda, or else use content to build and sustain coalitions by advocating for particular policy beliefs. However, what is not adequately explained, is how the form of these institutions relates to the form of policy change. Policy content notwithstanding, oligopolies are structured and stabilized in a way that allows them to manage agendas and shape reform outcomes during punctuations in the policy equilibrium. The policy oligopoly is a bounded concept, and its application is limited to scenarios with particular antecedent historical and institutional conditions, and certain contingent circumstances. However, this scenario is not exclusive to ‘post-PISA’ German education, so this dissertation begins to elaborate the conditions and circumstances where subsystem politics matter, and policy oligopolies have effect. As a contribution to our   26  understanding of changes to the instrumental logic of policy, the policy oligopoly is a concrete analytical tool which can illuminate the ways that meso-level policy-design institutions affect policy, and speculate on the ways that institutional and policy change might be related.  1.5 Methods and Evidence The purpose of this dissertation is to develop a middle-range theory for intermediate forms of policy change, by focussing on a narrow set of political phenomena: how a change in instrumental logic occurs when an insulating state comes under acute external pressure. Policies pertaining to compulsory education are a good place to begin, because this policy area – normally well-insulated and resilient – experienced acute pressure to change in a number of countries, during the 2000s. The combination of a powerful status quo, acute pressure for change, and unexpected outcomes in ‘deviant’ cases suggests that education policy would be suitable for preliminary theory-building for changes to instrumentation.  1.5.1 Research Design For education, there have been two general schools of thought for defining and measuring policy change. One strategy has involved a kind of ‘comparative statics’ approach, as used in the literature on the internationalization of education policy (Dobbins and Knill 2009; Jakobi, Teltemann, and Windzio 2010; Knill and Lehmkuhl 2002; Martens, Knodel, and Windzio 2014; Martens and Weymann 2007). This ‘spatial’ comparison assesses change by comparing snapshots of (national) policies to an external barometer, with cross-national variance indicated by the degree of convergence or divergence from some standard or model. It is useful for observing larger cross-national trends, such as the internationalization or globalization of education policy, but less instructive for assessing the particular (or surprising) dynamics of cases that converge/diverge against expectations (i.e., for explaining outliers). Another strategy involves internal comparisons of single cases over time (the ‘case study approach’), with variance indicated by an increase or decrease in some measurable feature – preferably a feature also commensurable across cases, such as spending. This ‘temporal’ comparison can be useful   27  for describing variance across education systems, especially for early institutional development (i.e., how the status quo was established). It is less helpful for explaining reform dynamics in well-established systems given common ‘external’ stimuli. Evaluating the mechanisms that cause cross-national variation in re-instrumentation requires a different kind of methodology: a kind of ‘comparative-dynamics’ approach where within-case variance (between status quos) is also compared cross-case , as a variance in degrees of policy change. A comparative-dynamic approach has yet to be applied to a cross-national analysis of education policy, and doing so allows the tracking of single cases dynamically (comparing status quos within-case) while also comparing variance spatially (comparing type of change in one case to type of change in another).  This qualitative and comparative research design has embedded methodological tools and logics that can help one draw causal inferences from small-N comparison. Within each case, I use process-tracing to analyze empirical evidence on patterns of policy development and conjunctures of events, hypothesizing on the causal mechanisms that could explain each outcome. Process-tracing is a kind of analytical narrative that captures within-case variation over time, in order to ‘test’ variation against competing theoretical explanations and encourage assessment of how change occurred, rather than just why it occurred.26 This method involves “the analysis of evidence on processes, sequences, and conjunctures of events within a case for the purposes of either developing or testing hypotheses about causal mechanisms that might causally explain the case” (Checkel and Bennett 2014, 7). By carefully analyzing processual links, and discriminating between alternative explanations, this within-case method                                                           26 Process-tracing is particularly effective for ‘combinatory and conditional’ accounts of change in which antecedent conditions, interactive variables and processes seem to matter for the outcome, and where the focus is on how change happened. This is distinct from ‘nomothetic and positivist’ accounts of change in which the variance between isolated (independent and dependent) variables seems to matter, and where the focus is on why change happens (Capano 2009). Moreover, process-tracing implies more than historical narrative. For example, the comparative-statics approach tends to treat history as linear and additive rather than iterative: events simply follow one another, with history as a sequence rather than a process. An ontology of change (rather than of just sequence) takes historical irreversibility seriously, and calls for a different methodological approach that recognizes irreversibility and periodization, one way being to treat policy reform episodes as ‘re-iterated problem-solving’ (Haydu 1998; Rayner 2009).   28  derives explanatory leverage from internal validity rather than from assessment of ‘treatment’ by exogenous (independent) variables (Schimmelfennig 2014). In practice, process tracing relies on historiographical forms of evidence, such as interview transcripts, primary and archival documents, and secondary analyses (George and Bennett 2005, 6). Yet it is more than just historical detective work. Rather, it involves the construction of hypotheses from the observable implications of various causal mechanisms, then the testing of alternative explanations against one other on the basis of observed intervening variables or indicators of causal process. One begins by considering all plausible explanations – some of which could be in extant literature – which make implicit or explicit theoretical claims regarding independent variables, or the causal roles of agents or structural factors. The researcher is then critical of all explanations, attempting to disconfirm biases towards preferred models. The object is to derive hypotheses which are brittle, and thus falsifiable by observable evidence that has probative value. Ultimately, evidence is sufficient when ‘Bayesian-inspired criteria’ are met: a smoking-gun test, coupled with failed hoop tests for alternative explanations, can provide the researcher with a high (though not absolute) degree of confidence in a particular theory. 27 Although there is always the risk of omitted variable bias in small-N research, the researcher overcomes this by examining diverse and independent streams of evidence, by corroborating sources and triangulating data, and reaching the point of evidentiary ‘saturation’. Process tracing is a key technique for capturing causal mechanisms in action, a research logic that functions beyond historiography and beside the logic of frequentist statistics.                                                           27 Notably, it is seldom possible to get 100% confirmation of a mechanism or model. Even a proverbial ‘smoking-gun’ is not conclusive, because causal mechanisms are rarely observable (except indirectly). However, our hypotheses about them can generate observable and testable implications, and from these, one can devise ‘hoop tests’ to disqualify certain explanations. Hoop tests, coupled with compelling (albeit circumstantial) evidence that positively supports a particular set of hypotheses, can provide a high degree of confidence in a particular model (Checkel and Bennett 2014; Van Evera 1997). Evidence can always be further probed – or new evidence discovered – which leads to the falsification or modification of a theory, yet the method yields clear expectations and indicators for falsifiability. The point here is that process tracing offers a robust method for evaluating rival explanations, yet also allows for amendment of models in light of more or better evidence.     29  An additional aspect of the research design used in this dissertation is cross-case analysis, in which the logic of Mill’s joint method of agreement and difference guides a kind of comparative process tracing. The comparative analysis adheres to a logic of Most Different Systems Design when selecting cases, and then a logic of Most Similar Systems Design when assessing the inner workings of each case. Most Different Systems Design (MDSD) affirms that a shared variable in two dissimilar cases can account for similar categories of outcomes (here, this refers to similar attempts at re-instrumentation).  Most Similar Systems Design (MSSD) accounts for cross-case variance between actual policy change outcomes, by establishing the independent variable or causal mechanism that mattered most for the realized degree of policy change.   Process tracing typically proceeds through a combination of induction and deduction – even within-case – yet comparative process-tracing further leverages these different research logics, in order to draw stronger inferences via a comparison of processual evidence across cases. Where cases are not well-explained by extant theory, inductive process tracing uses within-case evidence to develop hypotheses. These hypotheses can then generate observable and testable implications not only regarding other within-case phenomena, but also for the other case, allowing one to deduce the presence or absence of theorized causal mechanisms (Checkel and Bennett 2014, 8). Whereas process tracing renders a high degree of confidence due to internal validity, the comparative method reinforces these findings through tests of external validity.28    Thus, comparative process-tracing provides greater explanatory leverage for singling out one causal process over others. Process tracing addresses the limits of Mill’s methods because it can confirm the viability of mechanisms for each case even when outcomes differ or pathways diverge, while the comparative method can leverage external validity. Comparative process tracing also reduces the                                                           28 It is not necessary to have external validity to have a high degree of confidence in research findings, however. “Indeed, a theory or explanation derived inductively from a case does not necessarily need to be tested against a different case for us to have confidence in the theory; rather, it can be tested against different and independent evidence in the case from which it was derived” (Checkel and Bennett 2014, 13).   30  likelihood that omitted variables explain the variance in outcomes, vis-à-vis a close examination of the causal processes which led from independent to dependent variables.29 In practice, one begins with immersion in the details of each case, colloquially known as ‘soaking and poking’, in order to develop ‘proto-hypotheses’ that must be tested for their plausibility. These hypotheses can then be tested cross-case, using evidence that is fully independent from the evidence that initially gave rise to those hypotheses, and proceeding by a deductive method to eliminate (or modify) hypotheses. The ‘joint method’ used in conjunction with process-tracing reveals not only the most plausible explanation for why reform was attempted, it also explains how change actually occurred in each case. 1.5.2 Case Selection Given the methodology above, there were two dimensions which guided case selection: temporal and spatial. Temporal case selection refers to the ‘PISA moment’: why do we see these particular forms and processes of policy change after PISA shock, and not before?  Spatial case selection refers to a comparison between Germany and France following PISA shocks: why these countries and not others?  I deal with each in turn.   As a ‘temporal’ case, PISA was selected because it presents a test of theories about the relationship between internationalization and national policy change. PISA was a kind of exogenous shock with unique characteristics for contributing to the internationalization of education policy. It was also the discursive reference point for attempts at policy change in several countries. PISA appeared to be a critical juncture for new forms of policy change, such as re-instrumentation, and national actors identified PISA as a primary reason for attempting such change. It was thus a ‘most-likely’ case for dramatic policy change in response to internationalization. However, the variance between attempted and actual reform suggests that we must further assess the validity and scope conditions of competing                                                           29 MSSD rarely controls for all but one potentially causal factor, so process tracing can establish that other differences between the cases do not account for the difference in their outcomes (Bennett and Checkel 2014, 29)   31  theories of policy change. Some national policy frameworks encountered (or induced) acute pressure from PISA and changed, while others did not, and this invites closer examination of PISA as a case for theory testing (Eckstein 1975; George and Bennett 2005, 75).   As ‘spatial’ cases, Germany and France were selected because their outcomes were not what traditional theories would anticipate, and thus offer opportunity for theory-building.  Along with a subset of other countries that underperformed in international ordinal rankings, Germany and France were ‘an instance of a class of events’ (George and Bennett 2005, 17). They were comparable to one other in terms of attempted outcomes and general antecedent conditions (i.e., an insulated area of the welfare state deemed as failing by an international organization). However, they were also ‘deviant’ or ‘outlier’ cases because they did not conform to theoretical expectations of policy change. Given the pressurized circumstances, France had the structural and contextual conditions that one associates with rapid and significant transformation of a policy framework, yet they were incapable of affecting even significant change to instrumental logic. Germany had the conditions one would associate with a resilient status quo, maintained by traditional ideas and partisan stalemate, although had the surprising outcome of successful change of instrumentation. The scope conditions establish the universe of possible cases where successful re-instrumentation could be induced. Placing Germany and France within this universe, their cross-case dissimilarity regarding welfare states and policy-making regimes (with differences in policy legacies, structures of education, demographic shift, macro-political institutions, and so on), narrow the causal pathways to a handful of mechanisms common to both cases. The fact that the case with the most impediments could affect re-instrumentation and the case with the fewest hurdles could not, suggests that some process – as yet theoretically undeveloped – explains the outcomes for both. Individually, Germany and France could be analyzed as heuristic case studies, useful for identifying new variables or causal mechanisms (George and Bennett 2005, 75). Together, they point   32  to a common but as yet unknown process or set of mechanisms, which can induce – yet not necessarily produce – a significant change in instrumental logic.   1.5.3 Evidence The types of evidence used for comparative process-tracing were both intensive and extensive. This evidence ranged from expert and participant interviews, to close reading of primary policy documents, to analysis of secondary materials such as academic publications and media reports. Elite interviews are a key method of data-gathering in process tracing, allowing the researcher to accurately reconstruct sets of events. My method of identifying interviewees involved mapping the individual and collective actors directly involved with the policy-making episodes, and then contacting as many as possible to ensure a diversity of perspectives on each policy-making process. I also proceeded using a ‘snowball’ sampling technique, having interviewees direct me to other participants or experts. The 58 interviewees, some of whom were interviewed on more than one occasion, were comprised of educational experts, elected politicians, civil servants at the national and sub-national level, policy advisers with NGOs and international organizations, academics, representatives of teachers’ unions and associations, and specialized journalists. The interviews were semi-structured to ensure that I could direct the conversation towards the issues I was trying to explain, yet also allowing interviewees to focus on topics they felt were important.  I continued the interview process until I reached the point of saturation, in which interviews began to yield variations on previous responses (see Appendix G: Interview Methods).  Beyond interviews, the research built on primary documents, including legislation, official policy documents, archived meeting minutes, historical archives of media sources, and reports or white papers from sub-national and national ministries or international organizations (see appendix H: Primary Documents). Evidence from primary sources was further corroborated by secondary literature from French and German educational experts and journalists. The range and depth of primary and secondary evidence allowed me to construct an accurate analytical narrative for each case, highly substantiated yet   33  also reflecting a diversity of perspectives, which then ensured that I could make valid and reliable comparisons between the case analyses. Analysis of the evidence met the standards of current best practice in process tracing: consideration of alternative explanations; safeguarding against confirmation bias; amassing of considerable amounts of evidence from varied sources including archival research and interviews; and combining process tracing with cross-case comparison (see Waldner 2014).  Evidence gleaned from interviews and other primary sources has probative value for building and assessing alternative explanations, but the sources themselves must also be probed for reliability. Evaluating sources of evidentiary bias – such as adjudicating between competing claims made by interviewees – was guided by the standards suggested by Marc Trachtenberg (2006), namely: attention to issues of context and authorship, assessment of the instrumental motivations of sources, and consideration of availability and accessibility of evidence. In practice, this meant employing multiple strategies for verifying the reliability and availability of source material. For primary documents, the key was to avoid misunderstanding the context or purpose of a source. One strategy for overcoming issues of context and authorship was triangulation: using other data sources – elite interviews, especially – to corroborate information.30 Another strategy involved analysis of meta-data: ‘interrogating’ source material for descriptors of who produced it and why.31  The latter strategy also pertained when assessing the reliability of other types of primary sources, such as elite interviews. While preparing for interviews, I considered how problems of mistaken recollection or misrepresentation of role might distort my findings. When evaluating the instrumental                                                           30 Elites who were directly involved in the policy process can be asked to confirm, deny or elaborate on the context, intention, or purpose of ambiguous primary documents. Moreover, interviews also allow the researcher to move beyond written accounts, which might only suggest an ‘official’ version of events.  31 For key documents, analysis of meta-data was guided by the criteria established by George and Bennett (2005, 99), including who is speaking (for example, are the authors identifiable individuals, or a more ‘anonymous’ groups such as a ministerial department)?  To whom are they speaking (for example, is it for a private or public audience)?  For what purpose are they speaking (for example, is the document intended as descriptive or prescriptive)? And what are the circumstances under which this document was produced (for example, was it produced when issues were ‘politicized’ or ‘administrative’)?     34  motives of interviewees, I followed Bennett and George’s criteria for analyzing any primary source (see footnote 31). I also considered criteria suggested by Philip Davies (2001) for evaluating the reliability of interviewees: whether they were a first-hand witness, what level of access they had to the events in question (with senior elites deemed more reliable), and if they had a previous record of reliability.  Finally, as a ‘validity’ measure of interviews, I considered the degree to which an interviewee was representative of a larger population of elites.32 It is impossible to interview every individual involved in a policy process spanning a decade or more, yet one can ascertain whether the basic preferences or biases of a given actor align with the expected (and often publicized) positions of collective actors on those issues (Tansey 2007).   This last point – that it is impossible to collect every shred of evidence – points to the third standard suggested by Marc Trachtenburg: how the availability of evidence can truncate or distort one’s causal inferences. Ultimately, an evaluation of the absence (or sufficiency) of evidence is a practical kind of knowledge, not a theoretical judgment. The researcher must make decisions about if and when they have enough reliable data to generate plausible explanations for the political phenomena in question. There are always constraints to evidentiary access, not least of which is limited resources. Yet evidence can also be unavailable because key decisions were made privately or in camera, or because potential respondents are unwilling or unable to be interviewed. Two strategies guided my decisions for knowing when to stop gathering data: consultation with local experts and ‘saturation’. Regarding the former, informal discussions with experts and academics immersed in French and German education policy yielded important advice on which archives or primary documents were accessible, which potential respondents were likely to grant interviews, and where other sources of data might be available. Regarding the latter, evidence was gathered until the point of saturation, when adding new streams of                                                           32 This is not a check of reliability, but it does allow the researcher to assess the representativeness of a given actor compared across the same category of elites.     35  evidence or lines of questioning no longer yielded new information. 33 Guided by the evidentiary standards above, and utilizing the current best practices of process tracing, I was able to draw robust inferences about the causal processes in each case.   1.6 Conclusion In this chapter, I proposed a set of interrelated arguments regarding the relationship between institutions of policy design and forms of policy change, for insulated sectors under acute external pressure. Given these conditions, we must re-conceptualize policy change to better account for intermediate forms of change, such as re-instrumentation, which embed innovative means for governance or implementation into an existing framework of policy aims and objectives. Such attempts to control reform during a punctuation in the policy equilibrium can be explained vis-à-vis subsystem politics. The degree to which a managed reform was successful depended on the properties of the policy subsystem. A policy oligopoly – in which members shared political if not policy goals – greatly increased the likelihood of a significant change in the instrumental logic of education policy.    These arguments suggest that we should refine our understanding of policy internationalization and the domestic factors constraining policy change. Rather than continue to assume that partisan ideologies and/or a diffusion of ideas are the factors which drive policy change, and that macro-political institutions resist this change, we must examine the actual causal pathways that lead from policy design to policy inception. A close and comparative analysis of two cases with unexpected policy change outcomes reveals that much variance can be explained by the distinct properties and features of sector-specific subsystems. Under the right conditions, subsystem types can determine forms of policy change during punctuations in the policy equilibrium when the status quo has been disrupted and challenged. Institutions and institutional arrangements might be more important for change outcomes than                                                           33 “That is, a researcher should stop pursuing any one stream of evidence when it becomes so repetitive that gathering more of that same kind of evidence has a low probability of revising their estimate of the likely accuracy of alternative explanations” (Checkel and Bennett 2014, 28).   36  currently anticipated by existing theories of policy-making. Institutions at the meso-level can sustain the status quo and guide national processes of policy change (and internationalization), more so than variables at the macro-political level.     1.6.1 Dissertation Roadmap The rest of this dissertation unfolds as follows. In Chapter Two, I develop a theoretical framework which connects subsystemic institutional structures to a significant change of instrumental logic (given particular scope conditions). In so doing, I define key concepts such as change to instrumentation, subsystemic politics, and policy oligopolies, and then present a model of agenda-management which explains managed and mismanaged reforms during punctuations to policy equilibria. This chapter develops causal propositions for the agenda-management model, proposes alternative explanations for the cases analyzed here, and generates measurable indicators to test the plausibility of each explanation. Chapters Three and Four operationalize the model for the two cases under analysis: Germany and then France. Each chapter provides historical and institutional context for the case in question, describes the pattern of policy change, and explains the emergence of particular forms of policy subsystem. These case chapters then establish evidence for the type of policy design institution during the punctuation in the policy equilibria, analyzes the effects of said subsystem, and compares this account to alternative theories. Carefully tracing the causal pathways and mechanisms of policy-making, and testing propositions from each plausible explanation, allow me to draw conclusions about each case. In the final chapter, I summarize the dissertation, compare accounts of policy change, and draw conclusions about the veracity of an agenda-management model of policy change. In the concluding chapter, I also turn to questions of generalization and external validity, re-specifying relevant scope conditions and reviewing theoretical contributions. To conclude the dissertation, I suggest ways in which the model can be further corroborated as well as extended to other country cases and policy areas.       37  Chapter 2: Theoretical Framework for an Agenda-management Model of Policy Change  2.1 Introduction: Subsystem Politics and Education Reform This chapter presents a theoretical framework for the comparative study of policy change in  ‘insulated’ policy regimes, during a period of intensive demands and pressures generated by globalization. The phenomena to be explained is why, under certain conditions, some countries could significantly change instruments of educational governance and evaluation, while others could not. In Section Two, this question is framed in terms of the Punctuated Equilibrium Framework (PEF) and compares the conventional agenda-setting model to the proposed agenda-management model. The latter argues that there are alternative patterns of policy-making during punctuations in policy equilibria, and that these lead to more intermediate change outcomes which are not yet well-theorized by the PEF. Section Three picks up the argument by providing more nuanced descriptions of policy change, and re-conceptualizing a form of change which accounts for both policy continuity and reform: a significant change in instrumental logic. In the fourth section, I argue that institutions of policy design are the key variable for understanding ‘re-instrumentation’ during punctuations in education policy equilibria. One particular form of subsystem (and a novel concept introduced in this dissertation) is a policy oligopoly. An oligopoly is a stable form of subsystem that can manage agendas during periods of policy instability, by defining problems as technical, and solutions as instrumental and manageable back in the subsystem. In Section Five, I argue that the mechanisms by which subsystem politics get translated into managed reforms is vis-à-vis the framing of policy issues and the control of policy development venues. Framing and venue-control are stabilizing strategies used by subsystem actors, as well as stabilizing mechanisms induced by the institutional field of subsystem politics. Section Six lays out the model’s theoretical propositions. For cases where an agenda-management model pertains, the existence of a policy oligopoly is sufficient not only to secure a greater likelihood for significant change   38  in the instrumental logic of policy, but also explains the persistence of the subsystem itself during punctuation in policy equilibria. In Section Seven, I posit the scope conditions that pertain to an agenda-management model, and discuss why the political, policy and historical contexts of each case establish common antecedent conditions, yet also unique status quos. In Section Eight, I suggest that ‘meso-level’ institutions were more determinate for the forms of policy change seen in these cases, than plausible alternative explanations such as policy internationalization driven by ideational diffusion, or change driven by party politics and electoral interests. I conclude in the final section by arguing for the plausibility of an agenda-management model and the implications this model has for a Punctuated Equilibrium Framework. Finally, I suggest that we reconsider the power of the status quo, and how it affects the relationship between subsystem politics and policy change.  2.2 An Agenda-management Model within a Punctuated Equilibrium Framework An agenda-management model of education policy change builds on the Punctuated Equilibrium Framework (PEF), an analytical approach to the policy process that can account for both policy stability and policy change (Baumgartner and Jones 1991, 1993, 2002, 2009; Baumgartner, Jones, and Mortensen 2014; Jones and Baumgartner 2005). Amongst the major theoretical approaches for the making of public policy, PEF is most useful for three reasons. First, it can describe the dependent variable in terms of degrees of change (i.e., as a variance between one status quo to another), whereas Multiple Streams Approach (MSA) and the Advocacy Coalition Framework (ACF) place greater emphasis on the timing or content of policy change. Second, PEF tries to systematically account for why disruptive reform episodes occur, vis-à-vis selective attention and exogenous shocks. MSA and ACF do not proffer a systematic (cross-case) means for understanding the dynamics of the ‘why now’ question, because this question gets answered idiosyncratically and within-case (vis-à-vis windows of opportunity or moments of coalition formation). Finally, PEF incorporates attentional dynamics into a framework for understanding how change occurs. While agendas matter for other approaches, how they matter is comparatively   39  under-theorized. PEF is an appropriate approach for tackling these specific research questions, for the particular dynamics in each case, and to permit systematic cross-case comparison.  However, the above does not suggest that an agenda-management model is simply a stage within the ‘standard’ PEF model of policy change. Instead, agenda-management is an alternative pathway in PEF, leading to different policy outcomes. In short, the conventional agenda-setting model prevails where agenda-management fails. Key aspects of PEF shape both models: punctuated policy equilibria; the importance of agendas (and selective attention); the critical role played by institutions; and similarities in the mechanisms of change. However, due to the scope conditions relevant for an agenda-management model, some of the dynamics of the ‘standard’ agenda-setting model must be altered. The former describes less disruptive policy change, and explains this change using different patterns of policy-making. Relative to agenda-setting, the agenda-management model predicts agendas which are set ‘externally’ yet managed ‘internally’, attention within issues (not just across them), the influence of institutions beyond the macro-political, and causal mechanisms which stabilize policy as well as change it. In the section below, I briefly review the ‘standard’ model of agenda-setting in the PEF, before highlighting key amendments for an agenda-management model.  2.2.1 The Standard Model: Agenda-setting and Disruptive Policy Change The agenda-setting model suggests that an especially prominent (though not exclusive) pattern of policy change involves long periods of stability interspersed with brief and relatively unpredictable periods of transformation, during which the attention of macro-political actors has been drawn to a policy area. The outcomes predicted by this model were exemplified by postwar US nuclear policy: decades of policy stability before (and after) a brief yet highly disruptive period of change that transformed the orientation, aims and governance of nuclear policy. This punctuation in the policy equilibrium was triggered by the accident at Three Mile Island in 1979, which captured the attention of high-level decision-makers in government. Models in the PEF argue that agendas are critical: those who can draw   40  government attention to issues can also affect policy change. For agenda-setting, drawing attention to a policy area occurs via two mechanisms: framing and venue-shopping. Framing refers to how an issue or problem is defined. Positive framing suggests that a policy area is well-managed, with problems that are insignificant or well-managed. A negative image of crisis or failure is a call for widespread attention and immediate action, leading to repudiation of the status quo and transformation of a policy area. Venue refers to where policies are discussed. Framing issues in different ways allows actors to move policy discussions to venues more favourable for the realization of their preferences. Much of the time, policies remain stable because positive framing and venue-control keeps issues out of the spotlight. This is what a policy monopoly does: it conveys a positive image of the policy area, reduces attention and conflict, and restricts venues for policy discussion and development (Baumgartner and Jones 1993, 2009; Baumgartner, Jones, and Mortensen 2014). However, monopolies are not invulnerable forever. In an agenda-setting model, framing and venue-shopping are causal mechanisms which drive policy change and undermine monopolies, as high-level decision-makers react to information which can induce them to transform policies and select new policy designers. The key institutions (and actors) for this model are in the macro-political sphere (i.e., the ‘high’ executive, legislatures, and judiciaries), although these types of actors are also responsive to issue-attention in the public sphere (i.e., public opinion and media coverage). In sum, policy change has a punctuated dynamic because policy areas move in and out of the public and macro-political spotlight. Patterns of policy change are determined by the selective attention of high-level decision-makers in government.   2.2.2 An Agenda-management Model of Policy Change An agenda-management model suggests that some of the dynamics above are altered when reform agendas are the result of ‘external’ pressure (such as globalization effects), when a policy sector is well-insulated by the state, and when the political effects of policy change are ambiguous. These conditions shift the patterns and processes within PEF, such that reforms are less disruptive, and so that different   41  types of institutions begin to play a more prominent role for driving and constraining change. For one, policy change need not be disruptive nor transformative even during a punctuation in the policy equilibrium. The agenda-management model examines cases where there were attempts to significantly change the logic of instrumentation during a punctuation, policy goals notwithstanding. Second, the model assumes that the reform agenda is set by actors not in the policy sector itself: bureaucrats and implementers within the policy area react to an agenda set by others, and try to manage it.34 Third, these actors then attempt to draw attention to particular aspects within the problem, allowing them to gain or retain political control over future policy design. Framing and venue-control thus become strategies (and mechanisms) not only for change, but also for preserving the status quo. As such, certain subsystem-types (like oligopolies) will have a greater capacity to frame problems as technical and solutions as instrumental, with implementation best managed back in the subsystem. Other subsystem-types will present a conflicted interpretation of the issues, with subsystem actors ‘shopping’ their policy image in different venues to either affect greater change or block reform altogether (depending on their preferences). In the case of the latter dynamic, agenda-management effectively reverts to agenda-setting; institutions and actors in the macro-political sphere again become the agents, determining forms of policy change (if any). I argue in this dissertation that significant changes to instrumental logic are determined by stable subsystems (in the cases here, by an oligopoly). Oligopolies do this by framing issues as manageable, and keeping policy design discussions – if not all forms of debate – out of the macro-political and public spheres.   2.3 Patterns of Policy Change within a Punctuated Equilibrium Framework (PEF) In many respects, “all policy is policy change” (Hogwood and Peters 1983, 25). A primary purpose of government is to solve societal problems and affect change in areas where human intervention is                                                           34 The agenda-setting model does not specify particular characteristics for events that trigger punctuations.     42  possible and desirable, and this is accomplished via public policy. However, there is seldom a straightforward relationship between the emergence of societal problems and the political response to them, because the timing and form of public policy seldom correlates neatly with the emergence of problems. As such, longitudinal analyses reveal a ‘leptokurtic’ pattern to policy change: reforms are unevenly distributed, involving stable periods with incremental adjustment to the policy status quo, interspersed with short bursts of profound change (Baumgartner 2013; Jones and Baumgartner 2005, 111).35 These dynamics are captured in the Punctuated Equilibrium Framework (PEF). Education policy changes during the 2000s were largely consistent with the ‘disruptive dynamics’ described by PEF, although in the cases examined here, elements of policy change are in need of further elaboration.  Defining change has been a perennial issue for comparative public policy. Much like the dependent variable problem found in literature on welfare state retrenchment, analysts of policy change have often been more adept at conceptualizing explanatory factors than carefully describing outcomes of interest (Howlett and Cashore 2009, see also Green-Pedersen 2004).36 The current orthodoxy for describing policy change is based on the seminal work of Peter A. Hall (1989, 1993). Hall conceived of three orders of policy change: first order change in policy settings, second order change in policy instruments and settings, and third order change in policy goals, instruments and settings (also called paradigm shift because it establishes a qualitatively different kind of policy framework). In this reckoning, first and second order change involved incremental adjustments to the status quo, whereas third order change established a new status quo.37 Conceptualizing policy change in terms of degrees                                                           35 Leptokurtic refers to a type of distribution (in this case, policy change) which is characterized by high and sharp ‘peaks’ (where dramatic change occurs) and ‘fat tails’ outside the mean (in other words, little or no change much of the time).    36 “Paul Pierson argued that ‘it is difficult to exaggerate’ the obstacle the continued dissensus over the definition, operationalization and measurement of policy change creates for comparative research and theory construction on policy dynamics” (Howlett and Migone 2011, 59; see also Pierson 2001).  37 Hall’s ‘new orthodoxy’ was a response to long-standing debates and misunderstandings over what constitutes policy change. In the tradition of Lindblom (1959) and Heclo (1974), incremental policy change long meant minor adjustments that reinforced the status quo. Hall’s model – and indeed the Punctuated Equilibrium Framework – sought to explain other types of policy change, and draw distinctions between incremental adjustment and   43  (orders) is problematic because it is difficult to operationalize, and less clear about significant forms of change that do not entail total repudiation of the status quo and re-orientation of a policy area.38 One response has been to collapse all three dimensions into a single variable: change or no change (Baumgartner 2013). This might be helpful for comparing policies with highly commensurable units of measurement (such as education budgets), but less so for policies with qualitative differences, such as modes of governance. Another approach makes a distinction between orders and types of change (Cashore and Howlett 2007; Howlett and Cashore 2009). This suggests that there can be significant (though not paradigmatic) changes to policy, such as reform of governance or evaluation policies, which meaningfully alter the status quo without upending it. A profound change to the means of achieving abstract policy aims is transformative, even if the goals of a policy area remain substantially the same.  2.3.1 Forms of Policy Change I argue that we must adopt change in instrumental logic as a qualitatively distinct dependent variable, to accommodate a broader range of empirical phenomena. Howlett and Cashore (2007; 2009) have begun to develop a typology of policy change, in which six types of change are discernible. Policy change can vary according to content (i.e., as a degree of change at different levels of abstraction, much like Hall’s framework) and in terms of focus (change oriented either to goals, or to the means of implementing those goals). The finer details of this typology are not important for the cases here, except to say there is one form of change that can be quite significant yet need not repudiate the policy status quo. This is a                                                           profound policy change. However, there remains much ambiguity surrounding the relationship between paradigmatic and non-paradigmatic types of change, because it is not clear under what circumstances anomalies are just anomalies (with a policy response that re-adjusts back to the existing status quo) and what circumstances anomalies are sufficiently large or cumulative to discredit the status quo and engender a paradigm shift. Now, the idea of ‘policy adjustment’ is itself being reconsidered, perhaps inspired by the gradual transformation models of institutional change (Streeck and Thelen 2005; Thelen 2003). There has been the realization that ‘incrementalism’ was long used as an umbrella term for a number of quite different types of concepts and policy dynamics, which require extensive disambiguation (Cashore and Howlett 2007; Howlett and Migone 2011; Rothmayr Allison and Saint-Martin 2011).  38 In relation to the current orthodoxy, the dependent variable described here is not so much a degree of policy change as a form of change (i.e., it is like second order change in that it pertains to instruments, but also shares a characteristic feature of third order change in that it involves changes to the logic which undergirds policy).    44  change in the logic of policy instruments, such that the norms and cognitive theories which guide policy implementation are transformed, while the abstract aims for a policy area remain the same. Clearly, this represents a significant form of policy change – it requires a high level of abstraction, and results in further changes to specific mechanisms, techniques, or tools (and calibrations thereof). However, it is not totally disruptive either, because actors can use new instruments – which function according to a coherent logic – to achieve their original goals (be that equity, excellence, high expenditure, differentiated schooling, and so on).39 Public policy instruments, as understood here, are the object of analysis inasmuch as I am explaining a particular form of policy change rather than specific policy content.40 Here, the point is to understand why there was successful change in instrumental logic for one case, and not the other. In the discussion below, I draw on the work of Howlett and Cashore (2007; 2009) and Lascoumes and Le Gales (2007) to conceptualize significant and sustained change to the logic of policy instruments.  2.3.2 The Logic of Public Policy Instruments An instrumental logic in public policy has conceptual coherence: it is a mechanism of governance or implementation that can both monitor and steer a social system, “a device that is both technical and social” (Lascoumes and Le Gales 2007, 4). As a technical device, it operates according to a logic of                                                           39 At least over the short-to-medium term.  The long-term implications of a change to instrumental logic are not analysed in this dissertation. However, ‘instrumentation’ has been the subject of considerable analysis amongst public policy scholars, and evaluations of instruments as an independent variable demonstrates their powerful transformative effects (Hood 2007). For example, the ‘political sociology of instruments’ (also called the ‘institutions-as-instruments’ approach) has demonstrated that instruments have significant downstream effects on a policy area. An instrument creates a specific representation of an issue by problematizing (and privileging) particular aspects. Once enacted and established, instruments can also induce inertia and path dependence (Lascoumes and Le Gales 2007, 10–11).   40 Much of the research on instrumentation as a dependent variable has examined instrument choice, with instruments either as functional responses to political or economic requirements (Dahl and Lindblom 1953; Foucault 1977; Hood 1983; Scott 1998), as simply the implementation aspect within broader policy schema (Lowi 1972; Wilson 1980), or as the specific choices of actors engaged in cognitive or political processes (Linder and Peters 1989, 1992, 1998). How actors learn about or choose between instruments is an important question, but not one that I theorize here. Rather, I ask why instruments have been selected as a mode of change from amongst (or within) other forms of policy reform, and under what circumstances changes to instrumental logic are successful.    45  consequences, based on an implicit theory about cause and effect, and demanding certain forms of intervention when cause-and-effect relationships have been impeded. For education, this refers to measurement and assessment of the functionality of an education system, based on cognitive models (or mere assumptions) about the causal relationship between inputs (resources, individual intelligence, contact time, and so on), and outputs (credentials, grades, competences, rankings, etc.). As a device that organizes social relations, instruments also have a logic of appropriateness; they refer to a mode of governance which indicates the role of the state and influences the behaviour of stakeholders in a given policy area. For education, this means the quality assurance tools used to steer an education system towards its goals. Specific instruments cohere around a core logic regarding these relationships between cause and effect, and between the governors and the governed. Instrument ‘mixes’ are certainly possible (and often the norm), and this can entail the co-existence of specific mechanisms and calibrations at the operational level (Howlett and Cashore 2009). However, each ideal-typical instrument is undergirded by a dominant logic, and re-instrumentation occurs when this instrumental logic changes. Lascoumes and Le Gales (2007) discuss five instrumental logics: ‘regulatory’; ‘economic’; ‘agreement-based’; ‘communication-based’; and ‘standards-based’. I explain and operationalize the different types of instrumental logic below, but to give a preliminary indication of where change is expected to occur: it is a sustained transition from the traditional ‘command-and-control’ logic of regulatory instruments, to one of the newer instrumental logics of public policy – even though the specific choice of new instrument can vary.  2.3.1.1 Technical and Social Dimensions of Traditional Policy Instruments Regulatory and economic instruments are the traditional policy levers of the bureaucratic state. For public education systems in Western Europe, a regulatory logic prevailed for well over a century.41 This                                                           41 Some economic tools have been incorporated into the instrument-mix, but an economic logic of financial ‘sticks and carrots’ is generally ill-suited to the public investment paradigm of post-war French and German education (i.e., tax incentives are meaningless, ‘performance pay’ is anathema to le service public, and so on). Economic   46  logic assumes that the state can establish what is in the national or public interest, detect circumstances when that interest is not served, and directly intervene to compel actors to change behaviour via legal sanction. Regulatory instruments mandate the parameters for social behaviour, and derive legitimacy for such imposition by being representative of the general interest – sometimes accompanied by legislation from elected representatives. Operationally, regulatory instruments compel behaviour in the education system by focussing on inputs – the only aspect that the state can fully control. The government can measure and monitor outputs (and even ‘mandate’ certain objectives be met), but a regulatory logic is input-oriented because it asserts that positive outcomes are generated vis-à-vis the direct control of educational inputs, such as curriculum, resources, student-teacher ratios, and so on. A regulatory instrumental logic has been operating in Western European education since the secular state consolidated control over public education in the nineteenth century, wresting control away from church officials or local actors.   2.3.1.2 Technical and Social Dimensions of New Policy Instruments  This traditional regulatory logic stands in contrast to the new public policy instruments being introduced to many areas of public policy. Newer public policy instruments derive their force and legitimacy from less hierarchical relationships between the state and civil society, and are ‘output-oriented’ inasmuch as evaluators monitor outputs closely in order to govern by them. Rather than output ‘control’, one could think of these instrumental logics as being about output orientation, steering or accountability. Governments still adjust inputs, but they do so without trying to tightly control – or sometimes even understand – the processes that lead from the inputs they can dictate to the outputs they desire. There are three types of new instrumental logics: communication-based, agreement-based, and standards-based (Lascoumes and Le Gales 2007). In some policy sectors these instruments overlap, yet they                                                           instruments and market mechanisms seem to be more prevalent in liberal market economies like the USA, vis-à-vis school vouchers, privatization, and so on (Hannaway and Woodroffe 2003).    47  function according to distinct logics. I deal with agreement-based and standards-based instrumental logics below, as they are most relevant for the cases under analysis.42  An agreement-based instrumental logic functions according to a quasi-market rationale, perhaps most often associated with ‘New Public Management’.43 Rather than having the state prescribe objectives and intervene in order to achieve them, this logic assumes that societal actors can establish their own objectives and solve their own dilemmas, and can thus commit to contractual agreements established with state authorities. Agreement-based instruments are legitimized by directly involving civil society actors in their own governance. The logic here is that the interventionist state gives way to a more mobilizing state, using contractual agreements that allow actors to formulate their own (narrow) objectives and/or incentives. Operationally, this means policies which establish contractual relationships between the state and civil society, or between levels of administration. These contracts are ‘output-oriented’ in that schools (or teachers, or delinquent students, etc.) must establish their own objectives (and sometimes also the incentives for reaching them) within the parameters of general educational aims established by the state.  A standards-based instrumental logic functions according to a technocratic rationale, which assumes that there are evidence-based or scientific means for determining problems and their solutions                                                           42 A communication-based logic is a new public policy instrument not elaborated here. Communication-based instruments foster information-sharing between stakeholders and evaluators, with a logic that precludes the need for direct state interventions using a dynamic of ‘naming-and-shaming’, in that the provision of information about progress or problems can shame actors into better problem-solving or goal-seeking behaviour. There were communication-based tools and techniques introduced to French and German education, but they were not the dominant logic for quality assurance of compulsory education.   43 There is some debate over the degree to which all new public policy instruments are manifestations of ‘New Public Management’ (NPM) or ‘the new managerialism’. Broad definitions of NPM have included “the insertion of business management techniques into the management of public sector institutions” (Stone 2001, 26), as well as simply referring to it as ‘the principal-agent movement’ (OECD 2001, 54). Among educationists, the new managerialism includes any efficiency or output-related reforms emerging from a culture of corporate consultancy; things like ‘customer care’, over-emphasis on quality and excellence, ‘high trust’ forms of employee control, and new forms of surveillance like standards or comparisons (Apple 2001, 410; Ball 1998, 123; Lundgren 2011, 21; Morgan 2011, 49–50). Instruments which operationalize the latter (standards; new forms of surveillance) are not predominantly agreement-based, but rather, adhere to other logics of instrumentation.     48  in a given policy area. Standards-based instruments have mixed legitimacy: partly based on scientific rationality (helping to de-politicize the issue) and partly based on deliberation, negotiation and cooperation. Their force derives from the moral suasion of ‘rational’ and ‘negotiated’ legitimacy, often combined with mechanisms of competition (such as benchmarking). The appearance of a standard is not itself a clear operationalization of a standards-based instrumental logic, because states have long used standards as ‘tools of legibility’ to create categories and thus conformity in society (Scott 1998, 25). As a new public policy instrument, educational standards involve more than just a category of achievement. Standard-setting is about accountability and benchmarking, enforced not by state-selected (and typically rather arbitrary) criteria but instead by actors from ‘outside the state’, such as technical experts or international benchmarks. Stakeholders also become involved in standard-setting, as there is often some implicit or explicit agreement among state and civil society actors regarding how to set standards which are fair and reasonable (Lascoumes and Le Gales 2007).  The different instrumental logics are discernible in concrete formal policies because each encompasses an implicit theory of cause and effect (a technical dimension) as well as norms regarding appropriate governance in that policy area (a social dimension). Although ‘contractual’ and ‘technocratic’ logics are based on abstract cognitive schemata about causality and social relations, they are relatively clear in policy documents because instruments must be oriented towards practical action by the state and/or societal actors.  2.3.3 Education as a Policy Area: The Importance of Evaluation Policy During this period, the key transformations in instrumental logics occurred in evaluation policy, one of the core structural features of an education system. Evaluation policies define, measure and enforce ‘quality assurance’, however quality as such is defined. Evaluation policies are means-oriented policies about governance, distinct from many of the other structural features of education systems which are   49  goal-oriented or ‘programmatic’.44 Because evaluation policy is oriented to means and not ends, it is a sector-specific example of an instrumental logic that can change independently of educational aims. If significant and sustained, a change in evaluation policy is both destabilizing and re-stabilizing; it changes the status quo, yet also bridges elements of the prevailing status quo to a new logic of action. Education evaluation is an area of public policy for which we can analyze the causes for instrumentation and reconsider the distinctions between policy transformation and the politics of ‘muddling through’. During the 2000s, changes to evaluation policy in Germany were significant compared to France. From late 2001, policies of educational evaluation changed from regulatory to standards-based in Germany, with a transition to an output-oriented quality assurance framework which empowered subsystem actors to set and maintain national standards. This involved a significant yet stable change in the logic of evaluation, which could also be integrated into the prevailing status quo of goals, educational structures, and financing. Attempts to transform French education after 2007 were less successful. New instruments for evaluation and governance (i.e., ‘contractualization’) were presented within an omnibus of reforms, and most of the reforms – including the new instrumental logic – were revoked by the President when discussions over policy design began to resemble a street brawl. After a new consultation process, a number of more incremental policy changes were initiated, mainly in areas                                                           44 Comparative educationists distinguish between the core structural features of education systems, classifying them into seven different dimensions: ‘provision’; ‘public investment’; ‘vocational training’; ‘access’; ‘stratification’; ‘centralization’; and ‘evaluation’ (Brauns, Müller, and Steinmann 1997; Busemeyer and Nikolai 2010; Hopper 1968; Witte 2006). The first five could be deemed ‘programmatic’ features of education because they orient a system towards particular aims (typically via the introduction of policy programs). These features often invoke the ideological preferences of actors in the field of education because they have significant distributive or redistributive consequences. The last two are ‘governance’ features – they do not define the broader aims of education, nor necessarily invoke ideological preferences. Governance dimensions oversee the ways in which goals and objectives are to be met (whatever those might be) and which levels of government are ultimately responsible for achieving them. Because educational jurisdictions were largely settled in Western Europe by the mid-twentieth century, there have been limited opportunities for change in political centralization.  Administrative centralization is a different matter, as this pertains more to the operation of schools than it does regulation or provision (see Hurrelmann et al. 2009).  Here, there have been adjustments over the last three decades, especially in France since 1982 (more on this in the chapter on France).  However, it has been the final dimension of educational structure, ‘evaluation’, which has been the site of much debate and policy change, as countries reconsider their systems of educational quality assurance.   50  other than evaluation. The greater complexity of this ‘obstructed reform’ scenario is not fully theorized here, yet juxtaposing it to the managed reform scenario provides insight into which variables are critical for explaining the power of the status quo. In other words, what was it about the status quo in each of these cases that led to a significant yet stable change where it was unexpected, and unsuccessful reform where it seemed most possible? Both cases describe attempts to change the logic of policy instruments during a punctuation in the policy equilibrium – only one of which was ultimately successful. 2.4 Subsystem Politics and a Policy Oligopoly There is a broad range of organizations and actors which can influence policy, and they operate within particular institutional settings and according to certain rules and norms. A policy subsystem is one such institutional setting with the capacity to shape education reform, even when final policy decisions are ultimately made by others. Policy subsystems and subsystem politics are general terms that refer to institutionalized group-government relations. A subsystem is a cluster of actors from among elected government, the permanent bureaucracy, and representatives of ‘policy users’, who specialize in specific issue areas, make routine decisions, and implement policies on a day-to-day basis. Without suggesting that form follows function, there are logical reasons why subsystems develop. From the perspective of ‘generalist’ decision-makers at the macro-political level, subsystems are institutions that can facilitate parallel processing of specialized information, and reduce uncertainty regarding the implementation of policy decisions (Baumgartner and Jones 1993, 44). For their part, non-governmental actors also have an interest in creating or sustaining subsystems: “Every interest, every group, every policy entrepreneur has a primary interest in establishing a monopoly – a monopoly on political understandings concerning the policy of interest, and an institutional arrangement that reinforces that understanding” (Baumgartner and Jones 1993, 6). The conceptualization of subsystem politics was   51  developed “to show that ‘real’ politics was not parliament-centred” (Jordan 2005, 317).45 Parliamentary and congressional politics do matter for public policy, but politicians can be quickly overwhelmed under conditions of high pressure and high uncertainty, making subsystems more relevant. Subsystems that are inclusive, coherent, and stable can manage the reform agenda in macro-political arenas, and they do this by influencing policy deliberation during punctuations in policy equilibria.  The capacity of a subsystem to induce a change in instrumentation depends on their institutional properties. Drawing on accounts of policy communities and subsystems, I define the three major properties of subsystems as 1) a bounded and defined structure, 2) a shared image (or interpretation) of the policy area, and 3) stability, as determined by the norms and procedures for communication and conflict management (Cairney 2013; Grant 2005; Jordan 1990, 2005; Jordan and Cairney 2013; Jordan, Halpin, and Maloney 2004; Jordan and Maloney 1997; Richardson 2000; Richardson and Jordan 1979). As dimensions of a subsystem that co-vary together, these properties are not independent variables in their own right. Rather, they are indicators of the existence and type of policy subsystem. Below, I elaborate on these properties for subsystems in general, before discussing which values indicate a policy oligopoly.                                                            45 A similar concept has been used in British politics, a ‘policy community’, as part of an effort to reorient policy analysis towards insulated and sectoral group-government relations, and away from formal political institutions and procedures like electoral systems or the parliamentary process (Richardson and Jordan 1979). Terms such as ‘policy community’, ‘policy network’, ‘issue network’, ‘iron triangle’, ‘policy whirlpool’, or ‘corporate pluralism’ have distinct meanings based on particular ontologies or specific political settings (a literature bifurcated between analyses of the USA and the UK, with more some analyses of the EU). The British ‘Anglo-governance’ school (pertaining also to Australia and Canada as empirical cases) has preferred the terms ‘policy community’ and ‘policy network’ in order to emphasize that these relationships are based on exchange, often using methods that analyze the flow of different kinds of resources across networks (Marsh 2011; Marsh and Rhodes 1992; Marsh, Richards, and Smith 2003; Rhodes 1990, 1996, Richardson 1982, 2000; Richardson and Jordan 1979). In the USA, the terms ‘policy subsystem’, ‘sub-government’, or ‘policy monopoly’ are more frequently used, typically emphasizing relationships based on institutional position or coalitional power (Baumgartner and Jones 1991; Heclo 1978; Jordan 1990; Jordan and Maloney 1997; Ripley and Franklin 1984; Sabatier 1988, 1993). Notwithstanding the proliferation of concepts and terms related to sectoral relationships between government and pressure participants, I use the terminology found in the literature in the US: subsystems and ‘policy monopolies’.   52  2.4.1 Properties of Policy Subsystems The first property, structure, refers to the finite and defined membership of a policy community. This membership structure restricts access to certain types of actors in order to try and build a shared institutional culture, and a consensus around policy design. There are three types of actors involved in a subsystem: elected government, state experts, and a bounded set of ‘pressure participants.’  Government actors are those elected members of the government and their political appointees who focus on educational issues: Ministers and/or their advisors (including state secretaries, deputy ministers, ministerial cabinets, and other special advisors). This category of actor is vital, yet often transient. State experts are typically (but not always) part of the permanent bureaucracy, spending longer periods of time evaluating issues in a policy area and often deriving authority from their technical training or experience. For some subsystems, the state has ‘outsourced’ this expertise to researchers operating in civil society, but these actors are still expected to report to the state in some official capacity (rather than unofficially, as éminences grises). Each ministry segments its bureaucracy differently, and subsystems reflect departmental divisions by including only those state experts and officials who are relevant (and sometimes those who are particularly influential). Pressure participants are ‘users’ of policy.46 For some sectors these actors can be from the private sphere, but in a public sector like education, pressure participants have traditionally been corporate actors such as                                                           46 ‘Pressure participants’ is a term adopted by policy community scholars to get away from the term ‘interest group’, which can be misleading because it presents an image of organizations that lobby the government solely on behalf of narrow self-interest. Pressure participant denotes a wider range of organizations directly affected by policy, whose interests in the policy area can be simultaneously narrow, such as material benefit, and broad, such as concern about how to improve general performance (Cairney 2013; Jordan, Halpin, and Maloney 2004). Disentangling the motivations of these actors is a challenge for the government (and for researchers) because they often present their narrow interests as part of the common good. In the context of a national policy community, pressure participants are identifiable by what they do (i.e., they use an ‘inside strategy’ to influence decision-makers) and not who they are (i.e., their interests or motivations). In pluralist regimes or sectors w