Policy learning to reduce inequalities: the search for a coherent Scottish gender mainstreaming policy in a multilevel UK

ABSTRACT

Policy ‘incoherence’ describes a lack of joined-up government that contributes to a confusing mix of policy instruments. It is an inevitable feature of multilevel policy-making, in which many actors compete to set the agenda, and ‘inequalities policies’, such as gender mainstreaming, which contain multiple and often-contradictory aims. This insight may prompt policy-makers to learn how other governments have responded pragmatically, rather than seeking to design abstract mainstreaming policies with unrealistic levels of coherence. Yet, policy learning is a political process characterized by contestation. Many policy-makers compete to define the policy problem, set the parameters for learning and determine which governments should take the lead. Therefore, we ask: How can we use policy theories to facilitate research-informed policy learning under these circumstances? We describe the framework that we developed for the Horizon 2020 project IMAJINE to encourage policy learning in multilevel policy-making systems. We then illustrate its value in a case study of our work for the Scottish Government’s National Advisory Council on Women and Girls (NACWG), which asked us to identify lessons from gender mainstreaming policies in other nations. This framework and case study help explain the limited impact of research on policy learning.

KEYWORDS:

• policy learning
• policy transfer
• Scottish government
• gender mainstreaming
• inequality
• territorial

INTRODUCTION: POLICY LEARNING TO REDUCE TERRITORIAL INEQUALITIES

Our role in the EU Horizon 2020 project Integrative Mechanisms for Addressing Spatial Justice and Territorial Inequalities in Europe (IMAJINE) is to research and facilitate ‘policy learning’ to reduce inequalities. We define learning as the use of new information to update policy-relevant knowledge (Dunlop & Radaelli, 2013), and examine how national and territorial governments across the European Union (EU) try to reduce inequalities, then encourage them to reflect on each other’s experiences (Cairney et al., 2018).

This approach combines academic research and direct engagement with policy practitioners, which presents conceptual challenges and payoffs. The challenge is to produce concepts, research questions and methods that are attractive to academics and practitioners (Flinders et al., 2016; Oliver et al., 2019). Practitioners often gather evidence to define and solve a specific problem quickly and simply, focusing on ‘what works’ or positive experiences from a small number of relevant countries. Policy scholars may seek comprehensive research, and warn against simple solutions, but without offering a sufficiently quick and feasible alternative to their audience. A pragmatic response is for scholars to (1) see policy learning through the eyes of participants, to understand how they define and seek to solve this problem, (2) incorporate insights from policy research to construct a feasible approach (the IMAJINE framework), then (3) reflect on this experience to inform research. The potential payoff to practitioners is theory-informed policy analysis (Weible & Cairney, 2018). The potential payoff to academics is experience from direct engagement, to learn how to use and adapt concepts to inform theories and practice (Cairney & Weible, 2017).

Unfortunately, the policy theories underpinning our work suggest that this potential will remain largely unfulfilled. Studies of ‘evidence based policy-making’ (EBPM) and policy learning suggest that researchers represent one small – and not particularly influential – part of a highly contested process of political choice, which takes place in a complex system over which policy-makers have limited knowledge and control (Cairney, 2016; Cairney & St Denny, 2020). Therefore, it is not realistic to propose a technical, comprehensive, expert-driven model for policy learning. Nor should participants seek to transfer policy lessons or solutions from one government to another without considering their policy-making contexts (Dolowitz & Marsh, 2000; Dunlop, 2017; Dunlop & Radaelli, 2013, 2018). Rather, learning is a contested process in which actors combine evidence and beliefs to define policy problems, identify technically and politically feasible solutions, and negotiate who should be responsible for their adoption and delivery in multilevel policy-making systems.

‘Inequalities policies’ exacerbate these complications. Inequality is an ambiguous term that is subject to contestation, to define inequality, determine which categories and measures to use (and therefore which social groups to favour), and promote more or less state intervention. Further, policy problems associated with inequalities span many government departments, and responsibility for relevant policy instruments spans many levels and types of government. Supranational, national and subnational governments cooperate or compete to take the lead on defining and addressing inequalities.

The inevitable result of this contestation, policy ambiguity and policy-making complexity is ‘policy incoherence’, defined here as a lack of joined-up government that contributes to a confusing mix of policy instruments (Cairney & St Denny, 2020; May et al., 2006). For policy-makers, it contributes to a major gap between expectations and policy outcomes. For academics seeking to inform policy with research, it contributes to low certainty about which actors and ideas are influential, and low expectations for academic impact. In that context, our most general research question is: How can we use policy theories to facilitate research-informed policy learning under these circumstances?

To that end, we outline a framework that we developed for the IMAJINE project to facilitate theory-informed policy learning, and illustrate this approach by describing our work for the Scottish Government’s National Advisory Council on Women and Girls (NACWG). The NACWG seeks to reduce inequalities in relation to gender, and reports annually to the Scottish Government’s First Minister, Nicola Sturgeon. It asked us to identify lessons from gender mainstreaming (GM) policies in other nations, focusing on minimizing policy incoherence (Rae et al., 2019). Our remit was to produce a quick understanding of mainstreaming policies in a small number of the most relevant countries to help inform its report.

The IMAJINE framework guides such academic–practitioner analysis in four steps (which represent the structure of this paper). First, define the nature of policy learning in political systems. Take into account the interaction between: politics, in which actors contest the nature of problems and the feasibility of solutions; bounded rationality, which requires them to use organizational and cognitive shortcuts to gather and use evidence (Simon, 1976); and policy-making complexity in multilevel policy-making systems, which limits a single central government’s control over choices and outcomes (Cairney et al., 2018). These dynamics play out in different ways in each territory, which means that the importers and exporters of lessons are operating in different contexts and addressing inequalities in different ways. Therefore, we must ask how the importers and exporters of lessons: define the problem, decide what policies are feasible, establish which government should be responsible for policy and identify criteria to evaluate policy success.

Second, map policy-making responsibilities for the selection of policy instruments. The Council of Europe defines GM as ‘the (re)organisation, improvement, development and evaluation of policy processes, so that a gender equality perspective is incorporated in all policies at all levels and at all stages’ (Council of Europe, 1998a, emphasis added; Booth & Bennett, 2002). Such definitions help explain why mainstreaming approaches appear to be incoherent. To map the sheer weight of possible measures, and the spread of responsibility across many levels of government (such as local, Scottish, UK and EU), is to identify a potentially overwhelming scale of policy-making ambition. Further, governments tend to address this potential by breaking policy-making into manageable sectors. Each sector has its own rules and logics, producing coherent policy-making in each ‘silo’ but a sense of incoherence overall, particularly if the overarching aim is a low priority in government (Jordan & Halpin, 2006).

Third, learn from experience. Show how relevant countries have tried to reduce inequalities and produce lessons from their experience. We describe three exemplar approaches to mainstreaming, selected by members of our audience, according to perceived levels of ambition: maximal (Sweden), medial (Canada) and minimal (the UK, which controls aspects of Scottish policy).

Fourth, deliberate and reflect. Work directly with policy participants to reflect on the implications for policy in their context. In our concluding section, we suggest that an inadequate fourth-step process helps to minimize the impact of research on policy learning.

STEP 1: DEFINE THE NATURE OF POLICY LEARNING IN POLITICAL SYSTEMS

Our review of policy-making research highlights three general insights that should inform expectations for policy learning (Cairney et al., 2018; cf. Rose, 1991, p. 4; 2005). First, learning is subject to uncertainty, ambiguity and contestation. Uncertainty describes incomplete knowledge, when governments can only gather limited information on a limited number of country experiences. Ambiguity describes the potential to understand and define policy problems in many different ways, and therefore the possibility that borrowers and lenders of policy understand the problem very differently (Cairney, 2020, 2019a; Zahariadis, 2003). Individual policy-makers respond to uncertainty and ambiguity by using two cognitive shortcuts: (1) identifying a small number of priorities and good enough sources of information, combined with (2) their use of beliefs, emotions, habits and experience to narrow the lens through which to see their world (Cairney & Kwiatkowski, 2017). Policy-making organizations have more resources, but they also develop standard operating procedures to limit their search for information (Koski & Workman, 2018).

As a result, policy learning cannot simply be ‘evidence based’ or informed only by academic research (Dunlop, 2016). Rather, policy-makers treat evidence-gathering as consultation, combining their own experience and beliefs with information that may include peer-reviewed scientific evidence, the ‘grey’ literature, public opinion data and stakeholder feedback (Nutley et al., 2013; Phoenix et al., 2019). Other policy actors exercise power to promote their beliefs and aims to policy-makers (Dunlop & Radaelli, 2013, 2018; Witting, 2017). Consequently, epistemic learning (from experts) plays a part, but combines or competes with other forms of learning, including reflective, to learn by combining diverse forms of knowledge and encouraging cooperation, bargaining, to learn how to win or negotiate political outcomes, and hierarchy, in which powerful actors at one level of government learn how much control they have over others (Dunlop & Radaelli, 2018, pp. 259–263).

Second, learning takes place in a complex policy-making system over which no actor has full knowledge or control. In other words, there is no single centre of government, and therefore no single hub for learning. There are many policy-makers and analysts spread across many levels and types of government, each producing different evidence searches and learning processes (Cairney, 2016, 2019b). To some extent, the lack of single central control relates to the choice to distribute and share power across political systems, such as when the UK model was based on membership of the EU (until 2020) and devolving power to governments in Scotland, Wales and Northern Ireland.

However, policy studies emphasize the necessity of ‘multi-centric’ policy-making driven by factors that limit the coordinative capacity of any state’s ‘centre’ (Cairney et al., 2019a). They describe these coordinative limits in many ways, with reference to: the ‘path dependent’ nature of policy-making (Pierson, 2000) and need for policy-makers to inherit the organizations, rules and choices of their predecessors as soon as they enter office (Hogwood & Peters, 1983; Rose, 1990); the tendency of elected policy-makers to pay attention to a small proportion of the policy problems for which they are responsible, and delegate most responsibility to actors such as civil servants and public bodies (Baumgartner & Jones, 1993; Jordan & Cairney, 2013); and for policy outcomes to appear to ‘emerge’ locally from ‘complex’ or ‘polycentric’ systems (Geyer & Cairney, 2015; Ostrom, 2007).

These studies describe policy-makers engaging with a policy-making environment over which they have limited knowledge and control, and cooperating with many other organizations or governments to coordinate policy-making. This environment can be summed up with reference to five main elements (Heikkila & Cairney, 2018):

• Many actors making and influencing choices at many levels of government.

• A proliferation of ‘institutions’ across different levels. Institutions are the rules and norms maintained by policy-making organizations. Some rules are formal, written and well understood. Others are informal, unwritten and difficult to grasp (Ostrom, 2007).

• The pervasiveness of policy communities, or the enduring relationships between policy-makers and influencers, across different levels.

• A tendency for different ‘ideas’ or beliefs to dominate discussion in different venues.

• The policy conditions and events that reinforce stability or attract policy-maker attention. Social or economic ‘crises’ or ‘focusing events’ prompt lurches of attention, while elections provide routine sources of shifting attention.

As a result, policy learning may contribute to policy change in one venue, but it does not guarantee policy change across a political system. This is particularly true if the issue is of low salience and the proponents or policy learning and change are relatively uninfluential. For example, studies of feminist institutionalism show that GM may seem straightforward when related to the formal institutions in one government, such as a statutory duty combined with a strategic plan adopted across that government. However, it represents the first step in a highly uncertain process to address the informal, unwritten, ill-understood and everyday sources of inequality in the ‘hidden life of institutions’ (Chappell & Waylen, 2013, p. 608). Similarly, the term ‘velvet triangle’ describes the policy communities of ‘feminist bureaucrats, trusted academics, and organized voices in the women’s movement’ that develop partly because women are excluded routinely from the positions of power (Woodward, 2004, p. 78). These networks help provide a focus on gender in policy-making, to prompt policy change. However, their origin – a response to exclusion from positions of power – suggests that few people involved in policy formulation can ensure delivery.

Third, policy learning is often ad hoc and unpredictable, when driven by political imperatives and unequal power relationships. There are many policy actors spread across many governments, sharing ideas and importing elements of each other’s experiences in different ways. The result can be ‘uninformed’, ‘incomplete’ or ‘accidental’ policy transfer (Berry & Berry, 2007; Dolowitz et al., 2012; Dolowitz & Marsh, 2000), when:

• Some governments feeling obliged to mimic others, often without knowing what they did, why, if they were successful and if they can transfer success (Berry & Berry, 2007; Heikkila & Gerlak, 2013).

• Some governments import a broad idea, then transform it to fit local contexts (Stone, 2017).

• There are many ad hoc experiences across multiple levels of government, such as when supranational or national governments encourage harmonization, subnational and national governments draw different lessons from international experience, or ‘policy entrepreneurs’ sell new solutions in multiple venues (Benson & Jordan, 2011, p. 369; Dolowitz et al., 2012; Keating & Cairney, 2012; McCann & Ward, 2012, p. 327).

As a result, policy studies focus less on describing transfer as the export of a high-quality and well-packaged product, and more on the ways in which importing policy-makers make sense of solutions in local contexts (Stone, 2017, pp. 64–65). Our understanding shifts from ‘transferred’ to ‘translated’, ‘transformed’ or ‘mutated’ (Freeman, 2009; McCann & Ward, 2012, pp. 326–327; Park et al., 2014; Stone, 2012).

Combined, these factors warn us against treating policy learning and transfer as a technical exercise driven by research. Researchers could play an important role, but their impact depends on their ability to adapt to policy-making systems by identifying: how policy-makers deal with uncertainty and ambiguity, where the ‘action’ is within multilevel systems, and the extent to which policy-makers seek to learn about a well-packaged policy or a broad idea. This need for researchers to adapt to political reality informs our general approach, in which we seek to generate a story of policy success from the perspective of the exporting government, relate it to a story of a policy problem in an importing government, and identify the compatibility of their policy-making systems. When applying this framework to case studies, we use three broad questions to tell a story of policy learning to reduce inequalities.

What story do importers of policy tell about the problem they seek to solve?

Different governments have different ways to (1) understand issues as policy problems, (2) identify technically and politically feasible solutions, and (3) determine what policy success looks like. We need to understand this agenda setting process that leads policy-makers to define inequalities and narrow their attention to solutions, and consider the extent to which any new solution would relate to their existing policies (Cairney & St Denny, 2020).

For example, ambitious but ambiguous policy initiatives such as GM require high and sustained policy-maker attention and commitment to seem coherent and feasible. Yet, the broad topic of inequalities tends to be low salience and, when attention is temporarily high, it is focused on an ill-defined problem that can be interpreted in multiple ways (Cairney & St Denny, 2020, p. 9). An inequalities lens can focus on gender but also income and wealth, spatial justice, race and ethnicity, sexuality, disability, mental health or the cumulative impact of multiple factors (p. 9). This attention is influenced by debates on the causes of inequalities, from ‘structural’ causes to be addressed by the state, to personal responsibility to be addressed by individuals or families (p. 179). Further, attention to gender in Scotland often seems subordinate to a model of economic growth that is not conducive to many feminist social policy recommendations (Cairney & Rummery, 2018, p. 549). Overall, an importing government may have a range of ambitions, from seeking ideas on how to transform policy radically, to seeking ways to fit policy change within an existing approach.

What is the evidence that a government was successful in reducing inequalities?

Policy-making complexity suggests that central governments influence but do not determine policy outcomes. Rather, the actions of some governments receive disproportionate international attention when policy is perceived to be successful, and many governments encourage this image. In that context, research on policy learning involves engaging with debates on what counts as good evidence and whose story of success we should accept. One popular response is for importing governments to rely on a combination of evaluation sources, including independent evaluations, international benchmarks and independent scientific reviews subject to peer review. These evaluations may use a range of methods, from the routine use of evaluations of the policy solution, to counterfactual comparisons to ask if another solution would have been more successful. However, the evaluation of policy outcomes in complex systems is rife with uncertainty (it is rare to find conclusive evidence) and ambiguity (it is rare to find unanimity in the evaluation of success). Therefore, stories of success may rely on pragmatic categories of assessment, including Good practice based on initially positive experiences, or Promising approaches but unsystematic findings, rather than the more unequivocal categories such as Research based or ‘sound theory informed by a growing body of empirical research’, or Evidence based, when ‘the programme or practice has been rigorously evaluated and has consistently been shown to work’ (Perkins, cited in Nutley et al., 2013, p. 9). Or they may relate to a vaguer story that new policies represent a natural progression from existing good policies or a radical change from the past.

Do exporting and importing governments have comparable political and policy-making systems?

Comparability relates partly to political systems, including their formal rules, divisions of power, and the role of political parties, multiple levels of government and the courts. For example, Ireland is a small, centralized, unitary state, while Germany’s larger federal system places the Lander at the forefront of domestic policy-making (Collins, 2004, p. 610; Jeffrey, 2008). It also relates to the dynamics of policy-making systems, and we can use well-established concepts to think systematically about their comparability:

• Actors: identify which levels or types of government are responsible for this policy.

• Institutions: identify the formal and informal rules that are crucial to policy adoption and success.

• Networks: identify the role that policy communities play in making and delivering this policy.

• Ideas: gauge the extent to which policy-makers in the exporting and importing governments express the same beliefs about policy problems and feasible solutions.

• Socioeconomic context and events: identify the extent to which they share similar reference points, such as the extent to which they face the same substantive policy problem, or have similar stories to tell about the spark for policy reform.

STEP 2: MAP POLICY-MAKING RESPONSIBILITIES

The combination of multilevel policy-making and mainstreaming policies rules out the possibility of one government simply learning from another. Rather, the importing government is one of many actors with the power to influence policy, and it may need to learn from the actions of multiple levels of government in another political system. For example, in their study of energy policy, Cairney et al. (2019b) show that (1) we can provide a table containing the breakdown of legal or formal responsibilities, but (2) there remain overlaps in responsibility, and some powers are shared, either by design, and reinforced by intergovernmental relations, or in practice, and not subject to effective multilevel cooperation. This limit is an ever-present feature of multilevel governance, where formal and informal power is vested in many levels and types of government (Bache, 2013; Cairney et al., 2019a). It is also accentuated in policy agendas that cut across established policy sectors, and exacerbated when the responsibility for policy instruments relates more to political negotiations than a technical process to ensure holistic government (Cairney & St Denny, 2020).

In that context, policy choices often seem incoherent in relation to GM for five main reasons. First, if mainstreaming relates to ‘all policies at all levels and at all stages’ (Council of Europe, 1998a), it is difficult to identify a manageable number of achievable areas that ‘share a set of ideas or objectives’ (May et al., 2006, p. 382). Indeed, identifying a full list of relevant policy responsibilities is challenging. We demonstrate this challenge by producing a specific map of responsibility for GM policies across the UK. Table 1 describes the range of competencies relevant to GM in a multilevel UK.

Table 1. Gender mainstreaming (GM) policy competencies in the UK and Scotland.

Level	Direct competencies	Indirect competencies
European Union (EU)	European Employment Strategy European Women’s Lobby Analysis and dissemination of gender research Crime victims’ rights European law on trafficking Sectoral gender equality monitoring EU funding for gender equality initiatives Promotion of female entrepreneurship Gender budgeting	Fiscal policy Structural Funds Social security directives Citizenship policy Fundamental rights Research funding
EU agencies	Fundamental rights monitoring and promotion	Development of work-related policies Health and safety monitoring Asylum support
UK/Great Britain level	Implementation of equality strategy Sexual discrimination and equal pay laws Overall equalities strategy Public sector equality duty Statutory guidance for schools on equal opportunities Armed forces gender policy and recruitment Employment regulation (including equal pay) Voluntary equal pay agreements with employers Childcare and early-years provision Maternity and paternity leave allowances Women specific health issues and screening Reproductive health provision Tax credits Pensions Child poverty reduction Carers allowance Forced marriage policy, outreach and casework Domestic violence courts Training for justice system employees Rape support centres Funding for women’s organizations Anti-female genital mutilation(FGM) strategy and awareness	Criminal law Domestic violence prevention Asylum, refugee and immigration policy Careers guidance Personal, social and health education (national curriculum) Welfare and social security policy, provision Women’s empowerment programmes in international development Prison and detention service management Pensions Transport provision Funding for equalities training and awareness Trade union relations Labour market regulation Overall fiscal and budgetary policy and strategy
UK/Great Britain national agencies	Monitoring and recommendations on equal opportunities Support for women seeking political office	Media regulation Serious organized crime policing (human trafficking) Immigration, refugee, and immigration processing and management
Scottish Parliament and government	Implementation of equality strategy Equality research Implementation of the Equality Act Childcare and early-years provision Early -years framework Reproductive health provision Abortion provision Prisons and woman offenders Forced marriage protection orders Anti-violence strategy for women and girls Abortion provision Prisons and women offenders	Criminal law Trafficking victim support services and awareness Trade union relations Welfare eligibility criteria
Scottish government agencies	Prisons and sentencing Social security payments Promotion of awareness, understanding and respect for human rights Councillor codes of conduct	Public sector pensions management Charity regulation Courts and tribunals Housing regulation Safeguarding children’s rights
Local authorities	Implementation of public sector equality duty Anti-violence against women and girls strategies Women’s political representation strategy Social care provision	Recruitment, pay and conditions of local government employees (including gender wage gap) Women’s peer-support networks Targeted apprenticeships for women Demographic monitoring and data collection Funding for community organizations Jobs guarantees and employment schemes Equalities awareness projects Relations with local employers Housing allocation Trading standards Environmental health Licensing

Sources: Cain (2015), COSLA (2017), Glasgow City Council (2017), European Institute for Gender Equality (2020), European Union (2020), Hubert and Stratigaki (2016), Johnson Ross (2018), Scottish Government (2017), Shreeves (2019), UK Government (2011).

This list of instruments and competencies is only manageable when we focus on a small number of commonly identified mainstreaming initiatives. For example, the Scottish Government has a statutory obligation to deliver the UK’s Public Sector Equality Duty established in section 149 of the Equality Act 2010 (Equality and Human Rights Commission, 2019), and non-statutory commitments, such as to encourage ‘gender-based budgeting’ to identify the impact of policy choices on women and men (Scottish Women’s Budget Group, 2016). Even so, the overlap of responsibilities can exacerbate a sense of incoherence, such as when the UK sets overall targets and guidelines, but Scottish and local governments become responsible for delivery.

Second, governments tend to deal with the unmanageability of policy-making by breaking it into sectors (such as education) and subsectors (such as pre-school, primary, secondary, further or higher education). Further, all such policy-making ‘silos’ have their own rules and aims which are entirely logical in a narrow context, but they combine to produce the sense of uncoordinated policy-making (Jordan & Halpin, 2006). There is a substantial literature devoted to the ways in which governments try to ‘join up’ government in that context. It generally suggests that joining-up initiatives are expensive, often fail and failure can be demoralizing (Carey & Crammond, 2015, pp. 1022–1028; Molenveld et al., 2020, p. 9).

Third, this sectoral dynamic takes place within a multilevel policy-making system in which, for example, the UK devolves only some aspects of policy (primarily social policies including health, education and social work, as well as criminal justice and aspects of transport) and allocates the Scottish government budget (albeit with some powers on tax and social security now devolved). The UK government can, simultaneously, place an equality duty on the Scottish government and produce policies that undermine it, such as recent welfare reforms that have affected women disproportionately (Cain, 2015).

Fourth, a low tendency to define inequalities primarily with reference to gender helps explain the lack of policy coherence: low-priority issues seem relatively coherent in isolation (such as in a high-level strategy document), only to be undermined by a tendency to devote disproportionate attention and resources to other issues (Cairney & St Denny, 2020). Although policy-makers take gender seriously, most of their attention is focused on more general and higher profile economic and constitutional issues in which gender often plays a subordinate role (Cairney & Rummery, 2018). This limited attention undermines clarity about how to translate general mainstreaming aims into specific sectors such as education, and this uncertainty increases as we increase the number of relevant sectors (such as health, justice, transport, social work and social care).

Finally, a key aspect of GM strategies is to empower women through continuous policy co-production, but the confusing spread of responsibilities is not conducive to this approach (Engeli & Mazur, 2018). For example, the NACWG could help co-produce recommendations for the Scottish Government (and Scottish local government), but the process would be separate from UK- and EU-level initiatives.

These issues are reflected somewhat in the NACWG’s final recommendations to the Scottish Government (NACWG, 2020, pp. 21–23), summarized as follows:

• Improve leadership and coordination by (1) increasing the status and power of the Scottish Government’s Equality Unit, and designated leaders in each directorate, (2) establishing a senior management group, and (3) setting clear standards (backed by performance management) to gauge progress.

• Report annually to the Scottish Parliament to keep GM progress high on the agenda.

• Build on the so-called ‘Scottish Approach’ to policy-making to foster the meaningful co-production of GM policies across Scottish and local government, and in partnership with public and non-governmental bodies (cf. Elliott, 2020).

However, they only relate to the actions of one of many levels of government and tend to underestimate the coordination problems inherent in mainstreaming initiatives. In that context, the theory-informed mapping exercise helps explain why policy will remain ‘incoherent’ even when individual governments take positive strategic action.

STEP 3: LEARN FROM EXPERIENCE (UK, CANADA AND SWEDEN)

Policy learning allows participants to manage such complexity by learning from the experiences of others, such as the governments most relevant to our audience. The selection of relevant experiences differs in research and practice. In research, we may seek a systematic approach to case selection (described in Stake, 1995; Yin, 2014) and a comprehensive comparison of countries. This aim is challenging because most research on GM is based on the study of few cases (Daly, 2005; Hankivsky, 2013; Liebert, 2002; Rees, 2005; Woodward, 2003). There are large cohort case study comparisons of specific gender equality policies (Rubery, 2002; Rubery et al., 2005), but no recent systematic large-n comparative analysis of GM implementation (although see True & Mintrom, 2001). There are cross-national policy reports purporting to evaluate GM progress (Council of Europe, 1998b; Theisen et al., 2005), but many engage superficially with notions of gender and equality (Woodward, 2008, pp. 292–293).

In practice, the learning and transfer literature suggests that governments seek lessons unsystematically and only from a small number of countries (Dolowitz & Marsh, 2000; Dunlop & Radaelli, 2013). This tendency is reflected in the discussions we had with policy actors in Scotland, who had a notional list of countries from which to learn. A review of GM to identify and classify all existing approaches does not reflect the pragmatism of our audience, more likely to be guided by professional knowledge and interest in specific models. We therefore focused on three countries whose experiences (1) resonate with our policy audience’s interests and (2) can be analysed as part of a larger project (IMAJINE) to compare GM experiences on a greater scale.

In that context, we describe the three selected governments as exemplars of policy change in order of magnitude. The GM literature describes a notional spectrum, from: minimalist ‘tinkering’, with a focus on legal remedies to improve equal treatment, to medial ‘tailoring’, with a focus on positive action interventions to address group disadvantages or to ‘integrate’ pro-gender equality objectives into the existing system, to maximal ‘transforming’, with a goal of mainstreaming equality by radically changing existing social structures and policy processes (Jahan, 1995; Rees, 2005; Walby, 2005). We place our three cases on the notional spectrum as follows: UK – minimal; Canada – medial; and Sweden – maximal. In each case, we construct a historical narrative highlighting key policy moments and outcomes to encourage reflection on progress and next steps.

Minimal mainstreaming

In the UK, GM emerged from an ‘equal opportunity’ framework, designed from the 1970s to address the structural barriers faced by individuals from certain groups, especially in education and employment (Daly, 2005). By the 1990s, equal opportunities became associated with fostering the inclusiveness of groups traditionally underrepresented in positions of influence, such as to increase the number of women elected, and in Cabinet, and create government agencies to promote women’s interests (Durose & Gains, 2005; Miller, 2009; Woodward, 2005, p. 6). The New Labour government (from 1997) fostered practices associated with ‘new public management’: treating greater diversity as part of a drive to ‘modernize’ government, and using ‘network’ governance to widen opportunities for input from women and minorities than in the institutional hierarchies associated with centralized government (Newman, 2002). The problematic idea was that (1) a greater presence of women would (2) advance women’s interests in policy, but both objectives remained unfulfilled (Childs & Whitey, 2004; Lovenduski, 2005; Phillips, 1995; Pitkin, 1972). Further, the agencies designed to foster scrutiny into policies and departments – including the Women’s Unit (1997–2001) and Women’s Equality Unit (2001–02) were under-resourced and marginalized, while departmental appraisals of ‘policy for equal treatment’ were patchy (Beveridge et al., 2000, p. 392; Durose & Gains, 2005, p. 105; Miller, 2009; Hankivsky & Christoffersen, 2011; Squires & Wickham-Jones, 2004; Veitch, 2005).

The mid-2000s marked a shift from a passive framework, where individuals were required to prove discrimination, towards an active promotion of gender equality (Hankivsky & Christoffersen, 2011, p. 36). The Equality Act 2006 introduced formal obligations for public bodies. The Act’s Gender Equality Duty required public bodies to promote gender equality and stamp out discrimination within their own structures and services. The Equality Act 2010 consolidated previous legislation in Wales, England and Scotland, and created anti-discrimination Public Sector Equality duties spanning ‘age, disability, sex, gender reassignment, pregnancy and maternity, race, religion or belief and sexual orientation’. Feminist policy actors experienced periodic success in reforming key women’s policy issues, including domestic violence and pension and pay equality. Nevertheless, GM efforts in the UK have remained fragmented, involving impact assessments and broad equality duties rather than a fully integrated and implemented strategy (Daly, 2005). On Rees’ (2005) spectrum, the UK oscillates between ‘tinkering’ and ‘tailoring’ policy.

The story is somewhat different in Scotland and Wales following political devolution (1999). Devolution campaigns were shaped by national women’s movements (Brown, 1998; Chaney et al., 2007). They helped institutionalize the principle of equality in the Scottish Parliament (SP) and National Assembly for Wales (NAW), such as via a mandatory equalities committee to scrutinize legislation and government departments, and the NAW’s ‘absolute duty’ to consider the impact on gender equality of any policy it passes (Equal Opportunities Commission, 2003; 2004). There is greater consultation with, and integration of, women’s groups in the policy-making process (Beveridge et al., 2000). Key actors have expressed the ambition to become world leaders in gender equality. In 2018, the Welsh Government launched a gender equality review to understand how to become a ‘feminist government’. To do so, it sought to learn from Nordic countries, chief of which Sweden, via research and practitioner exchanges (Parken, 2018; Taylor-Collins & Nesom, 2019). In Scotland, the NACWG recommends the adoption of radical policies, including free childcare, two months’ paid paternity leave and the transformation of the complaints process for victims of sexual violence (NACWG, 2018). Examples of devolved innovation include: the Violence Against Women, Domestic Abuse and Sexual Violence (Wales) Act 2015, as the first in the UK to address ‘violence against women’ rather than ‘domestic violence’; and the Scottish Government scheme to combat period-poverty by supplying free sanitary products in all school, colleges and universities, since reinforced by the tabling of a bill in the SP to introduce free universal access to sanitary products. Both governments have also set up dedicated offices to scrutinize progress towards equality. Nevertheless, examples of major and sustained policy divergence – from the UK’s often-minimalist position – are unusual.

Medial mainstreaming

Elements of GM have been implemented in Canada since the 1970s. A report by the Royal Commission on the Status of Women recommended substantial reforms in areas including employment, education, and criminal justice (albeit not in relation to violence against women), and the establishment of an agency to oversee policy development and implementation. The process acted as a focal point for feminist groups (Bégin, 1992). It led to the creation of the federal agency responsible for overseeing efforts to improve women’s equality and participation in all spheres of life (Status of Women Canada – SWC), establishment of a Minister Responsible for the Status of Women, and increased involvement of women’s groups in policy debates (Brodie, 2008, p. 153). Over the next 20 years, legislative reforms sought to improve women’s status, through equal opportunities and anti-discrimination legislation, but without addressing systemic and structural disadvantages and violence.

The Liberal government’s 1995 Federal Plan for Gender Equality introduced ‘gender-based analysis’ (GBA) across all federal departments and agencies (overseen by the SWC). GBA represents a ‘systematic, integrative, planning-based’ approach to each policy-making step, to identify the implications for gender equality (SWC, 1995). The idea was derived from GBA experiences in New Zealand and British Columbia, which had already adopted the instrument. By the late 1990s, statements of commitment to, and guidelines for implementing, GBA were woven into most departmental policies (Hankivsky, 2012, p. 173). However, it remained a low priority. Initial attempts to operationalize GBA mostly amounted to an increased use of statistics, and the SWC did not have the power to ensure implementation across the public sector effectively (Grace, 1997). Part of the problem was a government narrative presenting gender and women’s equality as achieved (Brodie, 2008). Further, a period of public sector retrenchment reduced SWC’s reach, and undermined the ability of feminist groups to put pressure on federal and provincial policy actors to deliver (Brodie, 2008; Knight & Rodgers, 2012).

GBA success varied markedly. Territorially, provinces such as British Columbia stand out for their historically sustained commitment, with other provinces applying lessons (Hankivsky, 2008, p. 70; Skinner et al., 1998, p. 4). Federal public sector cuts reduced their institutional infrastructure and undermined their ability to enforce GBA, but they demonstrated an enduring will to analyse policy processes through a gendered lens (Teghtsoonian, 2000). Implementation audits (in 2009 and 2015) highlighted variance in whether and how GBA is applied across federal departments and agencies (Auditor General of Canada, 2009, 2015), despite the introduction of a government-wide Departmental Action Plan on Gender-based Analysis (Privy Council Office, 2009). In many cases, GBA was ostensibly carried out, but with no evidence that the analysis influenced policy design. Successful implementation was linked to ‘equality champions’ in some departments, whose role was not replicated when they moved (Brodie, 2008, p. 157).

Since 2015, the Canadian government has begun shifting to efforts to mainstream diversity. It has adopted a more explicitly intersectional approach – ‘GBA+’ – to examine ‘the possible effects of legislation, policies, and programmes on diverse groups of women and men, girls and boys, by taking into consideration gender and other identity factors’ (Standing Committee on the Status of Women, 2016, p. 3). Oversight of implementation at the federal level rests with Women and Gender Equality Canada (WAGE – previously SWC), with each federal department responsible for adopting GBA+ when developing policy, legislation or services. Implementation remains patchy, leading the House of Commons Committee on the Status of Women (HOCSW) to call for the federal government to introduce legislation to make it mandatory (HOCSW, 2016). The government has introduced mandatory training for federal employees, and systematized its inclusion in all submissions to the Cabinet and Treasury Board. Moreover, the federal government committed to applying GBA+ to the budget process (enshrined in the 2018 Canadian Gender Budgeting Act).

Overall, the relatively systematic inclusion of GBA and GBA+ across departmental policy, the government’s commitment to revising federal equality mainstreaming efforts towards greater intersectionality, and the passing of legislation to give gender budgeting a statutory footing, place Canada further along Rees’ spectrum than the UK. Still, Canada’s model is ‘expert-bureaucratic’ rather than ‘participatory-democratic’, precluding it from reaching the ‘transformative’ ideals associated with maximalist conceptions of GM (Paterson, 2010). It has failed to engage the country’s broad and active women’s movement to help rethink policy and policy-making process using feminist perspectives (Rankin & Wilcox, 2004). The result is a layering of GBA onto existing policies and processes. Further, national public sector retrenchment limits the ability of key agencies to delivery, coordinate or monitor GBA efforts. Efforts to reduce the size of the welfare state also produced detrimental outcomes for women’s equality and welfare (CCPA, 2014, p. 10; Hankivsky, 2006).

Maximal mainstreaming

Swedish efforts demonstrate the depth and breadth required for maximal GM (Daly, 2005). GM tools have been made routine across many elements of policy formulation, service delivery and programme evaluation. A gendered lens has been applied across most policy areas, including health (Ågren, 2003, pp. 20–21), transport (Polk, 2008) and the environment (Sida, 2016). Elements of GM have been embedded beyond government and the public sector, to the voluntary and private sectors.

Attempts to improve gender equality in Sweden reflect a tradition of ‘social engineering’ through policy, where governments seek to organize society and structure behaviour according to a particular vision of ‘the good life’ (Woodford-Berger, 2007, p. 127). GM was woven through a well-established equality policy framework and engendered state machinery, including at the local level, from the 1970s (Rönnblom, 2005). Policy-maker commitment to GM became ‘part of the normal policy process’ (Woodward, 2005, p. 8). A Cabinet position for gender equality has existed since the 1970s. Official statistics have been gender-segregated since 1994. Public officials – including ministers, special advisors, commissioners and local authority actors – receive training to sensitize them to their role (Verloo, 2000, p. 11). Responsibility for implementing national goals is increasingly decentralized to county councils, which have included gender-equality officers since the mid-1990s (Andersson, 2015; Olivius & Rönnblom, 2019).

Swedish Prime Minister Stefan Löfven (Löfven, 2015) declared the ‘world’s first feminist government’, which seeks to achieve gender equality by removing all systemic and structural barriers to women’s empowerment. This approach includes specific policy instruments, gender-responsive national budgeting, and gender-based analysis in policy and policy-making. It is intended to span all policy areas, including historically male dominated areas such as foreign policy. The Swedish Gender Equality Agency was created in 2018 to coordinate, monitor and support actors implementing national gender-equality policy, although criticism quickly emerged concerning its lack of mandate and resources (Olivius & Rönnblom, 2019, p. 77).

Sweden’s commitment to GM extends internationally. Sweden shares with other Nordic countries a commitment to egalitarian social values and welfare systems. They cooperate to improve gender equality region-wide – such as via the Nordic Council – by sharing policy insights, developing new programmes and producing joint research (Åseskog, 2003). Swedish feminist policy actors have a long history as ‘norm entrepreneurs’, exporting policy ideas and feminist values. They have built successful transnational coalitions and lobbied successfully for policy change, on topics such as domestic violence, across intergovernmental fora, including the EU and United Nations (Montoya, 2013, p. 31).

Sweden has pioneered high-profile radical gender-related policies which have been emulated internationally. It introduced gender-neutral parental leave in 1974, allowing men and women to share time off following the birth of a child. The values underpinning this policy have inspired shared-parental leave policies across the Nordic region and beyond, including the UK (in a less ambitious form). Its 1998 law to criminalize the purchase of sexual services – designed to ban prostitution without posing a detriment to individuals selling sexual services (treated as ‘victims’ and offered support to exit prostitution) – has prompted similar legislation in Norway, Iceland, France, Canada, Israel and Ireland. The law was intended to be showcased abroad to spur a critical mass towards regional or international prohibition (Dodillet & Östergren, 2013, p. 110). It was part of a larger ‘Women’s Peace’ (Kvinnofrid) bill introducing measures to combat violence against women and sexual harassment. The bill was predicated on feminist ideas concerning the unacceptable commercialization and objectification of women. As such, the prostitution law helped communicate more general norms and values about women’s equality to improve women’s status and welfare across society (Florin, 2012).

Nevertheless, many GM efforts have stopped short of being transformative. Resistance and low commitment within policy networks at all levels of government (Lombardo & Mergaert, 2013, p. 302) has resulted in the technocratic application of GM instruments, such as the collection of gender segregated data. Responsibility for designing and implementing policy has increasingly been outsourced to private-sector consultants (Olivius & Rönnblom, 2019). The persistent belief among certain groups and actors that gender equality has been achieved has reduced the salience and impact of GM in routine policy-making. For instance, over time, authorities have ceased funding the health and social care organizations responsible for assisting individuals to exit prostitution (Florin, 2012), leaving the policy little more than symbolic (Wagenaar et al., 2017, p. 45). Symbolic reforms are created to give a sense that a problem is being addressed, but without resources being dedicated to implementation (Mazur, 1995, p. 2).

DISCUSSION AND CONCLUSIONS

Our framework helps identify three main conclusions that should guide policy learning for GM, to help reduce inequalities relating to gender. First, there are major limits to policy learning in complex systems. Learning cannot be comprehensive and does not lead to the importation of a well-defined package of measures. Second, a lack of joined-up government is an inevitable feature of mainstreaming initiatives. GM is an amorphous aim in which the combination of possible measures is overwhelmingly complex, in a multilevel policy-making environment where there is minimal clarity on how each level of government can take responsibility. Mapping exercises serve initially to highlight the potentially overwhelming amount and spread of relevant instruments and responsibilities.

Third, although there is a spectrum of ways in which governments have responded – to seek minimal, medial or maximal policy change – a major gap between strategy and delivery is a routine feature of country experiences. The UK’s experience demonstrates the potential for minimal policy change when there is too-high reliance on substantive representation and simple performance measures. Canada’s experience demonstrates the potential for complacency, combined with public services and welfare state retrenchment, to undermine more ambitious strategies. Sweden exemplifies the level of transformation required for genuine GM and provides positive lessons from decades-long innovations. However, it also highlights the negative impact of the belief that gender equality has been achieved if it is enshrined in high-profile strategies. Such beliefs may cause reduced funding and a return to the sense that GM is largely a symbolic exercise. The Swedish case becomes an aspirational and cautionary tale. It demonstrates that one of the most comprehensive and coherent uses of GM has shortcomings, producing the risk that many of its flagship policies lapse into ineffective symbolism when attention subsides.

These comparisons suggest that many of the difficulties experienced in achieving GM’s transformative goals are not country specific. Some stem from policy ambiguity and contestation (Hankivsky, 2013). Or policy is deliberately ambiguous to generate initial support, contributing to confusion during delivery. Contemporary mainstreaming strategies could produce the risk is that if ‘gender is everybody’s responsibility in general, then it’s nobody’s responsibility in particular’ (Pollack & Hafner-Burton, 2000, p. 452).

However, the problem with these conclusions is that they seem more useful to researchers, relatively distant from policy-making and more accepting of the conclusion that policy-making is complex and not subject to government control, than practitioners, who need to act despite their lack of control. Therefore, researcher impact on practitioners is limited without Step 4 Deliberate and reflect. Put simply, our framework does not guide action directly. Rather, it combines (1) a detailed map and checklist of factors that participants need to take into account when engaging in policy learning, with (2) stories of relevant governmental experiences. Further, these stories exist to prompt reflection, such as when we learn that no policy change has any guarantee of clear and unequivocal success, and that the case for policies to reduce inequalities may rest as much on the belief that it is the right thing to do as the expectation of a substantive payoff.

In that context, it would be misleading to describe our experience as successful, partly because our timeline was too short (one month), and our engagement with participants too limited. Although we received some initial direction on the NACWG’s work, we were unable to engage in sufficient dialogue with its members to understand how they understood the policy problem, gauged the feasibility of potential solutions, and defined policy success. Although Step 2 shows that the NACWG’s key recommendations are consistent with the implications of our mapping exercise, these recommendations are broad and did not emerge from deliberation on specific aspects of the map. Although our mainstreaming spectrum represents a useful guide to deliberation, we had to use informal contacts to establish which countries were most relevant, and we were not part of further deliberation. Overall, our level of engagement with participants was too low to produce the meaningful co-production of policy learning from a complicated piece of research that requires translation to aid deliberation.

Our summary of the whole IMAJINE work package (Cairney et al., 2020) contrasts this experience with our subsequent work to foster public health mainstreaming. In the latter, we were able to engage more directly and frequently with policy participants to understand how the importers and exporters of lessons understood the policy problem and policy-making context. These factors make the difference between delivering a report with limited impact and co-producing a report with clear benefits to our audience. Put simply, in the case we describe in this article, we learned how not to facilitate learning.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the H2020 European Research Council [grant number REV-INEQUAL-07-2016].