From evidence to action: Applying gender mainstreaming to pay gaps in the Welsh public sector

Progress on reducing gender disparities remains painfully slow, despite efforts to identify the determinants of gender pay gaps and specify size and shape. Recent studies highlight the need for a more nuanced account of the way that public policy shapes organizational responses and insights into the types of organizational practices that diminish pay disparities. In response, this research reports on an action research intervention in three large Welsh public organizations, subject to a unique statutory equality duty. Data demonstrate how an evidence ‐ based gender mainstreaming approach facilitated the development of a ‘ no blame ’ strategy, which legitimized organizational proactivity through collaborative and empowering change management processes. The research contributes to the study of gender pay gaps by demonstrating that gender mainstreaming, with facilitative local conditions and supportive public policy, shapes action on gender segregation, with particular success in women's low ‐ paid employment. Conclusions highlight theoretical and policy implications arising from the research.

P u blis h e r s p a g e : h t t p:// dx.d oi.o r g/ 1 0. 1 1 1 1/ g w a o. 1 2 2 3 9 < h t t p:// dx.d oi.o r g/ 1 0. 1 1 1 1/ g w a o. 1 2 2 3 9 > Pl e a s e n o t e: C h a n g e s m a d e a s a r e s ul t of p u blis hi n g p r o c e s s e s s u c h a s c o py-e di ti n g, fo r m a t ti n g a n d p a g e n u m b e r s m a y n o t b e r efl e c t e d in t his ve r sio n.Fo r t h e d efi nitiv e ve r sio n of t hi s p u blic a tio n, pl e a s e r ef e r t o t h e p u blis h e d s o u r c e.You a r e a d vis e d t o c o n s ul t t h e p u blis h e r's v e r sio n if yo u wi s h t o cit e t hi s p a p er.
Thi s v e r sio n is b ei n g m a d e a v ail a bl e in a c c o r d a n c e wit h p u blis h e r p olici e s.
S e e h t t p://o r c a .cf. a c. u k/ p olici e s. h t ml fo r u s a g e p olici e s.Co py ri g h t a n d m o r al ri g h t s fo r p u blic a tio n s m a d e a v ail a bl e in ORCA a r e r e t ai n e d by t h e c o py ri g h t h ol d e r s .
whilst also identifying key determinants to be addressed.Knowledge of pay gap drivers at the societal level has become particularly sophisticated with important implications for global and national government policy and legislation (Jones, Wass, & Makepeace, forthcoming;Peruzzi, 2015).However, in recent years, there have been calls for attention to organizational responses to pay gaps, specifically in the context of human resources (HR) policies and practices but also in terms of organizational culture change (Conley, 2003;Davies, McNabb, & Whitfield, 2015;Dickens, 1998).
In particular, there is a need to better understand how organizations respond to public policy developments and to determine whether legislation prompts organizations to embrace or reject responsibility (Acker, 1998;Conley & Page, 2014;Deakin, Fraser-Butlin, McLaughlin, & Polanska, 2015).These questions are particularly pertinent to public organizations that are viewed simultaneously as role model employers in relation to equality matters, whilst also being subject to criticism for contributing to the maintenance of pay gaps (Eveline & Todd, 2009;Thornley & Thornqvist, 2009).
In this article, we demonstrate how an intervention introduced through an action research collaborative, comprising academics, organizational practitioners and change management consultants, promoted a gender mainstreaming perspective and generated proactive organizational responses to gender segregation and pay disparities in Wales.
The data demonstrate how an evidenced-based approach, based on a gender analysis of employment data, worked in conjunction with a 'no blame' strategy to legitimize organizational innovation by empowering change agents and easing the concerns of those that might resist and block progress.The research settinga set of Welsh public sector organizationsoffers a distinctive context for the intervention as the Welsh employment profile is characterized by high levels of part-time working in feminized areas of employment, an economy heavily distorted by the public sector and a persistent problem with gender segregation (Felstead, Davies, & Jones, 2013;Jones & Robinson, 2011;Parken, Pocher, & Davies, 2014).Further, devolution has prompted an alternative legislative approach to equality, signified by a set of Welsh-specific duties introduced to support the implementation of the Public Sector Equality Duties (section 149 of the Equality Act 2010).Critical to the mainstreaming intervention under study was the Welsh-specific equality duty on pay differences (from here on referred to as 'the Welsh duty') that required public sector employers to address gender pay gaps in Wales in systemic ways (Equality Act 2010 (Statutory Duties) (Wales) Regulations 2011).
To summarize, this research contributes to debates on pay disparities by reporting on proactive organizational responses to gender segregation, generated and supported through a gender mainstreaming approach, driven by a unique legislative duty.In the article, we first review the literature on gender pay gaps, with specific reference to public organizations, highlighting potential for gender mainstreaming interventions, before discussing the background to the unique public policy context in Wales.Second, the nature of the action research intervention is explained, with details of the process outlined in full.Third, research findings are presented and discussed, initially in the form of a high level, overarching analysis of gender segregation across three case study organizations, and then through an account of change management undertaken with each case.The data demonstrate the importance of organizational proactivity and engagement and shed light on the key elements of change management that facilitate progress, particularly in the area of low pay.
Specifically, they highlight innovations focused on countering assumptions, de-stigmatizing women's work and genderneutral job redesign.Specific contributions to debates on pay disparities and gender mainstreaming are identified in the concluding section of the article, along with a series of policy implications.The next section provides a context for the research through a brief overview of recent studies on gender pay disparities.

| THE GENDER PAY GAP
Commonly understood as a measure of the average gross median hourly pay of all men and women in employment, the 'unadjusted' gender pay gap (currently 18.1 per cent in the UK), captures the ways that gender influences labour market participation and organization, thereby demonstrating inequalities in the employment structure (Parken, Rees, & Baumgardt, 2009).Whilst there is some consensus on the influence of industrial composition, fluctuations in the business cycle, welfare incentives and recessionary effects (Rubery & Rafferty, 2013), potential explanations for the gap, the identification of its determinants and proposed remedies for its eradication tend to vary according to academic discipline as Rubery and Grimshaw (2014) convey in their most recent review.They identify four schools of thought beginning with economic theorists, who have provided important evidence on the decomposition of the pay gap, often driven by questions of choice, individual decision-making and the impact of human capital (Becker, 1971;Hakim, 1996;Hakim, 2006;Rubery & Grimshaw, 2014).Although, Rubery and Grimshaw (2014) also report something of a shift away from supply-side explanations for pay disparities, highlighting that economists are increasingly developing 'useful insights' from demand-side analyses (p.328).Second, Rubery and Grimshaw (2014) review sociological perspectives, often at odds with a human capital perspective, concerned with the family wage and the 'undervaluing' of women's work, highlighting the importance of gender segregation (Eveline & Todd, 2009).The European Commission's Advisory Committee Report on Equal Opportunities for Women and Men provides a recent illustration of these concerns, concluding that occupational segregation produces inefficiencies in the labour market by creating bottlenecks in labour supply and exacerbating skills gaps and mismatches (European Commission, 2017; see also Bettio & Verashchagina, 2008).This raises questions about the role and effects of trade union activities, arbitration and collective bargaining processes, which is the focus of a third disciplinary perspectiveindustrial relations and institutional theorists (Rubery & Grimshaw, 2014).Here, studies such as Arulampalam, Booth, and Bryan's (2007) evaluation of pan-European pay gap distributions, highlight the significant role of wage-setting institutions and childcare provision.
The role of government is significant here and attention has been focused on the efficacy of equality policy and legislation in particular (Conley & Page, 2014;Hoque & Noon, 2004).Finally, Rubery and Grimshaw (2014) highlight the work of organization and management theorists, which tends to focus on the ways that management practices and organizational cultures sustain and reproduce inequalities (Acker, 1990(Acker, , 1998;;Thomas & Davies, 2005).Here, studies have identified a series of factors that shape the effectiveness of equality initiatives, suggesting these are more successful when organizations are under pressure from government or competitors and want to gain external legitimacy (Baron, Mitman, & Newman, 1991).In addition, 'internal' determinants such as organizational size, the proportion of women employees and having women in leadership positions have been identified as being especially significant (Andrews & Ashworth, 2013).However, evidence on the impact of other factors, such as organizational culture, the presence of trade unions and HR management practices, such as job evaluation, is often much more ambiguous (Baron et al., 1991;Dickens, 1998;Hoque & Bacon, 2014;Koskinen Sandberg, 2017;Saari, 2013;Thornley, 2007).
In sum, successive academic studies and government enquiries have identified that gender pay gaps arise from the combined effects of gendered occupational segregation, the legacy of the 'family wage' in wage bargaining, an uneven distribution of paid and unpaid work, the 'motherhood penalty', the part-time pay penalty, organizational cultures and processes, and the undervaluing of women's work and skills (Eveline & Todd, 2009;House of Commons Women and Equalities Select Committee, 2016;Rubery & Grimshaw, 2014;Women in Work Commission, 2006).As this research is focused on public sector organizations, the next section of the article reviews evidence on gender disparities within public service workplaces.

| PAY DISPARITIES IN THE PUBLIC SECTOR
As 'role model employers', public organizations have a significant role to play in addressing gender equalities, including pay gaps, particularly given the dominant role of women within this sector of the economy (Bullard & Wright, 1993;Conley, 2003;Grimshaw, 2000;Jones et al., forthcoming;Rubery & Rafferty, 2013).Research on the detailed gender composition of public organizations demonstrates that whilst more women are working in public services, they are much more likely than men to be working in job roles below their qualification levels (Conley, 2003;Connell, 2006;Jones & Torrie, 2009;ONS, 2013).In terms of pay gaps, the academic evidence on public workplaces is mixed.On the one hand, Thornley's (2007) work demonstrates stable patterns of persistent disadvantage and undervaluation, linked to a predominance of part-time working, whereas a recent study by Jones et al. (forthcoming) evidences the impact of good quality public sector employment for women, underpinned by job evaluation.However, Jones et al.
(forthcoming) also conclude that the improvement in women's human capital has reached the limit of its capacity in enhancing women's pay as, despite higher qualifications, gender pay gaps cannot be further diminished while women remain clustered in lower paying occupations.They also identify a narrowing pay gap for public organizations, relative to private, although this ceases from 2010 onwards, raising queries over whether this represents a 'historical shift' or a transient phase linked to austerity (Conley, 2012;Conley & Page, 2017;Jones et al., forthcoming, p. 19).
Research continues to highlight obstacles to the introduction of equality innovations in public organizations, despite pressures to become more representative of society and promote inclusivity in the workplace (Andrews & Ashworth, 2013;Ashworth & Newman, 2016).Some of these constitute external constraints, such as the current austerity context, which is presenting severe budget, workplace and service challenges for public organizations (Kickert & Randma-Liiv, 2016).Sociological perspectives highlight internal barriers, such as an often superficial strategic engagement with equality policies for ceremonial legitimacy gains, whilst organizational approaches illustrate the embedded nature of traditional organizational sub-cultures and associated norms and practices (Ashworth & Newman, 2016).On this basis, there have been calls for further research on non-pay job characteristics in the public sector (Jones et al., forthcoming).
The slow progress on reducing pay gaps has led many to advocate the adoption of a gender mainstreaming approach, a strategic activist intervention deemed especially appropriate to the public sector due to its ability to 'link explanation to action' and stimulate a 'continuous process of analysis and response' (Eveline & Todd, 2009, p. 551).
Whilst there are few examples of effective gender mainstreaming (Moser, 2005;Rubery, Grimshaw, & Figueredo, 2005) indicate its potential for application to gender pay settings, due to its multifaceted and systemic approach and consequent ability to focus attention onto organizational practices and processes, rather than differences between men and women.As such, gender mainstreaming attempts to account for the myriad of social, cultural and economic factors that produce segregation in employment and pay and therefore has superior capacity, compared with alternative mechanisms of audit and monitoring (Eveline & Todd, 2009;Pillinger, 2005;Rees, 1998).The ongoing and iterative processes of analysis and action associated with gender mainstreaming suggest it may be particularly well-suited to addressing public sector pay gaps (Eveline & Todd, 2009;Pillinger, 2005;Rees, 2005;Rubery et al., 2005), although Eveline and Todd (2009) argue it is most likely to be effective when certain conditions of 'political will, intensive links between research and action and adequate resources' prevail (p.536).
Overall, research on gender disparities in public sector organizations highlights a series of barriers to the institutionalization of practices and innovations designed to improve equality outcomes.Further, analyses of the gender pay gap often fail to interrogate non-work characteristics and address gaps at the lower end of the pay scale, in areas such as social care (Jones et al., forthcoming).As a result, our understanding can often be 'overly simplistic', highlighting the need to give greater priority and 'voice' to low-paid women (Smith, 2009, p. 28).A gender mainstreaming approach, it is argued, might enable a shift beyond the decomposition of the pay gap and human capital explanations in order to address systemic barriers to employment, job quality and progression (Eveline, Bacchi, & Binns, 2009;Rees, 2005).
The evidence gathered to date prompts a series of questions, such as: to what extent can gender disparities, particularly at the lower end of pay structures, be addressed by public organizations?Given the institutional constraints, how do organizations respond to public policy demands and legislative duties?Does public policy prompt responsible reactions, provoke resistance or encourage cynical instrumental responses?Having established the need for a more nuanced account of organizational responses to the gender pay gap in the public sector, the next section of the article considers the importance of the policy context which, in this case, concerns the Specific Equality Duty that applies to public organizations within Wales.

| PUBLIC POLICY CONTEXT: ADDRESSING THE PAY GAP IN WALES
Devolution of central government decision-making within the UK in 1999 resulted in separate governance structures in Wales, Scotland and Northern Ireland and the establishment in Wales of a National Assembly, which acts as the central political decision-making body, working in conjunction with the Welsh Government, as the Executive decision-maker.Since its inception, the Welsh Government has set out a distinct equality agenda based upon the principles of gender mainstreaming and beyond this to equality mainstreaming (Chaney, 2009;Parken, 2010a).Under the Government of Wales Act (1998), the government has an 'absolute duty' (Chaney, 2004) to mainstream equality for all people through public policy.This exceeds a 'strand-based' equality policy making and early efforts were made to introduce an intersectional approach to mainstreaming equality in Wales (Parken, 2010b).Since the Government of Wales Act 2006, responsibility for mainstreaming equality rests with the Welsh Ministers who must report the Welsh Government's progress annually.However, this duty confers no additional positive rights and formal challenge can only be made through judicial review.Therefore, in day-to-day practice, it is adherence to the Public Sector Equality Duties (PSED) and the Specific Equality Duties that underpin them under the Equality Act 2010, that inform policy making for equality.The Specific Equality Duties are determined by the devolved governments and therefore vary in focus and scope.
Successive UK governments have declined to legislate in response to calls for mandatory job evaluation and Equal Pay Reviews (EPRs) (Equal Pay Taskforce, 2001;Women and Work Commission, 2006), although it can be argued that EPRs have become mandatory in local government and the National Health Service (NHS) through collective bargaining agreements (Deakin et al., 2015).However, the Welsh Government seized the opportunity to introduce a specific duty to address pay differences under the Equality Act 2010.In contrast to the English context where public organizations were only required to have 'due regard' to the need to set a pay equality objective, the Equality Act 2010 (Statutory Duties) (Wales) Regulations 2011 require public bodies in Wales with more than 150 employees to report pay differences amongst all protected characteristics.Employers are further required to have due regard to having an equality objective to address pay differences between men and women or crucially be able to evidence why they think they need not do so.Uniquely, the Welsh duty insists that public authorities analyse and publish gender employment information by job, grade, working pattern, contract type and pay in order to facilitate understanding of the causes of gender pay gaps within their employment structures (Parken, 2015).In emphasis, it is evidence-based and action-focused, requiring an action plan, annual reporting and updating, and is underpinned by recognition that gendered employment structures produce inequalities in pay.Thus, the legislation aims to address occupational structures to close gender pay gaps, rather than pay systems and equal pay.However, the Welsh duty is often erroneously referred to as the 'equal pay duty' demonstrating the common elision between policy designed to address discriminatory practices in pay systems and that which is focused on gendered labour market structuring.In essence, the Welsh duty requires employers to address gender pay gaps through a focus on the interplay between different forms of occupational segregation (occupational, horizontal, vertical, working patterns and contract type), in order to surface the often unreflexive use of part-time contracts for women in low grade work in the public sector and women's over-representation in temporary and casual contracts.This, it was hoped, would lead employers to question whether it was always a choice to work part-time hours, as well as to consider the impact these factors may have on progression in both low and high grade work (Parken et al., 2009).
It should be noted that the Welsh duty differs substantially from those introduced elsewhere in the UK.The English and Scottish specific equality duties in respect of equal pay were drawn through from the Gender Equality Duty 2006 into the Equality Act 2010, so until recently, public sector organizations in England were only required to have 'due regard' to the need to set an equality objective on equal pay.In March 2017, a duty for English public authorities, mirroring the new duty for the private and voluntary sectors, was introduced for organizations with more than 250 employees.This specific equality duty requires public authorities to publish mean and median pay and bonus gaps, and the proportion of men and women receiving bonuses and in salary quartiles (Equality Act 2010 (Specific Duties and Public Authorities) Regulations 2017).The Scottish duty requires listed authorities with over 150 employees (amended in 2016 to organizations with 20 or more employees) to publish the hourly pay gap between men and women every two years, in addition to a statement on equal pay policy and basic occupational characteristics for men and women, people who are disabled and those who are not, and people from ethnic minorities and those who are not, every four years (Equality Act 2010 (Specific Duties) (Scotland) Regulations 2012).In Northern Ireland, a separate duty applies under s75 Northern Ireland Act 1998, which requires equality impact assessment of public sector employers' job evaluation schemes, which are considered to be public documents.Having outlined the legislative duty and demonstrated the complexity of the analysis and understanding it was designed to create, the next section of the article sets out the methodology for the research-informed intervention programme that was designed in response to the legislation.

| RESEARCH DESIGN AND METHODOLOGY
This article aims to provide a more nuanced understanding of the gender pay gap through insights drawn from a gender mainstreaming intervention, driven by a unique legislative duty designed to support public organizations in their efforts to address gender segregation and disparities.This section reports on the action research approach and design adopted in this instance.As Eveline and Todd (2009) explain, a gender mainstreaming perspective involves moving beyond explanations for a pay gap 'towards a responsible plan of action anchored to a complex interweaving of research findings and gender analysis' (p.543).As such, gender mainstreaming necessitates an action research approach, defined by Eden and Huxham (1996) as 'an involvement with members of an organization over a matter which is of genuine concern to them' (p.75).Action research is typically characterized by an intervention driven by a desire to achieve change.It is especially consistent with a gender mainstreaming perspective as it provides an opportunity to engage in a type of research that 'challenges unjust and undemocratic economic, social and political systems and practices' (Brydon-Miller, Greenwood, & Maguire, 2003, p. 11).In this specific case, the objective was to support the implementation of public policythe Welsh dutyand provide research-informed advice that would prompt organizational efforts to address and avoid the reproduction of persistent gender segregation in employment and pay through an action plan.Typical of an action research scenario, here the researcher was not a neutral observer but rather an actor 'within a dialogical process structured by both the researcher and the researched' (Heiskanen, Otonkorpi-Lehtoranta, Leinonen, & Ylostalo, 2015, p. 6).
It is argued that action research methodologies have a series of strengths.For example, they provide nuance and insights that are difficult to obtain through other methodological approaches (Eden & Huxham, 1996).This is linked to a central assumption that research subjects possess the ability to interpret their own circumstances, understand the issues they face and have the potential to address them (Brydon-Miller et al., 2003).Further, the opportunity that action research provides for researchers to deliver benefits at the point of action is a significant one as the role involves participation in a joint venture designed to deliver democratic and social change.In this case, the intervention provided the research team with the opportunity to advance progress on reducing gender pay and employment disparities within three organizations through a gender mainstreaming approach.However, as Brydon-Miller et al. (2003) remind us, action research is not just about 'doing good' but about 'doing this well' (p.25) so there is a need to ensure that an action research approach generates data, provides insight that informs other contexts and contributes to theoretical advances (Eden & Huxham, 1996).In order to fulfil these requirements, we have measured and reported our research process against standards for action research provided by Eden and Huxham (1996) (see Table 1).In doing so, we seek to demonstrate not only the consistency of the process in relation to gender mainstreaming interventions, but also the empirical and theoretical insights that can be drawn from the research.

| The intervention
The research was initiated in response to the introduction of a unique and specific equality duty for Wales.As a gender mainstreaming perspective implies, public policy is a starting point but there is a need for policy instruments to be 'brought to life' and infused with evidence and understanding in order to inform practice change that is likely to succeed and endure.The intervention constituted a three-year gender mainstreaming research project resourced by European Structural Funds (ESF), between 2012 and 2015 and, designed in the anticipation that HR professionals and workplace analysts might struggle with a structural change interpretation of an employment and pay data.The resulting project was consistent with the objectives of a gender mainstreaming approachto change the cultural, social and economic structures that reproduce gender inequalitiesas outlined by the European Union (EU) [COM (96) 67 final from 21/2/96 (CEC 1996)] and academic experts (Rees, 2005;Rubery et al., 2005).So, the focus was not upon men and women as individuals but rather on the unequal gender relations that shape their involvement with power, income, decision-making and the division of labour (Rees, 1998(Rees, , 2005;;Verloo, 2005).The explicit aim of the intervention was to support organizational responses to public policy, by working in conjunction with three large public sector employers in order to redesign practices and procedures, through change management processes introduced on the basis of the evidence provided.Activities and outcomes delivered through the three-year ESF project were further embedded and supplemented by additional research support delivered in 2015-2016, due to the award of an ESRC Impact Accelerator Account.
Three case study organizations were included within the interventiona local authority, an NHS Board and a university (see Parken, 2015 for a more detailed overview).These organizations were selected on the basis that they represented variation across the public sectorthere are 22 local authorities in Wales, seven local health boards and 10 universities serving a resident population of 3 millionand all were subject to the Welsh duty as responsibility for local government, health and education is devolved to the Welsh Government.The geographical location of the case studies was determined by EU structural fund criteria so two case studies were based in the South Wales Valleys and one in West Walesareas still impacted by the deindustrialization of the 1970s and 1980s and characterized by high levels of unemployment and poverty, the loss of physical and human capital, and minimal employment opportunities Adapted from Eden and Huxham (1996).
PARKEN AND ASHWORTH (Davies & Parken, 2017).This is illustrated by the gross value added per head of population in Wales, which was 70 per cent of the UK average in 2015 and just 60 per cent of the UK average in West Wales and the Valleys (ONS, 2016, cited in Davies & Parken, 2017).In addition, Wales demonstrates a lower proportion of economic activity in the private sector, a higher proportion of jobs requiring low qualifications, more prevalent gender segregation and lower average earnings compared to other parts of the UK (Felstead et al., 2013).
The case study approach designed for the intervention was neither a technocratic nor citizen activist approach (Eveline & Bacchi, 2005;Eveline et al., 2009) but rather reflected a hybrid, design, employing Rees' (2005) principles for gender mainstreaming.For example, the plan for organizational engagement reflected a distributed and collaborative approach, involving Chief Executives and HR directors, but also trade unions and corporate management teams, including potentially sceptical stakeholders, such as Directors of Finance.Agreement to this level of engagement, and particularly the sharing of research findings, was a stipulation of case study participation within the intervention (Parken, 2012).This level of commitment produced a clear benefit in terms of organizations applying equality change processes transversally across service delivery areas, with a view to wider workforce planning and development, as opposed to actions being limited to HR teams or narrowed to the micro examination of pay discrimination between individuals.
Persuading organizations to engage in a gender mainstreaming project designed to analyse and reduce gender pay gaps was not an easy process.There was considerable nervousness around the legislation, the prospect of litigation and negative publicity on unequal pay.Hence, effort was expended to secure organizations' trust, which involved providing reassurance to employers in two ways.First, robust data security protocols were put in place to ensure that data and case study identities remained confidential during the research.Second, drawing upon the narrative of the enabling approach taken by the duty, researchers employed a 'no blame' strategy.This was designed to demonstrate that gender pay gaps are produced by a multitude of factors, many often external to the organization, but to emphasize that with research support, employers had the potential to reduce disparities through organizational practices (Parken, 2012).This approach potentially runs counter to suggestions in the literature that organizations should be encouraged to assume responsibility for inequalities in employment and pay, rather than absolve themselves (Acker, 1998).However, this initial alleviation of responsibility was critical as it not only enabled organizations to participate fully within the research, but was key to subsequent empowerment and an embracing of responsibilities.The approach provided positive motivation in the face of seemingly impossible change and prompted an enthusiasm which is deemed fundamental to the sustained transformative change required through gender mainstreaming (Eveline & Bacchi, 2005;Eveline et al., 2009).

| Components of the intervention
Phase 1 involved the three case study employers providing the research team with their employment and pay data as individual but anonymized records.A lengthy process then ensued as in many instances basic data had to be synthesized and aggregated by the research team, due to the number of separate employment and pay data repositories within organizations.From this, the research team constructed datasets incorporating gender, age, department, occupation, grade, working pattern, contract type, tenure and level of pay (hourly, weekly, annual), 1 and undertook basic descriptive data analysis.
Phase 2 was focused on the situation of data, relative to existing research, further analysis and the development of a series of problem statements for each organization.The problem statements were subsequently converted into an action plan with the support of change management consultants.Further detail of Phase 2 of the research is described in the relevant section below.
Taking the two phases together, from the initial agreement to collaborate in the research through to the end of the change management support phase, each case study took approximately two years to complete.The following section of the article reports a summary of overarching research findings from each phase.

| Phase 1: co-creation of evidence-informed action plans
This section provides an overview of the data used to 'set the scene', identify the extent of disparities in employment and pay, and support case study organizations in developing their action plans.Throughout, the discussion provides a descriptive comparative analysis of employment data across the three cases, highlighting similarities and differences.Finally, we explain how the research team's interpretation of the evidence provided a basis to begin action planning.

| Overall employment patterns
As the cases are public sector organizations and each had undertaken job evaluation and/or EPRs in the preceding five years, there was little evidence of in-grade inequalities.However, as few men and women worked in the same jobs on the same grades and in the same contract types and working patterns, all organizations had gender pay gaps.We do not report pay gap data here, for reasons of organizational confidentiality.Rather, we illustrate the underlying employment segregation patterns that rapidly became the focus of attention due to their impact upon pay disparities.Whilst these data were reported in depth in detailed case study reports (Parken, Pocher, et al., 2014;Parken, Pocher, & Sloan, 2013;Parken, Sloan, & Pocher, 2014), due to space constraints, we provide only a brief precis for this article.The analysis involved directly contracted employees (within the organization's direct locus of control) and was focused on positions within the main grading structures of local government, the NHS and higher education.This included National Joint Council (NJC) grades 1-15 for local government, 2 in addition to contracted medical and dental staff and Agenda for Change Bands 1-9 for the NHS. 3 In the case of higher education, Higher Education Role Analysis (HERA) single spine staff in grades 1-9, 4 Readers (a mid-tier academic grade positioned between Senior Lecturer and Professor) and Professors were included.Those falling outside of these standard categories were excluded from our analysis.For example, locums were removed from the dataset as these are usually replacement medical professionals standing in at short notice and on a temporary basis.
Table 2 provides a summary of the overarching employment features of each case by gender and indicates that women occupied three quarters and four fifths of positions in the local government and health cases, respectively, which are reasonably typical proportions that closely mirror previous analyses (e.g., Thornley, 2006).In contrast, the higher education case displayed more of an overall gender balance.However, Table 2 also demonstrates that men were overrepresented as holders of full-time posts in all organizations, despite being within the minority of employees overall.This finding is consistent with underpinning research evidence which indicates that part-time roles in personal service, sales and elementary occupations in Wales comprise, on average, 50 per cent of employment in those sectors, thus restricting women's choice to work on a full-time basis (Parken, Sloan, et al., 2014).As full-time working is the norm for those in higher grades throughout the occupational strata, the proportion of full- Percentages rounded for ease.PARKEN AND ASHWORTH time posts on offer has a clear impact on the ability of women to access full-time work and thus higher earnings.In local government, just 26 per cent of all posts were offered on a full-time basis, whilst in health it was 60 and 70 per cent in higher education.

| Employment and contract types
Table 3 provides an extension of the analysis through the provision of data on contract type and work pattern by gender.In the local authority, at just over a quarter of all employees, men held over half of all full-time posts (permanent, temporary and term time only), resulting in nearly two thirds of men working on a full-time basis.
In contrast, women held over 90 per cent of all available part-time jobs, with four fifths of women working in this pattern.Full-time posts were most associated with higher grades and, in all cases, vertical segregation was evident with men being over-represented in the highest positions.The exception to this rule was low grade work undertaken by men in estates and environment departments across the three organizations, which was organized on a full-time basis.Women in low grades were overwhelmingly contracted on a part-time basis.This is consistent with the characterization of low skilled work which provides little opportunity for training or progression (Felstead et al., 2013).
In local government, around one third of all posts were in the lowest three grades, but these posts captured only a fifth of men's work compared with a third of women's.In these grades, four fifths of men worked on a full-time basis, while almost all women worked on a part-time basis.So the legacy of gendered 'family wage setting' was very much in evidence in the organization of work.In clear contrast, the health board demonstrated a much higher stock of full-time jobs at 60 per cent of the total with nearly three quarters of women working on this basis in good quality work.
Consistent with prior research, Nursing and Midwifery accounted for one third of all staff, producing a large group of well-paid full-time women employees in the middle of the band structure (Andersson Back, 2009).This meant that overall gender pay differences at the median were much lower than in local government and the university.The university case showed the highest stock of full-time jobs at 70 per cent, but at just over half of the workforce, women held less than half of all full-time permanent posts and 74 per cent of all permanent part-time posts.Part-time work adhered to the lowest grades across all the case studies, again resonating with previous studies (Conley, 2003;Thornley, 2007).
There was, however, some evidence of higher graded part-time working in all the cases, mainly for women.
This was an encouraging finding, demonstrating that in this set of public sector organizations women had the potential to gain part-time employment in higher level positions.Nevertheless, as demonstrated in previous research (Jones & Torrie, 2009;Koskina, 2009), men were over-represented in all cases in the highest grades, so this prompted further qualitative work and a series of actions in line with the change management phase of the research (see below).Across the board, the majority of full-time and part-time jobs were offered as permanent contracts, although men were overrepresented in their share of these full-time and permanent premium employment opportunities.Table 3 shows that, with the exception of the health board, men held the majority of permanent full-time contracts, while women were the majority holders of permanent part-time contracts and had a much higher incidence of working in less secure contracts, both part-time and full-time.

| Action
The senior management team of each organization was provided with a data analysis and interpretation report that situated findings in the context of the literature on gender and employment.Through the discussion of the evidence, case studies were beginning to observe the ways that employment structures served to limit opportunities, progression and earnings.It became clear that, in the absence of intervention, existing patterns would endlessly reproduce, but also that women's talent would continue to be wasted to the cost of the organization.Thus, the 'problem' of gender pay fast became a good opportunity to engage in advanced workforce planning with a view to enhancing staff development practices.Consequently, the research escalated to a form of direct intervention designed to support a reduction in employment inequalities that produce pay gaps.
In taking this action forward, the three cases began to identify issues to address, in conjunction with the research team.Table 4 summarizes initial findings across the three cases but also highlights the distinguishing features of each to be addressed within change management action plans.For the local authority, a key issue was the low stock of fulltime work for women, in addition to patterns of vertical segregation.The health board was concerned about the extent of part-time (often agency) working by nursing staff and dominance of men within medical and dental grades, whilst the university was focused upon horizontal segregation and disparities between men and women in professional services, particularly in lower grades.Following the presentation of case study analysis, the main findings were reviewed again by the research team, in conjunction with an assessment of each organization's recruitment, training, development and progression policies, consistent with a gender mainstreaming approach (Eveline & Bacchi, 2005;Rubery et al., 2005).This section reviews evidence from Phase 2 of the gender mainstreaming intervention when data from Phase 1 was employed to inform actions designed to disrupt patterns of segregation and subsequent pay disparities.The descriptive quantitative reports were used by researchers and the lead 'Change Manager' from each case study to generate a clear set of research-informed 'problem statements' for each organization.At this stage of the intervention, change management consultants joined the collaborative, valuable due to their expertise in delivering change on the ground in public organizations.The agreed change management approach had three key elements.The first was to ensure collective 'buy-in' to the process from across the given the need to revise, not only recruitment, workforce planning, training and development and promotion arrangements, but also generic line management practices and organizational sub-cultures.It was imperative that the intervention was not seen exclusively as an HR responsibility.The second element concerned the need to establish clear and effective leadership for the intervention.
This led to the appointment of an 'active and visible leader' within each case study who would play a critical role, not only in championing the project, but in leveraging seniority in order to facilitate change.Linked to the first element, there was an explicit attempt to appoint senior leaders unconnected to HR departments.The third element of the approach was designed to ensure that change was achievable so the action plans contained three-dimensional objectives: thresholds, targets and stretch.'Thresholds' were measures that were immediately achievable (e.g., an evaluation of Performance Development Review (PDR) processes), 'targets' were more challenging actions to achieve within the programme (e.g., wholesale revision of role descriptions) while 'stretch' referred to the achievement of aspirational longer-term outcomes (e.g., reducing horizontal segregation in employment and pay within the organization). 5 In order to embed the change management programme, the team presented evidence to groups of staff, including heads of department or service delivery, HR, operational and financial lead managers, medics and academics, as appropriate.In seminar sessions facilitated by the change management consultants and researchers, heads of service prioritized activities, becoming 'action owners' and directing the intervention.Each case study organization formed an Intervention Management Group comprising the 'active and visible leader', change manager, service director, workforce director and 'action owners' that would direct communications, provide resources and monitor progress, whilst also championing the intervention within the organization.Throughout the process, the case study organizations were supported by the change management consultants and the research team.However, in addition, they formed part of a wider network of employers concerned with employment and pay disparities.Each of the three organizations disseminated regularly to the network, which resulted in an ongoing transfer of learning to a further 10 organizations beyond the research collaborative, in addition to the exchange of ideas and actions between the three case study organizations.
As it is difficult to capture the full scale and scope of the resulting activity within this article, Table 5 presents an overview of selected problem statements for each case and example actions resulting from the change management process for illustrative purposes.To provide some further elaboration of the types of actions instigated, a summary example from each case study is presented in the next section.

| De-stigmatizing women's work
Given the problem statements regarding the predominance of women in the lowest three grades working on a parttime basis resulting in low earnings and the proportion holding two jobs or more, the local government case study undertook a series of actions.Recognizing the stigma attached to part-time working, signified by demeaning references -'she's just part-time'-and working in consultation with the legal team and the trade union, the organization proceeded to remove the 'part-time' label from job descriptions.So positions such as 'Part-Time Cook' , 'Part-Time Administrative Assistant' were removed from the organization with a longer-term aspiration that this would detach the link between hours worked and employee status.This was achieved through a full review of flexible working and redesigned PDR processes that establish whether women desire additional hours, permanent contracts and Overview of example problem statements and actions

Case study Selected problem statements Actions
Local authority Grades 1-3 account for a third of all staff, almost all are women who work on a part-time basis resulting in annual earnings below the low earnings threshold.Eleven per cent of women hold two jobs or more demonstrating an effort to build earningspresumed by choice.Women comprise around a third of senior staff and are absent from the most senior positions.
Two phases of consultation with low graded part-time and/or casual staff and their line mangers, exploring degree of 'choice' in working hours, multiple jobs and aspiration for progression.opportunities to progress.In addition, the organization introduced unconscious bias training for 400 HR staff and, in the process, assessed advertisements for gender-segregated posts.This had an immediate impact as the exercise revealed male-dominated jobs were often entirely task-focused, requiring previous experience that was not strictly necessary.The exercise empowered HR managers to roll out unconscious bias training to a further 600 line managers, with an online version released to all managers.A wholesale revision of the recruitment process is now underway.

| Countering assumptions
The health board was confronted with problem statements indicating that there were issues of low pay and limited progression within Band 1 roleslaundry, cleaning and cateringin addition to an assumption that nurses and healthcare assistants 'chose' fixed term employment.In response, the organization was encouraged to reflect on its perceptions of part-time working and question notions of 'choice'.A subsequent 'pulse survey' with women in laundry, catering and cleaning demonstrated that whilst line managers assumed employees had no aspiration to develop, employees felt differently with 60 per cent stating an interest in progression.A further survey exercise found that the vast majority of female catering staff (90 per cent) welcomed additional hours and development opportunities, such as job shadowing, along with a third of housekeeping staff.As a result, PDRs have been amended to ensure discussion of progression and development opportunities, positions redesigned to allow progression in otherwise flat structures, whilst job shadowing and secondment schemes have been extended across the lifetime of employment within the organization and not just at the outset.A second problem statement referred to an assumption that healthcare and nursing staff opt to work on temporary contracts, and potentially for a number of employers.Further investigation revealed that a significant proportion of these staff had a preference for working with one organization if location base and flexibility could be ensured, ideally on a permanent basis.

| Gender-neutral job design
The university faced a range of issues to address through its problem statements.In particular, the organization was confronted with evidence that women were more likely to be employed in lower grade clerical and secretarial roles, with no career path through to higher graded administrative roles.Further, gender segregation of jobs and working patterns meant that positions such as portering were perceived to constitute 'men's work' and configured on a fulltime, permanent and reasonably well-paid basis, whilst roles within catering and cleaning, were often part-time, temporary and poorly paid.A major change to the university's estate in the form of a large-scale new building project on a new campus site allowed the organization to engage in a radical response to this problem through a redesign of campus support roles.This resulted in the creation of a new range of gender-neutral 'Campus Services' positions, including two new roles: 'Campus Services Team Member' and 'Campus Services Team Leader' that involved a mix of gender typed activities and flexible working practices for all.Recruitment into these positions was also distinctive on the grounds that it departed from regular practice based on experience and was oriented towards the organization's values -' We are professional', 'We work together' and 'We care.'This resulted in a gender balance across the new roles and a higher proportion of women recruited into positions, particularly as team leaders.
The degree of innovation in this case resulted in the university being nominated for, and receiving, a national diversity award.

| DISCUSSION AND CONCLUSIONS
This article makes a valuable contribution to debates on pay disparities by reporting on proactive organizational responses to gender segregation and pay, generated and supported through a gender mainstreaming approach, driven by a unique legislative duty.Our research findings indicate how a collaborative research approach, 'no blame' strategy and well-designed public policy combined to liberate employers and prompt action on the ground.The data highlight a series of specific innovations introduced by case study organizations designed to change employment structures and create new opportunities through the countering of assumptions, de-stigmatization of women's work and introduction of gender-neutral job design.The research provides a rare example of a gender mainstreaming intervention that demonstrates a powerful impact on gender segregation in low-paid occupations where women's work tends to be less visible and valued than men's (Eveline & Bacchi, 2005;Eveline et al., 2009;Moser, 2005;Saari, 2013;Toffoletti & Starr, 2016).
The research makes an important contribution to a small, but developing, evidence base on the implementation of gender mainstreaminga strategic interventionist perspective on gender equality in public policy, characterized by 'a responsible plan of action anchored to a complex interweaving of research findings and gender analysis' (Eveline & Todd, 2009, p. 173).Reflective of a gender mainstreaming approach, the duty required gender employment information by job, grade, working pattern, contract type and pay in order to facilitate understanding of the causes of gender pay gaps.This presented a platform for a programme of participatory action research that allowed case study organizations to view their employment profiles in the context of wider evidence, enabling them to move beyond questions of human capital and choice and recognize the historic undervaluing of women's work in collective bargaining (Deakin et al., 2015;Rubery et al., 2005).
On the basis of the evidence that women were merely responding to limited employment options on offer, researchers prompted employers to challenge a series of organizational myths.For example, by unpacking the assumption that women deliberately chose low-paid part-time work, the case study organizations became cognisant of the ways that employment structures might endlessly reproduce inequalities and, importantly, of their role in that process.Similarly, employers were disabused of the notion that women lacked confidence or did not aspire toward opportunities for promotion and advancement because the research approach facilitated a more detailed examination of assumptions and organizational processes linked to progression (Crompton, 1997;Sandberg, 2013).Through the subsequent change management phase, employers began to address the 'problem statements' resulting from the analysis and instigated processes of review, action planning and evaluation.Overall, the evidence suggests that these elements resemble the staged and viable implementation process envisaged by Moser (2005) and, as such, served to mitigate against the structural, political and cultural barriers that typically undermine gender mainstreaming initiatives (Eveline & Todd, 2009;Pillinger, 2005).
It is worth highlighting two specific implications for gender mainstreaming that arise from this research.Firstly, prior research has suggested that workplace-level participation is important to the success of gender mainstreaming (Pillinger, 2005), although studies have identified that power relations and social dynamics can often make participatory research on gender equality problematic (Benschop & Verloo, 2006).Our data indicate that the three main elements of change managementcollective buy-in, an active and visible leader, and phased objective settingwere central to the success of the intervention, but that the overarching 'no blame' strategy was especially significant.This strategy was not only distinctive, but also pivotal in securing authentic and enthusiastic participation.By situating the problem of gender pay gaps within a wider social and economic frame and alleviating organizations from direct and total responsibility, the research collaborative prompted high levels of energy, in addition to a sense of empowerment and ownership.Employers then grasped the opportunity and responsibility for making transformative change to gendered employment structures within their own organizations, consistent with a gender mainstreaming approach (Eveline & Bacchi, 2005).It suggests that in addition to 'political will, intensive links between research and action, and adequate resources', a fourth elementproactive organizational engagementis also likely to be critical to the success of any gender mainstreaming intervention (Eveline & Todd, 2009, p. 536).
Secondly, one of the key criticisms associated with gender mainstreaming centres on a concern that the central emphasis on gender becomes diminished as mainstreaming develops (Eveline & Todd, 2009).However, in this research, there were clear signs that key actors within the case study organizations were moving beyond discussions of 'the gender pay gap' toward a focus on 'gendering' their organizations (Benschop & Verloo, 2006;Eveline & Bacchi, 2005;Moser, 2005).In this case, driven by research evidence, organizations were focused on de-stigmatizing women's work, countering assumptions and gender-neutral job design, which suggests their actions far exceeded the usual "tinkering" with HR policies (Rees, 1998).As such, our findings support Saari's (2013) contention that gender pay interventions create an important arena for negotiation and learning that has the potential to ultimately transform 'gender blindness into a gender perspective' (p.51).
The research suffers from a series of limitations.We focus on just three case study employers, albeit with a coverage of just over 20,000 employees but, nevertheless, it may be that different patterns of segregation would occur in another set of organizations and that cultures, processes, mechanisms and motivation for action would vary too.In addition, the research was conducted within a Welsh political and economic context, one which undoubtedly influenced aspects of the gender pay problem, case study responses and the types of actions produced.Finally, prior studies have emphasized that gender mainstreaming is a 'process rather than a goal' which requires an ongoing and continuous need to sustain interventions in line with policy making (Eveline & Bacchi, 2005;Eveline et al., 2009;Moser, 2005).This research covers a four-year period and, whilst the relationship between the researchers, consultants and case study organizations remains a close and productive one, there is an evident need to sustain these connections.Longitudinal research would determine whether case studies realize their potential to deliver a longlasting impact on the overall gender pay gap but might also shed light on the prospects of 'a gender perspective' becoming established within organizations over time.As such, future research on gender disparities might facilitate an ongoing evaluation of the extent and embeddedness of change.It will be important to consider, for example, how recent turbulence driven by austerity impacts on the progress made by the case study organizations (Kickert & Randma-Liiv, 2016;Walby, 2015).
Our findings have a series of policy implications, not least the overriding conclusion that well-made public policy makes a difference to gender segregation within the workplace, particularly when a gender mainstreaming approach is adopted and employers can work in conjunction with supporting partners.In terms of national policy making, our research has particular relevance for the new legislative duty introduced by the UK government, given it prioritizes a minimal form of data reporting.Evidence indicates that this approach is unlikely to prove successful because it fails to encourage a systematic review of underlying structural explanations consistent with gender mainstreaming and focuses attention, and therefore potential blame, on the employer, discouraging an honest and open engagement with the problem at hand.Further, it is important to note that this research reports on an intervention that was reflective of a gender mainstreaming perspective widely promoted by the EU and underpinned by action research resourced by the European Social Fund, which enabled the collaborative to 'bring the duty alive'.As it stands, Brexit will severely limit the ability of the UK to deliver action research projects of this kind and learn from equivalent endeavours across the EU.
Finally, in terms of organizational implications, the analysis uncovered a considerable degree of occupational segregation, despite the fact that all three case study organizations had been through rigorous job evaluation processes.
Thus, in line with Koskinen Sandberg (2017), we would raise questions about job evaluation systems that provide organizations with a false sense of security and legitimacy and disguise the gendering of employment structures.This suggests the need for urgent work by academics and practitioners to develop systems that more accurately categorize, develop, support and recognize the work of women and remove barriers to progression based upon androcentric norms at all levels within contemporary organizations.

TABLE 1
Compliance with standards for action research a

TABLE 2
Gender composition and share of stock of jobs by working pattern a

TABLE 3
Gender contract type and working pattern a a Percentages rounded for ease.

TABLE 4
Summary of evidence informing action plan