The Treaty of Amsterdam marked the beginning of a new stage in the development of gender equality policy in the EU. Before Amsterdam, what was often termed the `women's policy' was extrapolated from a relatively `thin' constitutional reference point on equal pay for men and women in Article 119 EC. The institutional roles of the Court of Justice, the Commission, the European Parliament (especially its Women's Rights Committee) and the Council in promoting a policy of `equal opportunities' for women and men through legislation, soft law measures and case law are well documented in legal and political science literatures (e.g., Mazey 1995; Hoskyns 1996; Hervey and O'Keeffe 1996). A crucial variable determining the effectiveness of much policy-making has been the diverse national constitutional, institutional and legal terrains into which EU measures are nested once adopted, and this variable remains fundamental even after the shift to a mainstreaming paradigm. However, it is to the incapacity of the liberal norm of `equality' to deliver substantial socio-economic change in contemporary (social) market societies that the closest attention has been paid in critiques of sex equality law at both national and EU levels (e.g. More 1996; Fredman 1997; Hervey and Shaw 1998; Shaw 2000). The inadequacies of the `bare' equal treatment approach have long been recognised in the public discourse of the European Union institutions. Indeed, the Court of Justice was reminding its interlocutors of precisely this point when it declared in Marschall (a positive action case) that:
`... even where male and female candidates are equally qualified, male candidates tend to be promoted in preference to female candidates particularly because of prejudices and stereotypes concerning the role and capacities of women in working life, and the fear, for example, that women will interrupt their careers more frequently, that, owing to household and family duties they will be less flexible in their working hours, or that they will be absent from work more frequently because of pregnancy, childbirth and breastfeeding.
For these reasons, the mere fact that a male candidate and a female candidate are equally qualified does not mean that they have the same chance.'5
Interestingly, in this case the Court chose to refer to changes to be instituted by the Treaty of Amsterdam (at that time agreed and signed, but not yet ratified or in force), which had been introduced specifically to address unhappiness about the way in which the liberal equality norm was impinging upon national and regional positive action programmes as a result of the Court's own earlier judgment in the Kalanke case.6 The Treaty of Amsterdam introduced changes to what is now Article 141 EC (ex Article 119) to create a greater space for the adoption of positive action measures at national or regional level without contravening the equality principle, and also instituted a new `mainstreaming' principle in Article 3(2) EC, according to which `in all the activities referred to [in the list of Community activities in the previous paragraph], the Community shall aim to eliminate inequalities, and to promote equality, between men and women'. This can be said to 'constitutionalize' the guarantee of proactive policy-making to eliminate inequality, a markedly different guarantee to that of bare equal treatment or non-discrimination. The Treaty of Amsterdam has also led to the widening of the purview of equality policy, as a result of the introduction of the Article 13 legal basis for anti-discrimination measures, subsequently applied in two directives on race discrimination and non-discrimination more generally. These provisions in combination also carry the focus of anti-discrimination policy away from the workplace (and training) and away from the entitlements of the individual legal `victim' of discrimination alone towards other fields of Community competence such as labour market policy or development policy. The shift in focus also raises the possibility of wider societal responsibility for inequality and a concomitant duty on policy-makers to reconsider their practices.
To this end, it is crucial to note that, although the Commission's most recent `Framework Strategy' on gender policy for 2001-20067 continues to invoke the leitmotiv of equality before the law, its main approach is to combine the adjustment of policies or gender mainstreaming with the implementation of specific `positive' actions to improve the position of women in society. In its Strategy document, the Commission noted that:
`This situation [of persistent inequality] can be tackled efficiently by integrating the gender equality objective into the policies that have a direct or indirect impact on the lives of women and men. Women's concerns, needs and aspirations should be taken into account and assume the same importance as men's concerns in the design and implementation of policies. This is the gender mainstreaming approach, adopted in 1996 by the Commission which the Commission intends to operationalise and consolidate through this proposal.
In parallel to gender mainstreaming, persistent inequalities continue to require the implementation of specific actions in favour of women. The proposed framework strategy is based on this dual-track approach.' (emphasis in the original)
In keeping with the `beyond the workplace' approach which stems from the Treaty of Amsterdam, the Commission identifies five areas of intervention to all policies which must, in some way, be related: namely, economic life, equal participation and representation, social rights, civil life and gender roles and stereotypes. The Commission's Strategy is backed up by financial means provided for in a Council Decision establishing a funding programme,8 and other measures such as a Decision on gender balance in Committees.9 Gender equality policy is now the subject of an annual `work programme'10 as well as an annual report.11
The beguiling simplicity of this dual track strategy is belied by the continued under-specification and under-theorisation of the mainstreaming concept and of the strategies necessary to implement it (Scottish Office 1998; Verloo 2001). Gender mainstreaming as a policy tool originates in programmes of international governance especially in the area of development aid such as those of the UN Development Programme and the World Bank. It was endorsed in the Declaration resulting from the UN's Fourth World Women's Conference in Beijing in 1995 (Hafner-Burton and Pollack, 2002). It has also been the subject of attention at international level in Europe, within the Council of Europe and the OSCE, and it has been applied in varying degrees at national and regional levels of government in most of the Member States of the EU; as we have seen it represents a pillar of the EU's own gender policy. Even so, mainstreaming retains an uncertain relationship with other legal and constitutional principles, such as non-discrimination and equality. How does gender mainstreaming map onto established debates about the relationship between formal and substantive equality (if at all)? If the legal frameworks of gender mainstreaming are, on average, `softer' and more generalist, what does this mean? Will it imply, in the long term, a watering down of already relatively weak equality concepts enshrined in the existing equality directives? Or might it imply a strengthening of these concepts, at least in terms of substantive meaning, but a consequential weakening of enforcement and structures of enforceability? Its institutional implications both within the EU itself and for the Member States (nationally and regionally) have yet to be fully explored. The conditions necessary for effective law- and policy-making, and thus for implementing mainstreaming, are not clearly specified or understood (Beveridge et al 2000a and 2000b). The extent of policy learning and policy transfer from other fields where similar policy instruments are used (e.g., environmental policy) or other policy making arenas where a mainstreaming agenda is being used (e.g., global governance and development agencies or well-established national practices such as the Dutch Emancipation Impact Assessment (EER12) instrument (Verloo 2000)) can be made more explicit. The role of social actors and feminist lobby groups in the push towards equal opportunities policies is a well-established phenomenon, and the question arises as to whether the development of the mainstreaming approach opens new opportunities for social actor intervention, or changes the balance of social forces (Pollack and Hafner-Burton 2000a and 2000b; Hafner-Burton and Pollack 2001).
Alternatively, should the institutional adoption of an agenda of mainstreaming be seen as the `framing out' of the disruptive and radical aspects of feminism (Runyan 1999)? Has mainstreaming ever really left behind the neo-liberal context in which it originates in organizations such as the World Bank, where it can easily be constructed as an economic imperative, promoting strong economies, competitiveness and growth, and acting as an aid to social cohesion? One of the dangers with mainstreaming may be that with such apparent inclusiveness in policy-making, radical challenges to the status quo could, in reality, lose their capacity to disrupt received ideas about power and policy. It may involve as much `framing out' of interests as `framing in'. It may involve the co-option of feminist ideals into a soft-focus, family-friendly world in which choice and freedom are merely rhetorical devices, rather than real lived experience. Does mainstreaming in truth provide any better way of getting to the heart of a policy problem, simply because of its claim to inclusiveness? It may detract, in fact, from a better understanding of the policy problem because it suggests, wrongly, that the gender issues which are raised are relatively tractable rather than intractable. As the `what's the problem' approach to policy analysis advocated by Bacchi does, in fact, highlight, disputes over policy initiatives not only distinguish between those in favour of and those against a particular problem, but also help in `constituting the shape of the issues to be considered' (Bacchi, 1999). For example, issues about gender and development should not be assumed to be straightforwardly resolved because the boxes on the checklist of mainstreaming have all been ticked off. On the contrary, policy-makers should always question what representations have been assumed in the presentation of a policy issue and what alternatives there might be to the taken-for-granted solutions.
We can suggest that there a minimum of six key elements to a successful gender mainstreaming strategy as an element of public policy. The first point is to recognise the contested nature of the `gender problematic', namely, the fact that there are no simple solutions to allocative efficiency or fairness issues raised by the historical legacy of women's subordination or the persistent gendered division of labour in the family and household as well as in the employment sphere. No simple tick box answers will ever suffice. Hence, gender mainstreaming can, at most, be a strategy towards equality, justice and fairness, and not an end goal in itself. Closely related, the second point focuses on the diverse `gender regimes' applicable at national level which challenge the relevance of supranational policy-making in a field such as gender policy. Third (and fourth), there are two aspects of the `politics of presence' (Phillips 1998). On the one hand, there is the inclusion and participation of women in decision-making structures, e.g., in political and legal institutions including executives, legislatures and judiciaries, as well as women's involvement in other public bodies or committees, in NGOs and in private bodies; closely related are other `participation' issues such as the general question of consultation, access to information and the availability of appropriate institutions and policy machinery for the instrumentalisation of equality policies. The fifth element concerns the need for disaggregated and gender-sensitive statistics and data about women generally (including qualitative data about the experience of women household work, the family/work divide or `gendered' violence such as rape). Finally (sixth), there is `gender mainstreaming' in its narrow sense of gender impact assessment. What are the necessary practical instruments for effective implementation of mainstreaming? What expertise is required and available? Can there be effective participation of those affected by policies? What systems of accountability and monitoring of decision-makers (experts and officials) exist? Can target setting, benchmarking, and other mechanisms for policy evaluation at every stage from agenda-setting to policy implementation be used? What are the systems for policy-learning and policy-transfer which can be applied?
A number of possible theoretical frames for articulating how the gender mainstreaming approach works suggest themselves. One possibility is the feminist debate about `reason'. According to Minow (1991: 365), what matters in any given institutional setting is less composition than `reason'. Her concern is `points of view' and the relationships between power, privilege and seeing the other side. She suggests that `the more powerful we are, the less able we are to see how our own perspective and the current structure of our world coincide.' The challenge, therefore, is the construction of legal disputes and the recognition of points of view. For `otherwise, outsiders who become insiders simply define new groups as "other"'. Instead, she argues for highlighting the point of view of people labelled as "different" and `generating vivid details about points of view excluded from or marginalized by particular institutions.' She concludes:
`Seeking out and promoting participation by voices typically unheard are also crucial if equality jurisprudence is to mean more than enshrining the point of view of those sitting on the bench. The concerted and persistent search for excluded points of view and the acceptance of their challenges are equally critical to feminist theory and practice. Otherwise, feminists will join the ranks of reformers who have failed to do more than impose their own point of view.'
This suggests the value of the search for ways of framing into such disputes about law and policy, as they crystalise before bureaucracies and courts, factors and elements which comprise more than just legal reconstruction of those categories which the law's own system of definition - deems to be relevant. Arguably, the approach of gender mainstreaming with its `systematic integration of the respective situations, priorities and needs of women in all policies' and its attempt to achieve equality `by actively and openly taking into account, at the planning stage, [the] effects [of policies] on the respective situation of women and men in implementation, monitoring and evaluation' may be an avenue along which radical policy reform may proceed in the future. To put it another way,
`Gender mainstreaming is 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, by actors normally involved in policy-making'.13
This process cannot, of course, be guaranteed to deliver `gender equality', even if the reasonable disagreement of social actors about the meaning of `equality' could itself be resolved. What it provides, however, is an empowering frame of reference - a resonance point, to use the language of social constructivism in international relations theory (Schwellnus 2001) - for those who seek to take gender equality into the public sphere, as a matter of public debate.
Despite the rather optimistic note struck by this conclusion to the analysis, it would be wrong to exaggerate how far the mainstreaming agenda has already been taken in within EU public policy-making. The (limited) success stories have been rather well documented (Rees 1998; Pollack and Hafner-Burton 2000a and 2000b), e.g., in relation to labour market policy, policy on research and `women and science', development aid policy, and education and training policy. In the Framework Strategy, the Commission recognises the need for strategic action outside the `normal' cases of mainstreaming. A Commission Staff working paper of March 2001, published on the DG EMPL website, sets out a work programme for 2001 for each Commission service for the implementation of the Framework Strategy on Gender Equality, and is intended to muster support and effective action right across the Commission's bureaucracy. Furthermore, the Commission's ongoing activities have been subject to criticism for perceived inadequacies, for example, on the part of the European Parliament, which wishes to see further action on proposals such as the draft directive on equal treatment of the sexes in sectors other than employment and the assessment of policies such as those on childcare and the reconciliation of family life and employment.14 There are swathes of EU institutional activity, moreover, into which mainstreaming has yet to enter. The Court of Justice, for example, has not been touched by the mainstreaming agenda (Shaw 2001).
5 Case C-409/95 Marschall v. Land Nordrhein-Westfalen  ECR I-6363 at paras. 29-30.
6 Case C-450/93 Kalanke v. Land Bremen  ECR I-3051.
7 Commission Communication, Towards a Community Framework Strategy on Gender Equality (2001-05), (COM(2000) 335 of 7 June 2000). Clearly evident in the Fourth Equal Opportunities Action Programme of 1996-2000, it is said that gender mainstreaming can be traced right back to the Third Action Programme of 1991-1995: Beveridge et al, 2000b.
8 Council Decision (EC) No 51/2001 of 20 December 2000 establishing a programme relating to the Community framework strategy on gender equality (2001-05), OJ 2001 L17/22.
9 Commission Decision 2000/407/EC of 19 June 2000 to improve gender balance in committees and expert groups (OJ 2000 L154/34) accompanied by a Commission Communication (OJ 2000 C203/4).
10 European Commission, Framework Strategy on Gender Equality:Work Programme for 2001, COM(2001) 119.
11 European Commission,Equal Opportunities for Women and Men in the European Union. Annual Report 1999, COM(2000) 123; European Commission, Equal Opportunities for Women and Men in the European Union. Annual Report 2000, COM(2001) 179.
12 Emancipatie Effect Rapportage.
13 Council of Europe (1998), Gender Mainstreaming: Conceptual Framework, Methodology and Presentation of Good Practices, Final Report of the Activities of the Group of Specialists on Mainstreaming (EG-S-MS (98) 2), Strasbourg: Council of Europe, at p15.
14 Report on the Commission Communication `Framework Strategy on Gender Equality - Work programme for 2001', Committee on Women's Rights and Equal Opportunities, A5-224/2001, Rapporteur: Figueiredo.