Implementing Intersectionality in Public Policies: Key Factors in the Madrid City Council, Spain

Although feminist scholarship has discussed intersectionality extensively, few studies have addressed its implementation in public policies. This article fills that gap with an empirical study of the obstacles and enabling factors in implementing intersectionality in the Madrid City Council. We focus on the multiple meanings, actors, and structures involved in translating policy planning into concrete measures. Through a content analysis of policy documents, interviews, focus groups, and participant observation, our qualitative study identifies five key factors that hinder the implementation of intersectionality-informed policies: the absence of a legal framework and precise guidelines; the multiple and contrasting interpretations of intersectionality; the lack of training; the compartmentalized work structure andculture; andthe unavailability and misuse of data. These findings contribute to the scholarship on the implementation of intersectionality in public policies and provide empirical-based recommendations to overcome the identified obstacles.

"intersectionality" to indicate that racial and gender discrimination intersect and generate specific disadvantages for African American women. This concept allowed her to explain that the discrimination faced by African American women differs from gender discrimination against White women and from racial discrimination against African American men (Crenshaw 1989(Crenshaw , 1991 Today, intersectionality is a growing multidisciplinary field of research that explores the interactions among systemic forms of discrimination. It enables an understanding of how the interaction of multiple social structures produces simultaneous and co-constitutive inequalities (Cho, Crenshaw, and McCall 2013). Intersectionality draws attention to the power relations at the center of such complex social dynamics, which generate both privileges and processes of marginalization, depending on the contexts and groups affected (Hankivsky and Cormier 2011;Hill Collins 1990;Winker and Degele 2011). It is a conceptual tool of analysis but also a social justice framework aiming to dismantle the negative effects arising from the interaction of multiple forms of discrimination (Hill Collins 1990).
Intersectionality challenges single-issue approaches that address multiple discrimination as constituted by the sum of separable factors. It sheds light on how gender, race and ethnicity, class, religion, country of origin, sexual orientation, disability, and age interact and diversify individual and collective experiences of discrimination (Crenshaw 2000). For example, the interconnected experience of sexism, xenophobia, and class stratification affects the majority of migrant women in Spain in a unique and disproportionate way compared with Spanish women and migrant men (La Barbera 2013). To guarantee effective protection against all forms of discrimination, law and public policies need to consider such interactions.
For the past two decades, international human rights law has recognized that gender structures work interconnectedly with other social structures, generating a distinctively negative impact on women. For instance, the Committee on the Elimination of All Forms of Discrimination Against Women argued that Certain groups of women, in addition to suffering from discrimination directed against them as women, may also suffer from multiple forms of discrimination based on additional grounds such as race, ethnic or religious identity, disability, age, class, caste or other factors. Such discrimination may affect these groups of women primarily, or to a different degree or in different ways than men. States parties may need to take specific temporary special measures to eliminate such multiple forms of discrimination against women and its compounded negative impact on them. 1 Disregarding international recommendations and academic research, national legislations continue to address discrimination through siloed or additive approaches. In the absence of binding national legal norms, public policies generally obviate the interactions among multiple inequalities. Yet, some public institutions have recently sought to address the interconnected experiences of social inequalities through public policy, and an incipient literature is analyzing these efforts (Coll-Planas and Solà-Morales 2019; Hankivsky and Cormier 2011; Hankivsky and Jordan-Zachery 2019; La Barbera and Cruells López 2019; La Barbera et al. 2020). Our study aims to advance this emerging field of research.
In Spain, some local institutions, such as the municipalities of Madrid, Barcelona, and Terrassa and the province of Biscay, have introduced intersectionality into their actions. We focus here on the Madrid City Council's Human Rights Strategic Plan, which included human rights, gender, and intersectionality as interconnected, crosscutting principles of municipal planning (Madrid City Council 2017). Madrid, like all municipalities in Spain, is responsible for protecting and guaranteeing several human rights, including security in public space, equal access to adequate housing for all city residents, and equal access to cultural and sports activities (Articles 25 and 26, 7/1985 Act). 2 It also participates in the management of primary health care and offers services related to sexual health, pregnancy, nutrition, and psychological support. Municipalities have a key role in providing social services to all residents without discrimination, including migrant populations in irregular administrative situations. 3 In all these areas, where multiple and interconnected forms of discrimination are especially relevant, the introduction of intersectionality as a crosscutting principle of policy planning can make a difference in the protection and guarantee of city residents' right to equality and nondiscrimination. The Madrid Human Rights Strategic Plan foresees, among other things, intersectionality-informed measures such as launching an ethical committee to reduce the bias of the municipal police; promoting accessible and inclusive cultural, sports, and leisure events for people with disabilities, particularly children and teens; and encouraging the activity of civic organizations, especially those of women and discriminated groups (Madrid City Council 2017, 32, 55, 38).
In our study of this pioneering case, we address the following research questions: What factors determine the implementation of intersectionality in public policies? How does the technical staff translate intersectionalityinformed planning into praxis? Our aim is to investigate the challenges that the technical staff of Madrid City Council faces when translating policy planning into concrete measures and to identify the factors that might favor integrating intersectionality in the post-adoption phase. Examining the technical staff's discourses on such a process, we contribute evidence-based knowledge to the academic debate and provide recommendations for public institutions seeking to address interconnected forms of discrimination in the current context of growing inequality.
In the following sections, we first outline the theoretical framework and the methodological design of our study. We then analyze the results of our fieldwork and identify five key factors that determine the implementation of intersectionality in the pioneering case of Madrid: the existence of a legal framework and precise guidelines; the consistency of the interpretation of intersectionality; the availability of training, which works as the glue that holds the other implementation factors together; the existence of coordination mechanisms and a collaborative work culture; and the availability and use of disaggregated data. We argue that they work as hindering and enabling factors: their presence can facilitate the implementation of intersectionality-informed policies, thus accelerating social change towards equality, but their absence can impede policy implementation and the intended social transformation (La Lombardo, Meier, and Verloo 2017;Lombardo and Mergaert 2013;Mergaert and Lombardo 2014). While these factors are common to the adoption and implementation of other crosscutting policy approaches, our study shows their functioning in the case of intersectionality implementation.
Informed by our case study, we gather our conclusions and provide empiricalbased recommendations for public institutions. Intersectionality requires mandatory training for all the personnel with applied and specialized content, as well as disaggregated data on discriminatory social structures other than gender and age, such as ethnicity, religion, and sexual orientation. Relying on international recommendations and clarifying operative definitions can help overcome the specific problems related to the lack of a national legal framework and the confusion around a theoretically debated but scarcely applied concept. Finally, promoting intra-and inter-institutional coordination is a key structural change to implement an approach that requires multilevel and intersectoral policy planning.

Implementing Intersectionality in Public Policies: A Theoretical Framework
Our theoretical framework is interdisciplinary. We draw upon legal analyses of the benefits and problems that emerge when introducing intersectionality in the legal framework and praxis (Burri and Schiek 2009;Goldberg 2009;Hannett 2003;Jubany, Güell, and Davis 2011;La Barbera and Cruells López 2019;Radacic 2008;Satterthwaite 2005), as well as public policy studies examining constraints in the post-adoption phase (Bardach 1977;Derthick 1972;Pressman and Wildavsky 1973). We engage with the emerging body of literature focusing on the implementation of intersectionality (Christoffersen 2021;Coll-Planas and Solà-Morales 2019;Hankivsky and Cormier 2011;Hankivsky and Jordan-Zachery 2019;La Barbera et al. 2020). Since intersectionality, like gender, is a crosscutting and contested approach that aims to transform social structures, we rely on gender and politics studies to identify the factors that enable or hinder the adoption and implementation of gender mainstreaming (Lombardo et al. 2009;Mergaert and Lombardo 2014;Verloo 2001Verloo , 2007. We build upon these theoretical hints to analyze the post-adoption stage of intersectionality-informed policies, not only revealing connections with the existing knowledge on gender equality policies, but also identifying the distinctive effect of well-known implementation factors when putting intersectionality into practice. Studies on public policy show that policy implementation is more complex than it may appear. It is a process that comprises several stages starting from the adoption of the policy to the final outcomes for targeted individuals and collectives (Bleidt et al. 1985). While connected to the previous phases of decision-making and adoption and the following phases of monitoring and evaluation, implementation needs to be studied as a distinct phase of the policy cycle. Our study focuses on the implementation process and, specifically, on the translation of policy planning into specific measures. This translation is influenced by a series of institutional dynamics related to the allocation of resources, the involvement of key actors, and the meanings that the personnel participating in the implementation give to the concept of policy planning (May 2012).
Despite often being framed as a mere technical activity, implementation is a constitutive part of the political action (Anderson 2003). Pioneering studies in the 1970s demonstrated the political dimension of policy implementation and showed the complex dynamics of resistance, contestation, and negotiation among multiple actors and across different government levels that take place when putting policies into practice (Bardach 1977;Derthick 1972;Pressman and Wildavsky 1973). Specialized literature on gender equality policies also indicates that "implementation is a battle for power" (Engeli and Mazur 2018, 116) in which policies continue to be negotiated, contested, and even resisted. Public institutions are neither harmonious nor coherent organizations. They are open spaces for-overt or covert-confrontation and cooperation among different visions, interests, and objectives related to equality and nondiscrimination (La Lombardo 2017, 2019;Mergaert and Lombardo 2014). Understanding such overt or covert dynamics, which include the undermining of implementation arrangements, the erosion of accountability mechanisms, and the dismantling of equality strategies, is particularly relevant in times of antiequality politics (Graff and Korolczuk 2021;Roggeband and Krizsán 2020).
In this article, we explore the contested and disputed dimension of intersectionality as a policy concept (Bacchi 2017;Verloo 2007), including the divergent outcomes that result from disputes and negotiation (La Barbera and Lombardo 2019). Since implementation is a contested phase in policy making (Bardach 1977) and intersectionality is a disputed policy concept (Christoffersen 2021; Hankinviski and Cormier 2011), a discursive politics approach is well suited to examining intersectionality implementation. Our analysis of the multiple and contrasting meanings of intersectionality in practice also fills a gap in implementation studies, which have largely neglected the role of discourses in this policy stage (Beland and Ridde 2016).

Key Factors Determining Intersectionality Implementation
For the purpose of our study, we identify five key factors: the legal framework; the staff's interpretation of the policy concept; the offer of training; the institutional organization and work culture; and the availability and management of data. While the aforementioned elements act as a fundamental guide to approach the analysis of policy implementation, a cause-and-effect relation should not be taken for granted. Their relevance in the process of implementation depends on other variables, such as the context and the institutions involved and the stage of implementation analyzed (e.g., the translation of policy planning into measures or the impact of those measures on targeted populations).
The existence of a legal framework and precise guidelines constitutes a first key factor to put intersectionality into practice (Burri and Schiek 2009;Hannett 2003;Radacic 2008). Scholars discuss the impact of explicitly mentioning the notion of intersectionality in legal texts (Lombardo and Rolandsen Agustín 2012). While introducing an explicit mention in legislation might make implementation more feasible (Moon 2011), the reference to intersectionality is neither necessary nor sufficient (Hannet 2003;Verloo et al. 2012). The lack of a specific mention or definition in the legislation is not an obstacle per se for protecting against intersecting forms of discrimination (Burri and Schiek 2009) because antidiscrimination law already contains the elements that enable law practitioners to incorporate an intersectional perspective (Satterthwaite 2005). However, the predominant interpretation of equality and nondiscrimination in each context determines this possibility. For instance, mainstream legal studies in Spain define legal norms as neutral because they are general and abstract (La Barbera and Lombardo 2017;La Barbera and Cruells López 2019). The curricula of Spanish law schools do not include the work of critical legal scholars, who since the 1970s have argued that general and abstract norms produce a differentiated impact on individuals according to their social position (Unger 1983). The tendency of Spanish legal practitioners to use an outdated concept of formal equality represents a considerable obstacle for the incorporation of intersectionality (La , which instead requires a substantive approach to equality (Iyer 1993;McIntyre 2009;Razack 1998;Young 2010).
The debate on the characteristics of the legislative framework reveals the importance of a second key factor that determines the implementation of intersectionality: the interpretations of legal and policy concepts. New legal and policy concepts do not operate in a vacuum. Rather, public institution personnel interpret and operationalize new concepts in relation to already established frames (Goldberg 2009;Jubany, Güell, and Davis 2011;Lombardo and Mergaert 2013;Satterthwaite 2005). As a consequence, the meanings attributed to intersectionality tend to relate to established concepts in equality and nondiscrimination law and policies, such as gender equality and diversity (Christoffersen 2021;Hankivsky and Cormier 2011;. However, this relation is not neutral. For instance, gender-expert staff members are cautious of intersectionality because they perceive a risk of diluting efforts against gender discrimination (La Barbera et al. 2020;Verloo 2007). Moreover, intersectionality is often considered theoretically complex, ambiguous, and difficult to translate into practice (Hankivsky and  The contrasting meanings attributed to intersectionality reveal that neither legal frameworks nor policy planning come ready to operate by themselves, but rather require precise guidelines and previous models to enable a successful implementation (Coll-Planas and Solà-Morales 2019; CRIAW 2006; Hankivsky and Cormier 2011). Recent research on public policy shows that understanding the meaning of intersectionality and its social implications is a fundamental factor for a successful implementation (Hankivsky and Cormier 2011;La Barbera et al. 2020). The recurrent requests for detailed guidance, procedures, and checklists by the personnel in charge of implementing novel policy concepts (Verloo 2000) show that consensus around the meaning is a necessary step to develop appropriate implementation tools. Although no general parameter can be used for a successful implementation in all cases (Crenshaw 2011), the current lack of shared methodological tools for intersectionality-informed policies negatively affects a large-scale implementation (Hankivsky and Cormier 2011).
Because of the importance of a shared understanding of intersectionality and applied methodologies, training represents a third key factor in implementation. Training courses can enhance the staff members' sensitivity to consider inequalities as structural rather than individual issues and provide them with knowledge and tools to tackle the substantive and interconnected dimension of equality (Cavaghan 2017;Lombardo, Meier, and Verloo 2009;Verloo 2001). In addition, training increases the interpretative flexibility to identify cases of intersecting discriminations and fosters creativity in the use of existing tools (Hannett 2003;Radacic 2008;Satterthwaite 2005;Young 2010). It also contributes to neutralizing practitioners' racial, classist, and gender prejudices that bias interpretations and even nullify the efficacy of intersectionality-informed legislation and policy planning (Williams 2009). Previous studies on gender equality implementation highlight that training needs to permeate the entire structure of the institution-otherwise, new policy approaches remain at the planning stage or are reduced to anecdotal actions (Espinosa-Fajardo 2016;Meier 2006). Training is especially needed for legal services in contexts in which intersectionality is perceived as conflicting with the dominant formal approach toward equality and nondiscrimination (La Barbera and Cruells López 2019;Satterthwaite 2005;Williams 2009). Furthermore, since it fosters open debate, negotiation, and collective construction of disputed concepts (Ferree 2009; La Barbera and Lombardo 2017;Lombardo, Meier, and Verloo 2009), training affects the interpretation of new policy concepts, the second key factor explored in this study.
Training can also foster staff's commitment and thus function as an engine for seeking collaborative solutions and eventually reaching the required changes in the organization and work culture, a fourth key factor explored in this study (Espinosa-Fajardo 2018; Hankivsky and Jordan-Zachery 2019). Because tackling the interconnection of social inequalities requires crosscutting approaches, the institutional organization and culture also play fundamental roles (Engeli and Mazur 2018;Espinosa-Fajardo 2018;Lombardo, Meier, and Verloo 2017;Mukhopadhyay, Steehouwer, and Wong 2006;Rao, Stuart, and Kelleher 1999). Regarding the implementation of crosscutting policy approaches, gender and politics studies reveal the importance of creating specialized structures, allocating sufficient financial and human resources, and establishing intra-and interinstitutional coordination channels for intersectoral policy planning (Engeli and Mazur 2018;Espinosa-Fajardo 2018). However, the need for this transformation of institutional organization and work culture clashes with a generalized reductionism and incrementalism of public policies and the tendency of public institutions to demand rapid, short-term, and low-budget policies, which have little impact on reducing structural inequalities (Manuel 2006).
Finally, studies of equality policy implementation and international human rights organizations emphasize the importance of disaggregated data (Hankivsky and Cormier 2011). Since the Beijing Fourth World Conference on Women (1995) set the objective to "generate and disseminate gender-disaggregated data and information for planning and evaluation," 4 full breakdown of statistics is considered essential to plan equality policies tailored to the specific needs of women and men (Beveridge, Nott, and Stephen 2000). The United Nations Committee on the Elimination of Racial Discrimination explains that to collect "data which have been categorized by race or ethnic origin, and which are then disaggregated by gender within those racial or ethnic groups, will allow the States parties and the Committee to identify, compare and take steps to remedy forms of racial discrimination against women that may otherwise go unnoticed and unaddressed." 5 The Human Rights Office of the United Nations, within the framework of the Sustainable Development Goals, asks states to publish data in a format that allows identifying and analyzing multiple and intersecting discriminations, while ensuring the protection of individuals' privacy against data security threats at every stage of data management (United Nations 2018).
Disaggregated data allow making visible the outcomes of intersecting vulnerabilities-for example, racialized and gendered discrimination in the access to rights and freedoms-and planning measures to counteract the unbalance in the allocation of resources. Yet, the collection of disaggregated data is a necessary but not sufficient condition for identifying structural inequalities (Cavaghan 2020;Espinosa-Fajardo 2016). Quantitative data analyses risk replicating knowledge hierarchies and crystallizing categories (Cavaghan 2020, 688). Addressing the reproduction of intersectional discriminations also requires qualitative studies of lived experiences.

Methodology
Our case study is the Madrid City Council, a local institution that declared Madrid a "human rights city" in 2000. 6 Our unit of analysis is the Madrid Human Rights Strategic Plan (2017-2019). The plan was adopted under a coalition government between the citizen platform Ahora Madrid (Now Madrid)-linked to the Podemos party-and the Partido Socialista Obrero Español (Spanish Socialist Party) that lasted for four years (2015-2019). It was based on three mandates established in international human rights law: recognizing the local competences in the field of human rights, respecting the obligation to protect and guarantee gender equality and nondiscrimination, and including intersectionality in equality legislation and policymaking (Madrid City Council 2017).
In order to comply with international obligations assumed when Spain signed international human rights treaties, 7 the Madrid Human Rights Strategic Plan adopts a three-pronged approach that combines human rights, gender, and intersectionality and identifies human rights to be protected and guaranteed in the municipal territory within the framework of its competencies. For example, the Security and Emergencies Department is competent in the area of public health; the Urban Development Department is responsible for housing; the Department of Culture, Tourism, and Sports is in charge of cultural activities; and the Social Welfare Department provides social services such as municipal nurseries, day centers, and elderly home assistance. The plan focuses on the rights of women, the LGTBI collective, people from ethnic minorities, people on the move, and people with disabilities to a life free from discrimination and violence across all ages. Through the participation of City Council department heads, specialized technical staff, and civil society, it envisions the approval of municipal decrees, protocols, and the design of specific measures to be executed by the different departments and municipal autonomous bodies (Madrid City Council 2017).
We conducted our study from mid-2018 to mid-2020. Our research methods include documentary analysis (5 in total); participant observation at institutional events (13 in total); 25 in-depth, semistructured interviews with City Council technical staff and 3 with external specialists (28 in total); and 1 focus group with gender-expert technical staff. The documentary analysis of the main policy plans that incorporate human rights, gender, and intersectionality allowed us to understand the broad strategy of the City Council to put intersectionality into practice, as well as to identify the departments and key stakeholders involved. 8 The participant observation took place in a variety of institutional events, including public meetings with the participation of the municipal staff; the training courses offered by the City Council on the "human rights, gender, and intersectionality" approach; and the sessions of the Madrid Human Rights Forum, which the following municipal government terminated.
With the aim of breaking the traditional division between the research subject and the object of study (Hesse-Biber 2013), we involved stakeholders through a participatory research methodology. We conducted semistructured interviews to gather the meanings that the technical staff attributes to the concept of intersectionality and their reflection on the process of implementation. Interviews were structured in two parts. In the first part, we asked about the staff's understanding of intersectionality and the degree of awareness about their responsibility in its implementation. In the second part, we asked about the staff's perception of the obstacles and enabling factors to implement intersectionality.
Finally, through the focus group and participant observation, we explored the interactive dimension of discourse construction (Hesse-Biber 2013). The focus group with the gender-expert technical staff was tailored to reveal the construction of meanings attached to intersectionality in relation to other established policy concepts such as gender, diversity, and multiple discrimination. We asked staff members to identify the intersection among different axes of inequality in the measures adopted by the municipality. Calling upon their expertise, we also asked group participants to assess the challenges and chances of simultaneously addressing inequalities through policy action.
After obtaining participants' informed consent, we digitally recorded the interviews, focus group, and some of the events at which we conducted participant observation. 9 We used the Atlas.ti software to extract, compare, segment, and reassemble fragments of the transcribed material. We carried out content analysis of the material in two stages. First, we assigned transcribed text to small groups (two or three researchers) to identify close coding and suggest open coding. Then, all the members of the research group 10 discussed the close and open coding. We used concept-driven close coding because of our predetermined focus on both the key factors of implementation identified in the literature and the relationship of intersectionality with established concepts and approaches of equality law and policymaking. We organized the close coding around three concepts-"gender," "diversity," and "intersectionality"-and developed a codebook of proxies to reveal the usage and associated notions for each of these concepts. This methodological design allowed us to uncover how conflicting interpretations of intersectionality determine policy implementation, functioning "like magnetic resolution images, revealing the architecture of antidiscrimination law and the various preferences that are attendant to it" (Crenshaw 2011, 228).
We combined close coding with data-driven open coding to allow new and context-related issues to emerge in the analysis (Gibbs 2018). Our open coding relates to the staff's perceptions of obstacles and enabling factors in the implementation process. We used the codes "difficulties," "implemented measures," "resistance," and "opportunities." This analysis allowed us to disclose explicit and implicit meanings not only to make visible hidden norms and interpretations, but also to connect them to the discursive and material factors that determine implementation.

Key Factors Hindering the Implementation of Intersectionality
The Absence of a Legal Framework and Precise Guidelines The Madrid Human Rights Strategic Plan draws on international recommendations to introduce intersectionality as a principle for municipal policy planning. The plan focuses on the structural nature of discrimination and aims to overcome essentialist-based perspectives through intersectionality-informed actions (Source 22C19). It represents a qualitative leap to legitimize intersectionality in the praxis of the local institution (Source 05F19) and offers a crucial opportunity to guaranteeing equality and nondiscrimination. 11 However, the lack of national legal provisions is an obstacle to introducing intersectionality in the municipal regulation. While the national legislation mandates that public institutions analyze the gender impact of each piece of regulation, 12 no legal provision requires Spanish institutions to include intersectionality in bylaws, social clauses, protocols, or regulatory impact analyses. 13 In the absence of national legal provisions, the municipal legal services staff is not bound to analyze the differential impact of municipal bylaws on the basis of ethnicity, class, religion, country of origin, sexual orientation, disability, and age (Sources 08D19, 27B20). On the contrary, to consider these intersections in the elaboration of local regulation would require a differential treatment that, staff members argue, would constitute a form of discrimination (Sources 08D19, 27B20). They interpret intersectionality as a tool to deal with individual circumstances that are too specific to be included in general and abstract legal norms (Source 27B20). This position reflects a formalistic understanding of legal norms as neutral instruments "to be applied rather than interpreted" (Source 27B20).
The lack of precise guidelines for the application of the Madrid Human Rights Strategic Plan is another main challenge, especially for the legal services. The Strategic Plan foresees the application of the human rights, gender, and intersectionality approach in a series of new municipal regulations. 14 However, the interviewees manifested difficulties when required to include intersectionality in the elaboration of such bylaws (Sources 13E19, 05F19, 28D20).
What legal services do is to check that all proposals comply with existing legal provisions…. We have not received any concrete instruction to translate it [intersectionality] into practice. (Source 08D19) In line with previous studies (Coll-Planas and Solà-Morales 2019; CRIAW 2006; Hankivsky and Cormier 2011), we identify the lack of guidelines and applied models as a fundamental obstacle for implementing intersectionality (Sources 05D19, 08D19, 05F19, 01G19) because it generates resistance among the technical staff responsible for translating the policy concept into specific measures (Sources 13E19, 28D20). Participants in this study argue that the guidelines followed for the implementation of gender equality in the regulatory framework-such as directives for gender-inclusive language, the inclusion of gender equality in the social responsibility clauses of public administrations, 15 and the gender impact report-could serve as models for intersectionality (Source 08D19). For instance, the Regulatory Impact Analysis Report that all public administrations in Spain are required to elaborate before approving a legal norm 16 has generated a new work culture that facilitates the critical assessment of municipal activities from a gender perspective (Sources 22C19, 05D19, 05F19, 01G19, 27B20). With proper training, expanding the impact analysis to the intersections of other axes of discrimination beyond gender could enable the implementation of intersectionality.
Our study shows that policy planning can rely on international human rights law to call for action at the corresponding level of government despite the absence of national legal provisions. Yet, the obstacles encountered by the legal services staff of the Madrid City Council when translating intersectionality into praxis suggest that the existence of an explicit policy plan is a necessary but not sufficient condition for successful implementation. In the absence of a binding legal framework, policy planning must go hand in hand with the elaboration of implementation guidelines. Guidelines, including operational definitions and examples of applied intersectionality, are needed to translate key policy concepts into concrete measures tailored to the specific structure and competences of each department.

The Multiple and Contrasting Interpretations of Intersectionality
Participants emphasize the recentness of intersectionality in Madrid public policies (Sources 19B19, 16E19, 04F19) because it was introduced in 2017. Knowledge about the intersectional approach is uneven and varies across departments, activities performed, and levels of expertise on equality and nondiscrimination policies. While some departments lack information about the term "intersectionality" (Source 08D19), the gender-expert staff shows profound knowledge of intersectionality, although not without tensions or contradictions. The novelty of the approach and its disputed interpretations are key challenges to implement intersectionality.
Reflecting tendencies found in the literature (Christoffersen 2021;Hankivsky and Cormier 2011;, the Madrid municipal staff interprets intersectionality in relation to two main concepts of equality law and policy: gender and diversity. Most of the gender-expert staff links intersectionality to gender and understands gender mainstreaming as a preliminary step to intersectionality implementation. To tackle inequalities…, discrimination and power structures you need to understand that… gender is the pivotal factor. (Source 16E19) Thus, intersectionality is considered a tool to improve gender-sensitive municipal action. Some staff members interpret intersectionality as a way of "adding more complexity" (Source 04F19) that may put gender mainstreaming at risk (Source 01G19). As reflected in previous studies (Verloo 2007), some gender equality experts worry about the potential dilution of gender mainstreaming when intersectionality is introduced (Source 01G19).
For another group of gender-expert staff, understanding intersectionality as a subsequent-rather than a contextual and concurrent-step toward gender mainstreaming may reinforce the exclusion of people who experience intersecting forms of discrimination (Sources 19B19, 05F19). This group understands intersectionality as a tool to address gender discrimination linked to other inequalities and to design specific measures for the heterogeneous needs of women living in Madrid (Sources 19B19, 16E19, 05F19). These participants distinguish intersectionality from gender and recognize that the latter overlooks the interactions among different axes of inequality (Source 19B19). Yet, they only consider "gendered intersectionalities" (Hankivsky and Cormier 2011), disregarding that the combination of context-dependent factors of discrimination that put individuals and collectives in situations of disadvantage might not include gender.
Aligning with previous literature (Hankivsky and Cormier 2011), our study reveals that intersectionality is also conceptually and practically related to diversity. Interpreted as a tool to represent the heterogeneity of Madrid's population in terms of gender, ethnicity, class, religion, country of origin, sexual orientation, disability, and age, participants stress the importance of "taking diversity into account" (Source 04F19) when it comes to intersectionality. For instance, intersectionality-informed communication campaigns, such as posters announcing public services, diversify the image of beneficiaries and balance the equal representation of men and women, different ethnic profiles, elderly people, and people with disabilities (Sources 08D19, 04F19). Building upon such an understanding, some participants call for using "diversity" as an alternative concept, more intuitive and approachable than intersectionality (Source 24B20).
Other voices warn against identifying intersectionality with diversity because it implies adopting an additive approach and addressing multiple inequalities in a cumulative way, disregarding the intersections among different axes of inequality (Source 13B19). These interviewees recognize that equating intersectionality to represent differences in terms of gender, sexual orientation, ethnicity, religious denominations, and disability makes social diversity visible, but it does not allow grasping the interconnections among the systems that generate inequality (Sources 19B19, 05D19, 30D19).
The disputed dimension of intersectionality at a discursive level has material consequences. When intersectionality is linked to gender, other axes of inequality risk being disregarded. When, instead, the municipal staff equates it with diversity, the structural component of inequalities and their interrelations risks being lost. Depending on its association with gender or diversity, the technical staff expresses different degrees of resistance to or acceptance of intersectionality-informed policies, which affect decisions about resource allocation and organization.

The Lack of Training
The lack of knowledge and the disputed interpretations of intersectionality (Sources 08D19, 27B20) lead to questioning its relevance for municipal action or perceiving it as an added burden (Sources 14C19, 13E19, 05F19). To overcome such resistance, most of the interviewees point to training as a basic preliminary condition for successfully implementing intersectionality in municipal policies and legal instruments (Sources 13B19, 19B19, 22C19, 08D19, 16E19, 27E19, 04F19, 07F19, 01G19, 27B20, 28D20).
First you need training.… because it changes your attitude and perspective. (Source 05F19) The participants in this study consider that training not only contributes to "raise awareness" of new concepts among the personnel, but also offers theoretical and practical tools to translate principles of policy planning into specific measures of action (Sources 16E19, 27E19, 01G19).
We cannot apply an approach if we do not know what it is. Defining what it is and how we can implement it should be the first step. (Source 08D19) The need for practical instructions, ad hoc counseling, and operational definitions of intersectionality makes specialized training tailored to the different departments a fundamental factor for a successful implementation (Sources 19B19, 08D19). A good example in this direction is the training program offered by the municipality of Terrassa in 2018-19, which-counting on the active participation of the technical staff and the collaboration of specialized scholars-has been an opportunity to share critical reflections, guidelines, and good practices across European institutions. 17 Although the Madrid City Council offered training on "human rights, gender, and intersectionality" after the adoption of the Strategic Plan, this was limited in time, voluntary in nature, and counted with an uneven participation across departments (Sources 22C19, 16E19). Moreover, the content was general and introductory (Source 27E19) and did not satisfy the need for guidelines and applied models for operationalization.
A specialized and practice-oriented training would have contributed to fostering another crucial element for a successful implementation of intersectionality: the technical staff's sensitivity and commitment. Previous research has shown that the staff's commitment can function as an engine for changing perspectives, counteracting resistances, and facilitating policy implementation (Espinosa-Fajardo 2018; Hankivsky and Jordan-Zachery 2019). The staff's commitment can motivate them to surmount obstacles encountered on the ground and seek appropriate tools for solving practical problems (Sources 19B19, 04F19, 05F19, 01G19). For instance, because of their sensitivity to inequalities, the staff of the Social Welfare Department-in charge of social services, migrant population inclusion, social education, gender equality, and promotion of diversity-is keen to find ways to include intersectionality into their daily work. This sensitivity, combined with the staff's expert knowledge on equality and nondiscrimination policies, make them confident in the possibility of developing specific measures for that purpose (Source 19B19).
However, the preeminent role of specialized departments or decision makers entails a limitation for the implementation of intersectionality. Similar to the implementation of gender equality policies (Espinosa-Fajardo 2016;Meier 2006), if the intersectionality approach does not permeate the entire structure, horizontally and vertically, it is doomed to remain at the planning stage or to be reduced to occasional and uncoordinated actions (Source 07F19). While the commitment of decision makers and specialized departments facilitates the implementation of new policy approaches (Source 01G19), their instructions are void without the involvement of all the technical staff (Source 13E19).

If department directors do not focus on it [intersectionality]
, the Plan is useless. However, you also have to train the paper-pushers because they can detect a lot of things since they are the ones who truly deal with people. (Source 08D19) To ensure successful implementation of intersectionality, not only the technical staff but also the directors, the managers of different municipal services, and the personnel who interact with the citizenry need training (Source 04F19). Providing mandatory training on intersectionality across the entire municipal structure might also ensure its sustainability, counteracting the negative effect of changes in work teams and government (Sources 01G19, 13E19, 16E19, 28D20).
Training should transcend the logic of "applying a coat of varnish" by attending a one-day workshop and ticking a box (Source 05F19). It needs to promote a transformation of the compartmentalized work culture by encouraging collaborative dynamics and coordination among municipal departments and services (Sources 19B19, 06E19, 04F19, 05F19, 01G19). However, such a transformation requires time, reflection, and willingness to question established ideas and processes (Sources 16E19, 05F19, 27B20). Lack of time for such a critical reflection is a fundamental challenge to put intersectionality into practice (Sources 16E19, 05F19).
Training should be tailored to the different services and practical needs of each department (Source 28D20). For instance, in line with previous studies (La Barbera and Cruells López 2019;Satterthwaite 2005;Williams 2009), our analysis reveals that specific training is needed for the legal personnel, who otherwise do not consider themselves accountable for the implementation of intersectionality. Despite its origin as a category of legal analysis, legal services tend to understand intersectionality as an issue that can be relevant for policy planning but not for regulation (Sources 08D19, 27B20, 06C20).
Our findings show that mandatory and specialized training for all personnel of a public institution is fundamental for enabling the implementation of a novel and disputed policy concept like intersectionality. By promoting the understanding of the conceptual and practical significance of intersectionality, training can mitigate the staff's resistance and prompt their sensitivity and commitment. To provide such specialized training, public institutions should rely on the knowledge of scholars from different fields, the expertise of civil society organizations and grassroots associations working in different areas, as well as the good practices of other public institutions that apply intersectionality in their action.

The Compartmentalized Organization and Work Culture
Adopting intersectionality in public policy shifts the attention from single-issue approaches to the interconnection of social and institutional dynamics that produce inequalities. This multidimensional focus not only requires specialized structures of intra-institutional coordination among siloed departments but also calls for dismantling the compartmentalized, single-issue oriented work culture that dominates most public institutions, including the Madrid City Council (Sources 19B19, 22C19, 08D19, 13E19, 27E19, 04F19, 05F19 07F19, 01G19, 27B20).
One problem is the architecture of the institutions that organize work into 'watertight compartments'…. We also lack a [coordination-oriented] work culture…. The administration is like a house made of rooms, where people are stuck, never leaving their room nor thinking about creating a new space for meetings. (Source 19B19) The lack of effective internal communication about the work done by other departments leads to the design of parallel and redundant action plans (Source 07F19) and precludes the stable collaboration that crosscutting approaches require (Sources 19B19, 22C19, 05D19, 30D19, 16E19, 13E19, 04F19, 07F19, 01G19). In the absence of such mechanisms and structures, the occasional participation of staff from other departments generates mistrust and is perceived as an intrusion (Sources 19B19, 30D19, 01G19).
The case of the Madrid City Council suggests a series of strategies to promote changes in the compartmentalized organization and work culture. Regarding the importance of stable structures for intra-institutional coordination, participants mentioned the recent creation of the Gender Equality Unit. 18 This unit includes 44 equality agents and a gender mainstreaming department that elaborates strategic planning and facilitates coordination among all municipal departments and districts. With adequate training, such a coordination structure could become an engine for implementing intersectionality-informed policies (Sources 19B19, 16E19, 04F19, 27B20), granting them long-term sustainability regardless of staff and government changes (Sources 22C19, 27E19, 04F19, 01G19).
In addition, the territorial organization that relies on city districts 19 and collaborations with civil society (Source 07F19) exemplifies how multilevel structures facilitate the implementation of intersectionality. For instance, the Madrid City Council counts on several district-based projects that implement an intersectional and human rights approach, such as the Women Leaders Project. Considering the intersection of age and gender as a source of structural inequality, this project aims to increase the participation of older women in the Municipal Centers for the Elderly (Source 01G19). The proximity to the citizenry enables associations working in the districts to consider context-dependent intersectional inequalities and tailor the measures to the specific needs of the targeted population.
Inter-institutional coordination also emerges as a challenge that public institutions must confront to enable the implementation of intersectionality (Source 16E19). Our findings indicate that the different levels of governmentnamely, national, regional, and municipal levels-are poorly coordinated (Sources 19B19, 22C19, 05D19, 13E19, 27E19, 05F19, 07F19, 01G19). This puts local institutions in a paradoxical situation. The day-to-day nature of municipal services and their proximity to the citizenry situate municipalities in a privileged position to implement equality and nondiscrimination policies from an intersectional perspective. Yet, they need to collaborate with the other levels of government to pursue this goal (Sources 08D19, 30D19). For instance, while the Madrid City Council offers shelter for migrant women victims of gender violence, their residence permit is a national competence. Without inter-institutional coordination, it is impossible for the City Council to secure migrant women's legal status when protecting them from gender violence (Source 20A19).
The case of the Madrid City Council suggests that in specific areas of shared competences with national or regional governments, inter-institutional coordination is especially needed. Because of the multilevel structure of police corps in Spain, the benefits of inter-institutional coordination to put intersectionalityinformed actions into practice are especially manifest in the area of law enforcement. Pushing forward in this direction, the Human Rights Plan attributes to the municipal police's Unit of Diversity Management the responsibility to provide specialized assistance to victims of hate crime, especially LGTBI people, ethnic minorities, migrants, refugees, the homeless, and children and adolescents (Madrid City Council 2017, 33). Similarly, the Human Rights Plan includes the adoption of a protocol to provide adequate assistance to victims of violence on account of the type of violence and multiple discrimination that women suffer (Madrid City Council 2017, 33). As the plan highlights, the success of both initiatives depends on the effective coordination of national and local police (Madrid City Council 2017, 33).
The needed transformation of the organization and work culture is slow and difficult. The solution might lie in improving inter-institutional collaboration in matters of shared competence and repurposing existing intra-institutional crosscutting structures. However, such repurpose involves serious risks. On the one hand, relying on gender mainstreaming structures could end up postponing intersectionality indefinitely if the staff interprets intersectionality as subordinated to gender equality. On the other hand, a shared use of such structures could end up weakening both policy approaches, given the current political context of increasing attacks against gender equality (Graff and Korolczuk 2021;Roggeband and Krizsán 2020). Adequate training on the distinctive features of intersectionality and an institutional legitimation of both policy approaches could minimize the risks involved in the repurposing of existing gender equality structures.

The Unavailability and Misuse of Disaggregated Data
Another factor affecting the implementation of intersectionality is the availability and use of disaggregated data. The interviewees consider the availability of disaggregated data and the capacity of data crossing two fundamental steps for identifying intersecting inequalities (Sources 22C19, 27E19, 05F19, 01G19). They claim that existing data are insufficient to identify intersecting inequalities with statistical rigor (Source 22C19). Data collection not only on sex, age, disability, and income, but also on ethnicity, religion, country of origin, and sexual orientation, is deemed indispensable to design and implement intersectionality-informed policies (Sources 14C19, 28D20).
First of all, we need data.… Having data is like removing a veil, to begin to see [the reality] and from there to develop public policies that are more adequate to the real needs. (Source 05F19) For example, the absence of disaggregated data on ethnicity and religion obscures the inequalities suffered by Roma and Muslim women (Sources 22C19, 30D19, 28D20). Likewise, the reduced amount of variables in data collection-that excludes, for instance, possessing a residence permit or being responsible for children and other care-dependent persons (Sources 04F19, 05F19)-limits the identification of intersecting inequalities. In this regard, the Madrid Human Rights Plan contemplates some corrective measures, such as the collection of data disaggregated by gender, age, and disability to improve the promotion of accessible, inclusive and integrating cultural, sports, and leisure activities for children and teenagers with disabilities (Madrid City Council 2017, 55).
The main challenge to the collection of disaggregated data is the perceived incompatibility with data protection regulation (Sources 22C19, 30D19, 27E19, 04F19, 05F19, 01G19). Despite the United Nations' (2018) promotion of data management systems for the analysis of intersecting discriminations, the majority of the municipal staff worry about privacy breaches, echoing a widespread concern across all levels of public administration in Spain (Source 13E19). Nevertheless, good practices in Canada (British Columbia's Office of the Human Rights Commissioner 2020) and the United Kingdom (Human Rights Center Clinic 2013) demonstrate that it is possible to manage disaggregated data without violating the fundamental right to privacy.
The lack of specialization of the staff responsible for collecting and analyzing disaggregated data constitutes an additional problem. They perceive this task as an added burden, which results in resistance to intersectionality (Sources 14N18, 14C19, 30D19). Although increasing the budget allocation to human resources could be a solution, some participants argue that a successful implementation of intersectionality depends more on the reorganization and reallocation of existing resources than on major budget expenditure (Sources 13E19, 16E19).
It isn't about budgets, it's about perspectives; it's about the questions you ask yourself, the diagnosis you make, how you allocate the resources you have. You don't need a larger expenditure. (Source 22C19) Our research reveals a consensus on the misuse of already available data (Sources 08D19, 27E19, 28D20). Standardizing data collection and sharing are other changes that could rapidly improve the current use of resources at low or no cost (Sources 19B19 22C19, 08D19, 01G19). For instance, the City Council could systematically process the data that it routinely collects from municipal activities and share it across all departments. As discussed in the previous section, leveraging the territorial organization into districts could improve the knowledge of intersectional inequalities (Sources 22C19, 30D19). Thanks to the participation of civil society and the direct contact with the population, districts could collect and analyze quantitative and qualitative data on the combined effect of gender, religion, and ethnicity in the access to services, a combination of discriminatory factors that is otherwise invisible to the administration. In the long term, a digitized and standardized system would represent a significant enhancement of data management. The Municipal Data Office in Barcelona represents a good example of this type of centralized platform to improve policy diagnosis and design through the management of disaggregated data on gender, age, and territory. 20 Yet, disaggregated data management is a necessary but not sufficient condition for implementing crosscutting approaches, like gender and intersectionality (Cavaghan 2020;Espinosa-Fajardo 2016). Public institutions are called to combine quantitative and qualitative analysis, and be aware of biases in knowledge production to avoid replicating the exclusionary power dynamics that intersectionality-informed policies intend to eliminate. Both quantitative and qualitative methodologies are needed not only to elaborate more effective intersectional diagnoses and plan actions tailored to the specific population's needs, but also to increase the quality of monitoring and evaluation.

Conclusions
Using intersectionality in the pursuit of equality and nondiscrimination implies understanding human rights as interdependent, which is not only beneficial for those who are now in a situation of special vulnerability, but for all city residents.
The Madrid City Council is one of the first examples of a local institution relying on international recommendations to adopt intersectionality in policymaking. Studying how its staff negotiates a new policy concept within the existing knowledge, tools, institutional organization and work culture is fundamental to discern the specific obstacles to implement intersectionality-informed policies. The key factors identified in this study (Table 1) coincide with those found in the adoption and implementation of other crosscutting and contested approaches, such as gender equality. Yet, our study reveals specificities in the case of intersectionality implementation.
The Human Rights Strategic Plan shows that intersectionality can be included in policy action without an explicit mandate in national legislation. However, in the absence of national binding norms, intersectionality implementation faces a lack of legitimacy and theoretical and practical vagueness. The elaboration of guidelines, including operational definitions and examples of intersectionality, as well as protocols to evaluate the intersectional impact of implemented measures, can contribute to successful implementation.
The technical staff's multiple and contrasting interpretations discloses that intersectionality is an almost empty signifier (Christoffersen 2021), associated with preexisting equality and nondiscrimination frames and concepts. Making explicit those hidden interpretations, while connecting the discursive and material dimensions of the implementation process, allows us to delve into the disputed dimension of intersectionality and shows that its meaning cannot be taken for granted. Depending on its association with gender or diversity, the technical staff shows different degrees of resistance or acceptance of intersectionality. On the one hand, a specific resistance to intersectionality comes from gender equality experts, who perceive a risk of misappropriation of resources. On the other hand, intersectionality benefits from scarce ideologization so far and is welcomed when associated with the vague concept of "diversity." The great challenge for successful implementation is to find an operative formula that grasps the multiple and contrasting interpretations of intersectionality, while preserving its critical approach to power dynamics and the complexity of social phenomena.
Our research also indicates that, as in the case of any crosscutting approach, including gender mainstreaming (Engeli and Mazur 2018;Espinosa-Fajardo 2018), the compartmentalized organization and work culture of public institutions represent a structural limitation for the successful implementation of intersectionality. Channels of collaboration are needed to overcome the incrementalist fragmentation of public policies and prompt the practice of intersectionality. These might include the repurpose of existing crosscutting structures despite the risks of resource misappropriation and mishandling. Otherwise, the institutional change required to create an ad hoc structure might postpone the implementation of intersectionality indefinitely.
Resources-always limited-are one of the greatest challenges for public institutions, and data management in the case of intersectionality is not an exception. Yet, some specificities exist for intersectionality implementation. Intersectionality-informed measures require the collection, management, and analysis of an unprecedented amount of disaggregated data, which adds greater

Legal framework
International, national, and local legal binding provisions establishing intersectionality as a legal and policy criterion.
Lack of national binding legal provisions.
"What legal services do is to check that all proposals comply with existing legal provisions.… We have not received any concrete instruction to translate it [intersectionality] into practice" (Source 08D19).

Existence of international recommendations;
Specific guidelines and protocols.

Interpretations
Negotiation of meanings in relation to previous knowledge and experience.
Interpretations under different frames of reference, mainly gender and diversity.
"To tackle inequalities…, discrimination and power structures you need to understand that… gender is the pivotal factor" (Source 16E19); "[Intersectionality means to] take diversity into account" (Source 04F19).
Making explicit the disputed meanings of intersectionality.

Training
Knowledge provided to undertake institutional action.
Uneven, voluntary and limited in time.
"First you need training.
Open space for collective negotiation of meanings; Collaboration with specialized scholars, grassroots (Continued)

Institutional organization and work culture
Intra-and inter-institutional coordination and work culture.
Compartmentalized organization and work culture.
"The administration is like a house made of rooms, where people are stuck, never leaving their room nor thinking about creating a new space for meetings" (Source 19B19).
Repurpose existing gender mainstreaming machinery.

Data
Disaggregated data for adopting intersectionality -informed measures.
Lack of disaggregated data and misuse of already available data; Misinterpretation of data protection regulation.
"First of all, we need data.… Having data is like removing a veil, to begin to see [the reality] and from there to develop public policies that are more adequate to the real needs" (Source 05F19).

Sharing already available information;
Collaboration with administrative units with firsthand knowledge of the territory; Adopting mixed methodologies.
complexity compared to data disaggregation on the basis of gender. While a systematic and comprehensive data collection should be encouraged, sharing already available information would allow for intersectionality-informed policy planning at a low cost. The availability and standardization of quantitative indicators need to be combined with qualitative methodologies to assess the impact of implemented measures on people's lives. Collaboration with administrative units with firsthand knowledge of the territory is crucial to this end. Finally, intersectionality is perceived as a remarkably theoretical and conceptually complex novel idea that requires mandatory training for all the personnel. In our case, training works as the glue that holds all the key implementation factors together. First, discussing and clarifying multiple and contested interpretations of intersectionality and providing guidelines and applied models in training sessions can increase staff's commitment, reduce their perception of intersectionality as an added burden, and resolve potential conflicts with established policy concepts. Second, adequate training can enable the repurpose of existing crosscutting structures, clarifying the connections and specificities among established and new approaches. Third, training offers an intersectoral space for devising strategies to gradually transform the work culture of public administrations and to foster intra-and inter-institutional coordination. Finally, training is fundamental to learn about the combination of quantitative and qualitative data analyses. To reach such a goal, training needs to be comprehensive and targeted to every department, requiring the collaboration of specialized scholars, grassroots organizations, and other experienced institutions.
Future studies should consider that local institutions are particularly well positioned to effectively protect against intersecting discriminations, and should explore the impact that intersectionality-informed policies have in people's lives. Nowadays, this kind of research is especially needed due to increasing social inequalities. Yet, to undertake such studies, the implementation of intersectionality needs to progress further toward the translation of intersectionalityinformed planning into concrete measures.