“But We Just Need Money”: (Im)Possibilities of Co‐Producing Knowledge With Those in Vulnerable Situations

This article is based on the experience of carrying out research with young refugee women in Durban, South Africa. We reflect on the possibilities of co‐producing knowledge in a situation of widely asymmetrical power relations where the young women with whom we were interacting were located in situations of economic, legal, and social vulnerability, and when their major concern was to find money for basic survival. The premise behind our research was to produce data and knowledge that could be used to improve services for these young refugee women and to lobby for change in policies that would also improve their life situations. Our article reflects on this ambition and the possibilities of co‐producing knowledge that could improve these young women’s lives, our interactions with the young refugee women, and with the CSO that offers them support and with whom we partnered to organize our data collection. We also analyse the different positionalities of various members of the research team and how these impacted the data collection and knowledge production processes. The article aims to provide a critical assessment of the ways in which knowledge production may or may not be a liberatory practice and the conditions within which true co‐production of knowledge is possible. We ask whether it is, in fact, possible to co‐produce knowledge when working with people in vulnerable situations such as the women refugees in our project. As academics, how may we learn from our failures to try and move forward with more truly inclusive and equitable research that challenges epistemic oppression?


Introduction
"But we just need money!"These words, spoken by a young Congolese refugee woman in a restitution meeting organized in the context of our research, sum up in many ways the impossibilities of using research/knowledge production as a means of transforming inequalities that are so firmly ingrained in existing systems of power and domination.Whilst we had framed our project with these young women as producing knowledge that could be used to somehow improve their lives (albeit in a small way), we understood that the situations of vulnerability in which they found themselves were so extreme in many cases, that the only thing that would really improve their lives was a massive change in their legal and economic situation, which our research and knowledge production clearly could not provide.Under these circumstances, we asked ourselves whether or not our research had made any contribution, and whether we could have proceeded differently to have a more positive impact.As feminist researchers who are keenly aware of intersectional structures of inequality and domination, we clearly wish to pursue research that will contribute towards a reduction in these inequalities.
But we also had to recognize that various factors, including our own positionalities and the way in which research projects and funding are structured within academia, provide serious barriers to the possibilities of co-created knowledge and liberatory knowledge production.In this article, we discuss the ambitions and limits of our research and our interactions with the young refugee women participating in the project to provide a critical assessment of the ways in which knowledge production may or may not be a "liberatory practice."

The Research Project
Our research, which was funded through a joint France-South Africa research grant (we will discuss the constraints of this funding in further detail later in the article) sought to understand the experiences of young refugee women in South Africa and the ways in which they become vulnerable to gender-based violence and poor sexual and reproductive health (SRH) outcomes, as well as the barriers to access to services and their own strategies for resilience when faced with these situations of vulnerability.We start from an explicitly feminist perspective where we aim to produce "useful knowledge that will make a difference to women's lives" (Letherby, 2003, p. 4).We wanted to use feminist methodologies to expose the structures of inequalities that created situations of intersectional vulnerability for these young women, in order to make recommendations both for policymakers and for civil society organisations (CSOs) working with these young women to make changes that would improve their lives.We realized at the outset that these young women with whom we were hoping to do research and produce knowledge were in vulnerable positions, recounting experiences of violence and exploitation, and we were extremely wary of the impacts that our research might have on them.However, following Turton (1996, p. 96), we believed that "research into the suffering of others can only be justified if alleviation of that suffering is an explicit objective."Thus, whilst we realized that our research would expose the ways in which these young women were made vulnerable, and might lead to the production of knowledge concerning their experiences of violence, we kept in mind that the objective would be to create improvements for them and for other refugees, and that it was only with this understanding that we could engage ethically in the research.Indeed, ethical considerations were at the forefront of our thinking, and we were clear that we wanted to go beyond mere "procedural ethics" and a "do-no-harm" approach (Bilotta, 2020;Clark-Kazack, 2021;Jacobsen & Landau, 2003) to produce research that could actively involve partners outside of academia and challenge existing structures and hierarchies of inequality and discrimination.
The concept of vulnerability and the understanding of our different positionalities and thus differing levels of vulnerability were key to our research.We started from an understanding of vulnerability drawing on Fineman's (2010) concept of vulnerability as depending on the particularities of individual embodiment and positions "within webs of economic and institutional relationships" and "the quality and quantity of resources we possess or can command."Within this understanding, vulnerability does not attach itself to a particular pre-defined group, and is not an essential characteristic of some individuals; rather, it is produced through a range of social, economic, legal, and political structures (Reilly et al., 2022).In contrast to this, resilience is produced through the quantity and quality of social resources to which individuals have access within the societal structures where they are located.Rast et al. (2020, p. 857) discuss the concept of resilience in relation to refugees and argue that resilience cannot be disconnected from their rights and recognition in host societies and that "resilience capacity thus depends on economic as well as cultural and social resources on all societal levels."In the South African context where refugee rights are increasingly restricted, this resilience may appear hard to achieve and support (Kavuro, 2022).So we believed that young refugee women should not be labelled as vulnerable per se, and one of the objectives of our research was to understand how they could be rendered vulnerable, and how we could use our research to support their strategies for resilience.We combined this structural approach to vulnerability with a situated intersectional analysis (Yuval-Davis, 2015), which highlights the need to analyse the complex inequalities and power relations in societies without reducing these to a single social division such as gender, class, etc. Situated intersectionality, therefore, calls for particular attention to be paid to the "geographic, social and temporal locations" (Yuval-Davis, 2015) of the individual and collective actors who we seek to understand.As such it is especially relevant to the study of inequalities and vulnerabilities in contexts of migration.The adoption of a feminist intersectional perspective using Yuval-Davis' (2013) conception of situated intersectionality, we believed, would bring an in-depth understanding of the ever-shifting character of these intersectional axes of inequality, and would be combined with an analysis of mobilities to provide knowledge on dynamic matrices (Hill Collins, 1990) of domination, marginalization, agency, and resilience.A dynamic and situated intersectional approach can, we believe, serve as a "corrective to essentializing identity constructs that homogenize social categories" (Anthias, 2012, p. 107).That is to understand vulnerability as being situated and dependent on spatial, temporal, and social locations, which are dynamic and changing, rather than as based on a social category such as "woman" or "refugee."This understanding of situated intersectional vulnerability allowed us to understand better how young women refugees are made vulnerable but also to situate ourselves within the research teams and understand our different positionalities and potential positions of vulnerability.
Understanding that vulnerability depends not on pre-defined categorisations and identities but on locations within a dynamic matrix of power inequalities allowed us to think differently: We as researchers and civil society members could also, in some circumstances, be "vulnerable" and thus detach ourselves from a binary division between researchers and "subjects."It made us reflect more deeply on our different positionalities within structures of power and inequality and how these might also change throughout the research (see our discussion of positionality and power asymmetries particularly in Section 5), and which structures and inequalities had a particular impact on refugee women's positions of vulnerability, including possibly their participation in our research.

What Do We Mean By Co-Production?
One of the goals we had at the outset of the project was to engage in co-production of knowledge with young women refugees.The concept of co-production of knowledge has become popular in recent years but is subject to competing interpretations (Thomas-Hughes & McDermont, 2021).The origin of the term is usually traced to Ostrom (1996), who defines it as "the process through which inputs used to produce a good or service are contributed by individuals who are not 'in' the same organization" (p.1073).Since this initial definition though, the notion has been taken up and used in widely different circumstances.
Co-production in research has generally been developed as a response to the idea that research should be done not "about" but "with" marginalized subjects (Bell & Pahl, 2018).There is a general consensus that co-production should de-centre academia as a site of knowledge production and should acknowledge and involve those outside of academia as equally valid holders and producers of knowledge.In doing so it should create data that is more "representative of community needs" (Thomas-Hughes & McDermont, 2021, p. 292) and also greater opportunities for community engagement and capacity building.This is a particularly valuable approach when the research involves people who are generally marginalized and whose knowledge is overlooked or ignored, such as refugees, or CSOs working with refugees, for example.Co-production can be done through a wide range of research methods (and there is no particular method that is common to all those engaging in co-production) but should involve non-academic partners (CSOs, members of the community to be researched) as active research participants in horizontal partnerships (Lokot & Wake, 2021), where all participants contribute in a non-hierarchical manner to all stages of research-development, data collection, analysis, use of results.
Ideally, as Lokot and Wake (2023, p. 9) argue, co-production should mean research that "tackles unequal power dynamics, challenges existing knowledge production hierarchies, ensures more equal partnerships and shared decision making, emphasises reciprocity, promotes mutual capacity strengthening, ensures greater reflexivity and enables flexible ways of interacting and working across the research cycle."However, co-production is not without its own ethical challenges and should not be used as a way of masking or contributing to the maintenance of existing inequalities and systems of domination.Relations of power are often highly entrenched and also normalized/invisibilised, making them difficult to escape.Turnhout et al. (2020) argue that a dynamic of "depoliticization" in co-produced research can, in fact, lead to a reinforcement rather than a mitigation or reduction of existing power inequalities.Pincock and Bakunzi (2021) point to the specific problems in carrying out co-production and participative research with refugees, arguing that the power relationships between those refugees included in the research (peer researchers) and others are often not addressed.Whilst presenting itself as fully inclusive, co-production can mask real power inequalities leading to symbolic inclusion (Larruina & Ghorashi, 2020).In other words, using token inclusion of those from marginalized groups can occlude the ways in which these groups become marginalized.
The multiplicity of understandings of co-production means that there are many ways of conceiving and putting into practice co-production in research and no common understanding of what co-production should look like in practice (Facer & Enright, 2016;Thomas-Hughes & McDermont, 2021).One of the ways that can be envisaged for engaging in co-production is to create a team of researchers from both within and outside of academia.As Weiss (2016) argues, these "mixed" research teams may be seen as a "panacea for the ethical challenge of ethnography," but may also bring their own challenges.Given the marginalized and vulnerable positions of the young refugee women with whom we were seeking to work, we imagined that it would be very difficult, and might indeed create harm, to try and engage directly with them; intermediary organisations, which work daily with refugees and thus may understand their positions better than researchers from academic institutions, was vital.However, there are also power asymmetries involved between researchers and civil society activists as we discuss further.Our civil society partner pointed, for example, to the fact that she was constantly solicited by researchers for a "gatekeeper letter," which is a requirement for South African university ethics committees before research can start.She explained that this put her in an awkward position, not wanting to block research projects (and especially those of young researchers) but also not feeling that she was entitled to provide "gatekeeper permission" for researchers to engage with the refugees with whom she worked, without consulting these refugees and asking their opinion.
Our co-production thus required the building of trust as a principle of collaboration between academic institutions and CSOs.However, as we have also found in some of our other research projects, trust in researchers is generally low and CSOs are wary of those seeking just knowledge extraction.These low levels of trust are the result of researchers requiring initial "buy-in" from CSOs and other stakeholders as intermediaries and points of entry into communities of research interest where early introductions and discussions often include commitments from researchers to civil society partners and other stakeholders to collaborate and co-produce knowledge.However, civil society's experience has been that these commitments are often not honoured by researchers as soon as access is granted, and the research becomes distanced or entirely detached from the intermediary institutions.This is a problem that has been noted by other researchers seeking to build partnerships with civil society.As Hattery et al. (2022, p. 513) remark regarding research with communities of colour in the US, "many communities, especially communities of color, have a history of exploitation and abuse that manifests in a deep distrust not only of individual researchers who are external to their communities, but also institutions that employ and fund research." Mistrust goes beyond the personalities and processes of individual research projects and is anchored in the very structures of academic knowledge production which have been shown to construct a system and sense of epistemic entitlement that is often unquestioned by those working within these universities (Thapar-Björkert & Farahani, 2019).As Fraser and Taylor (2016) argue, these processes can be seen to have been reinforced by the neo-liberal marketisation of university research and knowledge (re)production.
Universities' continuing contribution to "epistemic oppression" (Dotson, 2012) can be played out in relationships with CSOs and other research partners (including academic institutions and researchers based in the Global North versus those in the Global South), and even when university researchers attempt to escape these structures, mistrust often remains.
We had also hoped that once we had built relationships with CSOs we could include young refugee women themselves as part of the research team.But for various reasons-which we will develop in more detail throughout the article-we found the limits of this strategy for refugee participation.

Putting Our Ideas Into Practice
As mentioned above, from the start of the project we realized it was vital to involve colleagues from outside academia, including members of CSOs working with refugees, and the young refugee women themselves.
We worked with two CSOs based in Durban, South Africa: a not-for-profit organisation based in the city centre that provides social services primarily for refugees and asylum seekers in the KwaZulu-Natal Province and a non-governmental organisation that provides healthcare services to the marginalized communities in Social Inclusion • 2024 • Volume 12 • Article 8818 the city.The involvement of these CSOs was key in understanding the situation and needs of young refugees and working to ensure that the results of the project were valuable to these young women and the CSOs themselves.Working with and through these organisations, we were able to get into contact with young refugee women and organize focus group discussions and interviews with these young women.We carried out four waves of qualitative research in the eThekwini city centre (formerly known as Durban) in the province of KwaZulu-Natal, South Africa, between June-November 2019 (Freedman et al., 2020), July-October 2020 (Mutambara et al., 2022), September 2021-May 2022 (Crankshaw et al., 2023(Crankshaw et al., , 2024) ) and January-June 2023.
We conducted in total five focus group discussions, 90 in-depth interviews with women asylum seekers and refugees living in the city, as well as interviews with key informants from organisations supporting asylum seekers and refugees in the city ( = 4).Ethical clearance for the research was obtained from the Humanities and Social Science Research Ethics Committee at the University of KwaZulu-Natal.

Positionality
As mentioned above, we are committed to conducting our research using explicitly feminist approaches and as a part of this abandoning the idea of the objectivity of a distant and neutral scientific method, to make explicit our own biases and positionality (Brooks & Hesse-Biber, 2007;Lokot, 2019).Starting from this position and from the belief that it is impossible to separate "the knower from what he knows" (Smith, 1974, p. 8), we begin here with a reflection on our own positions as researchers/civil society activists and the way that what "we know" and how we can produce knowledge is inseparable from these positions.Reflecting on the possibilities for emancipatory or liberatory knowledge production must thus start from a consideration of our positionalities as researchers/civil society activists, and the ways that our positions within structures of inequality and our asymmetric power relations with the young refugee women (and with each other) impacted on these possibilities.We believe that it is imperative to break the silence around the constitutive inequalities and power asymmetries during any fieldwork (Caretta & Jokinen, 2017), and that even if it is not possible to break down these power inequalities, to work towards reducing their impacts.
There were three of us who participated in the project as researchers affiliated with a university/research institution.The fourth researcher was the leader of a CSO working with asylum seekers and refugees.All of us identify as women, but we were differently positioned in terms of race, nationality, citizenship, and professional status.One "lead" researcher held a permanent position within a European university, identified as white, and was a European citizen.The second "lead" researcher had been employed for more than eight years within a South African research institute, identified as white, and was a South African citizen.The "junior" researcher on the project was a post-doctoral fellow at the same South African research institute, a black woman of Zimbabwean origin, with legal residence status in South Africa.She was also the youngest in the team.All three of us had obtained a PhD qualification but only one of us had achieved a full professor academic position.
The final partner was a senior member of a CSO, outside of academia, although she had considerable experience in engaging with academic researchers.She identified as a black South African (with Indian heritage), and had permanent employment within a CSO, although this was precarious due to the crisis in funding for civil society.She had obtained her undergraduate degree but had not completed her post-graduate studies and told us that she at first felt not qualified enough to fully contribute to the research.This feeling of "not being good enough" as a black woman and non-academic persisted for the first period of the research, and it was not until we had spent considerable time (indeed years) talking to each other and building relationships of trust that she said she felt free to express herself fully and contribute fully to the research.

Power Asymmetries Within the Research Team
Our different positioning within intersecting structures of inequality and domination clearly had an impact on our ability to engage in co-production and/or liberatory knowledge production.A first set of asymmetries amongst the researchers is important to note, even if these are secondary to the important asymmetries between researchers and refugees.The fact that the two lead researchers in the project were older white women with permanent/more secure employment played a role in their relations with the younger post-doctoral researcher who was on a finite employment contract under the conditions of a post-doctoral position.The post-doctoral researcher was employed specifically to work on this project, and thus, as well as being under the supervision of the senior researchers, her continuing employment in some ways depended on the "success" of this project.Although we attempted to mitigate these asymmetries by creating a supportive relationship and promoting our post-doctoral colleague as far as possible (for example, by making her first author in various publications drawn from the research), it is important not to dismiss these types of power inequalities within academic research teams, as we believe that they have an impact on the ways in which the researchers can engage with partners/participants and with each other.
However, the more marginalized position of our post-doctoral researcher within the academic research team, as well as her age, race, and nationality, also positioned her as closer to the young refugee women with whom we were working and thus enabled her to more fully engage in processes of co-production.The particular historical context of South Africa, and its racial divisions, as well as that of the city of Durban, where there is a very high level of urban violence, shaped our different abilities to participate in research activities with refugees.
For the two white researchers, for example, the areas where most of the young refugees lived were viewed by other members of the team as "too dangerous."As white women (and in one case a non-South-African white woman), going into these areas of the city would have put us at real risk of opportunistic crime and/or physical violence, and we were thus discouraged from going there.Our post-doctoral colleague who is a young black woman of Zimbabwean origin was a far less visible target in these areas and it was easier for her to engage with the young refugee women in their own areas where they lived.But even for this colleague, travelling to the areas where the young refugee women lived posed some risks; this meant that she had to take precautions, such as travelling in the company of a peer educator from a civil society partner organisation, dressing casually and not carrying anything valuable with her.Importantly, being closer in age to the young refugee women and being a black woman like them allowed her to develop much closer relationships with these women than the older, white female researchers could.Furthermore, although she did not share the same precarious migration status or social situation as the young refugee women, she had also experienced a migration journey that could situate her closer to the position of these young women.This proximity in age and sharing a migrant experience allowed her to build closer relationships with the young women and keep up constant contact with them as we discuss further below.
Finally, the positioning of our CSO collaborator was complex in that she was placed in a situation where she was the head of the CSO but not officially a leader of the project, as this was not possible within the academic funding structures, as we discuss below.She felt at the start of the project that she was not as legitimate a researcher as the academic partners.But at the same time, the academic researchers were dependent on her and her organisation for support involving young refugee women in the project.This position was also complicated by the CSO need for funding.Indeed, during the duration of the project, the organisation lost some of its core funding and its budgets were constantly under threat.At some point, this led to tensions concerning costs associated with the research.Our CSO partner remarked that organisations such as hers often "subsidise" research by providing space where researchers and refugees can meet, for example, with no compensation for personnel, rent, telephone call costs, etc.She also remarked that it was hard to ask for greater remuneration as she felt that research is perceived as "noble" and there seems to be an expectation that one should contribute voluntarily and without expecting remuneration because, ultimately, the research is going to add value the CSO programme.The costs involved in being the liaison between the academic research participant and the researchers as a CSO are therefore often hidden.
Division of budget is something that is rarely discussed in project methodologies, but both the structure of academic funding and the sometimes-differing priorities of research partners as to how this funding could/should be spent are in our view an obstacle to the success of co-production.This is magnified in the current context where both academic institutions and CSOs are generally suffering from funding cuts and thus budgets are ever more squeezed.

Structuring of Academic Research as a Barrier to Co-Production
One of the most immediate and evident barriers to processes of co-production of knowledge is the way that academic research is institutionally structured and funded.As Weiss (2016) argues, one assumption often made by academics is that activists with whom they may work in collaboration are "compromised" by the organisations within which they work.She argues that academics too must recognize the ways in which they are "compromised" by their institutional settings and structures and must engage in reflexivity, but also examinations of the way their own institutions function.Systems of academic promotion based on research output for example, clearly impact how academic researchers engage with their projects and how they collaborate both with fellow academics (in what Lemon, 2018, has named the "academic hunger games") and with non-academic partners.
Further, academic researchers have previously pointed to the ways in which funding systems impose a "top-down" process of knowledge production, where research questions and objectives must be framed by researchers who are recognized within the academic system in order to apply for funding, leading to inequitable relations from the start (Olivier et al., 2016;Phillips et al., 2013;Shuayb & Brun, 2021).
Only once the grant is received will researchers then have funds to put into place real participation from those outside of academia.In our case, the research project was funded by a joint French-South African funding scheme.This involved making a joint application submitted by two recognized academic institutions-one in France and one in South Africa-and headed by researchers who were engaged full-time within these institutions.This funding structure limited our co-production from the start.We could engage with our civil society partner and inform her that we wished to apply for funding for our project, but our colleague could not be a co-applicant.Further, any time she spent working with us on preparing the research project would be unpaid and without any tangible benefits if the project was not selected for funding.We subsequently had a similar experience working with this same colleague to try and obtain further funding from the European Union: In this case, all the work she put into preparing the application with us was unrewarded as the project was rejected.Spending time preparing research projects for funding applications is usually seen as an integral part of an academic researcher's role, but for colleagues in CSOs, the time spent is seen to be a risk if no funding is forthcoming.In our case, this meant that any real co-production only started after the funding had been obtained and thus the parameters of the research were already, to some degree, set.For our CSO partner, this meant that our initial meetings were to discuss Social Inclusion • 2024 • Volume 12 • Article 8818 how to carry out the research in a way that would be beneficial to all, rather than meetings about how to structure the project.
It was also clear that for the CSO partners, research was an "extra" to their everyday core work of supporting refugees.So, all their contribution to the research was on top of their everyday jobs.Our CSO partner said she was happy to be part of the team and felt that it helped her keep abreast of what was going on in research, as well as gain insight into the needs of refugees and how the organisation could better adapt to these.She was particularly interested in the policy briefs produced, which were useful to her organsiation, and also appreciated being invited to co-author research publications.Nonetheless, she was aware that, for the academic researchers, the research itself was their main occupation, whilst for her it was an extra, something she felt at times might be forgotten by her partners.This highlights the problem of the different priorities of academic researchers and CSO/NGO partners, which has already been noted by various researchers (see, e.g., Shuayb & Brun, 2021).
Following on from this question of research funding and an academic structuring of research that limits possibilities for co-production, this also impacted our ability to pay refugee women for their work in the knowledge production process, which provided a real barrier for our project.There is a continuing debate within academia around the ethical implications of paying research participants and how this might influence consent and lead to a "commodification" of research (Hammett & Sporton, 2012;Head, 2009).However, it seems that there is now a general consensus amongst those trying to carry out ethical research with refugees and others in vulnerable situations that this should involve some kind of financial compensation for the time that they take to participate and the material and emotional costs of participation (Warnock et al., 2022).For our research, we realized that the young women refugees would need to take time to participate and talk to us, and that they might also have costs related to transport, childcare, or missed opportunities for other work.It thus seemed vital to give them some kind of remuneration.This, however, was difficult in the context of our funding agreement, and also the university's own financial rules and South African labour law.
As many of the young women were also without legal residence status in South Africa, the barriers to payment were complicated, as it is formally illegal to employ those without legal residence status and a work permit.In these circumstances, we offered to reimburse them for travel expenses or packages of necessities/toiletries, but we were well aware that this was not enough, and this was brought back to us as we assessed our research project and its outcomes.

Overcoming Asymmetries to Produce Knowledge?
As our research project progressed, we constantly grappled with the questions and problems outlined above, which provided real barriers to the co-production of knowledge.Asymmetries both within the research team and between the research team and the young refugees were constantly made apparent and manifested in various ways.In some circumstances, the different positioning of the academic researchers and CSO partners was also a complicating factor as we talked to the young refugee women.In some of our interviews, for example, it became evident that the young women participating were in some ways unhappy with the services that they were receiving from two CSOs.This created a particularly difficult situation since most discussions were held on the different premises of the CSOs.For the women, their dependence on the CSOs for support clearly created a situation of conflict, and they talked about their grievances reluctantly, obviously afraid of sanctions from the organisation if they complained.They were most forthcoming about Social Inclusion • 2024 • Volume 12 • Article 8818 this when the discussions were taking place in French, a language that was common to one of the academic researchers and the refugees, but not to the respective CSO staff members.This kind of discussion revealed power asymmetries and put everyone in an unwanted position vis-à-vis the others: The young women refugees were scared to complain about an organisation that offered them support; the academic researchers felt uncomfortable listening to complaints about their respective research partner organisations and knew they could do nothing, really, to mitigate these complaints-which were generally about the lack of funds available for the refugees; and finally, for the CSOs involved, this situation was extremely difficult because they had limited funds and time available to provide support for all the refugees who were asking for help.In another situation, during a conversation with a refugee woman housed in a temporary shelter run by another CSO, one of the academic researchers was asked to intervene to prolong her stay.However, the researcher was aware that the time limit was one of the central rules that the CSO in question had established for the woman's accommodation; and although the researcher mentioned the situation to the person in charge, it was clear that she could not effect any change to the rules.In addition, the researcher was asked by a CSO staff member to verify that the women being interviewed indeed met the criteria for shelter, which placed the researcher in a highly uncomfortable position even though she voiced her ethical obligation to maintain the confidentiality of all research participants.
The real barrier to co-production, however, was evident in the huge disparities in the economic and legal situations of the research team and the refugees.The researchers made it clear to the refugees that they could not intervene to change their economic and legal situation and to gain residence papers for them.
As mentioned above, we were limited by funding and university regulations in the type of financial recompense we could offer to participants.At the end of the study, we held a restitution meeting with refugee women who had participated in our research to provide feedback on the findings as a way to acknowledge and show respect for participants' contributions (Mackenzie et al., 2007).Scholars have found that lack of reciprocity has been viewed by refugee research participants as "an extreme breach of trust and exploitation of privilege" (Mackenzie et al., 2007, p. 306).And although we, at all stages, were careful not to create expectations for longer-term support when we invited women to the meeting, they still came to us with the hope that there might be some financial advantages.One woman had left her young children in paid childcare in order to attend the validation meeting and another had passed up a few hours of financially remunerated work.At the same restitution meeting, when discussing what refugee women participants had gained from the research, the comment "but we just need money" from one of the young women brought this point home clearly.We were aware that the young woman in question had been engaging in transactional sexual relationships for economic survival, and we were powerless to effect any real change in her situation or to offer enough financial recompense to change this in a significant way.

Knowledge Production in the Interstices
One of the unexpected findings of our research and one which perhaps led to the most useful form of knowledge production, was that the young women refugees had little or no knowledge of SRH issues.
Several of the participants shared, for example, that they did not know how a woman becomes pregnant (Crankshaw et al., 2024).Whilst we had anticipated that lack of access to SRH services was a problem for many young refugee women, we had, as it turned out, overestimated what would be their basic understanding of female reproduction.This finding allowed us to question our assumptions, which we then understood to be based on an epistemology of "white ignorance" (Alinia, 2020;Bowleg et al., 2017;Mills, Social Inclusion • 2024 • Volume 12 • Article 8818 2007).However, once we had realized our ignorance, this exchange of knowledge around reproductive health allowed us, and particularly our post-doctoral researcher, to build relationships with the young women where there was a greater degree of reciprocity (Pittaway et al., 2010) and exchange.During conversations, whilst the young women explained their concerns over reproductive health, they were also able to ask questions and gain information that would be useful to them and could improve their reproductive health.
Having these long conversations about, for example, how pregnancy occurs, and what are the possible methods of preventing pregnancy, was thus a really useful and tangible outcome of our project, and led to the young women gaining useful knowledge which could give them greater agency in their reproductive choices.Perhaps this was an instance of real co-production of knowledge when all of the researchers and participants learned things that were useful to them and produced knowledge for each

Conclusion
We start our conclusion with a reflection from two other researchers working on a project of co-production with refugees, who conclude that "co-production is a process where tentative alliances are formed and necessary trade-offs are incurred, and that the value of co-production as touted in theory collides with the realities of power dynamics and the complexities of relationship formation" (Gibbes & Skop, 2020, p. 291).
We have outlined here some of the difficulties with the power dynamics and asymmetries that we experienced during our ongoing research.Despite these barriers, we do hope that we had some positive impact on the young women we engaged with, even if we are keenly aware that this is not enough, especially as the situation for refugees in South Africa becomes increasingly precarious (Mutambara et al., 2023).Within our research team of academic researchers and CSO partners we have, however, made some progress in understanding the pitfalls of co-production and how it might be done better to create real "liberatory practice"; we have used this knowledge to design new projects that will be more inclusive and participatory from the outset and we hope will enable progress in creating knowledge that could have a real positive impact on the situation of young women refugees in South Africa.
Our experience has taught us to understand better the difficulties and limits of co-production and our role within the latter.Perhaps, following on from the previous section where we talked about the most valuable knowledge being produced at the "interstices," what we have learned is to leave more room in our research for these interstices and unexpected moments of knowledge production and to understand that co-production is perhaps most relevant when it is about everyone-researchers, CSOs, refugees-learning from each other, before attempting to produce knowledge for an outside audience.We all have different types and forms of knowledge; the mutual exchange of this knowledge-and really taking the time and space to learn from one another-can lead to valuable outcomes for all.We understand the importance of time and of building long-term relationships with all partners to develop the possibilities of co-production and we are pleased that we have been able to engage in ongoing discussion with our CSO partners over future projects.
As Chu et al. (2014) argue long-term partnerships can be seen as key to building more equitable relations.Of course, we also wish to produce knowledge that will have an impact on policies and programmes for refugee women, and thus improve their lives.However, perhaps acknowledging that co-production is an ongoing process that can start from these moments of mutual learning is one of the major lessons of our research.Pahl et al. (2022) sub-titled a recent book on collaborative research "the poetics of letting go."Perhaps we should take inspiration from this title and aim to create more moments where we "let go" of our pre-defined ideas about collaborative research and co-production and create more room for the emergence of the unknown and more room for us and our various institutions to change.Returning to the title of our article, our project could not provide the economic stability that the young women refugees needed.
In acknowledging this we understand the limits of research but also leave open room for other forms of learning and evolving in our understandings and our relationships.