Representative co-production: broadening the scope of the public service logic

ABSTRACT

Although the public service logic (PSL) has been an important equipoise to the predominant goods-manufacturing logic, there is potential to broaden its scope. An explicit integration of social context may contribute to an enhanced conceptual understanding of the PSL, at the same time addressing a major challenge in healthcare: disparities among population groups. A ‘representative co-production’ approach is suggested. In such an approach, group representatives’ knowledge and skills are used in evaluating, designing, and delivering services with the purpose of supporting other group members’ value co-creation. A case is provided, demonstrating representative co-production in access to preventive health services.

KEYWORDS:

• Co-production
• value creation
• co-creation
• representation
• public service logic

Introduction

As in many countries, New Public Management (NPM) reforms were introduced to the Swedish public sector in the 1980s, borrowing ideas mainly from the industrial goods-manufacturing sector (Quist and Fransson 2014). Consequently, many public service organizations (PSOs) standardized their internal processes, an act that is argued to have contributed to the dominance of ‘one-size-fits-all’ solutions in healthcare services (Berwick, Godfrey, and Roessner 2002). Thus, the standardization strategy is ill-fitted to address one of the major challenges to many Western healthcare systems: group disparities in access to, and satisfaction with, provided services. Because many of these disparities stem from the different needs, expectations, and prerequisites of groups, standardized processes are likely to do more harm than good.

The World Health Organization (WHO 2017) states that health is a fundamental human right, but that all countries suffer from disparities in health among groups of people. As indicated in the report Closing the Gap in a Generation (CSDH 2008), the participation by disadvantaged inhabitants in interventions is important. Or, as emphasized by the report’s chairman (Marmot et al. 2012, 1024), we must ‘[e]nsure that the different needs, perspectives, and human rights of groups at risk of marginalisation and vulnerability are heard through their involvement in decision-making processes. Accompany this by effective mechanisms for adequate participation, engagement and consultation with all parts of civil society’. However, another idea brought from the manufacturing sector to the public sector during NPM was the so-called value chain (Porter and Millar 1985), in which the ‘product’ is sequentially refined within the factory and delivered to a passively waiting customer. Naturally, such a model offers poor possibilities of involving people, whether referred to as ‘customers’ or ‘disadvantaged groups’.

The public service logic (PSL) offers promising prospects in addressing the pitfalls caused by manufacturing logic (Osborne 2018; Osborne et al. 2015), not least by emphasizing the centrality in public service design and delivery of potentially active users as co-producers (Pestoff 2014; Radnor et al. 2014). However, to address inequities and disparities in healthcare, I argue that it is necessary to consider social context more explicitly than is the case today. Similar to previous research (Burke, Pasick, and Barker 2009), in this article, individuals and institutions (community, workplace, family, etc.) and social structures (formal and informal rules, traditions, norms, etc.) constitute social context. Whereas the former is rather straightforward, the latter needs further explanation. Lundquist (1984) defines social structures as ‘a pattern in the relation between actors’. Similarly, Giddens (1984) emphasizes rules – often informal – that guide social interplay as well as resources in the sense of capacity for change. Social structures have emerged because of interactions between people, but at the same time people may not be aware of the social structures, as they are abstract and not directly observable (Johnson 2001; Shanahan and Hofer 2005). Structures favour certain individual characteristics and disfavour others, thus enabling and constraining people’s actions and interactions differently (Lundquist 1984). Consequently, access to institutions may be closed for certain people, but not for others (Berger and Luckmann 1966). In a healthcare context, it is important not only to address the individual, but increasingly to put a focus on social context because it ‘directly and indirectly affect[s] health and behavior’ (Pasick and Burke 2008, 359).

Because social structures impact groups’ prerequisites of actions and interactions differently, it becomes pivotal to consider the different perspectives and positions of social groups as important resources (Young 1997) and that ‘particular groups are to be involved in their particularity – of situation, experience or identity’ (Martin 2008, 37). Involving representatives from a group’s social context, such as community members, that share a background with other group members is argued to be an important strategy to bridge such cultural gaps, and by so doing increase access to services for other group members (Allen et al. 2006; Reeb 2006). However, the term representation is not uncomplicated. In the present article, the representatives can be understood to represent both the PSO and the local women. They represent the PSO because they had taken over some of healthcare staff’s traditional tasks, such as informing women about healthcare services. They represent the local women because these women were invited to participate in the project because of their shared background with many of the local women. The distinction may not be as clear cut; at the same time as PSO employees carry out their daily tasks they may advocate their own social groups’ interests (Lim 2006). Thus, the representatives’ own values and interests should not be hidden, but rather used as a resource (Tahvilzadeh 2011). The question is based on what characteristics group representation should be understood as, whether they should be based on gender (AbouAssi and An 2017), income level (Wright 2015), or ethnicity (Atkins, Fertig, and Wilkins 2014). As in the present article, the distinction between these characteristics is not clear and more attention should probably be placed on the intersection between these and other characteristics (Mastracci and Bowman 2015). However, the complexity of representation implies that representatives may not only represent a social group, but in fact also the wider public or simply themselves (Litva et al. 2009).

In addition, the interests of the representatives may be included at different stages of a public service, such as identifying and alerting problems (Lundquist 1991), decision-making (Wright 2015), or the implementation activity (Kingdon 1995). The question is also what impact representation has. Whereas some studies indicate representatives’ positive influence on citizens from the same group (Atkins, Fertig, and Wilkins 2014; Riccucci, Van Ryzin, and Li 2016), others have found that it depends on factors such as the position of the representative (Suzuki and Avellaneda 2017). Other studies have found no impact, suggesting that the benefits of representation is policy-specific (Van Ryzin, Riccucci, and Li 2017).

A public service logic

Osborne (2006) suggests that NPM has been only a phase between the preceding Public Administration and the emerging New Public Governance, the latter focusing on inter-organizational and inter-sectorial collaborations and the engagement of citizens. Consequently, public sector management today is often a combination of traditional hierarchy, NPM ideas based on market and competition, and post-NPM organisation of networks (Lægreid et al. 2015). Inspired by the so-called service-dominant logic (SDL), Osborne, Radnor, and Nasi (2013, 149) introduced a public service-dominant logic, argued to offer ‘effective public management practice within the New Public Governance’. With the translation to a public sector, SDL’s criticism (Vargo and Lusch 2004; Vargo, Maglio, and Akaka 2008) of manufacturing logic’s intra-organizational focus on outputs of tangible goods and transactions remained prominent, as did recognition of the resources of multiple actors – essential to meet the reality of interdependent modern society (Quist and Fransson 2014).

Many of the recent developments of a public service-dominant logic have sought to address the specifics of the public sector as compared to the private sector, which is the focus of SDL. Osborne (2018) recently suggested dropping ‘-dominance’ to distance the concept from SDL, while retaining the emphasis on ‘service’. It has also been argued that ‘-dominant’ should be dropped because ‘service’ is described as a process through which value is co-created, a process in which tangible resources (traditional goods) and/or intangible resources (traditional services) are integrated and combined (Osborne, Radnor, and Strokosch 2016; Vargo and Lusch 2008). Because ‘everything’ is service, ‘-dominance’ becomes redundant. Thus, I have used public service logic (or the abbreviation PSL) here (Osborne 2018).

The engagement of citizens and public service users is important to the strategic orientation of public organizations in the PSL, not least by sharing their expectations and experiences of the performance of the service as well as information about their needs (Osborne, Radnor, and Nasi 2013). Indeed, the engagement and involvement of citizens and users ‘should occur at all phases of a (public) service lifecycle’ (Osborne, Radnor, and Nasi 2013, 142), not least as drivers of innovation (Simmons and Brennan 2017).

Osborne (2018) highlighted that the customer–provider relationship in the public sector often differs from that in the private sector. For example, in the private sector, getting a customer to return is important. In the public sector, a returning ‘customer’ may be a sign of service failure. Moreover, the reluctance or unwillingness among public service users is rather unproblematized in the private sector (Osborne 2018). Because of the risk or discomfort, patients may be reluctant to engage with healthcare staff (Batalden et al. 2015; Berry and Bendapudi 2007). Furthermore, public service users may not even want the service (but may need it); examples include patients in drug rehabilitation or inmates in prison (Nordgren 2009; Moore 1994).

Another difference between the private and public sectors in terms of importance for the developments of PSL is the value concept. In SDL, value is a highly individualized perception that is argued to be ‘perceived in an individualistic way’ (Grönroos 2011, 282) and to be ‘uniquely … determined by the beneficiary’ (Vargo and Lusch 2008, 7). It is increasingly being acknowledged that a sole focus on private value, which has been the focus of many NPM ideas as well as in SDL, is insufficient for the more complex public services (Alford 2016; Stoker 2006): public service users can not only create private value for themselves, but also contribute to public value; they can contribute to the ‘common good’ or ‘public interest’ (Beck Jørgensen and Bozeman 2007). Consequently, public value should not simply be understood as the sum of public service users’ satisfaction or private value (Alford 2016; Stoker 2006). As recognized by Alford (2014), ‘[i]n between these two are varying degrees of what might be called group value’.

Co-production and co-creation

A central concept in a PSL is co-production, defined by Osborne, Radnor, and Strokosch (2016, 640) as ‘the voluntary or involuntary involvement of public service users in any of the design, management, delivery and/or evaluation of public services’. Thus, co-production puts the service user to form the service delivery and its processes (Trischler and Scott 2016). Pestoff (2006) argued that, in the public administration literature, co-production typically refers to individuals’ or groups’ contributions to the public service; thus it ‘differs notably from the traditional model of public service production in which public officials are exclusively charged with responsibility for designing and providing services to citizens, who in turn only demand, consume and evaluate them’. There is disagreement about which activities should be included in co-production (Alford 2009a). To Bovaird and Loeffler (2012) and Loeffler et al. (2013), co-production activities are broad (Table 1).

Table 1. Co-production activities (Bovaird and Loeffler 2012; Loeffler et al. 2013).

Co-production activities	Examples
Co-planning	Deliberative participation
Co-design	User forums and customer journey mapping
Co-financing	Charges and fundraising
Co-prioritization	Participatory budgeting
Co-managing	School parent-governors
Co-delivery/co-performing	Peer support groups
Co-assessment	Co-monitoring and co-evaluation, such as user online ratings

Activities such as providing input, as in consultation, could be included in co-assessment, but only if citizens are truly able to influence decisions (Bovaird et al. 2016). Moreover, co-production is about the contribution of both professionals and citizens/users/volunteers (Bovaird et al. 2016; Loeffler et al. 2013); therefore, factors such as self-help (McColl-Kennedy et al. 2012) are not included in co-production. In the seminal work of Ostrom (1996, 1073, my emphasis), co-production ‘implies that citizens can play an active role in producing public goods and services of consequence to them’. As in the previous quote (Pestoff 2006), co-production is oftentimes considered optional in that people may be invited to participate in planning and production processes to improve public services (Osborne and Strokosch 2013).

In SDL, on the contrary, co-production is not optional: ‘the customer is always a co-producer’ (Vargo and Lusch 2004, 10). That is, an organization cannot – in a manufacturing fashion – produce and deliver value to a passively waiting customer. Because production and consumption cannot be separated for services – they occur at the same time – co-production becomes unavoidable (Parasuraman, Zeithaml, and Berry 1985). Later, SDL’s premises were updated and the industrial ‘co-producer’ replaced with ‘co-creator’: ‘the customer is always a co-creator of value’ (Vargo and Lusch 2008, 7). However, it is argued that co-production is a component of co-creation in addressing the participation in developing the core offering (Vargo and Lusch 2008). In this sense, co-production becomes relatively optional (Vargo and Lusch 2016). Another important distinction is between value creation and co-creation. To Skålén (2016), value creation implies that value is always realized by the user in her/his usage (value-in-use) and that co-creation implies the direct interaction between the user and the PSO.

The developing PSL literature’s view on co-production has borrowed from both public administration and SDL literature (Osborne and Strokosch 2013). In PSL, the service provider–user relationship is at heart, and ongoing interactions between them are essential for relationship development (Pestoff 2014). However, co-production in PSL is not only about increasing customer satisfaction and improving services, but also how these services can contribute to ‘the meeting of societal objectives or contribute social cohesion or well-being’ (Osborne, Radnor, and Strokosch 2016, 643). Indeed, some of these objectives include aspects addressed in co-production itself, such as social inclusion and citizen involvement (Osborne, Radnor, and Nasi 2013; Pestoff 2014). In PSL, public service users can create private value for themselves or contribute to public value, while also contribute to value for other users of the particular service (Osborne, Radnor, and Strokosch 2016).

Osborne (2018, 225) argued that, in PSL, co-creation rather than co-production should be favoured: ‘Co-production assumes a process where the PSO is dominant and where the logic is linear and based upon product-dominant conceptions of production.’ Co-creation, on the other hand, ‘assumes interactive and dynamic relationship where value is created at the nexus of interaction’ (Osborne, 225). Although these terms are often used interchangeably (Voorberg, Bekkers, and Tummers 2015), the distinction may be understood as linguistic: away from the producer and tangible goods lexicon (co-production) to a customer and service lexicon (co-creation) (Vargo and Lusch 2008). Thus, value cannot be produced within a factory – or a school, or a hospital – and delivered to a waiting service user. Rather, the provider can offer only potential value as value propositions, or offerings (Normann 2001; Vargo and Lusch 2008). The distinction may also be understood on different levels in which co-production is framed by co-creation and thus one of many co-creation activities (Hardyman, Daunt, and Kitchener 2015).

Co-production is still favoured over co-creation in public management theory (Voorberg, Bekkers, and Tummers 2015). Although I agree that co-creation should be more prominent than co-production in the public sector, both have their place. Co-creation may be used as an overarching concept, emphasizing the non-linearity of value creation for PSOs as well as the individual service users – that value is jointly created in interactions that takes place in the broader service system and people’s broader social contexts and life situations (Osborne 2018). Osborne, Radnor, and Strokosch (2016, 644) argued that such a service system includes ‘PSOs, service users and their significant others, the local community, hard and soft technologies and sometimes other significant stakeholders’ and presented four ideal types of co-production linked to the co-creation of value.

First, co-production is seen either as an intrinsic and involuntary element of the delivery process (the SDL view) or a voluntary, added on, activity that can improve public services (the public administration view) (Farr 2016; Osborne and Strokosch 2013). Second, public services are seen as either entities in their own right or as part of a broader service system. In the latter, the resources of the provider, the user, and other actors are integrated in the service system and may impact outcomes not only for individual users but for others as well (Farr 2016; Osborne, Radnor, and Strokosch 2016). The first ideal type is (pure) co-production, in which user and staff co-produces the service experience and outcomes. Here, the interactions (‘moments of truth’ [Normann 2001]) between the two parties become essential. The second ideal type is co-design and includes improvements of design and delivery of existing services. Here, co-production includes the voluntary involvement of users in evaluation, design, and improvements. Co-construction (of the service system), the third ideal type, addresses the broader service system and not the public service in isolation. Here, the users’ service experience integrates with their overall life situation – how the service impacts on their life. The focus of co-innovation, the fourth ideal type, is on the voluntary involvement of users in new forms of service delivery within broader service systems. In their typology, co-production may not only lead to co-creation of value, but also co-destruction (Osborne, Radnor, and Strokosch 2016).

In this article, the co-production activities undertaken by the representatives include activities that are typically addressed in the public administration literature, including evaluation, design, and delivery of the cancer screening program. However, those activities may lead to improvements of services that ‘spill-over’ on other ideal types of co-production (see Discussion section).

Individual, group, and public co-production

In the 1970s, co-production of public services was mainly considered an individual practice, but by the 1980s had started to also be recognized as a group or public phenomenon (Bovaird et al. 2016). Similarly, the seminal work of Ostrom 1996) increasingly emphasized groups of people. Bovaird et al. (2015) stated that group and public (or ‘collective forms’ of) co-production in the literature depends on whether inputs are collectively supplied or whether outputs are collectively enjoyed. Table 2 presents three levels of co-production (Brudney and England 1983).

Table 2. Levels of co-production (Brudney and England 1983).

Level	Explanation
Individual co-production	Activities undertaken by individuals for their own benefit
Group co-production	Activities undertaken by a number of citizens to improve service quality for that group
Public (or ‘collective’) co-production	Activities undertaken individually or in groups, but result in benefits enjoyed by the whole community

Rosentraub and Sharp (1981) highlighted the importance of the communality of benefits in public co-production, and was a reason why Brudney and England (1983) considered that collective forms of co-production had more potential to impact overall level of community services than individual co-production.

Alford (2002, 2009a) separated among users (or clients), volunteers, and citizens with different relationships with PSOs, and thus factors that influence them to co-produce. Service users typically provide input individually and receive private value or benefit from the service they consume individually but co-produce with the provider. Usually, no public value or benefits are rendered. Individual volunteers also contribute with input, but do not personally consume the service. However, volunteers may often act as a group and thus jointly provide input and also jointly enjoy public value. Whereas individual service users may express their preferences through market purchases or such as customer satisfaction surveys, citizens express their preferences through democratic processes, such as voting, and thus mainly contribute to public value (Alford 2002, 2016). Thus, group and public co-production may be criticized for contributing to the devaluation of citizenship because what governments and agencies should ‘produce’ is already a public choice, in that it is ‘a mediated process because it is articulated through the channels of representative government’ (Alford 2002, 339).

My interpretation of Alford (2009a) is that the individual service user is the most important locus of co-production because s/he contributes with input and also consumes provided services, something neither volunteers nor citizens do. The individual level is also important because an individual’s input may impact the PSO’s output and outcomes, thus providing private value for the user and also public value for the citizenry (Alford 2009b). Because it is often impossible not to include individual users’ time and effort of public services, understanding and encouraging their willingness and ability to co-produce becomes crucial (Alford 2002, 2009b).

Different from Alford (2009a), but similar to Brudney and England (1983), Bovaird et al. (2016) highlighted the importance of collective forms of co-production, particularly because it provides activities through which social capital can be built, and because it can make use of already established social capital to achieve valuable outcomes. Thus, trust between user and provider (Bovaird 2007) and among the cooperating individuals in collective forms of co-production becomes essential (Fledderus, Brandsen, and Honingh 2016). To Pestoff (2014), collective forms of co-production may be important for achieving sustainable and viable co-production. However, both types of co-production are important because they address different types of services: collective forms of co-production often concern more enduring welfare services, whereas individual co-production often concerns spontaneous and ad hoc services (Pestoff 2014). Somewhere in-between, Pestoff (2014) emphasized the importance of smaller group co-production, because such groups (unlike larger groups) are more likely to create collective action and interaction, with the potential to impact the co-producing participants (private value) as well as the addressed issue (public value).

I emphasize the ‘in-between’ level of co-production: group co-production. In addressing disregarded groups’ lack of access to services, public co-production would probably not be fit for purpose, especially because such efforts mirror the problem to be addressed: that accessibility is for certain people and may risk excluding those in greatest need (Brandsen and Honingh 2016; Cepiku and Giordano 2014). The research on citizen participation reported similar findings (Barnes et al. 2003; Church et al. 2002). As Rosentraub and Sharp (1981) noted, the differences in resources mean that co-production may indeed increase inequity in society.

The individuals who participate in group-level co-production in this article are not public service users, simply because the target group members do not use the screening program. Therefore, I argue that it is necessary to include representatives. However, representative co-production is a particular type of group co-production. Because the target group in this article is foreign-born women who do not participate in the screening program, it is difficult to attract members of this hard-to-reach group to participate in co-production. Consequently, co-production participants were women living in the same segregated area of the city, who share culture and language with those people the healthcare system fails to reach. Different from the target group, many of whom had never visited local clinics, many of the representatives did participate in the screening program. Thus, it would be an over-simplification to claim they were a perfect representation of the target group. However, because their commonalities in background and role in the local society, they can be said to represent the women of the target group. Thus, representation becomes a strategy to address inequities in public services.

Similar to Bovaird et al.’s (2016, 51) definition of collective forms of co-production as ‘the joint action of citizens to support public services and achieve outcomes’, I also focus on input that needs to be collectively provided by a group and not by single individuals. Moreover, co-production may benefit the co-producing individuals themselves, as well as contribute to public value, but the purpose is to provide private value to other service users. Paired with the linking of co-production with co-creation (Osborne, Radnor, and Strokosch 2016), as well as the adopted definition of co-production as the voluntary/involuntary involvement of users in design, management, delivery, and/or evaluation (Osborne, Radnor, and Strokosch 2016), I define representative co-production as follows:

The joint and voluntary involvement of group representatives in evaluating, designing, and delivering public services that enable value co-creation for other group members.

Extended service-dominant logic

As mentioned, the definition of social context in this article includes social structures as well as individuals and institutions. The latter is starting to gain interest in recent PSL developments, emphasizing service systems, and the importance of involving not only user and PSO, but also local community, technology, family, etcetera, in value creation (Osborne 2018; Osborne, Radnor, and Strokosch 2016). However, the former (structures) remain relatively unexplored in PSL. Because many public services address those worst off, and because public values include equity and equal access to services (Beck Jørgensen and Bozeman 2007), PSOs must recognize the impact of social structures. This section presents recent SDL developments that explicitly address the impact of social structures and are important to consider for the development of PSL.

During the last decade, two streams of SDL that emphasize social structures have emerged: service ecosystems and transformative service research (jointly referred to as ‘extended SDL’ in this article). Neither of these theories is deterministic, in a sense that the individual is entirely under the influence of structures (Johnson 2001; Lundquist 1984). Similar to Giddens’s (1984) structuration theory, the service ecosystems approach acknowledges that individuals’ actions and interactions both influence – and are influenced by – social structures, such as norms and (informal) rules (Akaka, Vargo, and Lusch 2013; Vargo and Akaka 2012). These structures are often positively described: they enable interaction and co-production, or they are intangible resources that aid collaboration between multiple actors (Vargo and Akaka 2012; Vargo and Lusch 2016). The other theory, transformative service research, also recognizes the possibilities of focusing on social structures to achieve change (Blocker and Barrios 2015; Skålén, Aal, and Edvardsson 2015); besides (in)formal rules, structures often refer to culture or environment (Anderson et al. 2013). In addition – and more so than in service ecosystems – transformative service research recognizes the potential negative aspects of social structures; for example, structures that through interactions may negatively impact people’s well-being (Anderson and Ostrom 2015; Anderson et al. 2013).

Because of social structures – such as informal rules, culture, and environment – perceptions of value, as well as the process of creating value, are considered to be intersubjective and shaped by social forces in extended SDL (Helkkula, Kelleher, and Pihlström 2012; Rihova et al. 2013). Because groups of people may share perceptions, transformative service research often treats value (or ‘well-being’, the term more often used within this literature) on a group level; for instance, the well-being of families or communities (Anderson et al. 2013; Rosenbaum 2015). Value is loosely defined in the earlier SDL/service logic literature and addresses private value, such as the customer being ‘better off than before’ (Grönroos 2008, 303), or that the customer’s ‘well-being has somehow been improved’ (Vargo, Maglio, and Akaka 2008, 150). Emphasizing groups, in transformative service research, well-being includes such items as literacy, health, access to services, and an absence of discrimination (Anderson and Ostrom 2015).

Expanding on the notion that value may not only be created but also destroyed (Alford 2016; Echeverri and Skålén 2011), transformative service research claims that destruction should be understood in relation to other groups: in efforts emphasizing one group’s well-being, another group’s well-being may be diminished or destroyed (Anderson and Ostrom 2015; Anderson et al. 2013). Thus, it is argued that a greater focus should be put on identifying disparities between groups concerning items such as well-being and possibilities to achieve well-being, and on actively improving well-being for disadvantaged groups (Anderson et al. 2013; Kuppelwieser and Finsterwalder 2016). Moreover, this should be achieved in a collaborative fashion with members of these groups; for example, ‘to find the best tools that help them reduce their own poverty’ (Fisk et al. 2016, 52).

Thus, by recognizing the impact of social structures, extended SDL offers a nuanced description of individuals’ prerequisites to take action and to participate in interactions; for example, that social forces impact areas such as access to public services differently for groups. Because group members may share perceptions and also value creation processes, the group level is an important level for transformation of public services. The emphasis on social context in extended SDL also implies the importance of such as community, family and other entities in efforts such as targeting disadvantaged groups (Blocker and Barrios 2015); an important feature that could also inform PSL developments.

Method and setting

In this article, data are derived from a collaborative and longitudinal research project launched to increase immigrant women’s understanding of, and participation in, a cancer screening program. As a researcher, I was involved as an action researcher, conducting research with rather than on or for staff and local inhabitants (Bradbury 2010).

Setting and participants

The Swedish screening program to prevent women from developing cervical cancer is commonly argued to have been successful, not least by contributing to decreasing the mortality of the disease by 60 per cent since the 1960s (SBU 2008). Today, 450 women are diagnosed with the disease and 140 women die annually (Andrae and Sparén 2017). Of the latter, over three quarters did not have a test taken within the recommended screening interval (Andrae et al. 2012). The standard practice of the program is to send out invitations to all women between 23 and 60 years old, every three to five years depending on age. Translations in commonly spoken languages accompany the invitations. However, this one-size-fits-all solution does not fit all – participation rates vary across the Swedish population.

In one multicultural district in one of Sweden’s largest cities, the participation rate in the cervical screening program was as low as 57 per cent, compared to 82 per cent on average for this part of the country (VGR 2010). This district also scored poorly when considering other health indicators, including smoking habits, obesity, and physical inactivity (Olsson and Panifilova 2009). Similarly, the district suffered from high unemployment and poverty levels (GS 2010). In this area, about 20 local immigrant women were involved in a project supporting other immigrant men and women during pregnancy and childbirth. These women had lived in Sweden for a relatively long time, spoke Swedish, and were well integrated into society. In supporting future parents, they offered information in their mother tongues, helped to explain the Swedish healthcare system, and participated in parental education together with staff at the local antenatal clinics. These 20 local women were asked to collaborate in the current research project as representatives of local immigrant women. Different from volunteers, but similar to other co-producers (Humphreys and Grayson 2008), the representatives in this project were paid on an hourly basis by the healthcare provider.

Data collection and analysis

In this project, data were collected both qualitatively and quantitatively. It was collected qualitatively through two semi-structured group interviews with the representatives not only to understand their experiences and expectations of the current screening program, but also to design a screening program that was more adequate in the local context. Four and nine of the representatives participated in the first and second group interviews, respectively. All 20 representatives were invited to participate in the group interviews, which were conducted at the local hospital, lasted 1.5–2 h, and were recorded and transcribed. The group interviews were conducted around three broad areas: barriers hindering women in the local context from taking the test; how the screening program could be redesigned to overcome these barriers; and how to involve locals in so doing. Qualitative data were also collected through continuous meetings with representatives and staff and thorough observations during the execution of the new program, such as during outreach activities and interactions with locals. In these meetings, participants shared prior working experiences, the particular challenges in the local context, as well as difficulties that emerged during the project. Accessible quantitative data (VGR 2010) of participation rates among the districts in the region were analysed to identify groups that the program did not reach. Quantitative data of tests taken were retrieved from the university hospital laboratory on a monthly basis during the intervention year to keep track of the number of tests taken, and were compared with the corresponding numbers from two years prior to the intervention year. The quantitative data were continuously shared within the project group, as well as with politicians, management, healthcare professionals, and other stakeholders to facilitate transparency.

Qualitative data were analysed using the content coding explained by Graneheim and Lundman (2004): the audio files and transcriptions were coded after being listened to and read through multiple times; based on similarities and differences, the codes were sorted into categories; the categories were compared and categorized under a number of emerging themes of barriers for participation: information, the Swedish healthcare system, practical issues, and the environment. A control chart (Bergman and Klefsjö 2010) was used to monitor and analyse the number of tests and was mainly used to understand if the new design made any difference in participation rates.

Finally, a note on data limitations is needed. Although most of the representatives participated in the group interviews, only two interviews were conducted, so the findings must be treated carefully.

Findings

The findings in this section are based on three of the co-production activities identified by Osborne, Radnor, and Strokosch (2016): evaluation (sharing experiences of the barriers of participating in the screening program), design (suggesting ideas to improve the screening program), and delivery (executing the new ‘locally adapted’ design). Thus, delivery includes to act as a bridge by providing cultural and locally adequate dissemination of information about the disease as well as the screening program to local women. The data are derived mainly from the qualitative findings based on the group interviews with the representatives. The process was not as linear as it appears in the following; for example, some ideas from the design phase were adjusted and improved based on difficulties that occurred when they were delivered.

Evaluation

One primary reason the screening program did not attract as many women in the local district as compared to other districts was simply lack of information; for example, that ‘far from everybody knows what it is’. The representatives argued that it was not only the cervical cancer screening program that was unknown to the local women, but also knowledge of the Swedish healthcare system; for example, the very idea of preventive healthcare services, which may not exist in their home countries. Many of the representatives said that some locals were not accustomed to seeking healthcare services if they were not sick, just as one only goes to the dentist when in pain: ‘They must get into the system of prevention … most of the ones that are new here don’t care about it, they think it is the same thing as going to the dentist’.

Similarly, many women had only visited a gynaecologist or midwife during pregnancy or ‘if something had happened’ or ‘if there’s something going on, gynaecological stuff, like something’s changed’.

In addition, women’s environment could constitute a barrier. For example, the representatives said it was possible that some parents would prevent their daughters from taking the test because they thought their daughters were not sexually active and mistakenly believed that the virus causing precancerous cells was only transmitted sexually.

Even though the invitations were translated to commonly spoken languages in the local district, many thought they were hard to understand. They described the invitation as complicated and containing difficult language: ‘… the words are very hard to understand, even for us who have lived here for a long time and been doing healthcare stuff, how is it for people who have only lived here two or three years?’

Practical issues also hindered women from taking the test. For example, representatives claimed that having children made it complicated to take the test because there was no one to look after the children, and the locals assumed they were not allowed to bring friends or family to the waiting room. Even if one understood the invitation, the procedures to reschedule appointment times were anything but easy, and paying the patient’s fee [roughly 10 euros] was complicated, even though the fee itself probably was not a barrier. Those who had taken the test thought that the staff did not provide enough information about the purpose of the test, which could be one reason women would decline future tests: ‘… there are many who take it [the test], but they have no idea what they have been doing … because there was no interpreter or nobody who explained what they had been doing’.

Design

The representatives suggested that the written invitations needed to be shortened and rewritten in easier language. Moreover, some translations were wrong. However, many of them argued that there was little use in improving the written invitations, and that the focus should instead be on spreading information about cervical cancer prevention face-to-face: ‘… when I stand in front of a woman and talk to her, then I and my words stick. But a piece of paper … you forget’.

Orally spread information was important because some women in the local area could not read in their mother tongue. However, orally spread information did not need to be disseminated face to face, but could also be spread through social media or radio. Because many of the representatives were already active in local associations and knew a lot of shop owners, they could identify a number of places in which the importance of participating in the cervical cancer screening program could be discussed. There were also local events that attracted a lot of people, which provided a good opportunity to inform. To communicate information accurately, the representatives proposed that they receive education by healthcare staff: ‘We want to learn more about this!’ However, some representatives thought that the topic would be ‘too sensitive’ to discuss in gender-mixed groups such as Swedish language classes.

The representatives thought that the idea of cervical cancer prevention was something very positive and something that should be emphasized because, to many women, ‘cancer’ was a frightening word, because they thought it could be neither prevented nor cured. They suggested some kind of campaign approach, or ‘happening’, to highlight the positivity.

Moreover, the representatives thought that it would be important to communicate that it was possible to bring a family member or friend to the clinic, to help look after children while a woman had the test. This family member or friend could also help with translation or comfort the woman, as many of the women had probably not had the test before. The representatives suggested that they themselves could help look after the children because ‘they [the mothers] will feel safe with that’.

Delivery

During one year, the screening program’s new design was launched in the local district. A midwife educated the representatives about the basics of cervical cancer and the screening program. The representatives then played an important part in delivering information to other local women and in participating in dialogues. Many representatives participated at meetings in various associations to spread information. At first, it was difficult to talk to people, but after some time it improved: ‘They have become more aware of the test … it is more discussion now, people ask more questions and approach us for more each time we are there’.

The representatives also informed women at local squares and shops. Here, too, they received a better response from the locals, possibly as a result of learning how to inform. Many representatives said that they talked to women who had lived in Sweden for a long time and never had the test:

I remember one woman who had lived here for five, six years and never had the test taken … she said ‘I don’t want to do it’. I sat down and talked to her for a while … Next Monday she showed up together with three of her friends.

The representatives were unable to answer some questions from the locals and were therefore provided with additional education by a midwife. Sometimes the representatives collaborated with midwives when informing women at local squares and associations; the midwives provided healthcare-related information, and the representatives translated and adapted the information: ‘we explain in their language, in our way, that we care about them’. Some women had been afraid to take the test, but the representatives knew how to comfort the women, and eventually convinced some of these women to take the test.

When the representatives informed women at public places, some local men approached hostilely and questioned what they did. The representatives dealt with these difficult situations by helping each other. However, the interest that men showed was largely positive and rather surprising to some of the representatives:

The most interesting thing is when men ask what it is and after a while bring their wives […] he was so afraid, came with his wife and looked after the children and said this project was great and ‘continue to support us’, I think that was what I liked the most with this project.

Some of the representatives participated in filming short films in Arabic and Somali about the importance of the test. These films were then spread on social media. The representatives also participated in radio interviews, recorded information in different languages for a phone service, and helped to improve the written information on the invitations.

During the one-year delivery phase, the representatives said, their role seemed to become more established in the local district, and their ability to inform appropriately increased. They argued that providing oral information was the best path because oral information spreads quickly and easily. They could also provide feedback to the healthcare provider about issues the locals wanted to know more about; for example, mammography or different vaccination programs. Some argued that focusing on only cervical cancer, the focus of the project, was probably too narrow and suggested a broader approach focusing on ‘women’s health’. Indeed, one representative perceived that the healthcare provider had ‘opened up’ through the project: ‘Sometimes one doesn’t need to follow the rules and those books and the things one have learned. Sometimes one needs to skip it, and they [the healthcare provider] did that through this [the project]’.

Representatives as co-producers

Through the representatives’ involvement in evaluating the existing screening program and identifying barriers, designing a more locally relevant program, and delivering many of the newly designed ideas, participation in the screening program increased by 42 per cent in a year (from 4141 to 5900 tests). Overall, experiences were positive, and the representatives said that they learned a lot by working together as well as with the midwives. They sensed that they had done something important in the local community, and felt that they had been strengthened and had gained self-confidence.

However, the central issue of their representation needs to be examined: the question is whom they represented. As mentioned, the representatives in this project were paid, while many of them also had other jobs or were studying. They had lived in Sweden for relatively long periods, were integrated into Swedish society, and all spoke Swedish. Thus, in some respects the representatives are likely to be quite distant from the target group in this project. For example, the representatives did not think the fee for taking the test would present a barrier because they thought it was relatively inexpensive. However, a midwife recalled meeting women for whom the fee was an issue. Moreover, the decision not to target gender-mixed group was the representatives’ idea, but was probably faulty because many men were interested in the information and passed it on to their female family members. Despite this, the knowledge and skills of the representatives were important in identifying barriers, redesigning the program, and executing it. Their participation was also important because they brought their own networks into the service system.

Discussion and implications

This article has sought to contribute to the development of PSL by explicitly recognizing social context. The definition of such context is twofold, as it includes social structures as well as individuals and institutions (Burke, Pasick, and Barker 2009). The extended SDL literature may be important in developing ‘social-context-sensitive’ PSL because it addresses both aspects of social context: the impact of social structures on service exchange, and the multiplicity of resource integrating individuals and institutions in value creating service systems. The latter is starting to gain recognition in PSL (Osborne 2018; Osborne, Radnor, and Strokosch 2016), so I have emphasized the former, social structures, that enable and constrain prerequisites for inhabitants’ actions and interactions differently based on their characteristics (Lundquist 1984). Such structures also explain disparities among groups in accessing public services; some people face barriers that others do not. However, the aspect of social context covering multiplicity of actors offers possibilities to address such inequities. Involving individuals to represent hard-to-reach group members may help include the interests of the disadvantaged in co-production, referred to herein as representative co-production.

However, the novelty is not co-production that expands beyond the individual user, but the representation, which becomes necessary as the target group is non-participants in a cancer screening program. Consequently, recruiting non-participants to participate in co-production may be difficult. Rather, those invited to co-produce were women living in the same segregated area of the city, sharing culture and language with those the healthcare system failed to reach. Because of their commonalities in background with those hard to reach and their well-established role in the local community, the invited co-producing women represented the women of the target group. Thus, representative co-production becomes a type of group co-production that includes a strategy to address inequities in public services.

The representatives’ co-production efforts are similar to those explained in traditional public administration literature (ideal type 2) in that they are optional and include their participation in assessing and improving a public service (Osborne, Radnor, and Strokosch 2016). The representatives’ co-production efforts may benefit individual women of the target group to participate in the screening program and to experience the service meeting (ideal type 1), in which their co-production is considered essential (Normann 2001). Moreover, the representatives also included their own networks in service delivery, integrating resources from actors new to the particular service, thus helping change the service system (ideal type 3). Thus, to enable local women’s interactions with staff, the representatives’ co-production activities were both indirect and direct. They were indirect in that they contributed with input to design services that better fit the local women. They were direct in that they worked along midwifes when delivering the service, thus enabling the ‘moment of truth’ by acting as a cultural bridge between local women and staff.

Figure 1 specifies the activities that the representatives undertook in the findings in this article, and connects co-production to co-creation.

Figure 1. The potentials of integrating social context with the PSL.

In Figure 1, PSL integrating social context recognizes the impact of social structures on individual’s actions and interactions. In an extended SDL fashion (Blocker and Barrios 2015; Kuppelwieser and Finsterwalder 2016), the starting point in the illustrative case was to investigate disparities in the screening program to identify groups not reached by healthcare services. Representatives were involved to increase understanding of what constrained disregarded groups’ value-creation processes (Anderson et al. 2013; Anderson and Ostrom 2015). From a co-production perspective, the provided case addresses the voluntary involvement of local representatives in the evaluation, design, and delivery of public services (Osborne, Radnor, and Strokosch 2016).

The overriding categories of barriers for immigrant women included lack of information about the purpose of the test, unfamiliarity with Swedish healthcare system and disease prevention, practical issues such as unavailable childcare, and the women’s environment, such as their own parents’ knowledge of the test. One could argue that the standard practice of the screening program was designed for ‘Swedish-born women’ who know the intent of the invitation and are socialized to preventive healthcare services. Based on the representatives’ accounts in this study, many immigrant women are not familiar with preventive healthcare.

I argue that a ‘representative co-production approach’ offers potential not only to identify shared barriers among group members in accessing services (evaluation), but also to integrate their intangible resources in designing services more relevant to meet the needs and expectations of other local users from the same population group. However, the representatives’ involvement may also include an active role in delivering the new design to other members of the group. As indicated by the arrows in Figure 1, evaluation, design, and delivery may be understood as interconnected phases in the representative co-production of public services.

An important reason for inviting the representatives to participate in the project was the relational aspect – their long-established relationship with the community in general, and specifically with women in the local district. The relationship between service provider and customer is crucial in SDL, not least for the provider to understand the customer’s process of creating value and thus adapt their value proposition (Payne, Storbacka, and Frow 2008). In the provided case, however, neither staff nor management had any relationship with many of the local women, simply because many of these women did not visit the clinics, and also due to barriers such as communication difficulties for those women who did visit. It would be difficult for service providers to establish a relationship with non-visitors of their service. A strategy for reaching disadvantaged groups could be to build on established relationships to understand group members’ value-creation processes, and also to convey information, and possibly to bridge the gap between the service provider and the groups. Another benefit of involving representatives with good relationships in the community is that their networks become involved as well. Indeed, the networked nature of value creation (McColl-Kennedy et al. 2012), rather than the linearity of the value chain (Porter and Millar 1985), was evident in the studied case, in which associations such as local shops became involved in disseminating information. It is doubtful whether the healthcare providers would have had access to these arenas themselves.

Figure 1 also incorporates the link between co-production and co-creation, as elaborated by Osborne, Radnor, and Strokosch (2016) and Osborne (2018). Because the PSO cannot produce and deliver value – only potential value as in value propositions (in this article: information and invitation to take a test) – representatives co-produce the value proposition. Real (private) value is realized by the individual user by combining the co-produced proposition with other actors’ resources – something that take place in interactions situated in her broader life situation (Eriksson et al. 2016; Osborne 2018). In this sense, co-production by representatives precedes value realization by the individual end-users. In other words, representative co-production may enable co-creation for the user by redesigning PSOs’ value propositions to better fit the value creation processes of the users.

Representative co-production may be a fruitful strategy to address inequities of inaccessibility for certain population groups to public services. Thus, the representatives’ efforts mainly focus on benefitting others (Elg et al. 2012). Although the representatives’ co-production activities (evaluation, design, delivery) elaborated in this article may not be new to co-production theories, public participation literature, or healthcare practice, it is important to recognize that it is potential value (proposition) that is co-produced, not real value. Thus, the impact the representatives’ co-production has on individuals’ value creation (or destruction) processes is important. In other words, the activities labelled co-production in this article are not an end in themselves, but improve services that benefit individual service users or inhabitants by taking place in their broader life world.

Conclusion

Oftentimes, the goods-manufacturing logic of many NPM reforms has proved unfit for the public sector. Standardized, ‘one-size-fits-all’ solutions have failed to address the diverse needs and prerequisites of some social groups. The emerging PSL offers an alternative, elevating the largely passive public service user of the goods-manufacturing logic to an active role as co-producer. However, to target disadvantaged groups in society, it is necessary to integrate social context into the PSL more than is the case today.

Social context impacts the possibilities for people’s actions and interactions differently. It also suggests that group members – through social processes – may have shared needs, expectations, and experiences of public services. Such an approach may be used to reveal disparities, such as poor access to healthcare services for certain groups. Such an approach may also enable co-production in the PSL to expand beyond the individual by focusing on groups of individuals in representative co-production.

By integrating the unique knowledge and skills of representatives who share features with other group members, problems such as shared barriers to accessing public services may be identified. For public service providers, a way to offer the best possible service for disadvantaged groups is to involve the representatives further – in designing and delivering a service more likely to meet the needs and expectations of other group members. Thus, by integrating social context, the PSL may explicitly identify, involve, and address groups of service users or inhabitants. Importantly, the representatives’ co-production effort may enable the individual users and inhabitants to achieve value creation processes in their broader life situations.

The implication of this study to public management is the importance of involving representatives of groups that the public management knows little about, such as the non-participants in the screening program. The often long-established relational ties between representatives and other group members may bridge the user-provider gap, improving not only access to public services for disadvantaged inhabitants, but also access to disadvantaged group members’ needs and expectations. Public management and researchers should address not only the core service, but also co-production activities impact on the broader service system.

This article has aimed to broaden the scope of the PSL and co-production by integrating social structures and a representative group-level with the potential to support individuals’ value creation processes. To enrich our theoretical and practical understanding of the PSL, future research could investigate how social context impacts on public service user–staff interaction, depending on user characteristics, the profession of the staff, the type of public service, and so forth. Moreover, research could also empirically investigate collective forms of co-production’s impact on private, group, and public value and whether these are in conflict. Because this article has focused on representatives, the way in which the PSO acts and reacts in these co-production efforts has not been thoroughly elaborated upon. Future research could investigate whether co-production projects similar to the present case have led to a greater role for service users (or the representatives).

Limitations

This article has mainly contributed to group-level co-production and in developing PSL by focusing on the representatives’ roles. Thus, the contribution to practitioners or the wider public participation literature (Barnes et al. 2003) may be limited.

Acknowledgments

The author wishes to thank all involved actors in the study and the anonymous reviewers for valuable comments on earlier versions of this article.

Disclosure statement

No potential conflict of interest was reported by the author.

Additional information

Notes on contributors

Erik M. Eriksson

Erik Eriksson, PhD, works as a researcher at Chalmers University of Technology, Sweden. He has over ten years of working experience in the public sector. His research interests include healthcare and public sector management. Erik mainly uses a collaborative research approach together with practitioners, with the aim to co-create knowledge relevant for academics as well as practitioners.