Bringing all clients into the system – Professional digital discretion to enhance inclusion when services are automated

1.1 Aim

Digitalisation in general and robotic process automation (RPA) in particular is changing the accessibility and usability of digital public services. The aim of this article is to analyse strategies developed among front-line public case workers to enable clients to access and use public online services. These identified strategies will be viewed in relation to public values and the discretion of front-line case workers.

Two overall research questions guide the article:

RQ1: How do front-line case workers support clients’ use of digital services and RPA to enhance inclusion? RQ2: How is the discretion of front-line case workers influenced by digitalisation in general and RPA in particular at the two studied agencies?

This article proceeds with a contextualisation of the case studies in relation to policies and the governance of digitalisation and RPA in public administration in Sweden. Thereafter we present the theoretical framework responding to the research questions. In the fourth section, the case study methods are described, and the case studies are presented in the next section. Finally, we conclude and discuss implications for research and practice.

2.1 RPA in public administration

There are expectations that the use of digital and automated public services, like RPA and artificial intelligence (AI), can generate value for government functions (de Sousa et al., 2019; Bannister & Connolly, 2020). RPA is an umbrella term for digital systems that can learn and identify patterns in decision-making, based on an analysis of different data, and draw conclusions without human intervention (Čerka et al., 2017). Within policy, they are hailed as a solution to multiple problems in new terrains ranging from global sustainability and climate change to unbiased automated case management in the public sector for welfare services (Swedish Ministry of Enterprise and Innovation, 2019). AI, RPA and digitalisation in general are associated with a multitude of different benefits and optimistic, future-oriented values (Gasser & Almeida, 2017). Thus, there are high expectations for how they can improve public services. RPA is often initially applied to automate more routine-based processes, for example when taking data from one system and adding it to other systems (Willcocks & Lacity, 2016; Madakam et al., 2019). RPA as a form of algorithmic decision-making system could be cost-effective in public organisations when it comes to structured decisions, such as in situations where frequent and repetitive decisions need to be made (Bannister & Connolly, 2020). Such arguments are frequently used in policies.

If automation is to be efficient in public administration, almost all clients should use the digital channels. Therefore, the policy principle ‘digital first’ is key in enhancing inclusion. The European Digital Strategy and the Swedish National Digitalisation Strategy are both based on these expectations and are guided by the ‘digital first’ principle. This principle states that digital channels should be the first choice in all relationships between public organisations, as well as among public organisations, citizens and businesses. Every citizen is supported and encouraged to use services through digital channels in the first instance. This points towards the importance of supporting and enhancing the discretion of case workers, in addition to supporting the design of systems.

2.2 Digital inclusion as a prerequisite to automate public services

There is currently a variety of aspects that can exclude different user groups from accessing digital public services beyond that of a lack of Internet access or knowledge. Considering the extensive role of technology in today’s society, digitalisation inevitably changes how equality is formed in new social settings (Robinson et al., 2015).

Research on digital divides was initiated as a descriptive approach for mapping digital access and use. The concept was introduced to a broader audience and given political attention at the turn of the century (Norris, 2001) but it has not yet been resolved since technology and practices continually develop. The concept now has several meanings and implications, and there is a need for more advanced models to describe and address digital inclusion challenges (Garcia-Garcia & Gil-Garcia, 2018; van Deursen & van Dijk, 2019). It is clear that ‘offline’ socio-economic stratification is reproduced in online contexts. There are new aspects in addition to these existing stratifications that can exclude social groups to varying degrees in digital contexts. Education, income and social networks are still strong predictors of digital inclusion and identity, particularly in connection to the labour market that forces people to use and learn new technologies in a contextual setting (van Deursen & van Dijk, 2019).

There is a complex plurality of inclusions in the digital context. There are so many ways to be digitally excluded that it is possible to see a digital diversity (Bernhard et al., 2019). Two simple examples from the case studies are enough to confirm the difficulty of predicting inclusion and the use of services. For example, the young gamer spending most of the day online playing interactive games can still be excluded when it comes to participating in democratic discussions or when applying for unemployment benefits as a public welfare service. The middle-aged man with a permanent blue-collar job and income, on the other hand, can feel completely excluded when his bank asks him to only use online services or when he has to submit his income tax declaration online. When people experience digital exclusion, they often turn to public libraries. A Swedish study of libraries’ practices for digital inclusion found that there is no single clear, specific strategy to support digital inclusion (Bernhard et al., 2019). Similar results have been seen in studies of digital inclusion at public libraries in the US, despite the different welfare systems (Mersand et al., 2019; Bertot et al., 2016). The research from the library context points at the critical role by front-line professionals which motivates our research design of this study.

When public services are delivered through digital platforms it requires that the client has sufficient opportunity, and competence to both access and use these technologies to obtain public services. New forms of digital inequalities follow in the wake of automation. Subsequently, there are new challenges regarding allocating staff at public agencies to support users and their needs, especially in terms of ensuring that non-users get guidance and practical help with using government services (Bernhard et al., 2019; Garcia-Garcia & Gil-Garcia, 2018; van Deursen & van Dijk, 2019). Clients must learn to use the public service system, even when using it on a single occasion, and this has been shown to be a learning process that not everyone finds easy (Hansen et al., 2018).

3.1 Professional front-line case workers

Front-line case workers in the public sector have a strategic function to translate and implement public policies and to balance different aims. We focus on the impartial and inclusive welfare service provision in contrast to demands for efficiency through RPA. Since public administration practice is set in a framing of specific institutional logic, digital discretion for front-line workers must also relate to the specific assignments given through political decisions and the logic where their practice takes place (Busch et al., 2018).

Public sector work involves digitalisation when restructuring work, re-organising public information and knowledge, and re-orientating official-citizen relationships. Digitalization has even been argued to transform street-level into system-level bureaucrats, working rather with and through the systems instead of the clients; the system analysts and software designers becomes key actors (Bovens & Zouridis, 2002). In this line, accountability also becomes transformed in digital government praxis, forming new, less personal relations between clients and front-line case workers (Petrakaki, 2018). The digitalisation of public administration has both enabling and constraining effects on front-line case workers that are often part of the street-level bureaucracy, as shown in a research overview conducted by Buffat (2015).

3.2 Values guiding professional front-line case workers

Core public values, such as transparency, efficiency and impartiality, are discussed in the field of digital government (e.g. Cordella & Paletti, 2018). However, there is a lack of specific and single common definitions and there are many taxonomies of public values (Bannister, 2017). The public values of digital government are instead related to their circumstances and given contextual meanings. From such a praxis-oriented view, public values can be grasped through three categories: duty-oriented – the duty of public administrators to the government and the state; service-oriented – to provide a high level of service to citizens as well as the resilience such service requires, such as respect for the individual citizen and transparency; and socially-oriented – broader social goals such as inclusiveness or fairness (Bannister & Connolly, 2014). This categorisation points at the implications of the values and how they are guiding the strategies and actions in praxis as opposed to the theoretical meanings and reasoning behind the values.

For a legitimate public administration, the systems which replace human decision-making with machine decision-making need to be designed with all public values in mind (Bannister, 2017). When it comes to the general policy principle of ‘digital first’, it is obvious that there must be a balance between the core values of efficiency/economy and equity/fairness. Properly designed and legitimate bureaucratic system has to have multichannel access, according to Bannister (2017). The public sector use of AI and RPA for supporting or making decisions is imbued with values and biases which may discriminate against both individuals and groups in society (Bannister & Connolly, 2020). A recent American study (Snow, 2021) of how social workers in daily practice work with RPA, concludes that a traditional and archetypal form of street-level bureaucracy is based on duty-oriented values. In addition, the they strive to maintain socially-oriented values.

In an extensive research overview, Twizeyimana and Andersson (2019) conclude that despite the limited numbers of studies on public values and digital government, three types of values are identified with the normative ambition to improve public administration, public services or social values. It is interesting how these closely relate to Bannister and Connolly’s more theoretically grounded analysis (2014). The values relating to duty do improve public administration and back-office organisations, the values relating to services do improve public services at the front-line, and the socially-oriented values relate to improving social values in general. Since Bannister and Connolly’s (2014) categorization is commonly used it will also frame our analysis.

3.3 Digital discretion among front-line public case workers

The concept of discretion is mainly used in law studies but it may also be useful for public administration. Discretion has emerged as a particularly useful concept in relation to various forms of automation, as shown in a systematic literature review by Busch and Henriksen (2018). Their study is based on 44 peer-reviewed articles on digital discretion over the last 20 years related to the traditional interpretation that street-level bureaucrats have broad scope to exercise discretion i.e., using personal competences in decision-making. Digitalisation is commonly suggested to reduce their discretion. By contrast, the authors show, using theories of public values, that digital discretion tends to strengthen ethical values and weaken democratic values, as well as professional and relational values. However, they note that there are strong contextual factors influencing the conclusions reached. The included studies show that the digitalisation of public services reduces human support for those clients who cannot manage the digital services themselves. They conclude that digital discretion, as expressed when using computers instead of interacting face-to-face with clients, gives less support for the clients and users of services. To conclude, Busch and Henriksen’s (2018) research overview indicates that automation and increased digitalisation increases the risk of digital exclusion among clients due to drastically decreased face-to-face contact with human public servants across all studies.

Case workers at the front-line of public administration are supposed to act in line with public values and translate them into their practice when meeting with clients and end-users. They make decisions and supervise citizens based upon the power delegated to them in their professional position. This form of power shapes the discretion of the public official based on professional competences, general legal frameworks and specific guidelines. Busch et al. (2018) use the term ‘public service values’, stressing that digital discretion can strengthen ethical and democratic values but weaken professional and relational values. When related to the categories discussed above, it can be seen as socially-oriented values being strengthened and duty- and service-oriented values weakened. When new technologies are introduced and implemented in an organisation, there is a need to change how work is organised and form new strategies involving RPA (Busch et al., 2018)

Busch et al. (2018) conclude from case studies at two Norwegian public agencies that front-line case workers consider professional discretion to be necessary to accomplish more complex tasks. This requires professional competences to make decisions with the discretion in regard to the rule of law and impartiality in particular, but also with the flexibility to adjust to and see each client’s specific situation. Professional public case workers act within an institutional logic that forms their organisation and to which they much learn to adapt.

When RPA becomes a co-worker in daily practice, there are implications for the degree of discretion amongst front-line case workers in terms of their capacities and competences to support users of public services (Hansen et al., 2018; Busch et al., 2018). Hansen et al. (2018) discuss whether discretion will disappear with the implementation of digital services since the implementation of digital services will lead to less face-to-face interaction and, accordingly, less discretion. In summary, digital discretion is formed in line with general values and assignments of the public agencies and set in the context of democratic governance striving towards inclusion.

3.4 Strategies among front-line case workers

To conclude, previous research implies that the professional front-line case workers in public administration use their discretion to form strategies within the specific institutional arrangement and that they continuously must adapt (Busch et al., 2018). It is argued that the discretion of professional front-line public case workers changes due to the implementation of RPA and subsequently have combine the demand of fulfilling the duty of providing services alongside that of social awareness. Additionally, they develop competences and discretion to be able to support those who cannot use the RPA systems themselves. This study focuses on these strategies developed among front-line case workers. Firstly, how front-line workers act to help clients to use digital services and automated systems is investigated. And secondly, we examine how digitalisation in general and RPA in particular influence the discretion of front-line case workers and how the strategies relate to public values.

4.1 Case studies

This article presents case studies at two of the main public agencies in the Swedish welfare state: the Swedish Public Employment Service (Arbetsförmedlingen (AF)) and the Swedish Social Insurance Agency (Försäkringskassan (FK)). These were selected as they both are national agencies, but also have the discretion to shape local practice and hence work in different ways at each local office to some extent. We have included case studies from two local offices. Furthermore, they are agencies that manage social insurance schemes for targeted groups in need of financial support when unemployed, on sick leave or parental leave, or in specific pension programmes. These agencies thus deal with issues regarding inclusion more frequently than agencies relating to businesses or other public organisations, as shown in reviewed research and that was guiding our field studies and analyses.

RPA is deemed to be a key tool in efforts to ensure more efficient and impartial provision of the extensive national welfare schemes as expressed in Swedish national policy documents (Swedish Ministry of Enterprise and Innovation, 2019; Government Offices of Sweden, 2017). The Swedish Social Insurance Agency (FK) uses RPA to manage almost all applications for short-term parental leave, which is one of the most commonly used social security schemes in Sweden (Försäkringskassan, 2019). The agency’s service for citizens is accessible via a simple online service to report the child’s and parent’s social insurance numbers. The payment is then processed and paid into the client’s personal bank account. The Swedish Public Employment Service (AF) has developed a more advanced form of RPA to suggest available jobs based on an algorithmic service that matches several types of personal data and preferences with relevant job opportunities. These systems demand a high level of information security and more advanced log-ins, making it more complex to access the services and requiring clients to have increased digital competences. All this information was gathered from policy documents, instructions for users at the agencies’ websites, and other types of information materials. These cases were strategically chosen based on international research and national policy documents to see the early impacts of RPA as general public services strive to include all clients. In line with the strong emphasis on the ‘digital first’ principle, both agencies have significantly reduced the number of local front offices and contact centres have now been set up to support clients in using their digital services.

4.2 Case study method, data collection and analytical approach

This case study focused on the daily interactions among front-line case workers and clients at the local contact centres in the western region of Sweden. This was selected as a convenience sample as it was accessible for research. Both contact centres were located in areas with generally lower social and economic status.

The case studies were based on direct observations of practices at the service centre, interviews with front-line case workers and managers, and document studies. Our qualitative approach focuses on the perspective of those involved and to open up for an analysis of strategies among front-line staff and their relationships with public values as was identified as critical points in the literature review.

The case studies began with four open field observations at the agencies in spring 2018 and built upon earlier studies at Swedish contact centres for public services (Bernhard, 2014; Bernhard & Wihlborg, 2015). The observations took place at contact centres and included informal discussions with front-line case workers and clients. In the field notes we focused on how front-line case workers dealt with clients who were unable or unwilling to use the automated services themselves, or who did not have the necessary competence. We observed practices and how the front-line case workers operated and dealt with their interactions with clients, which language they used, and how they instructed and engaged with the clients, the digital tools and the services. Most clients first tried to use the digital technologies (RPA terminals) themselves and thus the first points of contact with the front-line case workers were when problems were incurred. Our observations focused on these critical moments and the strategies of the front-line case workers in trying to bring all clients into the system. Based on these situations, our observations and follow-up questions focused on the design and position of the RPA-terminals, the additional printed instructions (which were provided in several languages with images), the organisation of the contact centre area, and the front-line service workers interpretations of their work. Field notes, both textual and images, were taken on general aspects and also in detail in relation to specific events, locations and individuals. Each observation lasted for approximately 1.5 hours.

Based on our analyses of the observations and our literature review, we formulated a guide for semi-structured interviews (Kvale & Brinkmann, 2009). The interviews focused on how the front-line case workers assist clients with digital and automated public services. To guide the interviews in line with the research questions we focused firstly on their strategies to enhance the inclusion of clients and their digital competences and secondly, we discussed their discretion based on certain situations mapped during the observations to reveal challenges, opportunities and underlying values. The interviews also included some questions regarding their personal professional background, education and internal professional training at the agency. Four interviews, one with a group of three participants, all lasting 1.5 hours were conducted in 2018 with strategically chosen front-line case workers at two agencies. A follow-up interview with one of the respondents was conducted in 2019, focusing on how to enhance inclusion for clients with special needs and disabilities. All five interviews were carried out in Swedish and fully transcribed by one of the authors. Both authors analysed the interviews using a grounded reading, categorizing statements in relation to strategies and discretion, and finally in the broader terms of public values.

Our first reading of the entire material provided general impressions and ideas we discussed in order to focus and strengthen our theoretical framing and define more detailed concepts, as described above. In accordance with the abductive research approach, theory reading was alternated with an explorative empirical analysis of the observations and interviews. Since the research in these fields develops and quickly and additional analyses have been included in the phase of analysis. Tentative analytical points has also been discussed and elaborated on with both representatives of the agencies and researchers at workshops and conferences. All presented quotations from the interviews and the field notes were translated by the authors and some editorial adjustments were made.

5.1 Public agencies in Sweden and their introduction of RPA

The government agencies in Sweden implement and deliver services in line with national policies. They are state-controlled organisations with a strong focus on national equality, in contrast to the domination of local autonomy that otherwise characterises the Swedish welfare state (Feltenius, 2015). All national agencies have a high degree of independence from the government (Esping-Andersen & Korpi, 1986), meaning that all government agencies must devise their own digital case management systems. The national agencies play key roles in relation to welfare policies regarding income maintenance through the extensive social insurance systems assisting vulnerable groups in the society. The agencies included in this study provide social insurance to citizens in the event of unemployment (AF), parental leave, sick leave and certain public pensions (FK).

Front-line public case workers support the agencies’ clients and are instructed to increase client’s usage of a terminal where an RPA system is supporting the digital services. It is argued that the digital applications and online services are more secure, save time and easier to access around the clock (FK website). Both agencies deal with clients in vulnerable situations of unemployment and leave of absences for different reasons. FK manages many types of social insurance within the extensive Swedish public welfare system and thereby has many more different user groups.

The Swedish Public Employment Service (AF) manages the public unemployment schemes, support and matching for new employment, and various training schemes. The ‘digital first’ principle is applied, and all services are provided online. There is RPA for administration of unemployment payments, checking certain criteria and making automated disbursements. There is also an RPA system based on extensive personal information to enhance matching for new employment opportunities. When this service was introduced, it requested higher levels of information security and the national standardised e-ID system was required to access all of the agency’s digital services. Our field study was carried out when this more advanced use of e-ID had only just replaced the PIN code log-in. The new RPA presented an additional barrier to the services.

The Swedish Social Insurance Agency (FK) uses RPA mainly in the standardised payment of short-term parental leave, which caregivers for children under the age of eleven can use to obtain compensation if they have to care for their child at home instead of working. Short-term parental leave is one of the most frequently used social security schemes in Sweden, with approximately 900 000 users being paid within the scheme yearly. RPA for short-term parental leave must be accessed using e-ID and the interface is provided as an app and on FK’s website. RPA checks the caregiver’s relationship to the child, their income and their employment status before making an automated payment to a registered bank account. RPA does not allow the caregiver to provide any additional information, although this is possible when using paper forms. The paper form is still available and used, but payments are delayed due to the manual administrative processes. These differences have been highlighted in an internal audit report at FK, arguing that impartiality has to be maintained regardless of the channels used for applications (Försäkringskassan, 2019). There are currently no other extensive RPA solutions at FK. In the same report, FK’s management argued that one reason for a reluctance to use more RPA services is the need for a high degree of professional discretion to manage individual assessments.

At both agencies, RPA has been developed to enhance efficiency in the agencies’ administration of welfare schemes. Since these changes require higher levels of security through a more secure log-in. The higher thresholds to access services, that are even more problematic for clients in vulnerable situations, that have fewer digital resources and lower competences.

5.2 Strategies to support users when RPA is a part of the services

When we came to the FK contact centre for our first observation on a weekday morning, there were around 15 people already waiting outside an hour before opening. When the doors opened, the clients were then helped promptly by four front-line case workers, but there were continuously new clients lining up to receive help at the front desk or to use one of the five terminals.

A typical situation that was seen on most observation occasions can be described as follows. One client was using terminal in the service area themself, while a case worker showed another client with a foreign background how to use the terminal to apply for a Swedish personal ID. Another client was being shown how to make an appointment for an upcoming meeting at FK via the computer. One front-line case worker spoke in English when guiding and instructing a client how to use the terminal – none of them had English as their first language. Three other people with foreign backgrounds, likely an immigrant family, were being helped by another front-line case worker who explained how to book a meeting digitally to apply for ID cards.

In this setting, the front-line case workers supported the clients. Their ambitions were to promote the use of digital services and to include all clients in the digitalised case management system. In the interviews, one of the team managers reflected on the increased requests for support when discussing user requirements:

… we also get a good work environment, for our employees. We have to make sure that not everyone, the whole city, needs to be in the office at the same time, as it were. (Interview at FK)

Even if the main strategy is to help as many clients as possible, the managers see the stressful situations for the front-line case workers. He points out the need to use their professional competence in a sustainable way to address different needs and how they collaborate with the clients to promote use of RPA and self-service in line with expectations.

The RPA is designed to be efficient for most clients but instead increased the need for support and manual administration for those who cannot use the systems themselves. The welfare schemes managed by both agencies target clients in vulnerable situations, who commonly have fewer resources and lower levels of competence. Therefore they require even more personal guidance and support. One front-line case worker said:

There is a need for a lot of human resources when helping clients to use this stuff. Now we are better at doing this. It looks a little different in different offices even though we use the same system, the same rules, the same everything, but there are differences in terms of how you organise things and which staff you put in which places. (Interview at AF)

This shows how work processes and arrangements have been adapted to meet the needs of all clients by focusing on their particular needs. This can be seen as social values guiding how service values are expressed, in line with Bannister and Connolly’s (2014) conceptualisation. However, there were still obvious challenges when meeting clients at the front office who do not have the ability, or the personal technology (tablet or mobile phone) needed to use the services. As one of the front-line case workers explained when describing the process of applying for unemployment benefits:

All unemployed people must submit a report every month, describing what they did last month. It has been digitalised for a long time and the log-in was a four-digit password. But they changed it to make it more secure a few years ago. So now you have to access it using e-ID. When we did this, the number of digital users fell dramatically. Today, there are people asking for the paper form again. They no longer know how to manage it digitally and with e-ID. Some of the regulars we know in person and we just give them the paper form. But we try to encourage some of them to learn. Some bring an old mobile phone and say, “I can’t use the e-ID since I don’t have a smartphone”. Then we ask them if they have a computer and tell them that it is possible to use the computer. Then they say “Aha!”. We also ask if they have a bank card and when they say that they do, we ask them to go to their bank to get an e-ID on the card. We cannot fix it, we have to send them away. (Recounted during observations at AF)

This indicates that the more advanced RPA, despite being internally efficient, left more clients digitally excluded. It also shows the limits of the front-line case workers’ strategies. The only thing they can do to help the clients is to send them away to get a e-ID somewhere else. The frustration expressed during these types of discussion showed that the professional front-line case workers had high service-oriented values framing their discretion and building on an social approach that they had to compromise to fulfil more duty-oriented values framing their strategies.

In one of the discussions during the observations, a front-line case worker raised another aspect of the advanced digitalisation in the risks of having all data exclusively online in the RPA. He said:

Yes, it’s amazing as long as it works and the day it doesn’t, I’m glad I have 30 years of experience, so I know how to do it manually. (Interview at FK)

This reflects Bannister and Connolly’s (2020) argument that although the use of digital and automated public services may generate value for government functions, new risks are also embedded into the structure that will require new strategies from staff.

Taken together, and to answer the first research question, we can conclude that the front-line case workers commonly developed strategies to be more service-oriented in order to complement the duty orientation of the systems. Additionally, and in particular at AF, workers showed clear socially-oriented values to work around the digital systems to encourage and include those clients who did not use the digital services. It was obvious that the more automated the systems became, the higher the thresholds were for the most vulnerable clients and the more the front-line workers developed socially-oriented strategies. However, it was also clear that the strategies we identified were formed within their discretion, in line with other studies such as that carried out by Busch (2020).

5.3 Professional digital discretion among front-line case workers to support digital inclusion

The introduction of RPA has been shown to result in a shift towards a digital mode of discretion, which can have a positive effect on civil servants’ discretionary practices. In a study of municipal social services, the positive effects were mainly seen in terms of their ethical, democratic, and professional values (Ranerup & Henriksen, 2019). Here, we analyse the strategies used to include clients in the systems at the national agencies when introducing RPA.

To help clients use the online services, AF provides public computers in a service area called Customer Square. Here, clients can get help turning on the computer and finding the website, but for reasons of personal privacy the front-line case workers turn away and leave the client alone when they log in using their e-ID. The identification stage is critical in order to access services, since the personal data is the basis on which the RPA system provides the services. The front-line case workers have learned how to keep these limits clear and support the clients without helping with the actual log-in.

The new organisation of a ‘customer square’ at the contact centre changed the discretion for the front-line case workers at AF since they were no longer behind the desk, but rather a co-worker together with the RPA-terminal. Their new interpretations of digital discretion were still in a formative phase and one of them described how they form new professional strategies alongside RPA:

I have a role called ‘customer resource’. I meet the clients at the Customer Square which was set up in 2019 and I guide them to solve their questions using e-services if possible. We (my colleagues and I) are instructed to help as many clients as possible to access the online services. This is because AF’s management wants to emphasise a new way for us to work. We are a resource for the clients, we are not employment administrators. This is new for everyone, we have to learn … (Interview at AF)

This description indicates that the front-line case workers are developing strategies to complement RPA. Before, they were employment administrators working in the front office and making back-office decisions which have now been taken over by RPA. They experienced that their main assignment now is to make the clients use the RPA-terminals which communicate directly with the back-office and receive a decision. The RPA here act as new co-worker, changing the discretion of the human front-line workers.

At the same time, there were new barriers limiting the discretion of the front-line case workers. The most common barrier was clients lacking national e-ID. The front-line case workers at AF and FK could not provide the e-ID. They had to send the clients to other places/websites to arrange that and then clients must return to the agency. Even when the front-line case workers have a high degree of discretion and high service ambition, they are not able to fully support the clients to use RPA to make the work at the agency more efficient.

As illustrated by the description of the situation at AF above, RPA demands more secure log-ins and this constrains what can be seen as the clients’ discretion. They also showed the clients at the contact centre how information security has increased and that personal information is no longer shared among the staff, instead remaining ‘hidden’ in the system. Since personal and case sensitive information is less transparent for the staff, there is also the potential to make case management more impartial with a focus on legislative duty-oriented values. The ambition to design more advanced and efficient systems is in line with the complex legislative framework and with the intentions of the welfare policies behind the specific social insurance scheme to be inclusive and impartial.

Nevertheless, the improved security at the log-in stage prevents the front-line case workers at AF from supporting the clients since e-IDs are provided by other organisations. Here, too, some clients’ refusal to even attempt to use digital devices and a personal e-ID further constrained the discretion of the front-line case workers. This was illustrated in one of the interviews:

Such arguments from clients directly contrast the demands from AF’s management and government policies based on the ‘digital first’ principle. The front-line case workers are trapped between the clients they strive to deal with using socially-oriented values and the policy and management demands based on duty-oriented values. The workers struggle with this as they know that most of their clients live under tough conditions and are not always “… able to afford a smartphone or you can’t afford this plastic card that the e-ID is on, or so on” (interview at AF). Instead, the front-line case workers developed alternative strategies to promote digital inclusion within the limited scope which their professional discretion allows them, as mentioned in one interview:

This demonstrates that the front-line case workers have the professional discretion to adapt to the situation and the specific needs of each client, applying more socially-oriented values to enhance inclusion. An additional challenge highlighted by the informants is that clients often lack Swedish language skills and knowledge about the Swedish welfare system. The online application systems at both agencies are designed and provided in Swedish, although there are general information pages and checklists available in other languages. However, some of the front-line case workers also have competence in languages other than Swedish and thus have other resources for coaching clients. The informants discussed this in terms such as:

The language … almost all of our e-services are in Swedish, and this is problematic. One might wish that there was more in English at least because it is an international language that most people understand. But the information you find when you are logged in to ‘My Pages’ is always only in Swedish. Our personal translation and guidance are not always correct and clear. (Interview at AF)

This shows how the front-line case workers coach clients to use RPA and related digital systems. They need competence not only as front-line case workers in their specific area, but also to support the clients’ digital competences and to coordinate and coach clients in more general terms about welfare policies and related issues within Swedish society.

6.1 Identified strategies and their relationship to public values

The strategies used by front-line case workers to help clients in using digital services and automated systems, and thereby become included, are mainly driven by socially-oriented values. Even if the front-line case workers were all clearly grounded in duty-oriented values by adhering to all the rules, they develop new strategies to support clients as individuals in their unique circumstances and the new challenges generated by the RPA. As in the case when RPA required more advanced log-in processes at AF, workers struggle to remain service-oriented by motivating the clients to use the systems and striving to help and encourage them. Only as last option do they offer the client a paper form to solve the problem and go around the RPA and their duties.

Basic duty-oriented values (inclusion, impartiality and the rule of law) were discussed during the interviews in a more service-oriented way as illustrations and arguments, rather than as formal legislative limitations. However, the front-line case workers clearly know the limits of their discretion and discussed, almost excusingly, when they switched to more socially-oriented values to help clients with special needs. Thus, they reinterpreted the core meaning of duty-oriented values to make all public services inclusive and impartial.

6.2 Discretion of front-line case workers when RPA is introduced

The front-line case workers are supposed to have the discretion and competences to enhance inclusion and to help guide clients towards self-help and empowerment. They have discretion in terms of how they relate to their assignments, the services they provide and how they support clients.

When clients encounter new thresholds to accessing public welfare services due to RPA, the front-line case workers develop new competences using their discretion to help the clients navigate and use RPA. We have shown that in addition to their duties relating to the institutional arrangements of the specific agency, the front-line case workers also use their discretion to be more social and see the client as an individual. The personal encounters that take place at the agencies’ offices are important in terms of trust in both the agency and the government in general. They are also important in providing unique situations for personal coaching regarding the specific digital systems and digital competences more broadly. The standardisation of RPA forces the front-line case workers to increase their personal service flexibility, referring to service-oriented values. There is an emphasised need for more social values and the ability to see the person behind the client and to help them in their situation. Furthermore, the focus on socially- and service-oriented values proved to be a central concern in supporting clients and compensating for their inability to use RPA. However, workers discretion to express and show such values was limited by institutional arrangements – both the specific agency’s regulations and the more general arrangements of e-ID. Similar limitations of discretion alongside increased digitalisation has been found in earlier research (Busch et al., 2018).

These Swedish public agency case studies show that the discretion of public front-line case workers is reframed when RPA is implemented. The case workers changed their strategies towards clients who cannot use the systems themselves. In corporating previous research, we have shown that case workers emphasise that they are fulfilling duty-oriented values by embedding services and social values into their strategies. The results imply that the service-oriented values in practice focus on inclusiveness and impartiality as front-line case workers formed strategies to bring all clients into the system and enhance digital inclusion. Consequentially, there is a need to embed knowledge on different forms of digital inclusion into the design of public digital services and in the training and organisation of front-line case workers’ daily practices working side-by-side with RPA.

6.3 Concluding remarks and openings for further research

The introduction of RPA in public services brings new client interfaces and changes the strategies of clients and front-line case workers. Both groups need to develop new competences and their relations are reframed around the RPA. The clients’ digital competences become more important. As their ability to obtain information about, access and use public services is a basic democratic need in a welfare state, they also need more competences to navigate a complex setting of welfare institutions.

Our overall conclusion is that front-line case workers strive to bring everyone into the system by finding strategies to show socially- and service-oriented engagement in spite of the new constraints from RPA. Their grounding in socially- and service-oriented values is clear and they have the competences to see their clients’ needs beyond the RPA systems. Additionally, it is crucial to focus on the introduction of RPA since it is changing clients’ experiences of public services. These have been shown to have an important impact upon the legitimacy of not only public services, but also the government in general (Rothstein & Teorell, 2008).

We conclude that the front-line case workers develop their competences and frame their strategies in line with service- and socially-oriented values. They interpretate their digital discretion in new ways in order to be able to work alongside the RPA to bring all clients into the system. In addition to other studies concluding that the discretion of front-line case workers is limited by the introduction of RPA systems (Busch et al., 2018; Ranerup & Henriksen, 2019), we have shown that it is mainly tasks based on duty-oriented values that are conducted by the RPA. Instead, the front-line case workers develop their competences in line with socially- and service-oriented values to support clients. It can be seen as an enhanced socially- and service-oriented discretion, compensating for the limitations given of digital discretion.

We note that this study has limitations. The small number of cases was caused by the limited access to respondents. However, by setting the results in a well-grounded theoretical framing we still contribute to the interpretations of public values through the strategies of discretion of front-line case workers in public administration when RPA is introduced. These findings can be tested in welfare state models other than the extensive Swedish one. Our results can also open for survey designed studies and comparative settings. This study also opens for further conceptualisation on how digital divides influence how we build inclusive and trustworthy public services bringing all clients into the system.