Value positions in the implementation of automated decision-making in social assistance

ABSTRACT

Automated decision-making (ADM) has been introduced into the public sector in many countries, including for decisions made regarding the provision of economic support to people in need. This article analyses the value positions (the ‘desired intentions’) for using ADM to determine such decisions in four Swedish municipalities. The value positions related to efficiency of the application process, its result and the improved service that ensues, including more time for helping people to become self-supporting. The core issue of helping people in need and the mundanisation of activities when discussed as ‘services’ is of paramount importance. Meanwhile, the value position of professionalism relates to providing well-designed technology to preserve the quality of decisions. The results also showed differences between municipalities regarding aspects of professionalism that influence the (digital and otherwise) discretion of caseworkers: user participation in design, length of experiences and so on. Divergences between instances of value positions were detected, such as the need to increase the use of information and communications technology as a part of service values, whereas a lack of trust in citizens may exist among caseworkers. Through the framework of value positions, this article contributes to the discourse on the use of theory in social work research.

KEYWORDS:

• Automated decision-making
• robotic process automation
• social work
• public values

Introduction

Why and how is automated decision-making (ADM) by robotic process automation (RPA) (Wirtz, Weyerer, and Geyer 2019) implemented in social work? The current study explores the emerging value positions (Rose et al. (2015), or desired intentions, behind the implementation of ADM in social work. The context involves application processes related to decisions about the provision of social assistance, such as economic support to cover basic needs like food and housing, for example, under the Social Services Act (2001:453) in Sweden.

Generally, Nordic countries have established their welfare structures similarly, with social assistance serving as the ultimate protection. (For a recent example from Norway, see Gjersøe 2021.) In Sweden, social services boards commonly oversee these activities for municipalities, together with other actors involved in social work. Professional caseworkers determine the kind of economic support to be provided based on regular meetings with clients and the paper applications the clients submit. At present, full ADM is not allowed in decision-making processes in Swedish municipalities, in contrast to the state administration; however, that may change soon (The Government 2021). Instead, ADM in the form of RPA can be used by caseworkers to support a decision in the case management of social assistance. Devices for submitting e-applications must be available and must be used by approximately 60–80% of clients for submitting applications to provide input to the RPA.

Aim and research questions

In the current study, our aim is to capture instances of four value positions that have emerged in the new arrangements with ADM through RPA – efficiency, service, professionalism and engagement – as suggested by Rose et al. 2015a. The value positions express how municipalities relate to using such technologies in decision-making regarding social assistance. Our research questions are twofold:

• What value positions, and divergence between them, exist in the implementation of ADM in the handling of cases of social assistance?

• What can we learn from the similarities and differences between municipalities regarding the value positions related to ADM?

Background

The use of information and communication technology (ICT) in social work has a long history. At present, ICT is used in several fields, such as in case management (Devlieghere, Bradt & Roose 2018) and interaction in care (Gavin and Steckley 2020), while the use of ICT in more specific and direct decision-making activities is somewhat uncommon. A more advanced type of ICT like artificial intelligence is sometimes used to provide decision-making support for caseworkers when analysing relevant data related to the care of children in need (Gillingham 2019). Whilst RPA might be used in more direct ADM, its capacity is normally limited to applying pre-programmed routines to structured data (Wirtz, Weyerer, and Geyer 2019) provided by clients and others. Therefore, RPA can suggest an outcome for such a process or what can be characterized as a ‘decision’; however, its actual status depends on the legal regulations in place.

The use of ICT in social work in the form of ADM through RPA has been investigated in a few studies. For instance, Petersen, Christensen, and Hildebrandt (2020) described caseworkers’ use of discretion or their right as public servants to decide with a degree of independence based on the legal framework in place (Lipsky 2010). They argued that decision-making in social work is a cooperative endeavour that relies on individual consultative skills. However, ADM depends on a certain degree of uniformity and simplification (Petersen, Christensen, and Hildebrandt 2020). Wihlborg and Gustafsson (2021) critically analysed ADM and found that technical problems and the lack of user involvement in its design may impede its successful implementation and may result in less time spent meeting with clients.

Ranerup and Henriksen (2019) examined the emerging value positions in the implementation of ADM through RPA in one innovative municipality in Sweden. They found that instances related to improved accountability, decreased costs related to providing economic support and enhanced effectiveness in the handling of applications were viewed as important. The issue of intentions in general ICT-driven design and adoption is a related topic. Some authors framed this as the risk of transforming from a traditional but functioning ‘bureaucracy’ to a ‘technocracy’ (Gillingham 2018) if accompanied by minimal reflection on the problems that ICT is supposed to address. Indeed, ICT must support desired intentions and play a concrete, positive role from the perspective of caseworkers and clients, as opposed to causing technostress (Scaramuzzino and Martinell Barfoed 2021). Furthermore, reflections by involved policymakers and others may be characterized by ambiguity in relation to the administrative, care-related and economic values behind the implementation (Devlieghere, Bradt, and Roose 2017).

Our study contributes to the very limited research on ICT in decision-making in social work. More specifically, our results contribute to contexts in social work in which ADM is on the agenda (cf. Petersen, Christensen, and Hildebrandt 2020; Ranerup and Henriksen 2019, 2020; Wihlborg and Gustafsson 2021). Moreover, the use of theories in social work research is increasingly discussed (Järkestig-Berggren, Järkestig-Berggren, Anand, and Bagga 2021). Applying the concept of value positions contributes with a theory emanating from the research field of digital government (Panagiotopoulos, Klievink, and Cordella 2019).

The next two sections present our theoretical framework and methodology. The results are described in terms of a general background and important value positions in our four cases. Finally, we provide a comparative discussion of emerging value positions, as well as conclusions, limitations and recommendations for further research.

Theoretical framework

‘Public value’ is a concept that has surfaced in the past decades as a way to discuss the public sector and its activities. The concept may be seen as an alternative to the dominant new public management concept (Bannister and Connolly 2014; Panagiotopoulos, Klievink, and Cordella 2019) and the leading business logics therein, thus offering a relevant perspective on the role of ICT in the public sector. An oft-cited example is a framework of public values that defines duty- and service-oriented and socially positioned values as representing a ‘mode of doing things or an attribute of a way of doing things that are held to be right’ (Bannister and Connolly 2014, 120). In a similar vein, Cordella and Bonina (2012) suggested that ‘public values’ can be used to discuss the implementation of ICT in the public sector. However, as they acknowledged, competing values need to be balanced.

Rose et al. (2015) performed a review of the public sector literature and suggested a framework of ‘value positions’, which they tested in an empirical study of managers in the Danish public sector. They found that value positions are connected to what the involved actors want to accomplish, which comprise ‘ends-in-view that are tied to assumptions about how information technologies benefit good governance or increase impact’ (Rose et al. 2015, 533). As formulated by Rose, Flak, and Sæbø (2018, 364), the so-called value positions refer to ‘the overall purpose, motivation and shared goals for stakeholder groups’ participation in the implementation’.

Rose et al. (2015) further defined the four ‘value positions’ as follows:

• Efficiency ‘concerns providing lean and efficient administration that minimises waste of public resources gathered from taxpayers’ (p. 539). Thus, technology provides automation that is generally viewed as a productivity tool to promote effectiveness and efficiency. The representative values are efficiency, productivity and value for money.

• Service ‘involves maximising the utility of government to civil society by providing services directed towards the public good representing values, such as public service, citizen orientation, service level and quality’ (p. 540). Thus, the role of technology is ‘to improve availability, accessibility and usability of services by providing them online’ (p. 540). This means that technology is an information-processing tool that changes and improves how users communicate with the service providers. The representative values are quality, citizen-centricity and service level.

• Professionalism ‘is focused on providing an independent, robust and consistent administration, governed by a rule system based on law, resulting in a public record that is the basis of accountability’ […] where ‘[the roles of technology are] to provide a flexible and secure digital public record and to support standardised administrative procedure’ (pp. 539–540) as a form of infrastructure. The representative values are accountability, equity and legality.

• Engagement ‘involves engaging with civil society to facilitate policy development in accordance with liberal democratic principles, thus articulating the public good’

(p. 541). The use of technology supports deliberation and networking. The representative values are participation, democracy and deliberation.

Some values may be seen as ‘official’ as expressed in policy documents, rather than as values in actual daily use in everyday activities (Rose et al. 2015). The framework is multifaceted and structured, with four clearly defined value positions for studying projects in which ADM is implemented that brings forward the potential relationship between instances of value positions (e.g. congruence, divergence). The role of ICT is related to attaining a certain value position, as described previously.

The framework of Rose et al. (2015) has been used in other studies in the public sector. For example, Persson et al. (2017) analysed strategy documents by applying the framework proposed by Rose et al. (2015). They found professionalism, service and efficiency as comparatively stable, while the level of importance attributed to engagement values was decreased. Sundberg (2019) investigated the emergence of strategies and views through a quantitative survey of municipalities and a database containing qualitative information and reported results similar to those of Persson et al. (2017). However, professionalism and efficiency were more distinct value positions, whereas service and engagement were interrelated through citizen-centricity. In a comparative study, Rose, Flak, and Sæbø (2018) investigated stakeholders, their value positions and the outcomes in the form of ICT in the public sector. Meanwhile, Ranerup and Henriksen (2019) studied a municipality and its implementation of ADM in the handling of social assistance cases. They found the value position of professionalism to be the most common, although instances of the other positions also appeared.

Methodology

This is a multiple exploratory case study (Yin 2009) of four municipalities in Sweden. Our methodology enabled cross-case comparisons, through which we take the first step in theory-building related to value positions during ADM implementation in social work. Multiple case studies pose several challenges related to the description and analysis of data. Compared to brief testing of a combination of theories, including, for example, value positions and stakeholder theory (Rose, Flak, and Sæbø 2018) where very limited descriptions of the included cases are possible, the strength of our approach was a more detailed qualitative description and analysis of emerging value positions.

The study included municipalities with longer practical experience with RPA than others: from 2017 and onwards (Trelleborg and Nacka), from 2018 (Landskrona) and from the spring of 2020 (Mölndal). The sizes of the municipalities ranged from 33,000 (Landskrona) to 100,000 (Nacka) inhabitants. Among the Swedish municipalities, 22% out of 290 states have implemented RPA for the handling of social assistance (The National Board of Health and Welfare 2021).

Data were collected through five to eight interviews with managers, politicians and caseworkers in each case. The study was presented as aiming to capture the joint experiences of involved caseworkers, managers and politicians regarding the use of ADM through RPA in the provision of social assistance. The interviews, which lasted from 30–60 minutes, were recorded and transcribed. The anonymity of the interviewees was communicated and secured. The interviews focused on the previous history of and reasons behind the implementation of RPA, as well as the results obtained in terms of implemented technologies and the routines performed, amongst other relevant points. What is the background to the introduction of RPA? Can you tell me about how people apply for social assistance? What is the role and effect of the RPA? are some examples of questions asked during the interviews. Official documents (project descriptions, policies and descriptions of application processes) were also used (see Table 1).

Table 1. Empirical data

Municipality	Data source(s) and number of instances	Time period for interviews or documents
Trelleborg	Interviews with politicians (1), managers (3) and caseworkers (3)	2018–2020
	Documents (8)	2018–2020
	Oral presentations (2)	2017–2018
Nacka	Interviews with politicians (1), managers (2) and caseworkers (2)	2019–2020
	Documents (11)	2015–2020
	Oral presentations (1)	2018
Landskrona	Interviews with politicians (1), managers (2) and caseworkers (2)	2020–2021
	Documents (5)	2018–2021
Mölndal	Interviews with politicians (1), managers (5) and caseworkers (2)	2019–2020
	Documents (12)	2017–2020

The analysis was conducted according to the following steps

1) The authors read and re-read the data sources to familiarize themselves with their content. The intention was to find and mark citations related to value positions and the implemented technology. Furthermore, notes were taken about ‘what happens here’.

2) The value position framework (Rose et al. 2015) was applied deductively to determine which category or categories could be considered relevant to the individual citation.

3) More fine-grained reading, re-reading and categorization of instances were completed with the help of the various sub-categories of the value positions. Related to these instances or citations, inductive analysis and sorting were also conducted based on the notes taken regarding ‘what happens here’. The intention was to put together a fair description of the instances of emerging value positions, their more direct empirical themes and important divergences between instances. Significant citations were selected to serve as examples, and an overview was created (see Table 2). The collected data were also used to write short texts about the background and the technologies applied.

Table 2. An overview of instances of value positions

Municipality	Efficiency	Service	Professionalism	Engagement
Trelleborg	Lower costs of the application process and economic support provided	Constantly improved digital services Human support, but no direct contact with decision-makers A focus on helping clients become self-supporting	Well-designed, further developed technology, making full or partial decisions Trust in applicants Using human competence in a better way	Working with a model supported by local politicians Taking part in the national debate about labour market policies
Nacka	Lower costs of the application process and economic support provided, but might be problematic	Digital services to help clients as customers and increase use of the system A focus on helping clients become self-supporting	Well-designed, more integrated technology, serving as a decision support system ADM in the future Trust in applicants, but might be problematic	Working with a model supported by local politicians
Landskrona	Lower costs of the application process and economic support provided	Digital services and increasing use A focus on helping clients become self-supporting	Well-designed technology Trust in applicants, but might be problematic Reorganization of caseworkers	Working with a model supported by local politicians
Mölndal	Lower costs of the application process and economic support provided, but might be problematic	Digital services Human support A focus on helping clients become self-supporting	Well-designed technology Improved transparency regarding rules, technologies and processes Trust in applicants Using competence in a better way, but securing access to caseworkers who make decisions	Working with a model supported by local politicians Increasing participation through higher rates of self-support

4) Thereafter, a discussion of the value positions in the four cases in line with our research questions was conducted.

Findings

Trelleborg

Background

The municipality began to change the organization of its social assistance provision process in 2010 (Ranerup and Henriksen 2019, 2020). This led to the introduction of a labour market board to oversee social assistance in place of the traditional social services board. The application process was streamlined to provide active help in the form of a meeting with caseworkers and to arrive at a decision about the economic support to be provided in just one or two days (Trelleborg 2016, 1). The new ways of working, called the ‘Trelleborg Model’, were discussed in a national debate in Sweden (Rakar 2018). An e-application for applying for assistance was introduced in 2015 (Trelleborg 2016). After being implemented in 2016 in a few other areas in the municipal administration, in 2017 RPA was introduced into simpler decision-making processes related to social assistance. Caseworkers participated in the design process; a plan for how to become self-supporting was made part of the ADM routine (Ranerup and Henriksen 2020).

In the new process, applicants must attend regular meetings with caseworkers dedicated to helping them secure employment. This documentation and the e-application serve as the foundations of the ADM. Applicants can also contact a help desk for general support, although this does not involve direct physical contact with the caseworkers handling the decisions. From 2018–2019, further work was done to improve the use of ADM in handling negative decisions, resulting in the provision of better explanations for decisions and the opportunity to file appeals to challenge negative decisions. According to municipal reports, 85% and 30% of all e-applications for social assistance are managed by RPA partially and fully, respectively (Trelleborg 2017, 2019b).

Value positions

Efficiency. The emphasis on efficiency related to the overarching aim for social assistance to guide users to become self-subsistent: ‘Irrespective of the political party one belongs to, [he or she has] the same will that people should become self-supporting’ (Politician T1). This will in turn hopefully lower the cost of providing social assistance. The combination of employing technology to reduce the resources consumed during the process and the application of a specific model, with its emphasis on helping people become self-supporting, was highlighted.

Service. The specific model of work and the role of technology herein were considered important aspects of the value position related to service, including a digital ‘activity plan’, improving explanations for decisions and enabling appeals against negative decisions through the platform. At the same time, the human and ‘citizen-centred’ component of the services was also emphasized:

I receive clients in the reception area […] Then, we can check their personal files if they have submitted an application. The information they must provide depends on the type of e-application made. (Caseworker T2)

Professionalism. A common value related to professionalism was that a well-designed digital process should manage the applications, either totally or partially, with the later alternative considering the most complicated ones

In the application process, we have the same process as before. [The system] calculates and exercises control according to the rules we have designed. Then, it makes […] a positive or negative decision. The negative decisions were introduced by June 1st [2019]. (Manager T2)

The actual decision was described as follows: ‘The activities of the labour market board makes up an administrative and partly automated process, pursued in close cooperation with the help desk in the municipality’ (Trelleborg 2019a. Normally, decision-making did not involve personal meetings with applicants, except for caseworkers specializing in helping applicants become self-supporting. Thus, the caseworkers contributed to accountability and legality in the process:

Well, I think that I can use my knowledge where it is really needed. […] But now, I can spend it on judgments and appeals and investigations about debts; [… the] problematic rent payment are processed more quickly than before. (Caseworker T1)

An important instance of professionalism involved promoting the use of e-applications, which entailed the necessary trust in applicants.

Engagement. According to representatives of the municipality, their model for new ways of working was supported in the local political debate. In the national debate, however, trade unions and journalists perceived the role of RPA as controversial (Kaun 2021). The municipality was active in the national debate regarding the role of municipalities in helping job seekers versus the service provided by the National Employment Agency:

One wanted the National Employment Agency [to be centralised to certain municipalities]. They started to talk about privatisation, etc. […] Now, our chief administrator is part of a national group looking at these issues. (Politician T1)

Nacka

Background

In 2016, Nacka changed the way it provided social assistance, moving from oversight by a more traditional social board to assigning that responsibility to a new labour market and business board. The changes involved a greater focus on users’ becoming self-supporting and the formation of a unit of caseworkers with that particular focus in mind (Nacka 2015). In April 2017, partially inspired by Trelleborg, e-applications were introduced, followed by the implementation of ADM through RPA in December 2017 (Flinckman, no year; Nacka 2019).

Under the new arrangement, the caseworkers do not meet with the clients, with certain exceptions, other than during their first contact (‘reception light’). Subsequently, regular contact was maintained by a ‘careers advisor’ (Nacka 2019, 3). A caseworker always considered all suggestions made regarding case decisions. Based on the local interpretation of the legal framework in Sweden, no decision-making processes were fully automated, although some expressed a desire to see that happen in the future (cf. Nacka 2019).

Value positions

Efficiency. The focus on lowering costs by helping people become self-supporting was emphasized in Nacka (Flinckman). The importance of the internal aspect of improving processes was also recognized:

The work related to investigating and improving work processes [to prepare for RPA] has also improved the internal process. We can already see that our work is more efficient because of the time that we saved. (Caseworker N1)

However, during the first phase of the introduction of e-applications, problems with double workloads were reported in relation to rewriting the information provided by clients into the case management system.

Service. The focus on providing services through digital and other means that can help people become self-supporting was a central aspect of the work. However, the use of technology via e-applications and the internal system for case management initially did not function properly. Furthermore, the work to increase the clients’ use of e-applications (68% Autumn 2020) was emphasized, even though some activities were put on hold to a degree due to the coronavirus.

Professionalism. Regarding professionalism, trust in clients was mentioned repeatedly and was emphasized, as was control in the form of spot tests (Nacka 2019, 1). However, the process was not entirely easy:

We had a few colleagues who were new and had worked in more traditional municipalities where one staff member checked everything. They said that the hardest challenge was not to [check] everything. This is only an application, and we must trust the customer. (Caseworker N1)

Several instances related to professionalism that focused on design were noteworthy, such as further involving caseworkers despite their potentially busy schedule.

Engagement. Ambitions from citizens’ points of view were addressed by politicians: ‘Reach out to the citizens in the way that is best for them, to learn about their needs and improve the provided service’ (Politician N1). Furthermore, the stability of the political majority has made expectations clearer.

Landskrona

Background

During the last few years, the municipality attempted to improve its ways of coping with the increased pressure involved in providing support for people in need, many of whom were unemployed (Landskrona 2019). In 2018, the municipality began implementing changes to its case management process through the application of ICT (Landskrona 2018; SALAR 2021). In response to a rising interest in ADM and RPA, the national organization for municipalities (SALAR) initiated a dissemination project that would run from 2019 through the spring of 2020. Here, 20+ municipalities would receive advice from qualified consultants on how to prepare for the implementation of ADM for making decisions related to social assistance. Landskrona took part in that project. During the spring of 2019, the municipality was already able to introduce an e-application system via RPA.

Moreover, preparations were made to change the organization of caseworkers into three new roles: the administrative handling of applications by economic assistants, the management of the formal decision-making process and a role dedicated to motivating and helping a few clients (SALAR 2021).

Value positions

Efficiency. In Landskrona, the work was driven by the value of reducing the costs of providing social assistance:

In order to meet the increasing needs of society, efficiency in the organisation of welfare has to be improved. This applies to the organisational arrangements and the resources that are being spent. Digitalisation is an important tool that the social services board can use in improving efficiency. (Landskrona 2018, 4)

Service. Digitalization and the introduction of ADM for handling social assistance cases were part of the goal of providing better services to clients in terms of helping them become self-supporting (Landskrona 2018). This was partially a consequence of civil servants’ strategies on how to handle ‘turbulent times’. This goal was also related to efficiency:

Social services in Landskrona were under heavy pressure in 2016, 2017 and 2018 […] and they looked a bit at what Trelleborg had done. Robotisation was not a political decision, but rather [an idea] from the civil servants, and the managers decided to look at it as a means of simply improving efficiency. (Politician L1)

Professionalism. The e-application and the ADM were seen as important aspects of the new way of working with preserved accountability. However, difficulties related to establishing trust in citizens were also highlighted:

The biggest challenge has been to let go of controlling [applications]. To stop doing things in a way that they have always been done, which might be natural, as social assistance has been handled in the same way for many years. (SALAR 2021)

A significant related issue was the new way of organizing caseworkers in groups to avoid the negative aspects of the present regulations in social work:

When they said that this was a shattered assignment, we thought that it had to be split up […] so that you apply for social assistance at the reception […], five caseworkers handle the formal decisions and new applications that extend the norm, then we have economic assistance working together with the robot as their colleagues and the rest (13–14) work on planning and coaching. (Manager L1)

Engagement. In this municipality, the issues concerning politicians’ agreement with civil servants regarding digitalization and ADM implementation as mentioned above (‘service values’) and the vision of using the individual’s own resources can be characterized as instances of ‘engagement values’.

Mölndal

Background

During the last few years, the municipality’s political intentions included a digital agenda with the aim of providing better services to citizens (Mölndal 2017b). This resulted in a decision to take part in a dissemination project run by the municipality of Trelleborg from 2017 through spring 2018 (Rakar 2018). Part of the project involved modifying the application process through the implementation of e-applications and RPA and helping people to become self-supporting (Mölndal 2017a).

In 2018–2019, the municipality participated in a public procurement process with 13 municipalities in the region targeting RPA. From February to June 2020, it used and evaluated the e-application and RPA in approximately 20% of recurring applications for social assistance. A separation between caseworkers managing the applications and those providing additional support to help clients become self-supporting was planned. A political decision made in June 2020 that, by October 2020, the municipality should continue to develop improved routines in using RPA in the application process.

Value positions

Efficiency. The transition aimed to decrease costs through the improved application process and to focus on helping clients to become self-supporting. The e-application and RPA were viewed as important aspects of improving the application process. However, according to many interviewees, attaining the value of efficiency was problematic: ‘Right now, it hasn’t worked as expected. Instead, we have more tasks to do than before’ (Caseworker M1). Thus, it is important to resolve this issue.

Service. The improved service in the form of digital applications and the implementation of RPA was important to many interviewees, with implications for different policies (Mölndal 2020a,b, 2017b). The new focus on being able to support oneself was also seen as an improvement of services. The preserved option to receive personal help was a way of attaining not only accountability in the application process but also the value of citizen-centricity:

They were very helpful in assisting clients in submitting their digital applications. […] We provided such help, but it isn’t possible now [during the pandemic]. (Manager M1)

Professionalism. In Mölndal, the new digital application process and its components were expected to strengthen the values of equity, legality and accountability:

The risk of mistakes as a result of the human factor is minimised. This contributes to improved legality in the decisions about the applications and helps us avoid mistakes in calculation. (Mölndal , 5)

Both caseworkers and applicants were involved in developing the new process, which relied on trust in applicants. However, difficulties related to these values were encountered, including problems with RPA programming, which was not sophisticated enough, and incomplete/incorrect information provided in the e-applications.

From 2019–2020, activities were initiated with the desired value of improving the organization and accountability of the application process and identifying various types of tasks and caseworkers:

[It] should be that some caseworkers handle the economic support, whereas another group handles the planning and coaching as it is called. (Caseworker M2)

In contrast to the municipality of Trelleborg, the planners in Mölndal recognized the importance of preserving direct contact with the caseworkers deciding each case: ‘You can meet your own caseworker if you want to. […] So we don’t take away this possibility’ (Manager M1).

Another theme was improving accountability in the form of documenting important decisions regarding technical changes and new routines. In fact, this was the main issue in an audit ordered by the municipality’s auditors during the spring of 2020 (EY 2020).

Engagement. This value position was mentioned in very few instances or citations:

When the number of people in need of economic support becomes smaller, this contributes to improved participation, integration and self-support (Goal no. 9 of the Municipal Political Board). (Mölndal)

Furthermore, the different strategies related to social assistance were decided by political boards at various levels.

Discussion

This study focused on the use of ICT in relation to desired value positions (Rose et al. 2015) in four municipalities that implemented ADM in the social assistance decision-making process. As shown in our results, all municipalities wanted to improve the efficiency and effectiveness of their provision of economic support using technology and a new case management process. However, less time was available with which to help clients due to time spent addressing technical problems with the new ICT routines during this early phase, according to reports from some municipalities. A reasonably well-functioning, basic ‘infrastructure’ consisting of a device for submitting e-applications, an internal case management system, the RPA itself and a device that permitted the client to follow his/her application are all necessary for this value position. To enable this the case management processes in all municipalities had to be somewhat reorganized and streamlined.

Concerning service values, the digital services themselves were an important desired value position for all municipalities, as well as increasing the use of e-applications. On the other hand, providing enhanced access to human support was also considered an important goal that was part of the activities. Many interviewees (managers, caseworkers, politicians) brought forward the issue of enabling more meetings between clients and caseworkers. The actual motives behind this might vary in terms of the desire to understand the personal needs of clients often related to the issue of enabling self-support or a political goal of lower costs of providing social assistance. This finding was different from the results of Wihlborg and Gustafsson (2021), who reported that the lack of involvement of caseworkers in the process design resulted in the failure to achieve such goals through the new method.

An interesting difference was observed concerning the provision of more human support (cf. Mölndal, Trelleborg, Nacka), but with the exclusion of caseworkers who worked on the actual decision-making tasks in one case (Trelleborg). However, there were different views about who should perform these roles and the necessary background (cf., caseworkers with an academic education in social work, a more general administrative background or a background with relevance to labour market issues). The municipalities had different solutions to this issue already up and running or under way. Thus, we identified that what can be characterized as one core assignment in social work in Sweden and elsewhere, such as helping people to become self-supporting (Gjersøe 2021; Wihlborg and Gustafsson 2021), was central, albeit not uncomplicated, to activities related to the implementation of ADM.

Regarding professional values, all municipalities focused on providing well-designed technology and having trust in applicants related to the e-application. Thus, the handling of paper documents, such as receipts, must be minimized. Except for Trelleborg, the ADM was used only for decision-related support. However, some additional differences existed in the implementation processes regarding the desired value of clarified rules of decision-making (Mölndal) in relation to the value of solving problems with transparency in digital and ‘traditional’ processes (EY 2020). A further difference involved the value of actually having trust in applicants, despite perceived problems in doing so, which then caused double the work (Nacka, Landskrona). Thus, this enlightened the role of controlling receipts etcetera when acting in line with professional discretion. This was an interesting take on the fundamental perception of social work as focusing on individualization, personalization and proximity (cf., Laurent 2008). Admittedly, if removed from the regular case management process a new routine for check-ups of a smaller number of applications must be established. This was the case in the municipalities in our study.

Finally, instances of engagement values were relatively few, often with a focus on local political strategy about digitalization and helping people become self-supporting. These issues seemed uncontroversial, but the question is: who should handle this work? The National Employment Agency in Sweden had been reorganized, and many physical service points have closed. This coincided with this expressed intention about strengthening issues related to self-support. The municipality of Trelleborg was active in the national arena concerning an expanded role for municipalities, even to the point that they would take over the role of the National Employment Agency. The activities related to self-support and RPA might be seen as a way for municipalities to take command of an important problem (Gjersøe 2021). This is important since in a society like Sweden the municipalities have the main responsibility for people that are not part of other programme for economic support.

Divergence or conflicts between instances of value positions also existed; municipalities expressed the value of increasing use of e-applications and the need for trust in applicants. However, interviewees from two municipalities (Nacka, Landskrona) openly admitted that difficulties may be encountered in showing trust, leading to more control and less saved working time. A further divergence pertained to the value positions of lowering the costs of the application process and providing the help and, on the other hand, the value of more human support for providing help, something that in itself is costly. However, money may, of course, also be saved by making direct budget cuts in the area. A final example of a divergence related to the appreciation for the role of humans for providing support and services, while some municipalities (Trelleborg, Nacka) expressed interest in full automation in future. Interestingly, in a very recent study of RPA in social assistance Gustafsson (2022) used the concept of tensions to discuss issues of control and trust between street-level public servants and clients. Her study also detected tensions between high expectations of quality and efficiency whereas the readiness of the digital infrastructure and the RPA-developers might be less than ideal. Our present study confirmed this with a similar focus on divergence or tensions between intensions.

A unique feature of our framework is that it includes the role of ICT in the realization of the different value positions identified (Rose et al. 2015). In particular, we found that ICT – in a more limited sense (e-application and RPA) – is not the only viable factor in the realization of individual value positions. Rather, the technological infrastructure as a whole, the length of experiences and the implementation process are important components to consider. This adds an additional dimension to the statement that the relationship between public values and ICT is not straightforward (Panagiotopoulos, Klievink, and Cordella 2019). Our analysis made obvious the fact that problems with design and implementation should be addressed gradually over time. These problems may pertain to involving caseworkers in designing the system, to integrating RPA with the case management platform to avoid double workloads and to increasing the use of e-applications for a higher use of RPA as a decision-related support system.

Conclusions, limitations and recommendations for further research

Based on this exploratory case study of ADM in social work, we conclude that the identified value positions at a surface level were surprisingly similar: making the application process more resource-effective and reducing the costs of the provision of social assistance using well-designed technologies, which entails having trust in the applicants. The multifaceted focus on addressing the more direct value position related to service through ICT was also important. This can be interpreted as a strategy of the mundanisation (Kaun 2021) of social work related to various aspects of improving the help provided for people in need through the value and role of digitalization.

The main differences between cases were related to the aspects of professional values: the actual design of RPA, its role and user participation in design, the distribution of work between various categories of caseworkers, the relationships with clients regarding trust and the difficulties that may exist, as well as the view on offering direct contact with caseworkers involved in the decision-making process. Thus, a conclusion is that the differences or controversies regarding ADM were mainly related to issues about the (digital) discretion of caseworkers in a context that includes relationships with clients (Petersen, Christensen, and Hildebrandt 2020; Ranerup and Henriksen 2020). Thus, caseworkers’ discretion and the applicants themselves were at least partly affected by the quality of the digital infrastructure and the length of experience with it in the municipalities. Last but not least, the activities related to ADM through RPA were all part of some local engagement value with the aim to enhance self-support and a political debate on who should handle such issues.

Through the theoretical framework applied, this article contributed to the emerging discourse on the use of theory in social work research (Järkestig-Berggren, Anand, and Bagga 2021). This was done through a theory emanating from the research field of digital government (Panagiotopoulos, Klievink, and Cordella 2019) that combines the focus on public values with the role of ICT. The applied theory brought forward the aspired intentions as expressed by involved actors and policy documents. This in contrast to using public values to provide advice in implementing digital government (Scholta, Halsbenning, and Becker 2022). Thus, the value positions and experiences indicated that ADM can be implemented without creating an outright ‘technocracy’ (Gillingham 2018) can be of a fruitful ‘pragmatic’ use. Combining different value positions with the respective roles of technologies as described by Rose et al. (2015) (see ‘Theoretical framework’ section) provided a nuanced picture of ICT, its relationship with humans and the organizational context. This is not to say, however, that concerns from caseworkers’ perspectives, such as technostress (Scaramuzzino and Martinell Barfoed 2021), were absent, especially during the initial stages of implementation.

A limitation was that the methodology we used to study the value positions in four municipalities made it impossible to make a separate analysis of the answers given by the groups involved (cf. managers, politicians and caseworkers). However, differences may exist between them, something which merits further research. As noted, caseworkers may encounter difficulties regarding the establishment of a (necessary) degree of trust in clients. A detailed study of the situation of caseworkers in multiple municipalities and their discretion in view of the role of ICT (Busch and Henriksen 2018) is relevant from various perspectives. This should include the situation and relationship with clients (Nordesjö, Scaramuzzino, and Ulmestig 2022; Ranerup and Henriksen 2020). Moreover, this situation may be influenced by the available legal frameworks in each country that may or may not allow the use of ADM for making important decisions (The Government 2021).

Our results on the value positions did not provide a detailed account of ‘what actually happened’. Rather, they are the outcomes of an analysis of the intentions expressed by important stakeholders and policy documents in four cases that were involved in early activities regarding RPA and related dissemination projects. Further studies can, therefore, also explore whether the value positions found in our study are the same in more recent cases among the 60+ municipalities that have also implemented ADM (The National Board of Health and Welfare 2021).

Acknowledgments

Ethical approval has been received from the Swedish Ethical Review Authority Dnr. 2020-02097.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Swedish Reseach Council for Health, Working Life and Welfare (FORTE) [2019-00710].