Digitalization and automation for the sake of IT? Insight from automation initiatives in Swedish municipalities

Digital strategies, or digital agendas, have become drivers for digital government everywhere. In Europe, this is reflected both in the EU digital strategy [1], as well as in locally developed digital strategies on the national level [2]. In Europe, the Nordic countries stand out as the most digitalized (ibid) and Sweden, which is the setting for this study, placed fifth in the recent UN E-Government Survey of the leading countries in E-Government development [3]. In 2017, the Swedish government published the Swedish national digitalization strategy – the Digital Agenda – stating that the overarching goal is for Sweden to become “the best in the world at utilizing the opportunities created by digitalization.” [4] (p. 6). In Sweden, as in many other countries, the public sector is currently facing challenges as a result of changing demographics, in particular an aging population [5, 6]. Digitalization in general, and automation in particular, has been suggested as a solution to these challenges. This affects government, public authorities, and the public sector on all levels. However, local government organizations, the municipalities, are particularly affected as they carry the responsibility for delivering welfare services to citizens [7]. As a response, the Swedish Association for Local Authorities and Regions (SALAR) has, in a series of reports and policy documents, encouraged Swedish municipalities to pursue automation technologies to face the challenging times ahead [8, 9, 10, 11, 12, 13]. As SALAR influences and guides Swedish municipalities [14], this has led to most of Sweden's municipalities launching initiatives for digitalization in general, and automation of administrative processes in particular [15, 16].

However, recent studies have shown number of unexpected consequences following the implementation of automation technologies, and that the efficiency gains fall short of expectations [16, 17, 18]. Research on automation in the public sector has also shown that, while automation does bring benefits, it also brings challenges, raising questions regarding responsibility of automated processes and how to adapt the human workforce around the automated systems [15, 16, 19]. As a result, several scholars have expressed the need for more research into this area to better understand how automation affects public sector organizations and how they operate [20, 21, 22]. A recent study focused on policy documents [23], such as the ones mentioned above, to see how they portray artificial intelligence and automation, and what the expected outcomes are envisioned to be. Using a public value perspective, applying the model developed by [24], the study [23] shows that the main focus of these policy documents is to increase efficiency, whereas other desirable ends-in-view for e-government were not as visible (i.e., increased professionalism, service, and engagement). Thus, there is a gap between expectations (policy) put on Swedish municipalities and the outcomes of automation initiatives in Swedish municipalities thus far (practice).

The aim of this study is to address the apparent gap between policy and practice by exploring the purpose of digitalization and automation initiatives as expressed by employees in three Swedish municipalities. Adopting a pragmatist lens on intentionality and the interplay between means-at-hand, ends-in-view, and long-term goals [25, 26, 27] we ask; what ends-in-view are present in automation initiative practices in Swedish municipalities and how do these ends-in-view compare to those expressed on the policy level?

Based on our analysis, we illustrate discrepancies between employees’ ends-in-view, in comparison with those expressed in policies on local government digitalization and automation. Despite the official Swedish digitalization strategy expressing that digitalization “is not a goal in itself” [4] (p. 9), we illustrate how compliance with directives is an important driver, as well as a feeling of having to keep up with others and showcasing a “modern” organization. These ends-in-view illustrate that digitalization projects are indeed being initiated for the sake of digitalization (without clear long-term goals), rather than for particular purposes, potentially creating a conflict with the stated long-term goals of digitalization.

The paper is organized as follows: First, we introduce the empirical cases we use for our analysis and describe how data was generated. Second, we introduce our analytical lens and explain how we conducted our analysis. We then proceed to present our findings with empirical examples from our data, followed by a discussion where we reflect upon our findings. Finally, we conclude our study and make some recommendations for future research on automation and the policy-practice gap.

Swedish local government is made up by 290 separate municipalities. The municipalities are self-governed to a large extent, although adhering to the Swedish Local Government Act. This act stipulates that all citizens, no matter what municipality they live in, have the right to obtain municipal services. These services include e.g., social services, care for children and the elderly, public schools, emergency services, environmental services, and building permissions. Local government organizations are not associated with paperwork without reason. An integral part of everyday work in local government organizations is administrative work related to public service delivery, such as information handling and decision-making for service eligibility (i.e. case handling). In most Swedish municipalities, case handling is still a manual and paper-based labor for many employees [11]. Oftentimes, a large part of the administration involves manual copying and pasting information from one digital system or paper form to another system/form. This manual copying and pasting is considered costly and is associated with a high risk of human error. For these reasons, digitalization and automation of administrative work is currently seen and marketed as a way of reducing both costs and errors in the information handling process [5, 6, 10, 11]. Due to the decentralization of local government, each municipality decides how, and to what degree, its operations and services can and should be digitalized and automated [18]. However, although self-governed, the municipalities are under pressure by “soft” governance initiatives to automate administrative work, such as the policy documents mentioned previously. As a result, there is currently a hype to implement automation technologies in Swedish municipalities.

Research on automation of administrative work in Swedish municipalities has illustrated that the development and implementation of automation is slow and has faced many challenges. For example, general interest and commitment has been lacking in the organizations and the development has therefore been run by a few enthusiasts [28]. Tensions between top management's abstract digitalization visions and the experience-grounded ideas originating from e.g., operating staff appear to be commonplace (ibid.). In order to understand these tensions better, it is relevant to investigate the underlying ends-in-view that guide automation initiatives in Swedish municipalities.

This study is part of an ongoing research project [31]. The aim of the project is to map current implementations of automated case handling in Swedish local governments and develop an analytical tool that can be used by researchers and practitioners alike to decide if, and to what degree, a case handling process can, and should, be automated (ibid). The research project builds on principles of engaged scholarship [32] and consists of several case studies. The ‘case’ being studied concerns how Swedish local government organizations organize for automation implementation, what processes are being automated, and what consequences automated processes bring for different stakeholders in the organization. The three municipalities were chosen based on maximum variation sampling. At the time of data generation, they were at different stages of developing automation solutions. One of the municipalities (Municipality West) had implemented over twenty automation solutions (automated case handling processes), while the other two municipalities (Municipality North and Municipality East) had recently carried out pilot projects and only had a few automation solutions implemented (less than five).

Research data has mainly been generated through semi-structured interviews, based on a qualitative and interpretive approach [33, 34]. When conducting our interviews, we used snowball sampling [35] to identify informants, with the aim of identifying and interviewing different stakeholders in relation to the automation initiatives. Thus, we have interviewed employees from different parts of the organizations, ranging from the strategical to the operational level, including directors, strategists, business process developers, IT personnel, administrators, and case workers. Table 2 presents an overview of the three municipalities, their size, and the number of interviews conducted. To distinguish and simultaneously anonymize the municipalities in this paper, we have renamed them Municipality North, East, and West, respectively. The disparity between the number of interviews and informants in the case of Municipality East is due to repeated interviews with some informants. In the case of Municipality West, two of the interviews were conducted with two people at a time.

The interviews were semi-structured and the interview guide was designed with the larger research project in mind (described above). The questions covered digitalization in general, automation in particular, the informants’ roles in their respective parts of the organization, their views on the recent push for automation, what the purpose of automation is and its perceived usefulness in relation to their respective areas of work. For this study, we focus on a small part of these interviews: the parts in which the informants were asked about what they consider digitalization and automation to be and their views on the purpose of automation. The interviews lasted approximately 90 minutes in length, and all but one of them were conducted via video calls. Prior to analysis, all of these interviews were transcribed. These case studies have been conducted in adherence with the ethical guidelines by the Swedish Research Council [36, 37] and the European Code of Conduct for Research Integrity [38].

When applying the [24] model to the extracts in which our informants express the purpose and ends-in-view of digitalization in general, and automation in particular, we observed instances of all four value positions. The Efficiency ends-in-view was the most prominent in our interview transcripts. In fact, almost all informants, 27 out of 34, highlighted that the purpose of automation in local government is to reduce costs, increase efficiency, and streamline processes. The Professionalism end-in-view was expressed by 14 out of 34 informants and was exemplified mostly in terms of increasing the quality of record keeping, and that the automation of processes may lead to an increased degree of fair treatment and adherence to legal requirements. The Service ends-in-view was mentioned by 13 out of 34 informants and was expressed mostly in terms of automation being a tool for increasing service levels and service quality towards citizens. In contrast, the Engagement ends-in-view was hardly mentioned. Only 2 out of 34 informants mentioned that automation serves to increase citizen engagement. However, Engagement was sometimes mentioned as a possible side-effect of fulfilling the other ends-in-view. Exemplary quotes are presented in Table 3.

In the introduction, we asked what ends-in-view are present in automation initiative practices in Swedish municipalities and how do these ends-in-view compare to those expressed on the policy level? In this section we discuss and reflect upon our findings in relation to this question. In the first section, we discuss how the purposes for digitalization and automation we observed in our findings correspond to those expressed in policy, based on ends-in-view provided by the model by [24]. We then proceed by discuss how the additional purpose statements that we found in our data correspond to ends-in-view and how these, and the implications they may have, can be understood.

As mentioned in the introduction, the Swedish government has issued a Digital Agenda expressing the aim of becoming “the best in the world at utilizing the opportunities created by digitalization” [4] (p. 6). The Digital Agenda clearly states that digitalization “is not a goal in itself” but rather a tool that should be applied in tackling larger societal challenges, leading to increased quality of life for Swedish citizens [4] (p. 9). The content of the Digital Agenda has trickled down to the local government level through policies formulated by SALAR. Also, many municipalities now have similar policy documents in place [39]. A content analysis of a set of these policies, showed that automation is presented as a means to increase efficiency in government [23]. Our findings support those by [23] in that, among the ends-in-view developed by [24] Efficiency is the most frequently mentioned, with 27 of 34 informants stating efficiency-related purposes as important drivers for digitalization and automation. Another similarity we see is that Engagement is the least mentioned, mentioned by only 2 out of 34 informants. Concerning the remaining two ends-in-view, Service and Professionalism, we observe that both of these are mentioned as the second and third most prominent, following the same pattern as in the study by [23]. However, we observe that Professionalism is mentioned more in practice than it is on the policy level. Hence, Professionalism seems to be more salient for people working in practice than for those responsible for the policies. Additionally, [23] showed that Professionalism is mainly mentioned in policies as related to risks and considerations when pursuing automation, rather than an explicit underlying purpose of automation. Here, we illustrate how those working in the local government context see Professionalism as a driving value of digitalization and automation. While our analysis does not provide an answer to why, we observe that this suggests that, in practice, Professionalism is considered more of a benefit derived from digitalization and automation than originally expected by policymakers. In summary, our analysis shows that practice and policy align in terms of Efficiency, Engagement and Service, while they diverge somewhat with regard to Professionalism. These results are thus confirming previous studies to a large extent.

Turning to the inductively generated ends-in-view, these emerged because they did not fit perfectly with the model presented by [24]. While [24] answers the question of “what ends-in-view should be pursued in E- government”, we ask “what ends-in-view are pursued in E-government”, and, even though we found the ends-in-view established by [24] in our data, we also found that the informants often mentioned purposes that did not correspond to either policy or the established value positions. We argue that there are additional ends-in-view that currently driving digitalization and automation initiatives in Swedish municipalities: (1) Complying with directives; (2) Keeping up with others; and (3) Being modern. However, we also claim that these should be regarded as emerging from policy and discourse on digitalization at large, rather emerging from perceived benefits of automation. The reason for claiming that these are stemming from policies and not from perceived benefits in practice, is that several of the informants acknowledge that these ends-in-view drive automation initiatives, but they are simultaneously criticizing these ends-in-views of being the wrong drivers of digitalization and automation initiatives.

Digitalization can be seen as a mean for pursuing certain ends-in-view, for example Efficiency or Professionalism. Although the Digital Agenda is clear concerning that digitalization is not a goal in itself, our three inductively generated ends-in-view illustrate that implementing digitalization and automation, indeed, has become a goal in itself. These three additional ends-in-view show how the municipalities are attempting to follow the directives presented to them, and that they are pressured, both vertically and horizontally, to digitalize and automate their processes. Our findings suggest that “digitalization” is not only providing the municipalities with the means for pursuing the ends-in-view of Efficiency, Professionalism, Service and Engagement. “Digitalization” has also led to the emergence of other means that directly influence which ends-in-view should be pursued. For example, Complying with the directives (end-in-view) can be seen as a response to top-down-pressure (mean), Keeping up with the others (end-in-view) as a response to peer-pressure (mean), and Being modern (end-in-view) as a response to techno-optimism and digitalization trends (means). This type of evolving intentionality [27] allows the actors, in this case municipalities, to disconnect the technology from the value it is supposed to bring. While it can be discussed whether these ends-in-view contribute to the realization of public values, the ends-in-view emerging from our data can be both congruent with those previously established (by e.g., [24]) as, for example, complying with directions can lead to initiatives that improve efficiency. But they can also be divergent if, for example, municipalities spend resources on automating processes that may not lead to significant efficiency gains, nor any other public value.

As ends-in-view are evolving, so are the long-term goals (e.g., [27]). Hence, even though it is generally agreed upon that digitalization should not be pursued for the sake of digitalization, as public policy and discourse frames “digitalization” as a goal (reflected e.g., in the Swedish Government's digitalization strategy), this creates means that influence ends-in-view that in turn influence long-term goals. Unintentionally, the goal of digitalization is then, well, digitalization.

The aim of this study was to explore what ends-in-view are stated as the purposes of implementing automation in Swedish municipalities to understand how digitalization policies are operationalized. We have done this by applying a model of ends-in-view by [24] as an analytical lens. During analysis, we created three additional categories inductively, in order to account for ends-in-view that did not fit into those provided by the [24] model. Our findings show that, regarding the [24] ends-in-view, practice generally reflects those public value positions expressed on the policy level [23]. However, taking into account those ends-in-view we created inductively based on our data, we can see that the operationalization of digitalization policy creates additional ends-in-view than those expressed on the policy level. Our data shows that digitalization initiatives are, in part, not treated as a means to an end, but rather an end in its own right. While some would argue that the establishment of such means are short term ends-in-view to allow for the fulfillment of long-term goals at a later stage, such an assumption of rational intentionality is, at best, naïve as the interplay between means, ends-in-view, and long-term goals allows for both ends-in-view and goals to develop and evolve over time [27]. Bearing that in mind, the conflation of means and ends can potentially move digitalization initiatives off on an unforeseen trajectory, resulting in local government digitalizing for the sake of IT, rather than for the fulfilment of public values.

We recognize several limitations of this study. For one, there are 290 municipalities in Sweden, and our analysis covers only three of them. Here we see opportunities for future research to compare a larger sample of municipalities in Sweden, as well as other types of public organizations, both nationally and internationally. Another limitation is that our analysis covers several concepts on a general level, e.g., studying digitalization and automation in general, and do not cover specific technologies and compare them. In the same vein, we also see opportunities for future research addressing nuances between different technologies and different stakeholders, by investigating if the ends-in-view discussed in this study differs between different types of technology and stakeholders. Finally, part of the findings in this study are inductively created, and while this is a natural part of research, it does create the need for further research to inquire the veracity of these results. Are public organizations digitalizing and automating administrative work for the sake of IT? This is an important question that we encourage the digital government research community to join us in investigating further.