Digital discretion: A systematic literature review of ICT and street-level discretion

1. Introduction

While policy makers expect policy making to be a top-down process where the intentions of the policy maker are pushed downward a hierarchy and materialize in actual outcomes, a different picture emerges through the implementation practices of street-level bureaucrats. Street-level bureaucrats refer to public service workers on the street-level who interact closely with clients and can exercise a substantial amount of discretionary power [1]. Discretion is the freedom street-level bureaucrats have to make decisions concerning individuals regarding the sort, quality and quantity of sanctions, and rewards during policy implementation including the possibility of no sanction at all [1, 2]. Public service provision is complex and policy implementation has traditionally required the unique expertise and skill sets of street-level bureaucrats acquired through years of experience. This competence is so exceptional that Lipsky [1] claims that “the nature of service provision calls for human judgment that cannot be programmed and for which machines cannot substitute” (p. 161). Given the diffusion of information and communications technology (ICT) in the public sector [3, 4], and the rapid technological development during the last two decades, this claim can be questioned. By ICT in this article, we refer to technologies that are used to register, store, edit, and handle client data as a basis for making case assessments or executing decisions [5]. The most frequently referred to technologies are variations of databases, case management systems, and automated systems.

In their seminal article, Bovens and Zouridis [5] observed that the structures of many traditional public agencies are changing due to ICT, and that this development can be traced as a change from the ‘street-level’, via ‘screen-level’, to a ‘system-level’ bureaucracy. They argued that street-level bureaucrats gradually disappear from the streets where face-to-face contacts are replaced by computers, and where computerized routines influence their discretionary practices (screen-level bureaucracy). Ultimately, street-level bureaucrats become replaced by fully automated technologies that make decisions about clients based on collected data and predefined algorithms (system-level bureaucracy). Although Bovens and Zouridis [5] claimed that street-level bureaucracies were changing, they acknowledged that the discretionary practices of traditional street-level bureaucrats such as teachers and nurses were more unlikely to be influenced by ICT. Inspired by their article, the purpose of this literature review is to take stock of research that has studied how ICT impinges on the discretionary practices of street-level bureaucrats. We are interested in studying how ICT has influenced street-level discretion in the past two decades given the rapid development of ICT and the wide variety of tasks that street-level bureaucrats conduct.

The role of street-level bureaucrats, their practices and particularly the discretionary power that they exercise, has been subject of much attention among academics and practitioners [6]. To ensure decisions according to intentions of the policy maker and to avoid a too strong street-level footprint, digital discretion has been introduced understood as the use of computerized routines and analyses to influence or replace human judgment. From this perspective, all use of discretion can in principle be mapped out by using algorithms [7]. The concept of digital discretion emphasizes the shift from viewing discretion as the street-level bureaucrats’ intellectual process to a situation where ICT replaces parts of, or the full intellectual discretionary process [5]. Digital discretion therefore describes how ICT has the potential to influence discretion by e.g., devising decision alternatives, or replacing discretion by fully automating decision-making. There are two main reasons for why digital discretion is desired. First, an administrative ideal is to treat clients equally and avoid that factors such as personal mood, bias, and coincidences influence decision-making [1]. Legal boundaries and principles of sound administration have successfully been implemented to curtail some of their leeway [5]. Whereas decisions made by street-level bureaucrats most often go unnoticed to the public, there are frequent examples of news articles stirring the debate about street-level discretion since examples of corruption [8], errors [9], and bias [10] continue to occur. ICT can be used to control the actions of street-level bureaucrats to close the gap between ‘policy as written’ and ‘policy as performed’ [1]. Research has shown that automation can strongly increase the quality of legal decisions in comparable executive agencies [11], and can thus be seen as “the zenith of legal rational authority” [5, p. 181].

Second, digital discretion can make decision-making more efficient. The discretionary practices of street-level bureaucrats are time consuming since each case needs some degree of attention. By automating parts of or the whole decision-making process, street-level bureaucracies may save considerable amounts of time and money, or spend their resources in a more beneficial way. For example, a public tax agency may allocate their resources to control tax embezzlements instead of handling individual tax reports manually.

Researchers adopting a bottom-up perspective will argue that digital discretion is not desired and that human judgment is necessary to adjust policies to real-life situations [12]. Without this adjustment, outcomes would be considered unreasonable by clients. Thus, discretion is a tool that strengthens a policy by ensuring that policy objectives are fulfilled. The role of ICT is merely to provide access to resources street-level bureaucrats can use in the decision-making process, e.g., by making more relevant information easily available [13].

In this literature review focusing on street-level bureaucracy through the lens of digitization, we use the concept of public values to categorize the overall goals embedded in the included articles. Digitization is the process of converting analog information such as text, pictures, and sound into a digital format [14]. In a public-sector context, digitization typically implies increased use of technologies that can assist street-level bureaucrats in handling cases by providing easy access to information about clients through channels such as online forms, and automating parts of or whole work processes. From the popular media and research literature, we observe that digitization has become particularly important in the public sector with the purpose of improving public services and yield cost savings. Based on this observation, we expect that digital discretion has increased its influence since Bovens and Zouridis [5] made their observations, and will continue to do so. Generally, a value is considered something good without further justification and something that is worth pursuing and of importance for an entity [15, 16, 17]. The concept of public value has received much attention in public administration and has been conceptualized as the successor of the New Public Management (NPM) paradigm [18, 19]. Public values have been subject to a number of classifications [20] and the concept has been highlighted as core pillar of public administration [21]. It is not the objective of this review to contribute further to classifications or taxonomies of public values but instead to categorize research contributions related to street-level bureaucracy within this core pillar of public administration. A broad and traditional categorization of public values is chosen for this review: ethics, democratic, professional, and people values [22]. Given its significant role in public administration [23], the categorization of the contributions along four generic parameters provides a measure for determining if digital discretion potentially leads to a value shift in street-level bureaucracy and highlight what the characteristics are. The first research question that this literature review addresses is:

2. Public values

One of the central and widely cited conceptualizations of public value was introduced by Moore [19] more than two decades ago in his book titled “Creating public value: Strategic management in government”. The book has received much attention but also critique for its pragmatic approach and non-empirical foundation [18, 20]. The work of Moore [19] along with the discussion of the work [18, 20] has led to an emphasis on public values in administration in general [22, 28], within the domain of ICT in public sector [21], and more specifically in relation to e-government [15]. Research contributions point to the ambiguity and comprehensiveness of the conceptualization of public values (see for example MacCarthaigh [28] and Bryson et al. [23]. Jørgensen and Bozeman [20] identify as many as 72 values and Bannister and Connolly [21] identify a number of classifications, taxonomies and subcategories of values which they synthesize to three orientations which are impacted by ICT: duty, service and social. Common for the classifications is the inclusion of core aspects of public administration, i.e. accountability, responsibility, equity, and democracy. As stated in the introduction the aim of this literature review is not to identify new classifications or taxonomies of public values but to present how the reviewed articles reflect digital discretion related to public service values recognizing that the term represents core aspects which generally are followed independently of specific context [23]. The more generic classification of Kernaghan [22] is used which serve as a common denominator for those more comprehensive classifications mentioned above [20, 21]. Following the classification of Kernaghan [22] the sources from the literature review are grouped along the four general parameters; ethics, democratic, professional, and people values. Each of the four parameters is briefly introduced in the following.

3. Methodology

A systematic literature review of scholarly research was conducted to get a comprehensive understanding of digital discretion, and more specifically to find out if digital discretion is causing a value shift in public service provision and if so; under what conditions this value shift occurs. The review was guided by recommendations outlined by Webster and Watson [33]. Apart from answering the initial research questions, the literature review also analyzed the attributes of the research community.

The review targeted articles on digital discretion published on or before January 31st, 2017. Five databases were searched to gain access to leading publications within social sciences: Web of Science, Scopus, EBSCOhost, IEEE Xplore, and the E-Government Reference Library (EGRL v. 12.0). The search was limited to article title, abstract, keywords, or topic, and conducted in two rounds. In round 1, the search term was ‘street*level bureaucracy’. Using this term only was considered an incomplete search since much research on street-level work has been conducted without using street-level bureaucracy as the analytical lens [34]. In round 2, the search terms were ‘e-government’, ‘digital government’, ‘information technology’, and ‘ICT’ combined with ‘discretion’. Combining ‘discretion’ with ‘digital government’ did not provide any results in the selected databases. All combinations of search terms and their respective results are listed in Appendix A.

Manuscripts were excluded based on the following criteria:

•

Recurring articles

•

Articles with anonymous author

•

Written in a non-English language

•

Non-research manuscripts

•

Research-in-progress articles

From the initial set, we excluded 251 manuscripts for the following reasons: 193 manuscripts were duplicates; 12 manuscripts had anonymous authors; 15 manuscripts were written in a non-English language; 31 manuscripts were non-academic; and one manuscript was a PhD dissertation. After this process, we ended up with 111 articles.

All abstracts were read, and articles were included if they discussed:

•

Street-level work in public agencies, and

•

The relationship between technology and street-level discretion, and/or

•

Managerial control of street-level discretion using technology.

After reading the abstracts, 70 more articles were removed because they did not meet the above criteria. Reading through the 41 remaining articles resulted in 21 more articles being removed because they were not relevant or accessible. As a result, 20 articles remained in the dataset.

Following recommendations by Webster and Watson [33], a backward and forward search was conducted using the identified articles as a basis. The forward search added 20 articles to the existing dataset. The backward search added four more articles to the final pool resulting in a total of 44 articles for our review. The fields of e-government and public administration are informed by a multitude of journal and conference articles as well as books, and we do not claim this review to be exhaustive. However, we believe that the selected databases contain leading e-government and public administration research, and that the review is representative of scholarly research on digital discretion.

We used a bottom up approach applying techniques from grounded theory to study the reviewed articles. This approach has been recommended for rigorous literature reviews [35]. The first step was to read through the articles. The initial coding was done by applying open coding techniques resulting in codes that represented the aim, focus, and reported findings of each article [36]. The codes were generated mainly from an analysis of the article abstract, introduction, findings section, and conclusion. Whenever necessary, the entire article was carefully read.

In the next step, we identified relationships between the initial codes (axial coding). The codes were reduced into a set of 13 subcategories [36]. When categorizing the number of codes into subcategories, simplicity was sought while at the same time making sure that the diversity represented in the initial codes were represented. In the third and last step, our objective was to identify how the articles aligned with overall public service values [22]. Public service values reflect an ideal type of public administration which generates trust and confidence in public sector decisions [28]. In this context, it serves as a reference to discuss how various aspects of public service provision are influenced by digital discretion.

The reviewed articles cover a diversity of street-level bureaucracies ranging from child care services to automated handling of student grants. Due to the diversity, it is not possible to dive into the details of all 44 articles. However, the purpose of this article is to provide researchers in fields such as public administration and e-government with an overview of the academic contributions within the area of what is here labelled as digital discretion.

4. Research on digital discretion

In this section, we outline the research area of digital discretion. First, we provide an overview of conducted research with information about publication authors, timeframe, and outlets. Furthermore, we provide an overview of theoretical foundations and research methods. The conceptual articles build their arguments on firsthand experiences, technological trends, extant literature, and example data. The empirical articles build their arguments on collected data. Second, article reflections about digital discretion are laid out using the concept of public service values [22]. The articles are categorized according to four public service values each representing the focus in the articles: ethics, democracy, professionalism, and people.

The key problem that Lipsky [1] addresses is the potential loss of democratic control of the public policy making process since street-level bureaucrats influence the outcomes that clients experience, and the actions they take can actually become public policy. This observation is made despite the many procedures that have been implemented to control their behavior. ICT has been introduced to ensure implementation of public policies according to the intentions of the policy maker. The diffusion of ICT into public agencies has caused structural changes in some street-level bureaucracies leading to computerized interaction with clients instead of face-to-face contact. While these observations can be made, Lipsky [1] holds that “the nature of service provision calls for human judgment that cannot be programmed and for which machines cannot substitute” (p. 161). In the digital era, we believe that this claim can be questioned. Digital discretion, understood as the use of computerized routines and analyses to influence or replace human judgment, can change what street-level bureaucracy fundamentally is. Guided by the literature, this review addresses two research puzzles: (1) whether digital discretion can cause a value shift in street-level bureaucracy, and (2) under what conditions digital discretion can cause such a value shift.

Addressing the first research puzzle, the review indicates increased diffusion of digital discretion which has implications for public values guiding street-level bureaucrats. The scarce number of empirical studies in this area makes it difficult to draw generalizable conclusions but the findings can rather serve as an indication of a general development. While digital discretion seems to be more suitable for strengthening ethical and democratic values, it generally fails in its attempts to strengthen professional and relational values. Street-level bureaucrats who willingly and knowingly act unethically are at the center of societal attention on a regular basis. The reviewed studies show that digital discretion can support ethical values by removing personal biases from decision-making [10, 38]. Traditionally when interacting closely with their clients, street-level bureaucrats became acquainted with them gaining knowledge about who appeared to deserve a better treatment thus leading to favoritism. Technology does not make individual considerations but focuses merely on objective predefined criteria. However, one study found that automation could lead to novel forms of corruption since system administrators had excessive control of system outputs [8]. Thus, the threat is not emerging from the use of ICT which in fact seems to prevent street-level bureaucrats from unlawful actions as intended, but rather from the design of algorithms and system control [5].

Democracy is founded on the principle that certain agencies or people are appointed to conduct tasks on behalf of the community. They are expected to do so according to policies decided by democratically elected representatives. Thus, the people controls the state. The responsibility given to others than those elected presupposes that those who implement policies such as street-level bureaucrats can be held liable for their actions. Democratic values are values that support this principle. A common characteristic derived from the studies is that ICT seems to be well suited for control functions. By limiting the scope of actions that street-level bureaucrats can take, democratic values such as the rule of law and accountability are enhanced [42]. However, other studies clearly indicate that ICT can actually obscure who is responsible for actions taken simply by referring to what the computer said [43, 44].

Concerning professional values, digital discretion seems to enhance efficiency but on behalf of the quality of decisions [12]. ICT, embedded with certain public values, seems to promote different goals, and thus often influence public service values [21]. While the attempt to make decision-making more time and cost efficient is legitimate, its conflicting consequence of reducing the quality of decisions is not. The reduction of discretion in street-level work is claimed to reduce service quality, demoralize street-level bureaucrats, and fail to achieve goals such as preventing errors to occur. However, the picture is not all black and white, and ICT is looked at with positive eyes when used as an enabling tool for street-level bureaucrats. In general, several of the studies conclude that ICT can be of considerable help to street-level bureaucrats by providing access to a vast amount of resources which enhances the quality of the decisions that street-level bureaucrats make. In addition, an increased diffusion of digital discretion is likely to influence the role of the street-level bureaucrat. While most street-level bureaucrats have gained a considerable competence within their work area through education and experience, ICT may turn them into mere ICT operators that simply follow computerized routines. Discretion is considered an award by street-level bureaucrats, and reducing or eliminating their discretionary power may result in reduced job meaningfulness [2] and ultimately in less attractive jobs which has consequences for recruitment [68].

In the last category of values, the classic tension in street-level work between concerns for individual treatment on the one hand, and commitments to efficiency and standardization on the other is salient. Street-level bureaucrats have traditionally considered how to balance these concerns. Insights from the review suggest that ICT has not changed this practice considerably. In fact, ICT is found to contradict people values by reducing the options street-level bureaucrats have to look at the specifics in each case and adapt policies to real-life situations. These concerns are important for street-level bureaucrats who are found to unequivocally prioritize clients in situations where they are confronted with this dilemma [60]. Furthermore, researchers argue that citizens actively seek human judgment due to their need for individualized treatment.

From the discussion above, we suggest that as long as the goals promoted by various types of ICT are known, the value shift caused by digital discretion is a matter of prioritization. In those cases where ICT seems to have emergent effects and promote public values that were non-intended, digital discretion is causing a value shift outside the control of public managers and public policy makers. However, the effects of ICT are most often a result of both IT strategic decisions as well as emerging from the use of ICT. Thus, it is difficult to unequivocally make conclusions about a value shift based on our review. We suggest that digital discretion more easily supports ethical and democratic values than professional values whereas digital discretion seems to contradict people values. Findings indicate that we may witness the beginning of a value shift towards values favoring standardization and equal treatment of clients on behalf of values emphasizing individualized concerns. We hold that ultimately it must be the outcome of decisions that is important and not the logic behind them [32]. If ICT can enhance public values that result in fair and robust decisions, clients will be satisfied with public service provision and the underlying logic will be of less importance. Furthermore, findings show clearly that the influence is dependent on the context within which ICT is implemented. This conclusion leads us to the second research puzzle; namely to explain the conditions under which digital discretion can cause changes in public values.

The review of academic contributions illustrates that researchers studying digital discretion agree on context as vital for understanding the implications of digital discretion. Beyond this shared understanding, there is a significant difference in what researchers perceive to be the effects of digital discretion and which contextual factors that facilitate these effects. While street-level bureaucrats share commonalities such as the ability to exercise discretionary power, close interaction with clients, and scarce resources at their disposal, they are nevertheless very different. Diverse types of public services differ in terms of clients involved, the seriousness and consequences of each case, type of policies and the formulation of rules, and the expertise of street-level bureaucrats. To explain how computers could take over tasks conducted by humans, Sheridan [69] developed a 10-point scale to demonstrate nuances in degrees of automation. The scale can be divided into two main categories: (1) Low-level automation where the computer leaves discretionary practices and decisions to humans, and (2) high-level automation where the computer is increasingly able to execute decisions by itself [44]. Whereas the computer will offer no assistance to a human at the lowest level in Sheridan’s scale of automation, the computer will act autonomously at the highest level. Cases of child abuse and neglect illustrate situations where computers can offer limited assistance to humans. In these cases, street-level bureaucrats are confronted with tragic situations that involve vulnerable children being neglected or even seriously harmed. Under extreme circumstances, the children die and sometimes due to professional errors among social workers [9, 51]. These cases are so complex and span over many years making it very difficult to utilize ICT tools, and certainly to influence discretionary practices which is associated with higher levels of automation. Tax reporting can serve as an example of public services that can be exposed to high-level automation. The reporting is based on schematic rules and numerical data which makes such cases ideal for automation. The criteria that decide the outcomes of a tax report are clear being merely dependent on collecting data from required sources such as the client employer and bank. While Bovens and Zouridis [5] claimed that street-level bureaucracies were turned into system-level bureaucracies which are driven by the logic of the information system rather than the individual judgement of the street-level bureaucrat, they made some important caveats as well. They argued that mass transactions were most prone for automation, and believed that the suitability of ICT to influence the discretionary practices of traditional street-level bureaucrats such as teachers and nurses seemed to be more limited. A tendency we observe is that the number of empirical studies is increasing whereas the number of conceptual articles is decreasing. Furthermore, the view of digital discretion becomes more nuanced as more empirical studies are published.

From the review, three main reasons emerge for why digital discretion is diffusing more rapidly in some cases whereas more slowly in other cases. First, the automation of mass transactions has been very successful in terms of rationality objectives, and politicians and government officials constantly experience pressures to find more efficient solutions for public service delivery [70]. The imperative for public managers to prioritize efficiency is illustrated through a study of Rose et al. [4] who found that municipal managers were under considerable pressure to prioritize efficiency, and that they often did go by rationality objectives.

Second, while work routines have been a focus of researchers in this area, powerful actors are in general more interested in how the established power relations in the organization can be either changed or maintained dependent on the agenda of these actors [71]. This means that a change in work routines is not a problem as long as street-level bureaucrats can adapt their practices to any constraints enforced by ICT. The reviewed articles are most frequently occupied with how digital discretion affects the street-level bureaucrat. What we observe from the articles is that street-level bureaucrats exclusively relate digital discretion to the traditional public administration paradigm and do not see digital discretion as an opportunity to transform government. For example, transparency is emphasized to reveal how governments work since the logic behind decisions is expected to reflect the way decisions are made today. The public administration literature has shown that public agencies are reluctant to change their practices. Existing structures are reinforced, and street-level bureaucrats continue to work as before [72, 73, 74].

Third, the technology in use has obvious limitations. The reviewed articles have studied several types of ICT but most often databases and case management systems. While several of the recent empirical studies have used student grant loans, electronic tax reporting, and social e-services as examples of public service provision that is automated [44, 47], we argue that new technology has begun to show evidence on its capability to perform tasks of traditional street-level bureaucrats. There is an increasing potential of ICT to transform non-routine street-level work such as teaching, nursing, and policing [32]. Artificial intelligence (AI) is a technology that has developed rapidly during recent decades describing computers that can act as autonomous agents and approximate the human brain. By time, they will have improved their own capacity to make accurate decisions [75] that overcome the many constraints of rule-based systems [76]. For example, IBM’s Watson can now better identify symptoms of diseases than experienced physicians [76], and AI has proved to conduct better assessments of English essays than teachers [77].

The findings and conclusions from this literature review point to a number of implications for street-level bureaucracy. While we observe that digital discretion is increasingly diffusing, that ICT is influencing the practices of street-level bureaucrats, and that digital discretion in some cases is substituting human judgment, can we still talk about street-level bureaucracy? Is it not turning into something else such as a screen-level or system-level bureaucracy [5], an infocracy [7], an e-bureaucracy [70], or a digital bureaucracy? From the research and policy literature it is observed that digitization is increasingly on the agenda of public policy makers, and from this we can conclude that digital discretion is likely to continue to increase its impact on street-level bureaucracies. The main characteristics of a street-level bureaucrat, as Lipsky [1] defined him or her, are (1) close interaction with clients during work hours, and (2) the ability to exercise a substantial amount of discretion. If digital discretion causes significant changes in how clients and street-level bureaucrats interact as well as limit the ability that they have to exercise human judgment, the two main characteristics of the street-level bureaucrat are changing due to digital discretion. While this development can be observed, it is acknowledged that certain types of street-level work seem to be unaffected by ICT which advocate for still talking about a street-level bureaucracy, for example in nursing and social work. However, the scope of street-level bureaucracy is decreasing, and opening space for a theory of digital bureaucracy where ICT is the core ingredient characterized by digital bureaucrats working in front of computers, who do not interact face-to-face with their clients, and who are limited to operating computers all with the intentions of improving public service provision in terms of enhancing ethical, democratic, professional, and relational values.

The most promising research avenue for digital discretion appears to be for public services where certain aspects of discretionary practices can be taken over by a computer. These are services that cover situations that are neither very complex nor straightforward. In between these situations, one will find that discretion is not “an ‘all-or-nothing’ phenomenon”, but rather a result of “gradations of power that exist in the relationship between managers and professional workers” [78, p. 881]. In this mid-position, many questions are unresolved such as how ICT can obscure the discretionary practices of street-level bureaucrats contrary to intended objectives [57], how street-level bureaucrats may create work arounds for computerized routines, and how discretion can be influenced by ICT without street-level bureaucrats being fully aware of it [12]. In some cases, street-level bureaucrats experienced to be empowered by technology having more information about citizens and being able to control information flows. What is more, technology-induced change is caused by street-level behavior emerging “from a dynamic interaction of external circumstances and internal motives or interest” [79, p. 585], thus making it hard to predict effects of ICT implementations. Like socio-technical arguments claiming that technology cannot be viewed isolated but rather related to a social context shaping its use [80], digital discretion researchers increasingly view ICT as one of many factors that influence the outcomes of the use of ICT.

6. Conclusion

In this study, we report from 44 scholarly articles on ICT and street-level discretion. Societal problems such as increasing and more complex demands on public service provision, and errors and corruption are creating pressures on politicians and government officials to provide services of higher quality and in a more efficient manner. The review shows that the environment in which ICT is implemented and used is vital for understanding why digital discretion diffuses and how the impacts of it are. For certain types of street-level work such as mass transactional tasks, ICT has reduced or even eliminated the use of human judgment. Examples of mass transactional tasks are the handling of student grant loans and tax reports where the data is numerical and readily available for government agencies, and where decisions are made based on schematic rule sets. In other types of street-level work such as social work, the discretionary practices of street-level bureaucrats are influenced by ICT to a lesser degree or not influenced at all. Contextual explanations for the prevalence of digital discretion can be attributed to factors such as the degree of professionalization, formulation of rules, computer literacy, and the level of information richness required. The impact of digital discretion is less explored in types of street-level work in between these extremes which opens avenues for future research. Another promising area for future research seems to be the increasing use of advanced technology such as artificial intelligence. This technology is now to a considerable extent able to deal with tasks of high complexity, and can thus address many of the shortcomings that the critics of digital discretion put forward.

The review further reveals that digital discretion has the potential to change the nature of public service provision. By using the concept of public service values, we found that digital discretion is strengthening ethical and democratic values but weakening professional and relational values for traditional street-level bureaucracies. Thus, digital discretion impacts the role of street-level bureaucrats and the work practices they perform in public service provision. Researchers in this area are mainly negative to digital discretion and conclude that street-level discretion is necessary to ensure values that are vital for public service provision such as making well-founded and fair decisions.

We conclude the literature review with claiming that the scope of street-level bureaucracy is decreasing. While certain types of street-level work seem to avoid extensive changes due ICT, it makes more and more sense to talk about digital bureaucracy and digital discretion since an increasing number of street-level bureaucracies are characterized by digital bureaucrats who operate computers instead of interacting face-to-face with their clients.

This literature review calls for more research. One third of the reviewed articles were conceptual, and the task now at hand is to evaluate the effects of digital discretion empirically, to explain under which circumstances ICT is influencing street-level discretion, and to explain how several types of technologies play a role in this influence. These are all under-investigated areas which provide good opportunities for future research. Other researchers are encouraged to join in exploring these questions and continue the research in this emergent area of significant practical importance.