How smart can government be? Exploring barriers to the adoption of smart government

1.Introduction

Digitalization is crucial if public administration is to thoroughly modernize. Sophisticated technologies such as the internet of things (IoT), sensor systems, big data analytics, and artificial intelligence (AI) have become increasingly important. Many digital initiatives that use these technologies in the public sector are launched under the umbrella term smart government, with the purpose of establishing new service delivery models by connecting and integrating physical, digital, public, and private environments (Gil-Garcia, 2012; Scholl & Scholl, 2014). These approaches go one important step further than past digitalization endeavors, asking how the relationship between public administration and its stakeholders could be rethought by using new technologies (Bright & Margetts, 2016). Thus, smart government is considered to be the next step of electronic government (e-government), mobile government (m-government), or open government (Gascó, 2015; Jimenez et al., 2014).

Starting with e-government, this first digitalization wave sought to create a digital environment in which public authorities provided services to their citizens electronically by utilizing the Internet and the World Wide Web (United Nations & ASPA, 2002). Terms such as Virtual State (Fountain, 2001) or NetState (Lawson, 1998) refer to this phenomenon. With the emergence of portable devices and the widespread availability of broadband wireless networks, the era of m-government has transferred such a “going online” of government from desktops to mobile device screens, without changing any other process parameters or service logics (Albesher & Stone, 2016). In contrast to e-government and m-government, smart government not only wants to digitalize processes, but to fundamentally rethink the way government works (Schedler, 2018). While e-government and m-government are forerunners of smart government, open government initiatives are a precondition for it (Jimenez et al., 2014). Without open data and open action in terms of transparency, collaboration, and participation – as Gascó (2015) describes the criteria and features of open government – no new and smart services are possible and, therefore, no smart government is possible (Guenduez et al., in press).

Smart government promises to success in further modernizing the public sector. However, many initiatives are still in early stages. Especially in early stages, there is a gap between rhetoric and reality, and between expected outcomes and achieved results. Moon (2002) showed this in his analysis of the introduction of e-government in public administration, pointing out that this gap is caused by widely shared barriers that hinder the successful uses of new information and communication technologies (ICTs) in public administration. Many other researchers – including Gilbert et al., (2004), Zakareya and Zahir (2005), and Savoldelli et al., (2014) – have documented barriers to the adoption of e-government. These studies have shown that many e-government initiatives to modernize the public sector did not reach their full potential. Thus, such initiatives are increasingly being questioned, and most fall well short of expectations (Anthopoulos et al., 2016).

We explore the perceived barriers to the adoption of smart government. With this research goal in mind, our study addresses the following questions: What are the specific barriers to the adoption of smart government and how can these barriers be classified? Following Moon (2002), we investigate the barriers in an early phase of smart government implementation, because these barriers especially affect this critical stage. We base our analysis on the perceptions of those involved in smart government initiatives in Switzerland.

We find this particular study objective worthwhile, for two reasons. First, pre-adoption attitudes can serve as an early indicator of the success of smart government implementation in the public sector. Ginsberg and Venkatraman (1992), for instance, showed that actors’ perceptions and interpretations of a new technology for the electronic filing of tax returns in 1987 predicted the introduction of the technology a year later. Similarly, Thomas et al., (1993) showed that hospital managers’ interpretations in 1987 predicted strategic changes over the next three years. Both studies showed that the involved actors’ pre-adoption attitudes and interpretations regarding an emerging technology or a strategic re-orientation predicted the introduction of the technology or chosen strategy in the following years. Thus, pre-adoption attitudes of actors involved in smart government initiatives may be an early indicator of strategic change and the success or failure of smart government implementation. Understanding what difficulties they face or expect can therefore be crucial to the successful implementation of smart government projects. Second, the early detection and understanding of these challenges is crucial to raise awareness about the complexity of smart government projects and to address them early on to successfully overcome them. Early detection enables proactive action, increasing the chances of success of smart government in the modernization of public administration. The remainder of this paper is structured as follows. In Section 2, we review the literature on smart government and on barriers to ICT implementation in public administration, drawing attention to the need for an analysis of the barriers to the adoption of smart government. In Section 3, we apply a mixed-method research approach by performing cluster analysis to coded qualitative interview data. Based on the evidence from interpreting the cluster analysis results, we then discuss our study’s primary contributions and delineate some limitations and our study’s implications for future research and practice.

2.Background

3.Methodology

We conducted 32 semi-structured interviews to explore the perceived barriers to the adoption of smart government. Because smart government is a new study field, and evidence of its (non)-adoption is fragmented, we chose to conduct semi-structured interviews, which facilitate insights into emerging fields (Gill et al., 2008). Semi-structured interviews have also been proven effective in gleaning broad-based information (Blasius, 1994). The total duration of the interviews (held in German or English) was 30 hours. Data collection took place from April to June 2017.

Sample. For participant selection, we defined three relevant actor categories involved in smart government initiatives: public managers, politicians, and private providers of smart government solutions. The categories provide three perspectives on the barriers to the adoption of smart government. Politicians bring the perspective of legislation, public managers that of executive authorities, and providers an outside view of the political-administrative settings in which smart government initiatives take place. We interviewed 13 public sector managers from different organizational departments located in different communities. Represented were executives from IT departments, organizational development departments, statistics departments, spatial development departments, and municipal utilities. Eight politicians with an interest in smart government initiatives participated; they are active in different parties (all the major Swiss parties were represented in the sample) and at different state levels (communal, cantonal, and federal). Thus, the politicians subsample represents a broad spectrum of observations and opinions. In the group of private providers of smart government solutions, the 11 participants also covered a wide range of perspectives. We interviewed hardware suppliers (e.g. sensors), software suppliers (e.g. applications), and business consultants. For every actor category, we assigned three known experts, each either strategically or operationally involved in smart government initiatives. Using a snowball sampling method, we then recruited new participants. This procedure ensured that the sample captured and represented as many different opinions as possible.

Procedure. The interview guidelines (see Appendix) for collecting raw data contained six topics: strategy, infrastructure (both existing and planned), possible applications, potentials for public service delivery, requirements for a transformation, and limitations. We evaluated the barriers to each of the six topics separately. Fountain’s (2001) model of technology enactment helped us to develop our interview questions, since we addressed organizational forms and institutional arrangements.

First, we wanted to know whether these actors had concrete strategies about smart government implementation. A strategy is often a means to guide an organization and its actors towards an intended direction (in our case, towards the adoption of smart government). Thus, we consider a strategy as an organizational form. The research indicated that conflicting or uncoordinated goals hinder the adoption of e-government; conversely, clear and realistic goals are key success factors for e-government implementation (Gil-Garcia & Pardo, 2005).

The next topic was infrastructure or “objective technology” (Fountain, 2001). Inadequate telecommunication infrastructures, which limit capacity, have been found to be significant barriers in early phases of e-government (Savoldelli et al., 2014). Technical constraints, such as a lack of platforms or meta-standards (an IT supply issue in a broad sense), have also been found to hinder the adoption of open data applications in the public sector (Janssen et al., 2012). Thus, we asked about existing IT infrastructure such as sensors, apps, networks, platforms, or data analysis tools and therefore about currently enacted technologies. We also asked questions about possible plans to extend this infrastructure and, if so, how and why.

We then asked about possible organizational forms or institutional arrangements that foster or hinder the adoption (and therefore the enactment) of smart government initiatives. We asked about possible applications of smart technologies, for examples of their own modernization projects, and about potential improvement of public service outcomes, taking a more normative perspective. These topics provide an overview of existing ideas, but also of many possible pitfalls. We then instructed the participants to consider all the opportunities smart government can provide, asking: Why has your department or public administration not yet implemented these applications? By asking this, we encouraged the subjects to share their thoughts about what hinders the adoption of smart government initiatives at a more implicit level. Further, we wanted them to identify possible limitations of smart government, asking if they saw aspects or application areas that required constraint, and encouraging them to explicitly share the barriers they perceived. We recorded and later transcribed all the interviews.

Analysis of the interview data. We analyzed the data in two stages. In step 1, we developed a codebook containing barriers to the adoption of smart government. To do this, we assessed the literature about barriers to the adoption of innovations in the public sector overall (De Vries et al., 2016), because the public sector has characteristics that play roles in the implementation of innovations (Borins, 2000). We also examined the literature about barriers to e-government initiatives (Savoldelli et al., 2014), since e-government initiatives were one of the most recent major innovation waves in the public sector. We thoroughly assessed the literature about difficulties of open data and open government initiatives (Conradie & Choenni, 2014; Janssen et al., 2012). The literature about open data and open government serves as orientation, because these initiative types contain some elements of smart government, for instance, big data or data analytics. This screening process revealed 17 barriers to the adoption of smart government: legal foundations, technical infrastructure, IT standards, political resources, plurality, silo thinking, the Swiss political system, scarce financial resources, efficiency, discomfort, citizens’ responses, skills and know-how, readiness for innovation, risk-aversion, long-term thinking, management support, and contested benefits. We validated these 17 barriers with the model of technology enactment in order to ensure that the three relevant factors – objective technology, organizational forms, and institutional arrangements – were captured by our data analysis.

In step 2, we conducted a cluster analysis, which meant that the interviews had to be coded in order to gain analyzable data. We used the variables developed in step 2 of the data analysis. To operationalize the 17 barriers, we defined keywords that served as decision criteria if that barrier was present or absent. For a complete overview of all barriers and their operationalization, see Table 1. We coded a barrier’s presence in the interview data as 1, and its absence as 0. Two coders evaluated the data. Before rating, both coders underwent a training session in which all the barriers and their operationalizations were explained. To assess interrater reliability, they coded 36 randomly selected interview questions. Conformity ranged between 86.1% and 100%, which was satisfactory. The remaining 156 interview questions were randomly assigned to each coder.

We then conducted a variable hierarchical cluster analysis using SPSS to examine whether these variables formed conceptual groups. Cluster analysis is an explorative and structure-detecting method that builds different groups or clusters. While the variables (or cases) in a cluster are very similar, the different clusters are distinct (Blasius & Baur, 2014; Backhaus et al., 2011; Mooi & Sarstedt, 2011). Cluster analysis is commonly used to classify different single cases into groups (Blasius & Baur, 2014; Backhaus et al., 2011; Mooi & Sarstedt, 2011). However, according to Blasius (1994) and Ek (2014), variables (not cases) can also be clustered according to similarities. We used the clustering method, because it allows one to group variables (here, barriers) based on their de facto similarities. Thus, the barriers’ configuration was not based solely on our interpretation, but on the clustering algorithm, which is a more objective criterion than our interpretation.

In the clustering method, we used the subcategory of complete linkage to analyze data. This algorithm, also known as the furthest neighbor method, can be used with binary data and tends to build fairly small groups (Mooi & Sarstedt, 2011), which fits our study aims. We used simple matching to represent the similarities between different variables. This measure is well established in analyzing binary data for a cluster analysis (Backhaus et al., 2011). Since there is no common decision criterion to identify the ideal cluster number, scholars recommend exploring different solutions and choosing the most readily explainable one (Backhaus et al., 2011; Mooi & Sarstedt, 2011). Because there is no appropriate number of clusters, we chose a solution that allowed us to reduce complexity without losing diversity and that is also easy to explain.

4.Results

Our first scan of the literature and our interview data revealed 17 possible barriers. We then applied a cluster analysis to the data to explore whether these barriers were somehow similar. The cluster analysis revealed six clusters containing between one and five variables. We display the results in a dendrogram (see Fig. 1). We will now explain the six clusters, what they consist of, and why these variables fit into the same group.

4.2Cluster 2: Technical infrastructure

Cluster 2 also contained only one barrier, technical infrastructure, which was mentioned 36 times. It contained two aspects: technical infrastructure (the hardware) and IT infrastructure (the software). We identified the barriers with these keywords: lack of technical infrastructure and lack of IT infrastructure. A sample statement was:

“We need new communication standards such as fifth-generation [5G] mobile networks or long-range wide area networks [LORA] in order to push smart government applications”.

Several studies have shown that technical infrastructure is a key challenge when implementing new technologies into the public sector (Savoldelli et al., 2014; Schwester, 2009; Wing, 2005; Zakareya & Zahir, 2005), especially in early implementation stages (Savoldelli et al., 2014).

Figure 1.

Dendrogram with the complete linkage method and simple matching as a similarity index that shows the six-cluster solution.

5.Discussion and conclusion

Smart government is the newest modernization wave in the public sector and promises to bring more customer orientation and effective administrative action via the application of data-driven technologies. Since smart government is still in its infancy (i.e. pre-adoption) phase, the perceptions and expectations of public managers and other actors involved in these projects are highly influential and can even determine their success or failure. We have explored and described which barriers and challenges these people perceive when initiating and adopting such projects.

To understand the perceptions of actors involved in these projects, we explored data from 32 semi-structured interviews. This procedure led to 17 barriers to the adoption of smart government. To explore whether and how these barriers interrelate, we conducted a cluster analysis, which revealed six different barrier groups: legal foundations, technical infrastructure, cost-benefit relationships, innovativeness, legitimacy, and policy coherence.

Fountain’s (2001) distinction between institutional and organizational features for technology adoption is also present in our data. Cost-benefit relationships, innovativeness, and technical infrastructure are organizational barriers, because they occur inside a public administration and its departments. The reported lack of clarity about cost-benefit relationships in smart government initiatives leads decision-makers to exercise restraint. If an organization is able to establish new business models and revenue structures, these concerns are likely to diminish. Innovativeness refers to an organization’s capacity to introduce new processes, products, or ideas into the organization (Hult et al., 2004), and is often studied as an attribute of an organization’s innovation culture (Pervaiz, 1998). Smart government initiatives also require innovativeness, since they seek to establish new models of public service delivery and new products. The innovation cluster combines aspects such as resistance to change, risk-avoidance, inadequate political-administrative management initiatives, and a lack of skills, which fit Pervaiz’s (1998) conclusions. Further, organizations usually decide themselves which hardware and software to use. Providing technical infrastructure that enables smart government initiatives forms part of an organization’s scope. Even if private organizations can decide more independently, this still applies to public organizations, since hardware and software use is rarely a political decision.

It follows that an organization must address issues concerning innovativeness, cost-benefit relationships, and technical infrastructure. Establishing new laws that enable smart government initiatives, overcome existing policy incoherencies, and increase legitimacy among citizens (their willingness to accept and support changes they consider to be beneficial) are also perceived as key for the successful implementation of smart government initiatives. However, an organization can do little about policy coherence, legitimacy, and legal foundations. Based on Fountain’s (2001) framework of technology enactment, we can say that these barriers exist outside the organization and form part of the institutional framework in which a public administration is located.

As the legitimacy cluster suggests, creating and maintaining legitimacy for smart government is a major challenge for project leaders and politicians. Skepticism about the use and abuse of behavioral data is widespread and steadily increasing. However, recent studies show that citizens accept data collection, connection, and analysis more readily if they understand specifically how their government will use such data (van Zoonen, 2016). Further, since citizens’ acceptance and uses of ICT-based smart government services are influenced by their social groups (Yeh, 2017), policymakers should find ways to communicate comprehensively with all these groups in their jurisdictions. These services must also focus on users’ primary needs: service quality, innovative delivery, and privacy.

Further, policy coherence suggests that there are many different interests and actors in smart government initiatives that need to be coordinated. Beyond the peculiarities of Switzerland’s political-administrative system, policy coherence has three aspects: First, collaboration across different entities in a public administration, for instance when inter-agency collaboration is required for specific tasks. Second, collaboration across government levels in cases where multilevel governance solutions are necessary, for instance, when state and city governments share responsibilities for specific issues. Third, collaboration over time, for instance, when smart government is implemented in sequential phases. Clear strategies foster the diffusion of e-government (Savoldelli et al., 2014; Wing, 2005). Thus, the policy coherence cluster describes the institutional setting in which a public administration is embedded. The same applies for legal foundations.

A glance at the research into innovation adoption in the public sector shows that the identified clusters are similar to past research findings. Legal barriers, financial resources, policy coherence, and aspects of the innovation cluster have been found to hinder the adoption of e-government (Gil-Garcia & Pardo, 2005; Savoldelli et al., 2014; Schwester, 2009) and the implementation of open data initiatives (Janssen et al., 2012). A lack of technical infrastructure, cost-benefit considerations, and legitimacy were found to be barriers to the adoption of e-government (Gil-Garcia & Pardo, 2005; Savoldelli et al., 2014; Schwester, 2009; Wing, 2005; Zakareya & Zahir, 2005). Thus, it can be concluded that the barriers to innovation adoption remain the same, regardless of the innovation type, indicating that these barriers are deeply rooted in the public sector and are therefore hard to address or eliminate. Considering this, and considering that these barriers were also present in the advanced stages of e-government implementation, it can also be concluded that they may not exclusively relate to the early stages of smart government adoption, but may also be present in later stages.

Taken together, this paper makes three main contributions. First, the 17 barriers give us clues about what actors involved in smart government have in mind when planning and implementing such initiatives. Our findings suggest that they must address 17 heterogeneous barriers summed up as legal foundations, technical infrastructure, cost-benefit relationships, innovativeness, legitimacy, and policy coherence more or less simultaneously. This indicates that the adoption of smart government is a complex and multifaceted task with many potential pitfalls. The six clusters can provide guidance on the factors that should be considered when implementing smart government projects. Second, the barriers we have found seem to be more or less the same as in the era of e-government. Thus, barriers encountered in the e-government era still strongly impact on new digitalization waves, of which smart government is one. In other words, the successful implementation of smart government is threatened by the same factors as in the case of e-government. Conversely, recipes for success in e-government may also hold true for smart government. Third, there are both organizational and institutional barriers to the adoption of smart government. Thu, the actors involved in the implementation of such initiatives must consider their own organization as well as the institutional framework when implementing them. Addressing organizational barriers (technical infrastructure, cost-benefit relationships, and innovativeness) may be within their scope of action, while overcoming barriers embedded in the institutional framework (policy coherence, legal foundations, and legitimacy) may go beyond their sphere of influence.