The struggles of smart urban governance

(1)

STUDENT: S.E.J. BOSMAN

(2)

Foreword

I would love to thank several people that supported me throughout the process of writing my master thesis. Without their help, it would have been a long road, but more importantly, a less inspiring journey. Firstly, I would like to thank my supervisor, Sarah Giest, who has guided me in conducting my very first qualitative research. It was nice having her as a supervisor, as she was always very kind and enthusiastic about my research plans. She provided me with excellent comments, in particular on the theoretical chapter, that helped me to refine my final work.

Furthermore, I would like to thank my supervisor at Accenture, who sharpened the practical value of my research. He provided me with various helpful contacts that helped me to get familiar with the complex world of the Internet of Things. The knowledge of the Internet of Things did not just help me conducting my research, but it will be useful in the upcoming years of my career as well. I am grateful for the help of Johan Stokking, who taught me a lot about what you can actually do with sensor technology and what the LoRa network is all about. I want to thank Ger Baron, the Chief Technology Officer of the municipality of Amsterdam, for helping me with arranging interviews within the municipality.

Lastly, I would like to thank my family, friends and boyfriend for the necessary distracting moments during the last five months. Furthermore, I would like to thank my fellow interns at the Technology Strategy department of Accenture, with whom I experienced hilarious moments at the office, in particular after a long day of literature research and designing research models. The comfort of having these three peers with the same daytime activities helped me working towards the goal of successfully finishing my master thesis.

(3)

Abstract

This Master thesis investigates the role of several institutional characteristics of the municipality of Amsterdam in the development of smart public policies. More specifically, this study attempts to uncover what obstructs the municipality to implement a LoRa network to enable technologically advanced policies. Although Internet of Things and smart governance verge on the edge of virgin territory, previous academic work reveals several important institutional mechanisms that tend to obstruct the acceleration of smart governance. Path dependency, known as behavioural loops inclined by established norms and routines, risk-aversion and incrementalism, tends to block policy innovation in a public environment (Howlett, 2009). Furthermore, new policies will inevitably have to cope with prior policy layers that may cause new policies to be less effective. In addition, IoT applications are not viable without being compatible with the established institution (Vlacheas et al., 2013). According to Head (2008), interaction with private parties tends to influence the development of smart policies, depending on the content of the agreements. This qualitative research represents a within-case study of three departments of the City of Amsterdam: the Water Management department, the Bicycle-Parking department and the Asset Management department. Based on twelve interviews, this study indicates that the impeding effect of path-dependent behaviour (H1) on smart policy developments is prevalent in all cases. Moreover, the interviews imply that the incompatibility of established policies and technologically advanced policies (H2) is likely to impede smart public policy development in both the Bicycle-Parking department and the Asset Management department. Finally, only in the Bicycle-Parking department, established agreements with private parties (H3) tend to slow down the development of smart public policies. The academic contribution of this thesis lies within combination of smart city literature and policy innovation literature in an IoT setting. Moreover, this thesis may enable municipalities to guard some institutional flexibility in order to spur policy innovation. Since the hypotheses are based on the assumption that the development of policy innovation and the development of smart, technologically advanced, IoT incorporated policies are influenced by the same mechanisms, the causal mechanism still requires thorough examination.

(6)

1 Introduction

Complexities in urban governance are perhaps of all times. However, each era may face different struggles. Nowadays, municipalities cope with ever increasing citizen demands within limited time and space constraints. Embedded devices respond to this issue as they provide efficient governance solutions. However, are governmental institutions capable of being front runners in implementing large scale technological innovations, and if so, would that even be desirable?

The point of departure

The acceleration of globalization and the technological pace have a vast impact on our business environment as well as our governmental facilities. The mutual reinforcement of globalization and technological innovations play an important role (Aggarwal, 1999). New technologies enable business to become more global, whereas simultaneously globalization allows for international knowledge spillovers, enhancing innovations (Keller, 2004). The government responds to the changing business environment by adapting policies on international, national and local level. Urban areas are turning into ‘smart cities’ and municipalities are increasingly facilitating public services through so-called embedded devices that are fully dependent on a continuous Internet connection. Services that provide real-time information on public transport or monitor water and air pollution are common examples. This study responds to this tenor by shedding light upon the facilitation of such smart public services in the City of Amsterdam.

The emergence and proliferation of ICT forms the foundation of the smart city literature. Without the Internet, a smart city would be non-existent. However, prior studies diverge in terms of progress in ICT development. For instance, literature on cyber cities or digital cities, which are the precursors of smart cities, dates back to before the emergence of ‘embedded systems’ (Domingue et al., 2011). That is, embedded systems can be considered the skeleton of Internet of Things, which is defined by the communication of devices with other devices over the Internet, without the interference of people. Such technologies are only recently developed. Hence, much prior literature often does not acknowledge the existence of the Internet of Things. This research contributes to the literature by exposing recent developments in ICT, regarding the prominence of embedded systems. In order to manage increasingly

(7)

complex urban problems, evidence-based policies and big data are introduced in governance. Evidence based governance relies heavily on the value of data, and may provide effective governance solutions through the use of data. However, this requires new policy instruments that must be incorporated in current institutions. This study aims to uncover which factors obstruct the implementation of smart public policy solutions. The research question that can be derived from this is:

“Which factors potentially obstruct the development of smart public services in the City of Amsterdam?”

This case study will be viewed in the light of evidence-based policy literature, smart city literature, network governance, historical institutionalism and policy layering literature. More specifically, technological developments regarding connectivity and sensor technology in the City of Amsterdam are being studied. Additionally, this study attempts to investigate how a LoRa network fits into the line of smart technological developments. A case study is conducted among three departments of the City of Amsterdam: the Parking Department, the Asset Management Department and the Water Management Department.

Thesis outline

The thesis is outlined as follows: Firstly, a literary background is presented in order to outline diverging academic perspectives on the topic. In this chapter, various aspects that concern the implementation of technological innovations in governmental institutions are brought together. Furthermore, this chapter provides a conceptualization of all variables. Secondly, the research methods will be elaborated upon. This chapter initially discusses the causal model and the composition of hypotheses. Thereafter, the research design will be outlined, extensively discussing the type of research, the choices that are made regarding the interview method and the selection of interviewees, the case study selection, the validity of the research and the research threats. Thirdly, the empirical findings that are obtained through interviews are discussed in the next chapter. The aim is to illustrate the results in a rather static manner,

(8)

2 Literature review

This chapter gives an overview of prior academic work on technological developments in a governmental context. More specifically, it elaborates on literature strands regarding smart cities, big data, network governance, policy layering and path dependency. Furthermore, a critical reflection of diverging academic perspectives is provided. Finally, the key concepts for this research are briefly elaborated upon. This chapter forms the framework for empirical research, as the causal mechanism and hypotheses are derived from the academic work that is presented hereafter.

2.1 Smart city literature

The term ‘smart cities’ was introduced the early 1990s, when technological innovation and globalization became important factors in the evolution of cities (Domingue, 2011). The link between smart cities and ICT is implied by the simultaneous emergence of ICT, achieving great international importance in the 90’s (Caragliu et al., 2009). Although the topic ‘smart city’ is relatively nascent, within the literature various forerunners can be identified, including ‘cyber cities’, ‘digital cities’ (Domingue et al., 2011) and ‘intelligent cities’ (Komninos, 2002). One may perceive these as separate or isolated concepts, however, the differences between the concepts seem to originate from some sort of sequential order (Meijer et al., 2015). Hence, cyber cities refer to the management of a growing amount of information and intensified use of computers. A few years later, the concept ‘digital cities’ refers to the momentous of virtual and online environments (Domingue et al., 2011). Thereafter, intelligent cities where introduced, which emphasise the creation of urban intelligence through the collaboration of the citizens, politicians and other stakeholders, enabled through online technologies (Komninos, 2002). Smart cities surpasses the intelligent city as it incorporates collaboration and interaction of the entire city through the usage of advanced technologies that include smart phones, embedded sensors, and Internet of Things (Caragliu et al., 2009).

Contemporary, the phenomenon smart city experiences a growing importance, as future economic welfare increasingly relies on the Internet. Daily lives of citizens are more and more shaped by online activity. Moreover, the role of cities regarding innovation is expanding. Additionally, the current emphasis on cities as opposed to countries is fanned by urban migration (Domingue, 2011). Cities will become more competitive and are therefore in need

(9)

of increasingly sophisticated facilities. Cities are considered “complex ecosystems”, and the governance of such ecosystems increasingly requires advanced approaches (Domingue et al., 2011; Kitchin, 2014). Demands are expanding in various domains, including safety, energy, environmental issues, healthcare, media and other public services. These functions are increasingly provided with an online application. Hence, the proliferation of smart cities can be perceived as a counter-reaction to the complexity of urban areas (Caragliu et al., 2009).

Tensions

A smart city is a broad concept that incorporates many aspects ranging from human and social capital, market-led governance, urban learning and sustainable development, enabled by advanced information and communication technologies. However, several tensions can be revealed within the concept. Meijer et al. (2015) refer to a smart city as a utopian perspective on an urban environment that controls governmental complexity with technological solutions. Although technological applications are often fit to support modern governance, solutions to wicked problems do not solely rely on technology but require a more comprehensive approach (Meijer et al., 2015). Within the smart city literature, there is a strong inclination towards technological and market-led solutions to urban problems (Kitchin, 2014). Smart cities have gained popularity among both governments and private firms. Governments increasingly rely on market consultation, since the industry provides technological applications that relieve or simplify governance. The industry, on its turn, invests in collaborations concerning smart facilities, as it taps a new source of income and may contribute to the competitive advantage of firms. Consequently, urban governance becomes increasingly technocratic, resulting in a greater gap between knowledgeable experts and legitimate politicians (Kitchin, 2014). Additionally, Meijer et al. (2015) acknowledge the increasing influence of large IT firms in shaping smart cities, but emphasise that, from a legitimacy perspective, the ultimate accountability of the government should be carefully guarded. Furthermore, urban life has become more complex and its governance is increasingly demanding and hectic. While consultation of private parties and technical solutions is evident, legitimate governance must be carefully maintained.

(10)

communicating with foreign cities and attracting people from far abroad. On the other hand, urban governance has received growing attention over other levels of government, due to the increased complexity of cities and extended local governance tasks (Domingue et al., 2011). Due to the increased competition between cities, it becomes more important for a city to profile itself in an attractive manner. As a result, the prominence of urbanity grows (Kitchin, 2014). The challenge here is to serve an expanded number of people ranging from international elite classes to social class citizens in a broader spectrum of facilities.

Conceptualisation

Due to both the diverging academic stances on smart cities and the variable nature of involved technology over the past decades, the concept ‘smart city’ remains somewhat vague (Meijer et al., 2015). However, an attempt is made to provide a decent conceptualisation of the topic. When discussing the concept ‘smart city’, several elements should be taken into account. Firstly, a city is considered to be smart when sustainable economic growth is driven by human and social capital (Caragliu et al., 2009). This mainly refers to the human value through knowledge and skills, and the network between groups and individuals (Coleman, 1988). More particularly, Giest (2014) elaborates on the importance of networks for smart cities. According to Giest (2014), network governance plays an important role, since the urban governance in a smart city is typically participatory, rather than solely directed by the municipality. Secondly, in a smart city economic growth and welfare is driven by ICT infrastructure (Caragliu et al., 2009). Innovation and ICT technologies happen to be highly sophisticated. Furthermore, He et al. (2014) emphasise the importance of smart devices to facilitate and monitor urban facilities. Thirdly, a focus on the quality of urban life is crucial for defining a smart city (Caragliu et al., 2009). Urban facilities are typically well developed and a broad range of services is provided. Fourthly, Coe et al. (2001) emphasise the ability of a city to learn, adapt and innovate. Hence, for a city to be smart, it needs to have sufficient absorptive capacity (Caragliu et al., 2009). In other words, modern technologies need not only be present, but bureaucrats, businessmen and citizens should know how to exploit them. In order to achieve this, human and social capital is required.

(11)

Table 1: An overview of the elements that define a smart city Smart city elements 1. Sustainable growth is realised by human and social capital, through the use of networks 2. Economic welfare is fanned by ICT infrastructure and technological applications 3. There is a strong emphasis on the quality of urban life 4. The urban environment has the ability to learn, adapt and innovate

2.2 Data revolution and evidence-based governance

In order to grasp the urban complexity, contemporary urban governance literature extensively relies on the absorption of enormous amounts of data. So called big data has made a revolutionary impact; some scholars even speak of a paradigm shift (Kitchin, 2014b). Big data are defined as “massive, dynamic, varied, inter-related, low cost datasets that can be connected and utilised in diverse ways, thus offering the possibility of studies shifting from: data-scarce to data-rich; static snapshots to dynamic unfoldings; coarse aggregation to high resolution; and relatively simple hypotheses and models to more complex, sophisticated simulations and theories” (Kitchin, 2014a: 3). According to Kitchin (2014a), among the three types of big data that can be distinguished – directed, automated and volunteered big data – particularly automated data is relevant for the understanding and governing of urban complexity. Such automated data is often obtained through the use of sensor networks and chips that are embedded on different products or devices, continuously absorbing information. Municipalities apply such smart technologies for registering traffic movement, lighting, energy, waste and temperature. As such, big data absorbed through smart devices enable municipalities to conduct both real-time analyses and adaptive forms of governance (Kitchin, 2014a).

Hand in hand with the data revolution goes the inclination of governmental institutions to rely on evidence-based approaches. During the last decennia, evidence-based policy has become a

(12)

of data. Consequently, data is used as an input for evidence-based policy designs. Furthermore, the complexity of modern governance plays a key role in the rise of evidence-based policy. In contemporary society, political decisions are legitimized through actual performance rather than through the democratic system, making public policy increasingly result-driven (Sanderson, 2006). While governance typically emphasises the development of policy through politics and bureaucratic mechanisms, currently, the focus has shifted towards policy delivery. Evidence-based policy claims to be more effective than traditional policy, since it focuses on how to improve society rather than understanding it (Solesbury, 2001). Furthermore, it lends methods from the private sector such as consultancy firms. Governments are increasingly paying attention to the practical usability and the communication of research results. Hence, the growing importance of collaboration between private and public parties in the development of governance is evident.

Evidence-based policies are implemented in an effort to limit policy failures. Howlett (2009) argues that governments need a certain level of policy analytical capacity in order to exploit data-based decision making. Evidence-based governance requires highly sophisticated learning capacity in order to perform, as it involves the development of new capabilities and the adoption of new technologies (Kim et al., 2014). Yet this is conform to the smart city literature, which implies that one of the features of a smart city is that it is capable of learning and adapting (Coe et al., 2001). However, many governments, even the ones with relatively sophisticated data analytical instruments, are not capable of processing and analysing policy input in a proper manner (Kim et al., 2014; Howlett, 2009). Governmental institutions often struggle with organizing data, and on top of that, they tend to behave as silos, operating rather isolated from other departments or abroad institutions (Kim et al., 2014). This hinders effective communication and information sharing.

Furthermore, since the evidence-based approach emphasises the practical value of policy, policies cannot succeed if they lack a proper evaluation on how well they ‘work’. A thorough evaluation enables policy makers to obtain insights into the pros and cons of a policy and to map adaptations if necessary. Additionally, since both evaluating and learning is required in order to adjust policy based on up-to-date information on prior performance (Sanderson, 2002), the time pressure on these tasks is rather high. This adds to the complexity of evidence-based policy. However, it must be noted that evaluation alone is not satisficing to accomplish policy improvement. After mapping and analysing policy outcomes, policy

(13)

makers must learn from the results, which requires memorization and intelligence (Sanderson, 2002), after which information must be organised and shared properly (Kim et al., 2014).

Concerns

Data management has become indispensable in modern governance. However, academics have pointed to several concerns regarding the proliferation of evidence-based approaches.

Integral vs. incomplete solution

Firstly, critics point out that data management methods cannot be perceived as an integral policy solution, but should be regarded as solutions for information processing (Uprichard, 2013). The value of big data is often overestimated by both scholars and practitioners; data alone cannot provide solutions to wicked problems. Smart city defenders unjustly imply that governance can be approached technically. However, too much emphasis on technical solutions may offer a limited governance approach, overexposing efficiency problems while neglecting social, cultural and political issues (Kitchin, 2014a). Rather than fully relying on evidence for performance measurement as policy input, Head (2008b) implies that policy makers should take into account political, cultural and ethical considerations. Furthermore, Bell (2004) argues that the discrepancy between human rationality and the complexity of social and economic dynamics is sufficiently large to make any attempt to filling this knowledge gap pointless. However, although evidence-based policy cannot solve governance problems solely, it may at the very least contribute to the efficiency of policy measures (Solesbury, 2002).

Secure and objective vs. vulnerable and subjective data

Furthermore, the emphasis on big data within urban governance makes governmental information both volatile and vulnerable (Uprichard, 2013). Namely, when data is excessively used as policy input, it probably becomes more valuable. This may result in hackers’ attempts

(14)

inevitably subjective, at least to some extent (Kitchin, 2014a). Nonetheless, rational approaches such as evidence-based governance can be improved by establishing institutionalised consultation of the community, making policies more reflective and robust (Bell, 2004). Head (2008b) points out that evidence collected by policy makers is never really objective, due to the fact that the process of collecting evidence and the stance you take as an evaluator involve choices.

Legitimate vs. technocratic governance

Furthermore, tensions may arise from the fact that the government heavily relies on private firms in order to cope with data management. Processing big data requires advanced systems that are often developed and managed by software firms. However, in order to maintain legitimate governance, the boundaries between the maintenance of governmental accountability and the exploitation of public-private collaborations must be carefully guarded. In this line of argumentation, critics elaborate on the self-interests of large software firms to intervene in smart urban governance, especially through technological assistance. Urban governance is increasingly accomplished through public-private partnerships. However, excessive reliance on the industry may result in both a “marketization of public services”, and a “technological lock-in” by powerful firms (Kitchin, 2014a). Furthermore, one may argue that the emphasis on fast and thorough monitoring, analysing, and adjusting policies creates too much focus on the technical aspects of governance and pushes away politics. Hence, although evidence-based policy is very promising and particularly contributes to the reactivity of governance (Sanderson, 2002), the boundary between effective evidence-based governance under political sovereignty and pure technocracy must be carefully guarded. There is a concern that dominance of instrumentality will overrule ethical and moral standards in policy making (Sanderson, 2006).

2.3 Network governance

Network governance refers to an increased level of interaction between various parties or organizations that connect with each other through some common factor such as an industry or a geographical area. The topic gained momentum in the 1990s, when Silicon Valley flourished (Giest, 2014). The growing prominence of the topic may be partially due to a

(15)

snowball-effect caused by the success of prior networks, however, a range of alternative explanations is proposed to clarify the rise of network governance. Several scholars argue that globalization and the accelerating technological pace form an incentive for organizations to collaborate more intensively (Chesbrough, 2003; Narula and Hagedoorn, 1999). The rapid pace of technological innovations increase competitive pressures, which stimulates organizations to collaborate and become interdependent. Likewise, globalization encourages international organizational activity, stimulating collaboration. Additionally, the mutual reinforcement of globalization and technological innovations encourage organizations to become more global while stimulating knowledge spillovers (Aggarwal, 1999).

During the 1990s, the idea rose that investing in human and social capital and innovation contributed to the economic growth of an area, which stimulated collaboration on a technological level (Giest, 2014). Furthermore, the transition from a welfare state to a regulatory state inclined private and public stakeholders to work together. For instance, where the government used to deal with public facilities by itself, now other parties are contracted and involved in various ways, leaving more room for interference of NGOs and (semi-)private organizations. Head (2008) proposes that network governance gained popularity both theoretically and practically, due to the fact that modern society relies more on the involvement of citizens and NGOs in governmental policy. Therefore, consultation of various types of stakeholders has gained importance. The competitive advantage of an urban area then depends on the strategic management of organizations within the network, so that collaboration maximizes the benefits of the organizations in the network. This is typically established by innovative technologies and advanced communication infrastructures. “The challenge is to make intelligent use of new inclusive techniques to foster innovative solutions and better outcomes”, (Head, 2008: 734).

Conceptualisation

In order to get a full understanding of the concept, network governance will be conceptualised first. During the past decennia, multiple academics have referred to the concept in several

(16)

of various collaborating entities within a network. Despite differences in nuances, authors agree that such a network or cluster encompasses a collaboration of different organizations, initiated to serve a common goal that is significantly different from a market situation or a hierarchy (Jones et al., 1997). However, this explanation refers to network governance in a broader sense, which may as well refer to collaboration between private firms, whereas this study particularly emphasises the governmental aspect of network governance. Although both derive from the same body of thought, they differ in their focus. For instance, network governance in the public sphere includes the contracting and monitoring of private and semi-private parties by the government (Provan & Kenis, 2008). Network governance elaborated upon by business scholars refers to the management of a network between autonomous, private firms, collaborating to achieve a common goal or competitive advantage (Porter, 1998; Sydow et al., 1998; Jones et al., 1997). Additionally, public network governance inevitably concerns the facilitation of public services whereas private network governance focuses more on private benefits.

In an effort to provide a definition of networks applied in the public context, Torfing formulates the following definition: “Governance networks can be defined as: (1) relatively stable horizontal articulations of interdependent, but operationally autonomous actors who (2) interact with one another through negotiations which (3) take place within a regulative, normative, cognitive and imaginary framework that is (4) self-regulating within limits set by external forces and which (5) contributes to the production of public purpose” (2005: 4). Within this definition, actors may be political actors, bureaucrats, business men, interest groups, citizens or other stakeholders.

Characteristics

Network governance typically emerges under certain conditions that establish easy accessibility of the network and cheaper development of products. More particularly, Head describes four conditions should be present to network governance to occur: “(1) demand uncertainty with stable supply, (2) customized exchanges high in human asset specificity, (3) complex tasks under time pressure, and (4) frequent exchanges among parties comprising the network” (2008: 918). Giest (2015) adds that facilitators are needed to form a link between organizations in the network, contributing to the enhancement of network capacity. Such facilitators or experts can be governmental and non-governmental and especially have a

(17)

communicative value within the network. They are likely to strengthen collaboration and cause knowledge spillovers. Although clustering has gained traction in the academic field, cluster policy also has a high risk of failure as a result of ineffective network arrangements (Giest, 2015).

2.4 Policy innovation

In order to meet complex citizen demands and deal with an ever growing data overload, governmental institutions are challenged to innovate their internal organisation as well as their public services. However, according to Nam and Pardo (2011), smart city innovations produce opportunities as well as risks. Advanced information technologies are of paramount importance in smart policy innovations, as they are required to process large amounts of data. But regardless of the paramount importance of an ICT infrastructure, it offers only a partial solution to smart policy innovations (Komninos, 2009). Broadband solely provides an innovative atmosphere, while human capital is necessary in order to create innovations. “An unsophisticated equivalence of intelligent cities and broadband networks has led many cities around the world to invest heavily in communication infrastructure assuming that broadband is a sufficient window to globalisation and competitiveness. However, broadband is not enough; intelligent cities are equally about knowledge exchange, human skills and innovation support institutions” (Komninos, 2009: 339).

Likewise, Nam and Pardo (2011) argue that despite the natural inclination to search for technical risks, the most common reason for smart city innovation failure is to be found within organizational matters. Such reasons include poor strategic conceptualisation and planning, and a lack of management support and leadership (Nam and Pardo, 2011). On top of that, Komninos (2009) argues, organisational skills are required in order to catalyse innovation. However, within a public sphere one deals with both legacy and sequences of prior policies (Kay, 2007). Therefore, future policies may be determined or constrained by

(18)

Path dependency

The elaboration of Giest (2015) and Head (2008) on the emergence of networks illustrates various incentive to collaborate, implying a rather functional approach. Contrarily, path dependency literature addresses the evolution of institutions and processes from a historical rather than a functional perspective. However, path dependency literature may serve as a complement rather than a supplement. As economists clarify how efficient transactions are achieved, path dependency may explain situations in which efficiency cannot be achieved due to the “lock-in” of mediocre technologies (Howlett & Rayner, 2006). Historical institutionalists argue that institutions evolve along fixed pathways and processes are determined by historical sequences (Peters et al., 2005). Path dependency emphasises the evolution of an institution as a trajectory of causal sequences rather than a rational process (Peters et al., 2005). As Howlett describes it, “Historical institutionalism, as is well known, is focused on the manner in which human behaviour is structured by existing norms, rules and organizations in order to promote predictable patterns of stability in the face of otherwise highly variable behavioural possibilities”, (2009: 242). Mahoney (2000) proposes three components that a proper conceptualisation of path dependency should entail. Firstly, there should be a causal importance of events that occurred early in the historical sequence. Secondly, events should be contingent to a certain degree. Thirdly, inertia should occur in the process, which means that events are deterministic to such an extent that patterns become more or less fixed (Mahoney, 2000).

Furthermore, due to its rigid and fixed nature, the continuance of existing paths may cause the failure of an technically efficient solution. Howlett and Rayner (2006) argue that this failure represented by path dependency is ascribed to either of three dynamics: so called network effects, increasing returns or standardization. Goldsmith and Crawford (2014) stress the resistance to change in governmental institutions in the context of digital innovations. Laws and procedures are often designed in favour of risk reduction, but to the detriment of flexibility and resilience. Furthermore, civil servants are rather timid and risk averse when it comes down to policy changes. Civil servants generally prefer the continuance of established routines, even though changes may entail more efficiency. In addition, policy makers may be afraid that their roles will change and they will lose sovereignty. Hence, Goldsmith and Crawford (2014) argue that policy innovations are most likely to be successful when aiming

(19)

at satisficing compromises rather than the best possible policy strategy. In addition, Zhu (2013) addresses the relevance of incrementalism in policy innovation. To illustrate, Zhu (2013) describes the housing system policy in China, which entailed a reform approach in the late 70s, but the policy reform was not completed until 1998. Furthermore, China has a tradition of distinguishing ‘old people’ and ‘young people’ with regards to housing subsidies. Continuing this path eventually lead housing equalities to increase, which was an unfavourable outcome.

(20)

Policy layering

Due to path dependent behaviour of policy makers, it is often difficult to realise change. Indeed, creating a flexible institutional setting and adapting policies to conduce innovation is rather complex. As Howlett and Rayner point out, “[…] new policy development is almost always constrained by previous policy choices which have become institutionalised.”, (2007:1). That is, each policy instrument is entangled in a broader spectrum of policy instruments which influence one another, resulting in a declined effect of the sole instruments. Hence, policy instruments typically evolve out of an incremental sequence of prior policies. As Kay (2007) denotes, new policies overlap rather than replace established policies. Hence, the activity of layering is described by Kay (2007) as a pertinent factor in the majority of public policies. When policies are layered onto prior policies and several policies are not compatible with each other, policy failure occurs (Kay, 2007). The extent of policy failure depends on the relative coherence among policy goals and the consistency of policy instruments. The incompatibility of old and new policy instruments is partly determined by the level of institutionalisation of the old instruments. The more institutionalised they are, the more likely it is that increasing returns, sunk costs and incremental learning will arise, which limits future policy options (Howlett & Rayner, 2007). Consequently, policy layering potentially results in inefficient processes and complex policy mixes.

The literature uncovers several ambiguous definitions of policy layering. Kay refers to policy layering as: “Institutions and policies are never wholly replaced but rather evolve by the addition or layering of new institutions or policies that may sit in a tense relationship with the existing institutions and policies and introduce functional instabilities into the overall policy system” (2007: 580). Hence, this policy layering is defined here as the patching of policies onto prior policies. Prior and new policies may be incompatible, but this does not have to be the case. Contrarily, Howlett and Rayner (2007) speak of policy layering when the stacking of policy inevitably results in policy failure, as both policy instruments and policy goals are incompatible with prior policies. Following this line of argumentation, three kinds of failure can be distinguished: policy layering, policy drift and policy conversion (Howlett & Rayner, 2007; Béland, 2007). Policy layering occurs when both the policy goals and the policy instruments of the old and new policy are conflicting. Policy drift refers to the situation in which policy goals are coherent but policy instruments are not adapted accordingly, resulting in ineffective instruments. Policy conversion addresses a situation in which policy instruments are altered in an attempt to achieve new goals, but the new goals are obstructed (Howlett &

(21)

Rayner, 2007). Throughout this study, Kay’s (2007) definition of policy layering will be used. Thus, policy layering refers to the stacking of policies onto each other, resulting in a layered policy mix that represents a historical evolution of policies (Kay, 2007).

Whether policy layering is perceived as problematic, depends not only on the incompatibility of prior and novel policies, but it is also influenced by policy paradigms. A distinction is made here between material and ideational aspects (Kay, 2007). Material aspects represents incompatibilities that result from competition, redundancy, or contradicting stakeholders. To illustrate, competition between layers can arise when both policies attempt to address the same financial resource or assets. Moreover, redundancy may occur when the new policy makes the prior policy irrelevant. On the other hand, the ideational aspect of policy layering refers to the prevalence of policy paradigms. Policy paradigms provide a cognitive map that frames policy makers’ perceptions of policy failure and success (Béland, 2007). Hence, the success of policies depends on the actual consistency between policy layers as well as the prevailing perception of policy performance.

In a highly dynamic environments such as smart cities, a tension occurs between the stagnating tendency of path dependency and policy layering, and the turbulent tendency of innovations. Urban innovations ask for flexible institutions and a regular adjustment of policy goals and instruments. Namely, a changing urban environment rises new demands and trends which require new policy goals. What is more, the technological pace stimulates innovativeness in policy instruments. To illustrate, contemporary public policy relies heavily on broadband and there is a growing use of mobile applications. Within several years, the government will probably be largely dependent on IoT applications (Vlacheas et al., 2013). However, established policies will remain even though new policies will appear. The paramount question is whether old and new policy layers are compatible. The compatibility of policy goals and instruments over time determine the level of constraint for policies. Moreover, the policy paradigm determines the level of infiltration of new policies.

(22)

3 Research design

This chapter elaborates on the research design that is used in order to investigate the causality between the different variables that are deduced from the literature. The first section presents the conceptual model that is studied. Thereafter, the causal mechanism and the hypotheses are elaborated upon. The following section illustrates an operationalisation of the variables, which form the key concepts of this study. In the fourth section, the case selection is discussed. This chapter ends with an elaboration on the data collection.

3.1 Conceptual model

This study elaborates on smart policy solutions in the City of Amsterdam. Developments in ICT technologies make urban public services more sophisticated and adjusted to the citizens’ needs (Caragliu, 2007; Head, 2008; Domingue et al., 2011). In order to cope with complex problems and diversified demands, governance increasingly relies on evidence-based policies and big data. However, such revolutionary governance methods must be incorporated into current governmental entities. The aim of this research is to uncover compelling factors in the implementation of smart public policy solutions.

Prior academic work uncovers several bottlenecks that hinder policy innovations. Deduced from the literature, three independent variables are identified that may impede a fluent implementation of smart policies. These include path dependent behaviour of civil servants within the municipality, the incompatibility of established policies and IoT systems, and constraining collaborative agreements with private parties. Hence, this study encompasses an internal and an external component in governmental processes. As such, it attempts to uncover both the governmental restraints regarding the development of smart policies, and the role of interaction with non-governmental parties – since those who develop supporting policy tools often reside outside the government. The research question is as follows:

“Which factors potentially obstruct the development of smart public policies the City of Amsterdam?”

(23)

This study investigates the causal relationship between potentially obstructing factors and the development of smart public services. This hypothesised causality is visualised in a conceptual model below. The development of smart public services forms the dependent variable and is positioned in the centre of the model. On the right side the applicability of the LoRa network is illustrated, which represents the smart service application that is thoroughly investigated in this case study. The LoRa network, which stands for Long Range network, is a network that particularly developed for IoT applications. It resides at a relatively low frequency, which leads to a long network range, transmission of small data packets and limited power usage. The LoRa network is only apt if the municipality is already familiar with smart public services. This is due to the fact that the network is particularly developed to connect embedded devices, and, by definition, smart public services involve embedded devices (Caragliu, 2009). Hence, IoT applications and sensor technology should be integrated in public policies in order to call a public service smart. The independent variables are illustrated on the left side of the model. The independent variables are composed based on the literature review and are expected to affect the development of smart public services.

Path-dependent H1: behaviour Incompatibility of established H2: policies and IoT systems Collaboration H3: and process integration with industry Applicability of LoRa network Development of smart public services

(24)

3.2 Causal mechanism and hypotheses

In globally competitive and highly dynamic urban areas, governmental infrastructure should be rather flexible (Coe et al., 2001). Evidence-based policy requires fast assessment and adjustment of policies. Moreover, public services are increasingly adapted to citizens’ needs, which are continuously changing (Domingue et al., 2011). There is a constant stream of big data providing officials with information on how the city behaves (Kitchin, 2014b). Hence, for many aspects of governance fixed policies are becoming obsolete and flexible policy solutions are gaining momentum. Therefore, the governmental focus is increasingly shifting from federal top-down to local bottom-up governance (Coe et al., 2001). However, flexibility may not always be within reach, as bureaucracy is prevalent in many governmental institutions. Path dependency, which refers to historical sequences and more or less fixed pathways in policy development, generally limits the ability to implement new policies (Peters et al., 2005). In addition, Howlett (2009) argues that civil servants’ behaviour is often constrained by existing norms and routines, which limits the flexibility of institutions. Resulting from this, civil servants become rather change-averse and the organisations in which they operate are often entangled in fixed processes. Since contemporary urban environments increasingly require the ability to cope with change, path-dependent behaviour is likely to form a compelling factor for developments regarding smart public solutions. Following this line of argumentation, the first hypothesis is formulated as follows:

H1: Path dependent behaviour of civil servants is likely to form a compelling factor in the implementation of smart public solutions

According to Vlacheas et al. (2013), the Internet of Things will play an important role in the development of smart cities. IoT applications are already being used to facilitate public services and its position in future governance is expected to grow exponentially. However, there are several implications that come with using IoT in governmental policies. New technologies have to be incorporated in policies in order for a governmental institution to use them. Vlacheas et al. (2013) points out that the difference in software between connected devices can form an obstacle in governance, since the functionality of such applications lies

(25)

within the connection of such devices (Vlacheas et al., 2013). Hence, the lack of resilience of IT services is harmful for the development of smart services, since it may create a rather reticent attitude among civil servants to use such advanced technologies. However, besides technical incorporation of systems and devices, such technological policy tools should be embedded into work processes and incorporated into people’s mind-set as well. Howlett & Rayner (2007) argue that policy innovations inevitably deal with prior policy layers, with regards to both established policy processes and instruments, and policy decisions. When taking into account that sensor technology is relatively unknown territory for many governmental institutions, it likely that prior policy layers affect the development of such technologically advanced policy tools. As established policies are probably lagging behind in terms of IoT applications, on top of the theoretical assumption that prior policies tend to constrain new policies, old and new policies are likely to be incompatible. Thus, a gap is assumed between established policy instruments on the one hand, and novel IoT-related policy tools on the other. A second hypothesis can be derived from this:

H2: The incompatibility between established policy instruments and IoT systems is likely to impede a fluent implementation of smart public solutions

Contemporary, municipalities are challenged to innovate their internal organisation as well as their public services. In doing this, urban governance increasingly relies on collaboration with private parties (Caragliu et al., 2009). It is needless to say then, that urban policies are often accomplished by a combination of parties (Head, 2008). Nam and Pardo (2011) point out that cross-organizational integration of information is required in order to sustain collaborative policies, especially those that heavily rely on data. This line of argumentation is supported by the evidence-based literature strand, since the increase of evidence-based governance simultaneously encourages the involvement of technological firms. Private firms perceive governmental issues as a new distribution market for their services, while governmental entities increasingly copy methods from the private sector including consultancy firms. On

(26)

governance. Such long-term agreements are likely to be found in governmental entities, which may overrule the stimulating effect of private firms regarding technological developments. Consequently, two contradictory hypothesis can be derived from this; one indicating that fixed agreements with private parties obstruct smart policy development and one that implies private party interference may spur the development of smart policies. Hypothesis H3b can be considered an alternative hypothesis. Hence, H3a and H3b are not expected to occur simultaneously, since they are conflicting.

H3a: Established collaborations and agreements between non-governmental parties and the municipality obstruct the development of smart public services

H3b: Intensive collaboration and integrated processes advance the development of smart public services

3.3 Operationalization concepts

In order to investigate if factors such as path dependency, incompatibility of policy instruments and collaboration with private parties affect the development of smart public services, these concepts should be operationalised first. The development of smart public policies represents the dependent variable. Smart public policies are public policies that fit into the frame of a smart city (Domingue et al., 2011). A smart city focusses on the quality of urban life (Caragliu et al., 2009); and is capable of learning, adapting and innovating (Coe et al., 2001). Furthermore, economic growth is fanned by social and human capital (Coleman, 1988;; Caragliu et al., 2009 Giest, 2014) and perhaps most importantly, by ICT technologies (Caragliu et al., 2009; He et al., 2014). The most crucial component that distinguishes smart public services from public services that are not smart, is the use of embedded devices (Caragliu et al., 2009). Consequently, within this framework, a smart policy indisputably involves sensor technology and IoT applications.

(27)

Table 2: An overview of the operationalised concepts

3.4 Case selection

This study aims to uncover relevant mechanisms that are likely to influence the development of smart public policies. Despite the fact that this research investigates a rather new topic, an explanatory approach is applicable here. That is, since this research as based on existing literature rather than the exploration and development of a new concept, the research can be considered deductive (Toshkov, 2015). Hence, hypothesis are formulated based on assumed relations derived from prior literature, which are tested through empirical research. A qualitative approach serves the research question best, since the study addresses a relatively

Name variable Type variable Definition Indicators

Development of smart public services

Dependent

variable A public service which is composed in a smart way conform to the smart city framework, whereas ‘smartness’ is determined by the use of embedded devices and sensor technology (Domingue et al., 2011; Caragliu et al., 2009). • Use of sensors • Number of IoT applications • Applicability of LoRa Path dependent behaviour of civil servants Independent

variable Path dependency encompasses the evolution of an institution as a trajectory of causal sequences. Human behaviour is fixed by established norms and rules, which creates a certain predictability in institutional patterns (Peters et al., 2005; Howlett, 2009). • Receptiveness for new policies or technologies • Attitude towards change • Pace of adaption • Behaviour resulting from organisational culture Incompatibility of policies and IoT systems Independent

variable When policies are layered onto prior policies and several policies are not compatible with each other, policy failure occurs (Kay, 2007). The extent of policy failure depends on the relative coherence among policy goals and the consistency of policy instruments (Howlett & Raynar, 2007). • Presence of advanced technologies • Knowledge of tech. tools • Conflicting instruments • Conflicting goals/ demands leading to viscous decision-making Intensive collaboration and fixed agree-ments with private parties Independent

variable Stable horizontal articulations of interdependent, but operationally autonomous (public and private) parties who interact intensively within a regulative, normative, cognitive and imaginary framework that is self-regulating within limits and contributes to the production of public purpose (Torfing, 2005) • Contractual agreements • Level of collaboration • Level of outsourcing

(28)

Additionally, the academic field underexposes the relation between contemporary technological developments regarding IoT, and current, potentially conflicting policy processes. Consequently, obtaining in-depth information on factors that are likely to hinder the development of smart public policies can best be accomplished by conducting interviews rather than surveys. Furthermore, a qualitative approach enables one to obtain intimate knowledge on the research topic (King et al., 1994). However, it should be noted that qualitative research leads to a limited generalizability of the research results (Brutus et al., 2010).

This thesis provides a case study, which investigates impeding factors regarding the development of smart public policies in the City of Amsterdam. More particularly, this study encompasses a within-case analysis of three departments of the municipality of Amsterdam. A within-case analysis can be defined as “the in-depth exploration of a single case as a standalone entity” (Mills et al., 2010). The unit of analysis is the department within the City of Amsterdam. For each case, this study provides a thorough analysis of the level of development of smart public policies, and the seemingly influential constraining factors with regards to the development. This vertical analysis represents the most fundamental aspect of the research. Additionally, a brief horizontal analysis is provided that compares the three departments in terms of development of smart public services and the visible presence of the hypothesized constraints.

The geographical focus of this study is Amsterdam. The urban area of Amsterdam is selected, as it has positioned itself as a forerunner in smart city applications. Amsterdam established multiple smart city initiatives and is ranked as the second smartest city of Europe after Copenhagen, due to its open data, technological and infrastructural developments, and livability (Amsterdam Economic Board, 2016). Next to that, Amsterdam is the only city in which the LoRa network is implemented to such an extent that it can be used by the municipality. Hypothetically, if there were two cities like Amsterdam, similar with respect to urban governance and the development of smart public services, it would have been most appropriate to compare these two cities by conducting a cross-case analysis. However, in case of the absence of a similar city, a within-case analysis of Amsterdam provides the best fit.

(29)

LoRa network

The research encompasses a case study that concerns a technological innovation called the LoRa network. This network was recently created and implemented in Amsterdam and concerns an innovation regarding Internet facilities. The distinctive character of this network is twofold: it is exclusively designed for the online communication of things, so called

embedded systems, and it is completely crowd funded. Embedded systems are systems or

devices that can communicate with each other through Internet, independent from human intervention. The Internet of Things, which refers to a dominant position of embedded systems over people on the Internet, will potentially be fanned by the LoRa network.

LoRa technology literally means a network with a long range, which implies that the network resides at a lower frequency than 3G internet or Wifi and can therefore cover a relatively large area (LoRa Alliance). A lower frequency also allows for less data to be transmitted over the network. However, since embedded systems transmit significantly less data compared to internet users through mobile phones or computers, this is not an issue. In addition, since devices that are linked to the LoRa network would search for connection less frequently compared to devices that are linked to 3G or Wifi, the battery will have a longer lifespan. The lower electricity demand provides the largest competitive advantage of the LoRa technology, since wireless embedded devices do not need a battery replacement for up to six years.

Multiple private firms including KPN are also developing a LoRa network in the Netherlands and abroad. As such, a new commercial product would be created in the form of network subscriptions for embedded devices. To illustrate, a LoRa subscription would be comparable to 3G subscriptions for your mobile phone. In August 2015, a Dutch start-up called The Things Industries launched a LoRa network in Amsterdam under the brand name The Things

Network. Unlike commercial providers, The things network decided upon a community

owned, crowd funded network that is freely accessible. In order to obtain city coverage the mere amount of twelve thousand euros of crowd funding was required. This entails the installation of twelve transmission stations throughout Amsterdam. In order to really use The Things Network, a chip should be embedded to the device that allows connection to the

(30)

subscriptions are rather expensive, because every embedded device needs a subscription. On top of that, the limited battery lifespan creates costs, as devices will need a check and a battery replacement every couple of months. A LoRa network may spur innovations in smart public services. It may stimulate the development of new smart services and make established services more efficient and customer friendly, due to opportunities created by cost efficiencies. The cons may include system transformation and implementation costs. This case study will offer insights on the potential of smart public policies and the relevance of a network such as LoRa.

Regarding this research, three departments in the City of Amsterdam are selected, including the Water Management Department, the Asset Management department and the Parking department. These departments are selected through a brainstorm session that was held among several civil servants with significant IoT knowledge. The next paragraph elaborates on the selection procedure during the brainstorm session. Within the selected departments, smart policies are taken into consideration since they may provide the municipality with several advantages. This research investigates which processes are obstructing the implementation of smart policies in each department. Additionally, the compelling factors in implementing the LoRa network in these departments are investigated. Furthermore, the applicability of the LoRa network can be compared across the three departments.

3.5 Data collection

The sources of evidence are twofold: a brainstorm session is organized and interviews are conducted. A combination of different sources is used in order to obtain a rather comprehensive view and to avoid bias by relying solely on one source. Each source of empirical data will be elaborated upon hereafter.

Ethnographic research

As a part of the empirical data collection, a brainstorm session was organized. The goal of the session was to uncover the most relevant public services that will be subject to a policy innovation. A total of three public services represented the use cases that are investigated in

(31)

this study. After the use cases were selected, civil servants involved in these public services were interviewed. During the interviews, the policy adjustments and the implementation of the LoRa network are proposed to the stakeholders. The response of stakeholders is central to this study since the aim is to unfold the obstacles in implementing smart public policy adjustments.

The brainstorm was organized at the municipality and a total of nine people were invited, among which seven civil servants. The invitations were distributed via the CTO of Amsterdam, in an attempt to realize a higher attendance. Five people attended the brainstorm session, three of which civil servants involved in smart policy solutions, the entrepreneur of The Things Network, and myself. Among the civil servants were the CTO of Amsterdam, a co-worker Information Management at the CIO office and a Program Manager Innovation. The entrepreneur of the Things Network and I occupied a special role in the brainstorm session. The role of The Things Network’ founder was to provide technical insights regarding the possibilities of the LoRa network. My function entailed leading the brainstorm session and uncovering which use cases feel most comfortable when implementing smart policy applications.

Since I was involved in the brainstorm session which makes me part of the data source, I can be considered an ethnographer. The concept ethnography encompasses the activity of conducting research in a social context while being part of the social context. Hence, a tension exists between ‘participating and analysing’ (Hammersley, 2006). An ethnographic stance was chosen in order to obtain the benefits of a group discussion. Conducting an ethnography allowed me as a researcher to both observe the empirical context and steer the discussion toward concrete outputs (Wolcott, 1990). There are several ways in which a researcher can behave as an ethnographer in research, depending on the level of involvement of the ethnographer in the social world that is being studied (Bryman, 2012). Regarding this research, a minimum level of involvement was chosen, since the aim was to participate in the brainstorm solely by steering the discussion towards concrete use cases.

(32)

explained the aim of the research and how it related to the brainstorm session. After that, the founder of The Things Network discussed the LoRa-network, introducing some technical implications and several opportunities. This introductory process took approximately twenty minutes, after which the brainstorm could commence. During the brainstorm, the civil servants proposed ideas that were elaborated upon immediately after that. Several ideas where brought up in the discussion, among which cycling management, parking management, real-time information on public transport, water level and water nuisance, emergency healthcare, tracking Canta invalid cars, tracking municipality-owned art, and renovation alerts for municipal assets. The brainstorming proceeded for approximately 60 minutes. The ideas that were considered most fruitful by the attendees were: bicycle parking and cycling routing; water management; and asset management. Chipping bicycles was considered interesting since it may provide a solution for bicycle registration in parking lots, besides the fact that the municipality is interested in tracking transportation by bicycle. Water management already relies on sensor technology, which is why this department was considered interesting with regards to the application of the LoRa network. Asset management, on the other hand, does not involve any sensor technology currently. However, providing bridges and quay sides with sensors can realise more timely notifications for renovation requirements. These three use cases were investigated in more detail by conducting interviews with the relevant stakeholders that are involved in each use case. Particularly, this thesis studied several operational layers of each public service in order to obtain a comprehensive view of potential obstacles regarding the implementation of the LoRa network.

Interviews

Representing the second source of evidence, empirical data was obtained through interviewing. The interviews revealed information on the three cases that are investigated: Parking, Asset Management, and Water Management. Although Canta automobiles had a good fit with this study and would have provided a decent case, it proved to be difficult to arrange interviews that would support this case. The interviews were conducted amongst civil servants with an inclination towards technological and IT expertise, civil servants with some decision making power on the particular cases, and the CTO of the City of Amsterdam. A total of twelve interviews were conducted. Generally, the aim of the interviews is to unfold obstacles for civil servants, in implementing smart public policies and supporting

(33)

technologies such as the LoRa-network. Interviews particularly focused on compelling processes, instruments and knowledge gaps in the public services, rigid or change-averse attitudes of civil servants, and public-private collaboration, which are referred to as common obstacles regarding smart cities and IT innovations (Caragliu, 2009; Nam and Pardo, 2011; Heard, 2008). Based on the hypothesis, compelling factors were expected to find their origin in path dependency, incompatibility of novel and prior policies, and support from the industry.

Civil servants

Interviews were conducted amongst civil servants within the City of Amsterdam. More specifically, civil servants were expected to have a sufficient level of technological knowledge, operating in the divisions City Districts (‘Stadsdelen’), Service Provision (‘Dienstverlening’), and Business Practice (‘Bedrijfsvoering’). Within these divisions, civil servants from the departments Parking, Asset Management, and Water Management were interviewed. The interviews among civil servants can be distinguished into two categories. The first category encompasses civil servants with some decision-making power, who carry responsibility for specific programs, assets or manage a department. These interviews indicate how decisions are typically made and which parties and what process implications are involved. The second category interviewees entails civil servants that either have an advisory role or a technical expert role. These interviews indicate the bottlenecks in the practical implementation of technological innovations. The aim is to uncover which factors may form an obstruction in implementing new smart public policy solutions supported by IoT tools, in this case the LoRa-network. Hence, interviews also indicate what restrains these civil servants to use new IoT tools to provide smart public services.

The municipality’s Chief Technological Officer

The Chief Technology Officer of the municipality was interviewed due to the fact that he is the most knowledgeable person within the municipality with regards to technological