Big Data for Sustainability: Is Big Data the Cure-All for 'Wicked Problems' in Policy-Making? - Carbon Emission Mitigation through Smart City Governance

Big Data for Sustainability

Is Big Data the Cure-All for ‘Wicked Problems’ in Policy – Making? - Carbon Emission Mitigation through Smart City Governance

Leiden University, the Netherlands Faculty of Public Administration

11th _{June 2018}

Author: Jana Weidinger (s1954164)

Supervisor: Dr. Sarah Giest, Dr. Maarja Beerkens Second reader: Elena Bondarouk

ABSTRACT

In climate change governance cities are increasingly relying on the implementation of so-called Smart City initiatives that use sensors and other ICT implementations to collect data to report on complex problems in policymaking and to ultimately solve these problems. This thesis will look at which conditions make these Smart City initiatives successful and uses the theoretical

framework of policy capacity as well as evidence-based policymaking for its analysis. It wants to argue that the data collected does not present a neutral reflection of the city’s reality but is infect evidence used for policymaking. The necessary implementation of this evidence requires an adequate skill-level for civil servants which goes beyond simple analytical capacity but also looks at data strategies and evaluation mechanisms within the cities.

OUTLINE

2.1. Definitions and Debates 10

2.2. Policy Capacity 13

2.3. Evidence-Based Policy Making 16

2.3.1. Neutrality 18

2.3.2. Indicators and Dashboard Technology 20

3. Concepts 23

4. Research Design 26

4.1. Key Concepts and Variables 27

4.2. Operationalization 29

4.3. Case Selection 32

4.4. Confounding Variables and Interferences 32

5. Data Collection and Case Description 35

5.1. Explanatory Variable: Policy Capacity 36

5.2. Moderating Variable: Communication and Visualization of Data 44

5.3. Outcome Variable: Mitigation of Carbon Emission 49

6. Data Analysis 52 6.1.Academic Implications 54 6.2.Limitations 57 7. Conclusion 59 8. Bibliography 62 9. Appendix 72

Appendix 1 Frequency Analysis of Civil Servants in Chicago (2018) Appendix 2 Analysis of Civil Servants in San Francisco (2015) Appendix 3 Salary Distribution in Austin (2017)

Appendix 4 Frequency Analysis of Civil Servants in Boston (2017) Appendix 5 Chicago Dashboard Energy Benchmarking

Appendix 6 Austin Dashboard Carbon Emission Levels Appendix 7 Data Management Plan

1.

INTRODUCTION

Urbanization leads to a growing percentage of the world population living in cities. According to the World Bank, by 2016 over 54, 2% of the world’s population was living in in an urban

environment. So-called ‘Mega Cities’ like Shanghai or Sao Paulo have shifted climate and sustainability governance to a city level and provide cities with more responsibilities regarding sustainability, economic development and social policies (Segbers, 2007). Cities thus need to “produce wealth and innovation but also health and sustainability. Cities are to be green and safe but also culturally vibrant” (Meijer and Bolívar, 2016, 393). This trend of urbanization however brings policy problems on its own, such as “waste management, scarcity of resources [and] air pollution” (Chourabi et al., 2012, 2289).

This development has led to so called smart cities that use Information and

Communication Technology (ICT)1 _{to address complex policy problems, such as sustainability}

and environmental, social or economic policies. A central component of these smart cities is ICT and the use of data collected mostly through sensors implemented in the city structure. Barns (2018) thus argues, “the widespread uptake of smart city strategies around the world is

provoking attention towards the governance challenges and opportunities of cities that are ‘run on information’” (p. 6).

Scarce economic resources, inadequate and deteriorating infrastructure, energy shortage and price instability drive many municipalities to develop smart city plans (Washburn and Sindhu, 2010). To ensure that cities become smarter in order to address these problems, a central component of the Smart City 2.0 is the analysis of high volumes of real-time data. However, as social scientists point out the data cannot “speak for itself” but has to be interpreted, analyzed and visualized so that it is also accessible to a non-expert audience (Wesselink et al., 2014, 341). The ability of the public-sector workforce in charge of databased policymaking, decides which policy problems make the political agenda and which variables and factors are included in the analysis. Knowledge and skill levels as well as the use of ICT implementations are therefore at the center of smart city strategies and this study (Head, 2013).

1 _{ICT is defined by the OECD as manufacturing as well as service industries “that capture, transmit and display data} and information electronically” (OECD, 2002, 81).

The use of ICT in smart cities often results in the implementation of so-called dashboards that aim to visualize and communicate data. Theorists (e.g. Head, 2008; Kitchin 2014) contest the assumed neutrality of this data. While advocates of ICT implementation in governance argue for a cost-effective and time-efficient way to speed-up bureaucratic processes, to make

governance more transparent and accountable and believe that ‘big data’ can solve society’s most pressing problems, others see the establishment of a real-life panopticon (Gabrys, 2014). This study argues that while technology in governance and policy-making is not simply good or bad, it is also not just a neutral enabling tool and a problem-solver but has to be treated with caution (Dalton and Thatcher, 2014, Leenes, 2015).

This research will conduct a small-n comparative case study, consistent of four cities located in the United States of America that have implemented smart city governance in order to mitigate their levels of carbon emission. These cities are Chicago, Austin, Boston and San Francisco. All of these cities have implemented smart city strategies with a focus on data

analysis in mitigating carbon emission levels. They vary in their level of policy capacity and two of them, Chicago and Austin have implemented so-called dashboards into their urban data platforms that visualize and communicate the collected data. The dashboard is defined by Kitchin and McArdle as “dynamic and/or interactive graphics, maps, and 3D models to display information about the performance, structure, pattern and trends of cities” (2016, 2).

The study aims to analyze the causal relationship between the outcome, carbon emission mitigation, the individual policy capacity of civil servants in smart cities and the use of

dashboard technology. Chabouri et al. (2012) have identified ICT implementation and a lack of skills and knowledge of IT and data analysis as challenges of smart city initiatives.

Independently from this human resource aspect, appropriate IT implementation poses a challenge to these initiatives in the interpretation of complex pieces of evidence (Head, 2013). Research in how these two components interact however is scarce. This study aims to fill this gap and look at how technologies such as dashboards depend on individual policy capacity of the workforce using them. The research aims to answer the research question:

RQ: “Under which circumstances is smart city governance successful in mitigating carbon

After positioning this study in the scientific research of sustainability and e-government, I will explain the contextual factors that have to be considered in smart city research. The second chapter then looks at the theoretical framework and includes current debates and definitions of smart cities. Two main theoretical strings, policy capacity and evidence-based policymaking (EBPM) serve as a basis for the concepts used to analyze the variables in the case study:

Individual policy capacity and communication and visualization of data. The fourth chapter will

then establish the methodology and the small-n most similar comparative case study. Chapter five introduces and describes the cases and contains the data collection. The data analysis in chapter six will then look at the theoretical framework with respect to the empirical analysis and will sum up the findings.

1.1SCIENTIFIC RELEVANCE

The following chapter will position the research focus in the existing research of smart cities, sustainability and e-government and will establish its scientific relevance. The unique structure of the city as governing mechanism on an increasing number of people makes it a laboratory and an innovation center for the prevention of climate change. The experimental setting of cities has thus been proven beneficial to address broad and complex policy problems on a local level (Laakso et al, 2017). Cities provide the platform for ICT innovations and the potential for economic growth through technology. Based on this, low carbon initiatives in smart cities hope to increase sustainability without decreasing quality in cities (Vassileva et al., 2016). The field of e-government connects research on sustainability and on smart cities. E-government and data analysis have so far found use in Smart Energy and Smart Grid applications. According to Beier et al. e-government is used as a monitoring and evaluation mechanism to “distributed energy generation and reduced energy dependency” (2018, 20). The literature mainly focuses on

improved energy management and efficiency as well as ICT tools for consumers to monitor their energy consumption. How and under what conditions ICT and data analysis achieve this

improved management of resources and efficiency has however not been studied extensively yet. The literature shows the importance for ICT implementation in the field of smart city innovations (Chourabi et al. 2012, Kitchin, 2014, Meijer and Bolivar, 2016). Neirotti et al. (2014) have classified hard and soft domains according to how important ICT systems are in the smart city initiative. They place sustainability initiatives within the ‘hard’ domain. Hard domains

thus benefit from “city settings in which the vision of a city that senses and acts can be the most applicable, thanks to the use of sensors, wireless technologies and software solutions to handle ‘‘big data’’” (Ibid. 27).

Smart cities use ICT to evaluate policies and to uncover new causal mechanisms in complex problems. Aichholzer et al. point out that ICT in carbon emission reduction presents a useful tool in uncovering the details in which the cities actual emission levels defer from the target levels (2013). Chatfield and Reddick (2016) show how a smart city in Japan uses e-governance techniques to “to generate insights into localized energy demand patterns at the smart community level” (p. 766). Data collected through sensors in the smart city of New York City can monitor and evaluate air quality, according to Anthopoulos (2017). Bibri and Krogstie (2018) furthermore highlight the intended evaluation of smart city strategies themselves in the policy field of sustainability. The literature shows that big data analytics and especially its volume and velocity has transformed policymaking as a whole and has enabled policy-makers to incorporate more evidence into the decision-making as well as the evaluation process in order to make policies more tailored to specific problems and to evaluate their effects. Höchtl et al. (2016) see big data analytics as a transforming power of policymaking. Different stages of policymaking use evidence collected through data analytics for the purpose of decision-making or evaluation. The authors conclude that “potential to significantly shorten the decision-making process and lead to better decisions as more valuable information can be derived from data otherwise declared as noise” (2016, 164). The authors also link the potential of data analytics in the public sector to the technological capacity and infrastructure challenges in the organization. On an organizational level, the inadequate use and interpretation can be traced back to ‘data exhaustion’ that describes data piling up at organizations and they are not equipped or willing to make use of it and analyze it for policymaking or other purposes (Giest, 20171). This highlights

the broad scientific consensus that the ability of governments and especially government employees plays an important role in the interpretation and evaluation of the data collected.

Similar to Höchtl et al., Larsson and Grönlund see technology not only as a tool but also rather as an implementation in everyday operations. They suggest a socio-technical perspective of technology in sustainability, where technology is treated “not as a mere tool or resource but rather as an important factor in relation to society, even an actor alongside humans” (2014, 139). They see technology in sustainability and e-governance as a crucial part, not only for the

implementation of a smart city project but also for information infrastructures, citizen engagement and organizational issues.

To sum up, while existing research has established a connection between skill-levels of staff and performance of smart cities as well as between ICT implementation and performance, the connection of ICT implementations and policy capacity of civil servants in smart cities present a research gap that this study wants to contribute (Beier et al., 2018).

1.2 CONTEXTUAL FACTORS

Although smart cities are a relatively new topic for public administration research and only emerged in an academic context in the late 1990s, the topic has a high-level of

interconnectedness to other disciplines, such as e-government, human resources or city

government (Meijer and Bolívar, 2016). This chapter will attempt to position the debate around smart cities and smart urbanism in the academic debate of New Public Management (NPM) and e-government.

The literature identifies two main aspects in the debate around smart city governance. First, there is the introduction of evidence that is rooted in the presumption that every aspect of policymaking can be measured with enough evidence (Head, 2014, 2015, Kitchin, 2014, Barns, 2018). The second aspect refers to the expectation that government has to run cost-efficient and effective in order to be managed successfully. Although NPM and evidence-based policymaking (EBPM) are not directly related, several aspects link them together in the academic debate about smart cites.

To describe and analyze the city, Amin and Thrift propose three metaphors that help to understand this new urbanism that they describe as an” organism […] underneath the clamour, clutter, confusion and disorder of city life was felt to lie a certain organic integrity” (2002, 8). The first metaphor is transitivity, similar to “the tradition of flânerie to read the city from its street-level intimations […] and to encounter the idea that the city [is a] lived complexity” (Ibid. 11). The second metaphor describes the rhythm and rythmanalysis, as “the rhythm of the city coordinates through which inhabitants and visitors frame and order the urban experience” (Ibid, 17). Lastly, there is “urban footprints and namings” (22), that defines the city as “localized time and temporalized place” (Ibid, 22). This includes the networks and communications within the city as well as the different communities and their interaction with each other. Even though

Amin and Thrift didn’t have smart city governance in mind when they developed these

metaphors, this form of modern urbanism stands in sharp contract with the ‘urban technocracy’ that is proposed by other theorists in the context of smart city governments.

Contrarily, Urban technocracy, the regulation of a city through information and analytics systems is described by Mattern (2013) as ‘instrumental rationality’, or by Morozov as

‘solutionism’, which defines a complex social issue that can be “disassembled into neatly defined problems that can be solved or optimized through computation” (Kitchin, 2014, 9). The

underlying assumption here is that problems can be solved with enough real-time data and information for policy-makers and governments to use. Hill describes this as “[smart city thinking] betrays a technocratic view that the city is something we might understand in detail, if only we had enough data—like an engine or a nuclear power station—and thus master it through the brute force science and engineering” (Hill, 2013, cited after Kitchin, 2014, 9). Furthermore, the increasing use of ‘science’ shows that the hope in these big data applications is that enough data can uncover causal mechanisms that would otherwise not have been visible to policy-makers (Mattern, 2013). Others also point out the accountability issue when it comes to big data applications, as

“Employing an evidence-based, algorithmic processed approach to city governance thus

seemingly ensures rational, logical, and impartial decisions. Moreover, it provides city managers with a defence against decisions that raise ethical and accountability concerns by enabling them to say, ‘It’s not me, it’s the data!’” (Kitchin, 2014, 9)

Technological solutions in this technocratic approach mark a sharp contrast to the concept of

Wicked Problems that was initially coined by Rittel and Webber in the 1970s and describes a

policy problem that has ten different characteristics. The most important in regard to evidence – based policy making are the interdependency of those problems, the absence of a definite solution and a “‘one-shot operation”, which means that “the results cannot be readily undone, and there is no opportunity to learn by trial and error” (Head, 2012, 665). Many problems that smart city governance is trying to address are parts of these so-called wicked problems, in that they are very complex and interdependent, like examples from climate-change-governance show. Kitchin also argues that these complex, structural problems are not easily solved by

technological solutions in smart city governance (Kitchin, 2014). Especially in the case of

neatly packaged way to meet these generalized challenges, thereby ensuring that future cities— whether retrofitted or new—are more sustainable and efficient than ever before” (Gabrys, 2014, 3). It is thus not surprising that she links the concept of smart urbanism and sustainability to the rhetoric of Michel Foucault’s interpretation of governance as a disciplinary power. She sees the monitoring of smart cities as “urban citizens becoming sensing nodes - or citizen sensors – within smart-city proposals” (Ibid. 7).

This collected data is used by cities as evidence for policymaking. EBPM stands in contrast to other forms of non-evidence based policymaking or political policymaking that can be characterized as “effort on the part of policy specialists to ‘speak truth to power’’’ (Howlett, 2009, 156). It has resurfaced in the light of trends and new practices, such as NPM, to ground policies in evidence rather than in norms and values and to “ground policy making in more reliable knowledge of ‘what works’ “(Sanderson, 2002, 1). This evidence is supposed to be neutral and uncovering new causal mechanisms in order to bring new insights to complex problems (Sanderson, 2006).

The basis for EBPM is the principle of rationality. Sanderson describes the use of evidence in policymaking as “[the] quest to understand and explain what works for whom in what circumstances” (Sanderson, 2002, 2). More evidence is believed to lead to better

information around policy problems and therefore to improved policy outcomes. Across OECD countries, reforms in line with the NPM approach have thus aimed at quantifying performance measures and the OECD itself has argued for “results-oriented management’ provides a new management paradigm” (Ibid. 2). Based on the increasing complexity of the world and the objects of inquiry for social science, the extent to which evidence and research can grasp this reality and provide valuable insights for policymaking has been called in question.

The macro theory of NPM connects the need for evaluation with an increased

disaggregation of institutions and of accountability. This angle helps to understand the focus on information and evaluation in smart cities. NPM is often described as the introduction of business-like management tools into the public sector, such as competition, incentives and disaggregation (Lægreid, 2014). While competition introduces “purchaser/provider separation into public structures so as to allow multiple different forms of provision to be developed” (Dunleavy et al. 2005, 470), it aims to increase quality and diversify suppliers while decreasing bureaucratic structures in the administration. Disaggregation of the public sector attempts to

break hierarchies and split organizations into smaller entities, thereby also diversifying information supplied from the organization itself but also to make management efficient and individualized. Lastly, incentives aim to reward performance and to “diffuse public service or professional ethos and moving instead toward a greater emphasis on pecuniary-based, specific performance incentives” (Ibid.). NPM furthermore wants to increase accountability from the managerial as well as the customer side and therefore offer individualized, tailored ‘products’ to citizens in the public sector. The fragmentation of governance and policies to the city level thus, according to Meijer and Bolívar, explains the managerial and organizational focus of smart city research well and includes new collaborations in governance with outside stakeholders instead of a traditional government-centric view (Meijer and Bolívar, 2016, Lombardi et al. 2012).

These new forms of governance collaboration often supply cities with new forms of evidence and often even implement ICT infrastructure themselves. The focus on ICT and data collection to acquire evidence and to monitor and document a city’s performance thus diversifies the evidence collected in order to use it for governance (Giest, 20172). In the logic of NPM, this

can increase the city’s accountability to its citizens and let them directly observe insufficient performance. It can however also lead to a principal-agent framework where the city is not able to adequately verify the quality of the evidence it is given by these other stakeholders. Lastly smart city governance is often used as a tool for scarce resources and to do ‘more with less’, this cost-efficient method of solving problems is thus closely linked to the goal of effectiveness in the rhetoric of NPM (Gailmard, 2012).

A second research angle smart city research is closely related to, is the field of

e-government. The theoretical framework of e-government is based on this quest for information and ICT systems. Zhang et al therefore argue that implementing elements of e-government changes the way governments supply information for their citizens and deliver services to the public (2014). E- Government can be defined, according to the OECD as “the use of ICTs (Information and Communications Technologies), and particularly the Internet, as a tool to achieve better government” (2013, cited after Beier, 2018, 4). Several aspects of the

e-government literature can thus be found in smart city concepts, such as the importance of ICT infrastructure for e-government diffusion and the legitimacy and acceptance of ICT innovations in the government context by its citizens. The connection between ICT and societal structures and dynamics, according to Meijer and Bolívar (2016) therefore addresses

“the question of designing – developing, facilitating, nurturing – synergies between

social structure and new technology has been at the heart of e-government studies in recent decades” (p. 394).

The fragmentation of governance as well as the perceived transformation, increased transparency and accountability that are at the heart of NPM and e-government debates can thus serve as a basis for the theoretical framework of smart cities.

As Bertot and Choi (2013) however also point out, governments’ e-government strategies and digitalized services do not always make government more effective. Instead, they can raise privacy and accountability issues and the storage und use of data for e-governance initiatives can present organizational and governing problems. Bekkers and Homburg (2007) furthermore point out the ‘myths of e-government’ and conclude that e-government technologies are not the cure-all for a better functioning and more transparent government but have several downfcure-alls. Instead they see a “significant chasm between sublime rhetoric of e-government and the muddy practice of actual e-government implementation” (Ibid, 380). The implementation if these practices into institutions is thus an important factor.

2. THEORETICAL FRAMEWORK

Meijer and Bolívar have identified three main routes in smart city research: technological, human resources and governance focus. While technological focus mainly examines the use of ICT in city governance, it is often focused on business-led initiatives and public-private partnerships. Human resource focused publications have “identified by their focus on human capital and/or human resources as the key feature of a smart city” (2016, 397). Government-focused

publications center around the transformation of the government through smart cities, which will be later discussed as a central point of debate. There is not one single definition for ‘Smart City’ but rather reoccurring concepts and components to what constitutes a ‘Smart City’ (Lombardi et al. 2012, Dameri, 2013, Chourabi et al., 2012). The definition, according to Dameri (2013), “is used to identify a large spectrum of heterogeneous solutions and city programs, involving different types of technologies and aiming to reach a very large set of different and not well-defined goals” (p. 2545). Terms like ‘wired city’ or the public as well as the private sector use ‘technocity’ interchangeably. Several authors have identified the use of ICT as the main driver and component of smart cities (Meijer and Bolívar, 2016, Neirotti et al., 2014, Dameri, 2013). Another reoccurring concept in smart city definitions is the transformation of the government. At the center of the smart city debate is the transformation of the position and the mode of governance towards its citizens. Similar to the concept of e-government, Smart City initiatives often aim to increase transparency or accountability for citizens. Meijer and Bolívar (2016) have identified four ideal-types of smart city governance “(1) government of a smart city, (2) smart decision-making, (3) smart administration and (4) smart urban collaboration” (Ibid. 398). These ideal-types present a continuity according to the degree of transformation of

government. While the first type is a mechanism of governance and the city approves and adapts the ‘smart’-aspect, the transformation level is low. In the fourth type however, ‘smart urban collaboration’, the city has created a network with actors and partnerships with its citizens. The collaboration aspect of the smart city is the most important factor of the smart city and the government has achieved a transformation to a “pro-active and open-mined government structure” (Ibid. 400). Albeit the degree of transformation of government is not a factor that determines the success of the smart city, it shows that the smart city governance concept is rooted in the assumption that governance needs a minimum transformation in order to implement

smart city features. The role of government is furthermore also discussed controversially, while more conservative approaches see the government as an institution that can simply adopt smart city elements, others see the government changing as well, as it is pursuing smart city

governance and the smart city strategy is more of an exchange than a strategy externally implemented on the city. The aspect of government transformation links smart city governance back to the debate around government legitimacy and offers a post-materialistic value of participation and inclusion. Dvir and Pasher (2004) state that smart city governance derives its legitimacy from engagement with its citizens. They conclude, “governments should provide its citizens with the enabling conditions which foster knowledge creation, knowledge exchange and innovation” (Ibid. 402).

Other than transformation of governance, definitions also include the organization and participation of its citizens in Smart Cities. Dameri (2013) defines the concept of smart city not as a “top-down” phenomenon but as a “bottom-up” one, which interacts with the strategy provided by the government (p. 2545). The ‘smart’ concept, Dameri argues, is connected to different other concepts, such as sustainability, digital or intelligent city. It is thus important to include the environment, the citizens, the technology as well as the governance, as the state or local city administration are important parts of the smart city and determine on what level it operates and governs. Meijer (2016) also identifies different key concepts related to the local participation aspect of smart cities: “local cooperative knowledge potential and the nature of the problem” (p. 73). His definition follows the ‘form follows function’ - principle and he concludes that there is no ‘one size fits all’ but that unique conditions like skills and partnerships that foster knowledge are essential for a smart city to succeed.

Other debates focus on administrative and managerial processes and distinguish between different types of smart cities. They have emphasized the evolution from smart city 1.0 to Smart City 2.0, which aims to use urban sensors and other ICT and web technology to “de- materialize and speed up bureaucratic processes and help to identify new, innovative solutions to city

management complexity, in order to improve sustainability and livability” (Chourabi et al., 2012, 2290). Legitimacy and transparency also connect the smart city debate to partnerships with non-governmental and private sector actors (Janssen and van den Hoven, 2016). Outside, often controversial, partnerships with the private sector as well as other non-governmental actors aim to make bureaucratic processes more efficient with ICT implementations or alter the supply of

knowledge in the public sphere and the evidence the government has available to make informed decisions (Kitchin, 2014). Other theorists, such as Peled (2009) also see the monopoly of the government as a sole supplier of knowledge challenged here in some way bureaucratic systems cannot cope and over-exhaust in the decision-making process. Smart city and smart urban projects also often include partnerships between an outside contractor and the government, resulting in multiple public-private partnerships.

Other definitions indicate a forward-looking perspective that conceptualizes smart city as a governing mechanism that has incorporated “awareness, flexibility, transformability, synergy, individuality, self-decisiveness, and strategic behavior (Chourabi et al., 2012, 2290).

While these definitions also seem to originate from a ‘self-fulfilling prophecy’ when a smart city is “striving to make itself ‘smarter’” (Ibid.), the components included in this definition are difficult to measure empirically and direct the concept in a normative direction of what ought to be. The multitude of definitions does thus tie certain concepts and aspects to smart city

governance, but these definitions only vaguely define the phenomenon and make a delineation of concepts difficult.

To conclude, smart city developments and strategies seem to be the ‘necessary evil’ to cope with developments of urbanization and transformation that makes cities “enormous and complex congregations of people [that] inevitably tend to become messy and disordered places” (Chourabi et al., 2012). The concept shows a high level of interconnectedness with aspects like sustainability or participation. The use of ICT is a central component of smart cities and its importance has been. This paper will follow the definition by Harrison et al. (2010) and defines smart city as “connecting the physical infrastructure, the IT infrastructure, the social

infrastructure, and the business infrastructure to leverage the collective intelligence of the city” (cited after Chourabi et al., 2012, 2290).

The different angles of research and definitions present a complex and diverse set of factors that influence smart cities and complement their performance. The next chapter will therefore elaborate on policy capacity and EBPM to build a theoretical framework for the analysis.

2.2 POLICY CAPACITY

The literature reviewed so far established that technology and the use of data and ICT

applications stand at the center of smart city governance. To enable these applications and to analyze their output, the public-sector workforce as well as the organization as a whole have to possess certain qualifications in order to make these ‘smart’ applications beneficial for

policymaking. Policy capacity collects these qualifications and prerequisites under an umbrella of different theory streams and angles, which reach from the individual civil servant to the organization’s potential to collect and incorporate evaluations and process data as a whole. Newman et al. (2016) argue that policy capacity is often mistakenly treated as “an end result rather than as a means to an end” (p.160). The literature shows three dominant angles on how to conceptualize policy capacity:

On the government level, Newman et al. (2016) offer two main ways to conceptualize policy capacity in EBPM: The first is “government’s autonomy and capability to create its own desired policy objectives [and] to generate some course of action to address those objectives”, (Ibid. 160). This refers to governments’ ability to set the policy agenda, frame policy problems and decision-making and in general control its own policy objectives. Daugbjerg and Halpin (2010) furthermore state that policy capacity is important for governments to develop a strategy for policymaking and to “make intelligent choices” (p. 142). They do not only see the choice of the right policy instruments as important but also the involvement of external stakeholders, as governments need to possess the ability to fit political instruments into a “particular political and economic context that are consistent with norms and expectations among the target group” (Ibid., 142). These external stakeholders and target groups are especially important in the context of environmental performance and sustainability, as they often require partnerships with the corporate sector, according to the authors.

A prerequisite for policy capacity is thus according to Daugbjerg and Halpin (2010) as well as Newman et al. (2016) besides state or government capacity is associative capacity. They characterize state capacity as the provision of skilled staff, resources as well as “political

willingness to commit itself to deploy these capacities actively and engage in industry

development” (Daugbjerg and Halpin., 2010, 143). Associative capacity on the other hand refers to other non-state actors and interest groups that can influence the policy problem and decision-making process. The authors thus conclude that policy capacity is an interdependent concept

between different interest groups in the policymaking process. Government thus needs to possess the necessary skills represented on various levels of the administration to incorporate knowledge in the bureaucratic apparatus and to form informed policy-decisions. Howlett has criticized this managerial focus of policy capacity and has coined the term “political analytical capacity” that focuses more on the process of evidence and information than on the managerial aspect of leadership. It describes the ability of civil servants to “produce and disseminate policy-relevant information and advice” (Ibid. 160). Howlett (2009) links this capacity back to the demand and supply side of EBPM and highlights the importance to acquire high-quality evidence as well as the ability of sufficient policy analysis. He stresses that

“organizations both inside and outside of governments require a level of human, financial, network and knowledge resources enabling them to perform the tasks

associated with managing and implementing an evidence-based policy process” (p. 157). He sees the difference between ‘political or policy capacity’ and ‘political analytical capacity’ as moving beyond strictly administrative capacity of governments to “knowledge acquisition and utilization in policy processes” (Ibid. 162). He contrasts the highly skilled and costly workforce to achieve this task with the organization’s ability to process data and knowledge, as both of these components can achieve the goal of a high level of ‘political analytical capacity’ (Howlett 2009, Howlett, 2015). Lavertu (2016) points out that the lack of qualifications in general that civil servants have can minimize the effect big data analytics and evidence-based policy making can have on policymaking because they lack the skills of analyzing and interpreting these data collections.

The second conceptualization of policy capacity refers to “technical capacities within the formulation stages of the policy cycle” (Ibid, 160).Technological capacity of governments describes the flow of information through the organization and refer to the individual level of the public-sector employee. This can reflect on the skills and the training of civil servants but also on the “quality of infrastructure and technology that helps information flow vertically between bureaucratic levels and horizontally between administrative units” (Newman et al., 2016, 160).

Wu et al (2015) offer a conceptual framework that analyzes policy capacity primarily as a primarily managerial problem. They have established a three-level framework with skills and resources according to which policy capacity can be found in the public sector, the levels included in the framework are individual, organizational and systemic. While the model aims to

go beyond the government-level in its analysis of political capacity, it does not only look at skills of the workforce but it “defines policy capacity as what results from the combinations of skills and resources at each level” (Wu et al., 2015, 167). The authors have identified three different dimensions of political capacity: analytical, organizational and political capacity. The analytical dimension looks at the skill level of civil servants, and it “help[s] to ensure policy actions are technically sound in the sense they can contribute to attainment of the policy goals if carried out” (Ibid., 168). Based on Howlett’s assumption that governments need capable individuals that have the “technical and scientific knowledge and analytical techniques” to make government work in a cost-effective and efficient way (Ibid.). The different components also affect each other as Pattyn and Brans (2015) suggest when they see a connection between the ability for evaluation of government strategies and policies and analytical capacity of civil servants. Furthermore, institutions must possess the technical and organizational skills to collect, store and analyze data for the purpose of decision-making and evaluation. The term “technical capabilities” thus refers to the skill-set and ability of individual civil servants to process information and to use evidence according to the objective of the government in policymaking (Newman et al., 2016, Wu et al. 2015, Tiernan, 2015).

Howlett and Ramesh (2014) as well as Tiernan (2015) have however criticized that the primarily managerial focus still leaves issues of accountability and public leadership unsolved. Furthermore, the focus on the managerial of policy capacity does not emphasize “the critical importance of relationships” (Tiernan, 2015, 2010).

Individual political capacity, according to Newman et al. is an important factor in how the policymaking process makes use of evidence. When the workforce is not equipped with a sufficient level of policy capacity, it may not be able to interpret and analyze the evidence collected in order to use it for policymaking. Furthermore, policy-makers might, as Höchtl et al. (2016) note, not be able to distinguish between important and irrelevant information, and thus not address the complex problem of climate change in an appropriate way.

Researchers have linked a post-secondary education in the public workforce as well as additional working experience in the non-profit sector to a higher capacity of EBPM in the policy-process. To conclude, while the organization as a whole has to find an adequate way to harvest, analyze and incorporate data into their decision-making and policymaking process, the

individual employee or civil servant plays a crucial role and individual qualifications can thus determine the success of data-driven policymaking.

2.3 EVIDENCE-BASED POLICYMAKING

The previous chapter has established that policy capacity reaches from the government’s ability to set the agenda in policy – making to the involvement and management of different interest groups and the incorporation of evidence in the policymaking process. The contextual

framework by Wu et al. (2015) links individual policy capacity to managerial components of leadership, analytical capacity as well as political capacity of civil servants in order for

organizations to achieve a high level of policy capacity in their decision – making and analysis process. Head (2013) furthermore argues that the demand for efficiency in NPM reforms, an increase in evidence provided from outside public bureaucracy, as well as the promotion of “greater contestability in provision of policy advice” (p. 401) has increased the demand for political capacity of governments and civil servants. The following chapter will look at the evidence used in these decisions and will further elaborate on the concept of EBPM.

EPBM marks a sharp contrast to ‘regular’ policy – making as it bases policy on facts and rationality in contrast to values and norms. The concept of EBPM first emerged in light of

health-related policies and the quest for science and reliable data to base decisions on. Head links this development to the early stages of policymaking in the 19th _{and 20}th _{century and the}

connection between knowledge and power, more evidence thus meant the creation of better programs in a lot of Western societies (Head, 2010). According to Head (2010) and Bielak et al. (2008), there is a demand and a supply side, which have different interests in the ‘evidence’, consistent of governments and legislative bodies on one side and a variety of outside actors like research institutions on the other. The diversity of research available from different institutions and requested from different levels of government thus also creates competition in certain policy areas (Ibid). This competition is significant in research on climate change, as multiple different actors and research institutes on different levels of governance are producing evidence, which can be contradicting (Giest, 20172).

Quality and reliability of evidence used by governments are crucial for their accountability in policy – making. According to Sanderson (2002) and Head (2015)

argues that governments see in this scientific evidence “an intrinsic authority” (p. 401) that does not take into account the specific context of the public sector. He thus argues that while evidence is used to evaluate policies, the quality of evidence itself is rarely evaluated. Sanderson (2002) also sees a government focus towards the fulfilment of the tasks rather than a critical evaluation of the evidence utilized.

Smart cities offer a new volume, velocity and flexibility of data as machine learning technology is capable of collecting real-time data that is combinable in different datasets and offers a wide range of data to analyze. In smart cities, data is collected through sensors

throughout the city. Kitchin (2014) classifies the data collected by big data initiatives, in three categorizations: “directed, automated and volunteered” (p.4). Especially automated data is controversial evidence as it is collected automatically from websites and other devices and thus used without the explicit consent of its ‘owner’. This causes a debate about smart cities and privacy rights as well as transparency of governments’ policymaking practices.

Data platforms and the collection of data in smart city governance aims to make cities independent from other data sources and to create own, individualized data to unique urban problems (Giest, 20172). The use of ICT and technology thus do not only affect the quality of

knowledge itself but also the volume and the type of information. There is more and more information available on detailed steps of the individual citizen, but not all of that data is useful to the cause or the policy-problem it is trying to improve or solve. It needs to be interpreted and analyzed, which requires “as the capacity and skill to handle, understand and utilize it” (Giest, 20171, 369).

While in the age of NPM evidence is mostly used for evaluating policy programs and public administration initiatives in a business-like manner, big data allows to create real time data analysis from people’s everyday life to patterns of their energy use or how likely they are to commit a crime. EBPM creates a challenge for policy-makers to translate results into workable and implementable solutions. Head et al. distinguish in the use of EBPM between “instrumental, conceptual, and political uses” (Newman et al., 2017, 158). Political use is the most important adoption in EBPM and it is characterized by the use of evidence for the legitimation of political decisions. Kay (2011) also argues that while political framing of evidence is an “inevitable indeterminacy of the policy world” (p. 238) EBPM should incorporate different forms of rationality to reflect on these different kinds of use of evidence. This form of use is also often

referred to as ‘evidence-informed’ or ‘evidence-aware’ decision-making. Head refers here to the basic difference between academics and policy – makers that is often pointed out in the debate about EBPM, as “according to one school of thought, the majority of academic research is simply not relevant to policy-makers, because professional practice and theory are not always aligned” (Ibid. 159).

2.3.1 COMPLEXITY AND NEUTRALITY OF EVIDENCE

Smart cities collect data from sensors and application around the city as well as data on transportation, environmental and energy or communal data. City’s often frame the data as neutral evidence that depicts the city’s reality as if experienced first-hand. This links data back to the three metaphors of ‘New Urbanism’ that were introduced in the beginning in the contextual framework and how the cities interconnectedness, its transitivity and its rhythm is being

portrayed through sensors and the collection of data (Amin and Thrift, 2002. Kitchin, 2015). Smart city governance usually targets complex societal and environmental problems. The ‘solutionism’ that is represented with ICT and urban data platforms uses indicators and benchmarking systems to monitor and eventually improve those problems (Innes and Booher, 2000). This reduction of performance to an indicator or a benchmarking system contrasts the complexity of the collected data. This oversimplification of problems the smart city strategy tries to address is represented in the evidence that is used to solve these complex problems. The characteristics of data analytics and machine learning produce high volumes of real-time data that in its ‘raw’ form include many variables and thus increase the complexity of the evidence available (Head, 2008, Edelenbos et al., 2017).

The data collected through smart city infrastructure is often mistakenly characterized as a neutral reflection of reality. Another component at the center of the theoretical framework of EBPM is the perceived neutrality of data. It is, as Kitchin (2014) points out, often characterized as “being benign and lacking in political ideology”, this “simple data” is supposed to be neutral as “sensors and cameras have no politics or agenda (p. 8). Head (2008) argues that while EBPM wants to establish causal relationship similar to scientific cause and effect, the complex decision-making mechanisms of policydecision-making cannot just be observed through one lens and decisions do not only rest on one base of evidence but several other bases and components. He proposes three lenses of evidence and knowledge, which can be used to analyze the type of information that is

used in big data initiatives: Political knowledge, Scientific or research-based knowledge and practical implementation of knowledge. While the first lens refers to political tactics and beliefs, it is described by Head (2008) as “data-proof or knowledge-proof” (p. 4). Scientific or research-based knowledge comes closer to high-quality evidence, as there is a possible scientific debate around this kind of knowledge. Practical implementation knowledge, includes managerial

decisions or unpublished, undervalued knowledge. He thus argues that there is no neutral form of evidence and that the term ‘evidence’ cannot be used interchangeably with the term ‘data’. While big data analytics gives the promise of efficiency, effectiveness and new insights, Head also points out that the messenger determines the lens of how evidence is presented. Thus, while data collected in smart city governance might be collected in an attempt to uncover new causal mechanisms to solve complex problems in policymaking, the data cannot speak for itself but has to be interpreted and has to be ‘observed through a lens’.

Moreover, theorists point out that politics cannot be based entirely on evidence but instead is also influenced by norms and values. Head (2008) identifies this as one of three challenges for EBPM. Secondly, there is multiple types of relevant evidence and its relevance depends on the messenger as well as the recipient, its meaning thus refers back to the three lenses of EBPM. Lastly, the complex nature of policymaking and decision-making cannot be reflected exclusively in the rational nature of evidence but consists of different networks and outside partnerships.

Additionally, since data analysis presents vast amounts of unstructured data in its ‘raw form’, the interpretation and analysis of this evidence by data analysts is almost inevitable and functions as a pre-selection and interpretation of different variables. Wesselink et al. (2014) support this claim when they argue that data “cannot speak for itself” (p. 342). This links EBPM to the concept of policy capacity and the civil servants’ ability to interpret the data. It has been established that policy capacity plays an important role in how governments are able to use the evidence that is conducted for policymaking. Furthermore, the neutrality that data is often said to speak with has to be seen in light of the three different lenses of knowledge that Head has

identified.

Lastly, there is a scientific consensus that technology is not a neutral tool for analysis. When algorithms or machine-learning technologies are used for data analysis, Kitchin and Doyle see the rhetoric as “neutral and non-ideological in their formulation and operation, grounded in

scientific objectivity” (Kitchin, 2014, 8). Leenes however argues that technology cannot be grouped within ‘good’ or ‘bad’ categories, but that it “always embeds certain scripts and enables certain behaviour and inhibits other behaviour, thereby normatively affecting those using it” (Leenes, 2015, 144). Gitelman and Jackson also break with the myth of ‘neutral technology and data’, when they say “raw data is an oxymoron, data are always already ‘cooked’ and never entirely ‘raw’” (Gitelman and Jackson, 2013, cited in Kitchin, 2014, 8). The way and the variables according to which data is processed is thus influencing its message profoundly. The data and what Kitchin describes as “data exhaust”, a by-product of data analysis, can result in biases or misinterpretations (Ibid).

2.3.2 DATA PORTALS AND DASHBOARD TECHNOLOGY

Smart cities use several technologies to make decisions based on the data that they are collecting. One technique to visualize and communicate data to citizens and other stakeholders such as non-state actors, companies and other government departments, are dashboards. Dashboards visualize measurements and assessments of the city according to indicators and other control tools, such as security or transport systems and process the data collected through smart city governance. The can be defined as

“visual analytics that are dynamic, interactive, inter- linked, and use traditional graphs, charts and maps, as well as more innovative visual presentations such as gauges, 3D models and augmented landscape images made possible by advanced computer graphics” (Kitchin et al. 2015, 94).

The indicators used for dashboards are aiming to “inform the public and assist analysts and decision-makers, who will consult the reports, learning from whatever information is relevant to the problem they happen to face” (Innes and Booher, 2000, 175). These indicators then represent and communicate policy issues. The problems arising here are the insufficient interpretation and knowledge of the audience that uses these indicators for decision-making (Lavertu, 2016). The literature distinguishes between different kinds of indicators that city governance uses. More complex indicators are compositions of different indicators and thus further reduce complex policy problems to the measurement of an indicator. When the indicator is constructed and calculated by an outside actor, the measurements and underlying algorithms are often not

transparent and therefore not reliable for analysis. This does not only reduce performance in a certain urban aspect from a complex policy problem to a number but also creates an information asymmetry between the constructor of the indicator, especially if it is an outside actor, and the civil servant or citizen using the indicator to evaluate a certain measurement (Lavertu, 2016).

Reducing a phenomenon or problem to an indicator can be problematic as it may not measure the intended effects or may not acknowledge the multilayered nature of the problem. Lavertu for instance illustrates, how the oversimplified measurement for school quality in the ‘No Child left behind’-act in 2001, can lead to policies not addressing the quality of education but simply ‘punishing’ schools in less prosperous neighborhoods (Lavertu, 2016). Innes and Booher (2000) stress that the development and the evidence the indicators are produced on are crucial for their reliability and their interpretation. Dashboards furthermore often connect these indicators to certain locations and districts and break down the performance of a district or maintenance task to one single number.

Ultimately, while dashboards and other urban data platforms provide a form of visualization of its data that is accessible for non-data experts as well and thus helps to

communicate the results, they are only a useful tool f they are aligned with overall policy goal. Sokhn et al. identify a “clear lack of operational tools that public administrations can use

everyday for monitoring and driving their decisions” (2015, 342). Dashboards can thus be linked back to the simplified visualization and communication of complex data that has been processed according to certain indicators and thus already has been interpreted and selected.

The underlying data and thus a level of transparency and accountability on how the data was processed is not always available. Thus, while the dashboard aims to reflect the city and its reality in these numbers, the city government pre-selected and analyzed the indicators and has already made an interpretation and selection or chosen a ‘lens of evidence’, according to which evidence can be presented, as mentioned above, for the user to see. Kitchin et al. furthermore argue that the dashboard

“decontextualizes a city from history, its political economy, the wider set of social,

economic and environmental relations that frame its development, and its hinterland and wider interconnections and interdependencies that stretches out over space and time” (2014, 19).

Due to the over-simplification of a complex problem into a number or an indicator, the cities specific characteristics or historic specifics are ignored. He furthermore argues that these indicators are not reflective of urban developments and processes, as they “enables longitudinal analysis but presumes that patterns and trends can be redirected through quick acting policy levers, often ignoring the temporal register of urban processes” (Ibid). The comparison between cities in their performance aims to make a comparison between different cities and thus often fails to represent their characteristics, forgetting that a ‘one size fits all’ approach is not reflective of how cities operate around the world (Ibid).

To conclude, the tasks in smart city governance range from organizing and controlling ICT platforms and data collections to interpreting data and then using these interpretations for policymaking. In the geographic space of the city, data analysis becomes increasingly important, as cities concentrate the majority of the population with its challenges of governance,

sustainability and coordination. Dalton and Thatcher thus point out that “most digital information now contains some spatial component and geographers are contributing tools, maps, and

methods to the rising tide of quantification (2014). The evidence collected and used for smart city governance thus has to be treated critically, as it does not represent a reflection of the city’s reality but a mere image of components. Furthermore, the use of indicators relies on the quality of data used for their calculation and has to be seen as an oversimplification of complex societal problems. The dashboards thus used in smart city strategies to either inform the public about progress or status quo of certain projects or to help civil servants with policymaking, the evidence used is interpreted through a different lens of knowledge. The data’s objectivity and neutrality are also contested by the fact that data needs interpretation and thus cannot ‘speak for itself’ (Wesselink et al. 2014, 342). The theoretical framework around policy capacity of civil servants reflects on the different skills and organizational as well as technical preconditions the public service workforce in order for it to establish evaluation techniques and to formulate informed decisions on complex political issues. This also includes upholding partnerships with outside stakeholders and to manage different streams of knowledge and evidence in order to use them for policymaking.

3. CONCEPTS:

The thesis will use two main concepts as a basis for the empirically tested hypotheses. Derived from the theoretical framework of political capacity, the first concept will look at individual

political capacity of civil servants in the city administration. As established above, political

capacity can be defined as the capacity of governments to make use of evidence collected through multiple sources and to understand and analyze that evidence in order to make informed political decisions. On an abstract level, this means that an organization or a political institution is able to use evidence supplied by different actors and to distinguish and decide between quality and reliability but also between perspectives. This is especially relevant for complex problems like climate change governance, as there are many competing forms of evidence but also many different actors supplying it, from national research institutes to non-governmental organizations to regional organizations. Head points out that there is also an increasing unreliability in this multitude of evidence presented that makes it difficult for policy-makers to judge and select evidence, which becomes even more relevant in times of collection and interpretation of a large volume of real-time data. (Head, 2015). Thus, “In complex programs with multiple objectives, […] where the clients are subjected to many sources of influence beyond the scope of the program, the challenges of accurate understanding are compounded” (Ibid. 286).

While the cities’ supply of knowledge is similar due to the location within the same country, the same political system as well as culturally similar debates, a myriad of differences emerges. These differences include the ways in which underlying values can change political discussions and what evidence is presented to legitimize political positions and decisions, the civil service workforce of these cities varies in their level of skills, their education background as well as their organizational structure. Furthermore, the city level allows this research to look directly at the level ‘above the individual citizen’ and see how evidence collected in big data analytics from these cities is used for policymaking. The concept used for the first hypothesis is therefore individual political capacity of civil servants.

Wu et al. have established a three-level framework with skills and resources according to which policy capacity can be found in the public sector, the levels included in the framework are individual, organizational and systemic.

This study will look at the ‘individual level’ of smart cities and ‘individual analytical capacity’, the ability of civil servants to apply and access technical and scientific knowledge that

is collected by the smart city infrastructure and is then used to lower carbon emission levels in the city. The authors have identified three different dimensions of political capacity: analytical, organizational and political capacity. The analytical dimension looks at the skill level of civil servants, and it “help[s] to ensure policy actions are technically sound in the sense they can contribute to attainment of the policy goals if carried out” (Ibid., 168). Based on Howlett’s assumption that governments need capable individuals that have the analytical and technical skill-set to analyze evidence. The use of ICT and dashboard technology furthermore requires an understanding of data visualization in order to derive adequate conclusions for policymaking from these benchmarks, indicators and 3D-maps. The organizational dimension looks at the organizational capacity of civil servants to practice and to promote leadership. On the individual level, leadership requires the management and facilitation of information and of skills and is an important tool for accountability of information in the public sector. In the smart city context, the term ‘smart leadership’ often empirically describes an institution or a department that

coordinates the data collection and strategy of smart city governance. Centralized data analysis departments play a crucial role in determining the outcome of the data analysis and could bridge the findings and interpretations so that a workforce with a lower skill level in data analysis could also use this evidence to form policy decisions. Since data cannot speak for itself but has to be interpreted, information asymmetry applies when data is analyzed by experts and then presented in a report or reduced to a number. A person not educated in data analysis might therefore interpret this report or number incorrectly or insufficiently (Lavertu, 2016). The last dimension, political capacity describes the ability of city government to clearly identify a policy issue as a problem and seek evaluation and communication with other stakeholders in the process. Political Capacity sees the coordination between different partners and opinions as “essential traits of successful public managers, as is the understanding of the political trade-offs necessary for an agreement among contending actors and interests” (Wu et al., 2015, 169). Another aspect of political capacity is learning from these different opinions and pieces evidence and changing or adapting policies or decisions accordingly. Dunlop furthermore sees communication between citizens and the administration apparatus as crucial and stresses the importance of two-way communication in the process. This also includes sufficient evaluation mechanisms in city administration and the collection and coordination of data in order to establish a evaluation and

communication strategy with the public (Ibid.). The first hypothesis tested for the main explanatory variable in this thesis will thus me:

H1: A high level of policy capacity will lead to civil servants successfully using evidence

collected with smart city innovations and thus mitigate carbon emission.

The second concept used for the moderating variable Z is derived from the theoretical framework of EBPM. Two of the main aspects that determine how and why data is used are the lens through which we see data and evidence and the supposedly neutral use of it to uncover causal

mechanisms in policymaking in order to solve complex problems. A physical manifestation of these two components are first of all the data analytics strategies in smart city governance that aim to collect data within the city in order to use it for measuring and evaluating the performance of the city, describe the urban environment through indicators or offer a real-time monitoring system through dashboards (Barns, 2018). While the government aims to enable users of the dashboard to observe the city in a rational manner close to its reality, the argument made behind the theory of EBPM is that the collection and the communication of data is crucial for its

interpretation and thus use in the smart city strategy. The theoretical framework will follow Brian Head here, as he sees different kinds of lenses of knowledge being used for policymaking but contradicts the understanding of knowledge and evidence as ‘neutral’.

A city dashboard thus indicates that the city has interpreted, communicated and

visualized the data it has collected as evidence and has drawn certain conclusions from the data. Furthermore, when a city uses the dashboard technology, but its workforce does not possess a sufficient level of skills to interpret the ‘raw’ data, the dashboard could decrease the negative effect of this missing analytical capacity. The dashboard could furthermore function as a more cost-efficient way, in line with evidence-based policy making to ‘achieve more with less’ in an attempt to answer austerity measures in public administration. The second hypothesis will thus be:

H2: Mitigating carbon emission in smart cities will be more likely if a successful communication

and visualization of the data through a dashboard has occurred and can therefore balance a lower level of policy capacity.

4. RESEARCH DESIGN

This chapter will establish the methodology and the case selection of the thesis. I conduct a small-n, comparative case study. This section will introduce the variables. Then the causal mechanisms and the conceptual model will be discussed as well as the limitations and possible confounding variables of a most similar case design. Lastly, this chapter will introduce the operationalization of the variables and the four different cases for this research.

Even though generalizability is limited in Small – N studies and due to the small number of cases it cannot rely on the logic of large – N studies to avoid errors. However, it offers a deeper and more detailed understanding of the cases presented. A complementary within – case analysis is used to further avoid measuring errors. The study has the goal to uncover the causal mechanisms that influence the successful implementation of sustainability governance through smart cities. (Toshkov, 2916, 261). The positivist approach of the design refers to empirically based data as well as hypotheses testing. The testing of the hypotheses will follow Van Evera’s definition of a hypothesis as “a conjectured relationship between two phenomena” (Van Evera, 1997, 9). The hypotheses tested will be formulated as causal hypotheses: “I surmise that A causes B.” (Ibid). The comparative case study will be conducted in a deductive logic, as it “start[s] with theory, get data to evaluate hypotheses” (Toshkov, 2016, 260).

While theory testing is limited with small-N case studies, deductive research, according to Toshkov it is still possible. The design will thus focus on the Most Similar Systems Design I, where cases show variation on the main explanatory variables, “while the values of all other possibly relevant variables remain constant across the selected cases” (Ibid, 262). The outcomes of the research at this state is not known yet to the researcher and is therefore irrelevant in this process. The goal of the research is to uncover causal mechanisms between the conditions of smart city governance and the outcome, the effects of smart city governance for environmental sustainability. The unit of analysis for this study will be the individual level of the public service workforce in the case studies.

4.1. VARIABLES AND CONCEPTUAL MODEL

This thesis aims to answer the research question: Under which conditions is smart city

governance successful in mitigating CO2 emission? The outcome variable will be the mitigation of carbon emission. The study will test one moderating and one explanatory variable: The individual policy capacity to integrate evidence into policies and the communication and

visualization of data collected through dashboard technology. The conceptual model proposed is the following:

Figure 1. Conceptual Model

The independent variable is based on the theoretical concept of individual policy capacity. The theoretical framework in the previous chapter has established the importance of individual policy capacity of the workforce. The qualification of the workforce is an important concept when we analyze an organization’s political capacity, however with respect to the interpretations of data through a dashboard, the individual political capacity of civil servants is an important factor in smart city governance that this analysis will focus on. It is expected that a high level of

individual policy capacity has a positive effect on the performance of a smart city and thus will be successful in mitigating carbon emission levels.

Derived from the conceptual framework by Wu et al. (2015), the analytical dimension looks at the skill level of civil servants. This is an important factor in smart city governance. The use of ICT and dashboard technology especially requires an understanding of data visualization in order to derive adequate conclusions for policymaking from these benchmarks, indicators and 3D-maps. EXPLANATORY VARIABLE Individual Policy Capacity OUTCOME VARIABLE: Mitigation of Carbon Emission MODERATING VARIABLE: Communication and Visualization of Data