Urban transitions: scaling complex cities down to human size

Journal of Cleaner Production 112 (2016) 4273e4282

Contents lists available at ScienceDirect

Journal of Cleaner Production journal homepage: www.elsevier.com/locate/jclepro

Urban transitions: scaling complex cities down to human size Frank Schiller University of Surrey, Sociology, ERIE, 37BC02, Guildford, Surrey GU2 7XH, United Kingdom

a r t i c l e i n f o

a b s t r a c t

Article history: Received 9 December 2014 Received in revised form 4 August 2015 Accepted 10 August 2015 Available online 20 August 2015

Complexity science has become prominent in studying cities as concepts like “smart city” and “big data” indicate. In particular network analysis has allowed to studying various aspects of cities in new ways. As such these analyses are often disconnected and subsequent business models often remain disembedded. However, complexity science can also compare various patterns extending over different scales (scaling) if they belong to the same entity (allometry). Such relationships pertain to cities too suggesting that buildings, infrastructure and traffic amongst other things develop interdependently and, that across specific city systems scaling phenomena can be compared according to cities' population size. The article argues that while many scaling phenomena of physical and social networks can indeed inform urban transition research the proposed central role of cities' population size is highly ambivalent. This is particularly true for economic indicators like GDP, which do not reflect the need for sustainability. Still, network and scaling analysis of the built environment can contribute to transition theory if explanatory social mechanisms relating human behaviours and social institutions to existing scaling phenomena are provided. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Urban transition Infrastructure Retrofit Mobility Smart city Networks

1. Introduction We like to think of cities as unique places, which we may associate with night life (Bangkok), coffee (Vienna), or art (New York). Yet cities manifest remarkably universal features that render them quantitatively comparable. The spatial sciences have come a long way to arrive at such methodologies for spatial analysis (Batty, 2012). In this article we will outline the analysis of scaling phenomena as a means to progress transition theory with respect to cities and urbanisation. Scaling refers to regularities across different hierarchical levels of the same entity. Typically, indicators of city development are treated independently: indicator X measures x and indicator Y measures y. Both establish causal relationships in statistical terms and both are considered to be independent of each other (Economist Intelligence Unit, 2009). This is for instance expressed in indicators that measure quantities per capita e.g. the number of schools per 1000 habitants or the unemployment rate of a city. The implicit assumption of per capita indicators is that an average increase of a given characteristic is linear proportional to the increase in population size. However, scaling stands for non-linear relationships: the per

E-mail address: [email protected]. http://dx.doi.org/10.1016/j.jclepro.2015.08.030 0959-6526/© 2015 Elsevier Ltd. All rights reserved.

capita measurement systematically increases or decreases faster or slower than population size (Bettencourt, 2013). Scaling analyses have shown a variety of such statistical relationships. Ultimately, knowledge of such regularities may lead to better management of cities with some writers proposing even predictive theories for the social realm with an air of selfenforcement (Anderson, 2008). Yet, the research agenda has shortcomings: so far these relationships have only been shown for some phenomena while ignoring others. For example, a key research gap relates to diseconomies of cities and the quest for sustainability. In order to move beyond these limitations the article will address end-use technologies that closely relate to final energy consumption to present a more complete picture of cities' future. Although scaling patterns are being analysed structurally in great detail, the social processes bringing these patterns about are not well understood. This is particularly true for those phenomena that are reproduced by conglomerations of interdependent social institutions (Conte et al., 2012). Perhaps surprisingly, such analysis can have immediate influence as for instance the applied concepts of smart cities and big data show. Complexity science has rarely provided theory explaining social self-organisation (for an exception see Pumain, 2006) but it has influenced evolutionary theorising in several disciplines working towards explanatory theories, e.g. in economics, geography and transition management. This article argues for explanations since we need socially robust

4274

F. Schiller / Journal of Cleaner Production 112 (2016) 4273e4282

theorising to manage the transition of cities (Nowotny, 2005). In this regard, transition theory offers a theoretical framework capable of integrating the methods of complexity science and bringing it in line with democratic decision-making. Particularly, the article proposes to elucidate social mechanisms for transition management. Identifying and specifying social mechanisms that can be associated with sustainable or unsustainable pathways promises the added value of envisioning possible and implementing concrete futures with stakeholders. The next section will introduce the analysis of urban scaling phenomena and summarise the key insights of this research strand. The discussion will particularly focus on the work of Bettencourt who has e as an exception to the general observation e proposed a theory of urban scaling. He suggests that cities show similar growth properties as found in nature. Thereafter the article will look in more detail at diseconomies of cities already analysed by sustainability and complexity science. To bring complexity science closer to application the following section will make an argument for developing complexity scientific questions within transition theory. The article argues that complexity science can help to establish the built environment as a fourth field of transition theory but that it is necessary to develop mechanistic explanations (Section 4). This is due to ever-present immergence pressures in the social realm. A discussion of the findings and a conclusion follow. 2. Cities on a scale In the past decades or so network analysis has made tremendous progress in analysing network typology and evolution. Scientific progress has not been restricted to the domain of nature but has been applied to society and its artefacts too (Andersson et al., 2006). Complexity science has studied the properties of networks across all domains not least in cities. By contrast, most network research in the social sciences has focused on institutionalised networks (Powell, 1990). Instrumental to progress has been mathematical graph theory (Caldarelli and Vespignani, 2007). Complexity science's analyses have primarily focused on the structure and secondly on the evolution of networks. Key to analysis is the degree distribution that is the number of links of individual nodes to other nodes. Some networks show a normaldistribution while others have a logarithmic distribution: The latter's degree distribution shows an invariant relationship between one factor and a second variable signified by a power law (see Appendix). These networks are called scale-free since the same structure prevails across all hierarchical levels. Their median reveals little about their behaviour. Instead nodes above the median influence the system disproportionally. There are two more particularities associated with scale-free networks: they may scale in different directions according to different power-laws, a phenomenon that is called self-affinity, whereas networks are self-similar if their scaling pattern follows the same power law in all directions. In natural systems scaling relationships allow prediction of how changes of one variable will impact on a second in the network's evolution. In the social sciences such correlation-derived prediction may amount to a natural fallacy. Nevertheless, scaling has been observed in social artefacts and social structures too. From a complexity science perspective the independent examination of urban indicators is considered inadequate for analysing and comparing cities since it ignores for instance emergent agglomeration effects. It has been shown that characteristics such as economic growth and mobility vary systematically with population size (Bettencourt et al., 2007; Batty, 2008). This concerns the changes of cities' attributes relative to their size. It is a remarkable feature that increases of population size show the same statistical relationships for a range of phenomena, across large sets of cities.

These relationships follow power-laws as various studies show: physical infrastructure studied so far grows less than proportionally €mmer et al., 2006; Jiang and with city size, e.g. road networks (La Claramunt, 2004; Samaniego and Moses, 2008), subways or train networks (Roth et al., 2012; Louf et al., 2014). This might also apply to other systems of provision (Kühnert et al., 2006). These are instances of sub-linear scaling and the power in this case is smaller than 1 standing for negative allometry. Other factors scale more than proportionally with population size such as innovations, GDP and income (Bettencourt et al., 2007, 2010; Bettencourt and West, 2010; Lobo et al., 2013). In contrast to the previous scaling behaviour the power is greater than 1. Thus, they scale super-linearly indicating positive allometry.1 These long-term relationships are not necessarily static but may change over time. In an analysis of French cities Pumain et al. (2006) show for instance how the scaling relationship of various professions changed over time also falling below 1. In conjunction with in- or decreasing returns explanations (cf. Arthur, 1994) the group of analytical relationships scaling superand sub-linearly provide explanations for a variety of phenomena in cities. Increasing returns have been applied to the dynamics and organisation of the socio-economic networks not least in cities (e.g. Louf et al., 2013). The main characteristics are: socio-economic activities become faster and lead to higher productivity, they also increasingly diversify while they become more interdependent (Bettencourt and West, 2010). The super-linear scaling of GDP and innovations relate to the diversification of the socio-economic processes with cumulative effects over time (Bettencourt et al., 2014), i.e. economics of scale and scope and agglomeration effects emerge (Pumain et al., 2006). However, we also find decreasing returns for other phenomena such as diseconomies from high land prices, congestion of infrastructure, increasing emissions or higher infection rates. The study of increasing and decreasing returns to scale has been pursued by several schools of evolutionary theorising in different disciplines, including, amongst others, evolutionary economics (Nelson and Winter, 1982), new economic geography (Krugman, 1991), evolutionary economic geography (Boschma and Lambooy, 1999), ecological economics (van den Bergh, 2007) and evolutionary transition theory (Foxon, 2011). Within their disciplinary contexts these theories show how increasing and decreasing returns explain economic and urban growth and explain it. Bettencourt (2013) has consolidated much research in urban complexity science and radicalised some of the aforementioned evolutionary theories by proposing an endogenous theory of urban growth and shrinking: cities emerge as diseconomies from transport and social interaction are overcome by social and physical networks transporting goods, people and information. These are more efficient than direct, unstructured paths enabling increasing returns in socio-economic activity (super-linear scaling) and exploitation of the economies of scale in urban infrastructure (sublinear scaling but faster than city area). Accordingly, urban form and social networks feed-back on each other and we find geographic and social characteristics of cities aligned in the theory. Furthermore, the theory proposes a threshold where cities are most productive. Beyond this threshold social costs overcome benefits and cities decline, e.g. their GDP may shrink and the maintenance of infrastructure may become increasingly difficult. Despite the theoretical openness to diseconomies of urbanisation complexity science has scarcely engaged in the study of such diseconomies or thresholds.

1 Allometry designates the relative growth of a part in relation to the whole entity. It was originally developed for the taxonomy of biological organisms.

F. Schiller / Journal of Cleaner Production 112 (2016) 4273e4282

Nevertheless, network analysis provides an analytical tool to compare cities along some specific functions. Yet only if a logarithmic distribution (scaling) has been proven to exist across the whole network or a coherent sample thereof (cf. Cristelli et al., 2012), can we assume that scaling actually exists, which then constitutes the basis for a theory of allometry (see Appendix). Bettencourt hopes that “cities may actually be quite simple as their average properties may be set by just a few key parameters” (Bettencourt, 2013, 1438). The added-value of this reductionism is seen in establishing base-lines and null models (Bettencourt et al., 2010). One step in this direction towards applicable indicators is the Scale-Adjusted Metropolitan Indicator (SAMI) (Bettencourt et al., 2010; Lobo et al., 2013). A SAMI allows the ranking of cities along a particular functional characteristic. These functional relationships are persistent even as cities gain or lose population. The functional performance is relative to the base-line for a city's size, which is defined by the scaling law. Based on urban allometry this new indicator separates dynamics of different features at different scales to provide measures for local settings. Bettencourt et al. observed differences between phenomena that occur in small numbers and those that occur in large numbers. Whereas small quantities such as murders and patents have relatively large deviations of as much as 30%, economic properties have smaller deviations with variances below 10%. The SAMI can be applied to any urban indicator that scales systematically with population size (Bettencourt et al., 2010). For all practical purposes it is the deviations from normal scaling that are the most interesting phenomena. 3. Scaling phenomena and the empirical complexity of cities Bettencourt's theory has been developed on the basis of a national data set. It describes scaling according to population size of cities in one country. As already Bettencourt et al. indicate population size might rather be a “proxy aggregate variable […] than a causal force” (Bettencourt et al., 2010, 6), which is immediately supported by the greatly varying deviations from the normal scale observed for some functions (SAMI), e.g. murder rates. Arcaute et al. (2015) support this view with their research. They have developed a method to test the population size of cities and the delineation of cities with respect to commuters. For this end the authors created statistical units and aggregated these in this way constructing different population sizes. Finally, they tested different parameters with respect to population size only to find that the observed scaling pattern is sensitive to the delineation of cities and hence population size itself. This sensitivity to cities' demarcation has also been observed by Oliveira et al. (2014) who employ a similar method. They suggest that allometric studies based on official administrative boundaries of cities suffer from an endogeneity bias. Arcaute et al. conclude that any urban scaling theory must better address variety and diversity. In particular super-linear scaling of income and patents is found to be highly sensitive (Arcaute et al., 2015). Causes seem to be more varied and other factors than population size might be more relevant for putting the theory to practical use. A generic indicator like “patents” is for instance only partly reliable for assessing specific innovations such as for buildings since their number can depend on the policy instruments applied (Noailly, 2012). Likewise, total factor productivity as a derivative of a CobbeDouglas production function is an insufficient indicator for future development. The CobbeDouglas production function is effectively an artefact of macro-economics; and energy-related expost observations based on total factor or CobbeDouglas productivity are unsuitable for forecasts since they omit important energy related economic mechanisms (Ayres, 2001).

4275

The need for a more differentiated analysis also concerns the relationship of cities to the surrounding territory and the political system they are part of Buhaug and Urdal (2013), Finka and nkova  (2015). Urban scaling phenomena of cities may Kluva indeed relate to cities in single countries only due to administrative traditions, political priorities and other factors (Cristelli et al., 2012). Some indication for this can be found in Europe: Baur et al. (2014) show that a scaling relationship does not exist between population size and energy consumption for cities of the European Union but can instead be confirmed for the member states. While sometimes dismissed as merely complicated formal institutions obviously have long-range effects on national economies (Rodrik et al., 2002). This speaks for differentiated analyses. Apart from that it also suggests multiple legacies prevailing over long periods. Some of these legacies have developed path-dependently governed by specific mechanisms over centuries, or even millennia  le my, 2014a). There is including road networks (Louf and Barthe also increasing evidence that the urban form has direct (Salat and Bourdic, 2011) and indirect influence on the economic (Redfearn, 2009) and metabolic flipside of development (Rickwood et al., 2008; Grazi, van den Bergh, and van Ommeren, 2008). This is even more important since increasing energy-efficiency in one sector is regularly offset by growing energy-use in another representing so-called rebound effects (of consumption). These are known to be particularly relevant for mobility (Antal and van den Bergh, 2014; Creutzig et al., 2015a). Following the known biologic scaling relationship between metabolic rate and the body mass of organisms (‘Kleiber's law’) one might expect a sub-linear relationship between population size and urban metabolism. However, the empirical evidence is less clear. Indeed, if the methodological challenge of defining population size functionally (and not spatially following administrative boundaries) is taken into account (Arcaute et al., 2015) only one study may be considered reliable so far. Oliveira, Andrade, and Makse (2014) find super-linear scaling in US cities indicating increasing urban inefficiencies: doubling the population of US-cities results in an average increase of 146% in CO2 emissions,2 rather than the expected efficiency gains from economies of scale in infrastructure. Indeed, we seem to be witnessing strong rebound effects in US-cities.3  le my (2014b) Against the simplicity of allometry Louf and Barthe have pointed out that Bettencourt's theory implies mono-centric cities with a single centre. This may lead to an underestimation of the decreasing returns associated with increasing mobility. In particular, the authors propose that delays due to congestions scale super-linearly with population (even in polycentric cities). Because CO2-emissions correspond to commuting time rather than simply distance, also the per capita emissions of daily commuting scale with the size of the city. They validate this hypothesis against limited data and call for further research on this crucial aspect of urbanization. Other studies reveal national, regional and global mitigation wedges (Creutzig et al., 2012; Baur et al., 2014; Creutzig et al., 2015b) that will be more effective with policies controlling rebound effects (Antal and van den Bergh, 2014).

2 By contrast, Fragkias et al. (2013) show that for U.S. metropolitan areas city size and CO2 emissions scale proportionally with urban population size. They analysed 930 cross-sectional observations over 10 years but rely on urban ‘core based statistical areas’ (CBSAs) that include Metropolitan Statistical Areas (MSAs) and Micropolitan Areas. Even though U.S. census updates and revises the delineations of the CBSAs periodically it would require further methodological review to compare it with the City Clustering Algorithm (see Appendix). 3 According to the innovative study by Brockway et al. (2014) the US has experienced an “efficiency dilution” in the past 50 years, which has resulted in a stagnation of energy-efficiency. Efficiency dilution signifies structural shifts to less efficient consumption that outweigh device-level efficiency gains.

4276

F. Schiller / Journal of Cleaner Production 112 (2016) 4273e4282

There are other systems of provision with supply and disposal functions, e.g. water, electricity, district heat, sewage, waste infrastructure (etc). Their eco-efficiency is also closely related to climate change mitigation and is not limited to primary energy use but also may include associated material flows and up-stream emissions (Hodson et al., 2012; Brattebø et al., 2009; Sartori et al., 2008). Pauliuk et al. (2014) employ a fractal model to analyse water and wastewater networks and find that both total network length and mass of the network scale sub-linearly. They identify a gap between maintenance and costs of network length. Maintenance costs are dominated by small pipes at the backend of the system indicating economies of scale. These can be considered in planning. Efforts to adapt the infrastructure of cities to environmental constraints concentrate on delivering existing services more efficiently, combining different services, terminating superseded services and recycling obsolete infrastructure or constructing new energy and resource-efficient infrastructure (Müller et al., 2013). These efforts are combined with the search for new business models for (energy) infrastructure in the form of energy service companies (ESCO) or multi-utility service companies (MUSC) (Sorrell, 2007; Steinberger et al., 2009; Roelich et al., 2014). Batty et al. (2008) have studied allometry in buildings for correlation functions to compute the variation of particular properties of a building with respect to other buildings. First, they find scaling in the height of global buildings and the respective number of buildings of any particular height. More importantly, they show scaling in buildings of London pertaining to building area, perimeter, height and volume for nine buildings types. For the relationships between area and perimeter, volume and perimeter as well as volume and area they can show allometry. However, they acknowledge that they had to omit building height despite the realisation that building height is a major variable with respect to the fractality of cities. Steadman, Evans, and Batty (2009) demonstrate allometry with respect to the ratio of wall surface-to-volume of building blocks in London. These various scaling phenomena translate into distinct hierarchical levels of urban morphology and form that allow to separating factors. For instance, density can be discriminated from behaviours of inhabitants or the building shape factor from passive volume for natural ventilation and natural lighting (Salat, 2009; Bourdic et al., 2012). Thus, while economies of scale, typological network effects and agglomeration economies of the past are important to determine environmental impact future-oriented transitions will have to employ agglomeration economies such economies of scale, increasing returns, knowledge spill-overs in order to overcome existing diseconomies of cities, such as carbon lock-in. 4. Making complexity science relevant for urban transition theory Complexity science may eventually consolidate theory building in various disciplines. In more applied fashion it can be utilised for mapping pathways through rugged, complex landscapes of uncertain co-evolving processes (Edenhofer and Kowarsch, 2015). With respect to situated policy-making these have to be translated into tangible policy-advice (Flanagan et al., 2011). In particular transition management theory lends itself for this task. Scaling signifies output complexity that emerges from the input complexity of a system. Input complexity refers to the heterogeneity of “empirical rules” and “initial conditions” that shape the spatial and temporal evolution of a system. Systems with long correlations of key dynamical variables in space and time exhibit output complexity (cf. Irwin et al., 2009). With the help of descriptive models complexity science has shown that already simple rules (low input complexity) can bring about output complexity, i.e.

si, 1999). scaling emerging from self-organising processes (Baraba Unlike localised physical systems with little heterogeneity, however, human and biological systems are characterised by far greater input complexity with numerous interactions at multiple temporal and spatial scales and substantial heterogeneity (Mitchell, 2009). Despite the interest in reproducing scaling in social systems with models of low input complexity, i.e. simple rules (e.g. Axtell and Florida, 2001); the hypothesis that multiplicative growth processes result in stationary power laws will have to cover growth dynamics in greater detail (cf. Andersson et al., 2006) e not least by taking the differentiated nature of social systems as a result of informal and formal institutions into account, e.g. by providing explanatory mechanisms. Transition theory offers guidance here since it has been interested in intentional action discriminating purposeful transitions from contingently emerging technological changes of the past (Geels, 2011). Due to their environmental relevance three technological fields have been of particular interest for the theory: energy, transport systems and agriculture. Complexity and transition theory thus share already an interest in transport and energy. In addition, transition theory has understood socio-technological regimes as systems (Geels, 2011). More clearly, transition theory has embraced the historical task of unlocking our carbon-based economic systems by showing in numerous empirical studies how investments sunk in machines, infrastructures and competencies can be overcome. It has also contributed to the study of end-use technologies, e.g. photovoltaic or solar heating or alternative fuels in cars (cf. McCormick et al., 2013). A multi-level and a multi-phase strand of transition theory have been distinguished. The multi-level perspective (MLP) has become the most prominent framework in transition research distinguishing niches, regimes and an overall landscape with respect to the rise of alternative technologies to locked-in carbon intensive technologies (Geels, 2004, 2011). Niches signify laboratories of these new technologies that confront existing regimes of old technology supported by the landscape exerting immergent pressures on the niche, e.g. as respending rebounds in emerging economies (Antal and van den Bergh, 2014). A transition starts with new technology forming in niches and ends with a relatively stable new socio-technological regime. By contrast the multi-phase perspective (MPP) has specifically addressed the dynamics of transitions. The rise of new technologies is shown to be a non-linear process starting under a stable regime followed by a period of accelerated change, which then stabilises with the arrival of a new regime at an advanced technological level. Four phases have been identified: pre-development, take-off, acceleration and stabilisation-phase (Rotmans et al., 2001). Marrying transition theory with complexity research on the built environment would extend the fields studied under the transition framework by adding cities. This would bring a sector under the framework that promises large savings of carbon emissions (Edenhofer et al., 2014 pp. 671). From the perspective of transition theory the relationship between the multi-level and multi-phase perspectives concerns a) the interdependence between technological trajectories and b) the interactions between the different levels. At a secondary level explanations concern individual trajectories involving evolutionary explanations and particular events (Geels and Schot, 2010). It is at this level where complexity science can best facilitate the transition of cities. Studying cities in sustainable transitions with complexity scientific methods would add to this new field of transition theory (see Hodson and Marvin, 2010; McCormick et al., 2013 for ealier proposals to integrate cities in transition theory). It can add building stock and urban form (the morphology of cities) to the existing fields of traffic and energy or, from a subject-centred, consumptive perspective, integrate mobility, electricity-use and heating/cooling.

F. Schiller / Journal of Cleaner Production 112 (2016) 4273e4282

Transition theory has employed mechanisms when addressing the temporal dimension of transitions (Geels, 2011); but mechanisms might also be identified with respect to governance (Biesbroek et al., 2014). Complexity science too has used explanatory mechanisms but its models have regularly relied on statistical mechanisms of low input complexity, e.g. preferential attachment simulating the growth of scale-free networks independent of their si, 1999). Vasileiadou and Safarzyn  ska (2010) context (Baraba recognise that in complex systems the evolution of variables, i.e. their dynamics, is particularly relevant. They propose to empirically investigate the degree of complexity of a system, to assess it at the level of the policy-making system and to design longitudinal and retrospective research identifying transitions.4 Explanatory mechanisms signify social processes that reproduce similar outcomes. They can be found at all levels of society and affect the relationship between social stability and change. Unlike the concept of emergence used in “generative” sociology (Epstein, 1999) there is no reason why explanatory mechanisms in the social realm should be restricted to endogenous explanations based on individualised accounts. As a matter of fact social systems exert downward pressures on actors all the time (immergence) (Mayntz, 2004). Transitions involve emergent and immergent social processes, i.e. mechanisms, that persist over long-time. When studying mechanisms social research may rely on contingent generalisations (Scharpf, 2002; Mitchell, 2002) or casing (Ragin, 2008), both rooted in systems epistemology although each in  ska (2010) recognise that different ways. Vasileiadou and Safarzyn in systems under transition the evolution of variables, i.e. their dynamics, is particularly relevant. They propose to empirically investigate the degree of complexity of a system, to assess it at the level of the policy-making system and to design longitudinal and retrospective research identifying transitions. While development of the urban form, for example, might be described in institutional terms as an emergent process with nearly self-sufficient path-dependencies (Redfearn, 2009; Sorensen, 2015) attempts to explain changes in housing policy as path-dependent processes may fall short (Malpass, 2011). Similarly, long-term strategies for retrofitting and refurbishing the existing building stock will have to be designed to unlock existing technological pathdependencies (Kohler and Hassler, 2012). Unlike in the natural domain in the social realm, but particularly in politics, pathdependency turns out to be a multifaceted concept where effects may quickly become new causes (Pierson, 2004). Indeed, many policies may actually represent sequences because individual decisions are reversible (while many environmental consequences are not). Transition theory reflects these contingencies of social life by constructing windows of opportunity for various actors to redirect existing socio-technological pathways (Foxon et al., 2013). If the observation is taken serious that policies are difficult to design in complex situations (Scharpf, 1997), i.e. high input complexity, policy-relevant studies and models of urban transitions will have to be attributable not just to agents and emergent phenomena at system-level (micro-macro explanation) but also to the wider socio-economic-political context (cf. Vasileiadou and  ska, 2010). Mechanisms bringing behavioural and instituSafarzyn tional research together and drawing on (exogenous) macro-micro

4 While the need for a universal theory of sustainable cities is obvious, it is surprising that claims to universality have sometimes been based on single-country data. As a rule universal claims should be based on global data even if such data sets are more difficult to assemble. This will take more systematic efforts in the future and methodological challenges such as the delimitation of cities remain. Nevertheless, the task promises insights that could better guide sustainable urban transitions globally.

4277

explanations recognising immergence increase the chances of producing useable, socially robust knowledge. 5. Discussion Scaling based theories integrate analyses of key social artefacts and processes in cities most of which are directly relevant for sustainable urban transitions. Bettencourt's theory describes the interdependency between urban form and socio-economic networks of people. Its Scale-Adjusted-Metropolitan-Indicators promise new, very precise tools for planners with the capacity to project patterns to functions and locations for which or where such data are not systematically collected. They might get extended to include existing and future work on scaling in buildings and infrastructure (above and below ground). At the same time the allometric theory does not (yet) account for the decreasing returns of the interdependencies that have materialised, amongst other things, as carbon emissions from producers and consumers ‘lockedin’ in a given urban form and morphology. Moreover, the debate around city sizes also highlights the methodological challenges of aggregated factors that involve human agency. This is particularly true for social scaling phenomena associated with high input complexity. There has been considerable work on traffic networks while fewer network studies have addressed the built environment with the majority focussing again on the typology of infrastructure networks. Only a small number of these studies have been explicitly devoted to the embodied resources and emissions. Despite tangible successes in modernising the building stock (Kalmykova et al., 2015) even less research has examined scaling patterns in the building stock. In sum, the slow feedback mechanisms of urban morphology have received considerably less attention than the fast feed-back loops associated with the urban form and traffic networks. Despite this bias complexity science promises further integration of currently still disperse fields. The integration of income into Bettencourt's theory (Bettencourt et al., 2010) is also relevant with respect to the diffusion of technology but might not be sufficient for two reasons: firstly, also income distribution is characterised by scaling that influences the diffusion of technology but is e like income generation e largely governed at the national level. While discussing this macro-perspective is beyond the scope of the article it opens up a host of research questions for economics (cf. Mandelbrot, 2009). Secondly, while personal income enables private investment in energy-efficient end-use technology income is of course also being spent on unsustainable activities and goods. Numerous studies have shown a correlation between income and energy consumption (e.g. Reinders et al., 2003; Schipper, 2004; Lenzen et al., 2008; Pachauri and Jiang, 2008; Wiedenhofer et al., 2013; Baur et al., 2014). By linking physical and social networks complexity science could contribute to existing micro-perspectives on rebounds effects (Sorrell and Dimitropoulos, 2008). Direct rebound effects might be attributed to network effects, e.g. from congestion (Louf and le my, 2014b), or to scaling in the demand for mobility Barthe (Noulas et al., 2012). Further research might eventually come to consider income in a macro-perspective allowing also integration of indirect rebound effects, i.e. income-induced shifts between specific end-use technology to arrive at socio-economic theories that better forecast behavioural and economic outcomes of innovation diffusion. By contrast, patents and GDP are less straight-forward indicators. Besides the discussed problems with patents in regard to eco-innovations there are stronger indicators for competitiveness available (cf. Tacchella et al., 2013). Similarly, GDP is widely considered inadequate for today's policy needs (Stiglitz et al., 2009)

4278

F. Schiller / Journal of Cleaner Production 112 (2016) 4273e4282

and has been amended by the indicator “well-being” (Kubiszewski et al., 2013). Fairness in accessing urban resources is a core aspect of sustainability (Broto et al., 2012) and well-being relates as much to the functions provided by cities such as infrastructure and housing as to the capability to access these. In practice a variety of indicators are therefore being used (Schwarz, 2010; Yetano Roche et al., 2013). Despite the evidence for self-organisation of the urban form and morphology (see also Bristow and Kennedy, 2015) it is not clear what scaling phenomena beyond human artefacts mean and how to address them. One problem in this respect is complexity science's inherent social reductionism (e.g. the neglect of law) that may fail to embed actors and to address them differentially. Although networks in the social realm can be the result of selforganisation, self-organisation does not automatically bring about social networks in which individual actors change their behaviours from unsustainable to sustainable life styles. Social networks normally require additional formal institutions providing information, incentives, constraints etc. to change behaviours. We therefore need to provide empirically explanations of socio-political mechanisms enabling urban transitions and we should expect a range of such mechanisms to exist even within a unified allometric framework of urban form and morphology. These should address the specific path-dependencies that have resulted from individual behaviours as well as formal institutions in a given social context and identify mechanisms that allow managing urban transitions, i.e. facilitate agency in niches to change regimes. More differentiated societies can run several niches in parallel, which they may coordinate through competitive markets and in different policy arenas with interdependencies and hierarchies between them. This reminds us that we do not simply observe complex social systems but re-design live social systems capable of reflexive re€rnberg, and To €rnberg (2014) termed social sponses. Andersson, To systems “wicked systems” because they have both, complex and complicated features. For the governance of sustainable transitions explaining complicated institutions is equally important as providing complex models of aggregate social phenomena. The social realm is thus generally best characterised by high input and wicked complexity involving multi-causalities, non-ergodic processes and complicated mechanisms. Bellow a descriptive system-level that may be constituted by scaling phenomena and allometry various mechanisms specify any model within its local context. This may include hierarchical relationships as for instance represented by the landscape concept of multi-level transition theory. Accordingly, mechanisms do not have to directly represent emergence but may also stand for immergence, i.e. contingent generalisations. Indeed, high input complexity might be crucial to move from descriptive to scenario modelling and from policy advice to implementation. While scaling phenomena are scaleinvariant agency is not; e.g. democratic policy-making is about the legitimate reduction of complexity (Nowotny, 2005) with individual interventions signified by a start and an end date. This is why policies need to be specifically designed for different scales of the hierarchical political-administrative system (cf. Clifton et al., 2008). Recognising the plurality of valid mechanisms will lead to bespoken models that are adapted to the local context but share generic properties defined by scaling. The growing knowledge provided by network analyses should enable more integrated general models with standardised interfaces that will make specification increasingly easy. This can be exemplified for the transition of the built environment. While the IPCC identified the building sector as a potential main contributor to energy efficiency (Edenhofer et al., 2014) energyefficient technologies that do already exist for the built environ€ hmer et al., 2010) are diffusing slowly (e.g. ment (e.g. Lechtenbo International Energy Agency, 2013). Diffusion is slow not least due

to disadvantages of eco-innovations compared to ordinary innovations (Rennings, 2000), a general neglect of energy-efficient end-use technologies by regulators and investors (Wilson et al., 2012), existing path-dependencies in planning (Sorensen, 2015), and, for instance, the local and small nature of energy-efficiency developments in mature cities that regularly involve heterogeneous groups of decision makers (Kohler and Hassler, 2012). Faster diffusion of energy-efficient end-use technologies in the built environment will require broad coordination across markets, policies and society. Establishing explanatory mechanisms involving actors from these three social spheres can specify theories and subsequent models addressing for instance:
how to integrate heterogeneous stakeholder groups with possibly diverging cognition and conflicting interests (e.g. Sp€ ath and Rohracher, 2014), i.e. creating synergies rather than competition between technological niches;
how to integrate end-use technologies consumers invest in (i.e. prosumers) with existing systems of provision (e.g. Galvin and Sunikka-Blank, 2014), i.e. managing competition between technological niches and existing regimes;
how to decrease the costs of energy-efficient innovations (i.e. frugal innovations) to support diffusion to low-income groups and countries (e.g. Tiwari and Herstatt, 2012);
how to govern business cycles in real estate markets (e.g. Wu, 2014; Schwartz, 2010) to enable a continuous energetic modernisation of building stocks (Ürge-Vorsatz et al., 2009; Hassler, 2009; Lützkendorf et al., 2011), i.e. protect niches against encroachment of the existing real estate regime. In providing explanatory social mechanisms pertaining to these and other questions context-dependent institutional differences will come to the fore. They need to be identified (despite the fact that generic models can be contextualised by imported local data) since institutional differences can be key for successful implementation. Models of transitions should use scaling analysis in ways that enable rather than pre-empt participation, to arrive at scenario models that empower agents to practise sustainable behaviours. This implies that mechanisms are preferred which reflect agency and support deliberative decision-making. Indeed, this may be key to overcome the marginalisation of end-use technologies in energy innovation for climate protection and is particularly true for the built environment where stakeholders are numerous and heterogeneous. While this is a challenge for home owners and private enterprise as well as public research and public planning, regime change in the built environment will require targeted and managed long-term transitions pertaining particularly to urban form and morphology. Transition theory provides a tested framework for this challenge (Jong et al., 2015). While it was introduced without incorporating social mechanisms integrating these is becoming necessary as global up-scaling of energy-efficient end-use technology for cities is needed (Eames et al., 2013). 6. Conclusions Urbanisation processes pose a key challenge for a sustainable transition. Due to the global nature of the problem we need to identify and explain cities' universal features. Bettencourt's theory is an important step in this direction by revealing the universal selforganising nature of the built environment. Urban allometry suggests efficiency potentials of an urban organisation but it also embodies socio-economic drivers of unsustainable growth processes. The article noted the methodological and empirical challenges of this approach and argued for alternative indicators to GDP and patents. It highlighted several fields where network analysis is

F. Schiller / Journal of Cleaner Production 112 (2016) 4273e4282

successfully being used to study energy- and resource-efficiency potentials of buildings and infrastructure. It emphasised the task to integrate these analyses to initiate the sustainable transition of cities considering their urban form and morphology worldwide. The built environment with its allometric features is connected to social networks that drive economies and diseconomies of cities. The article argued that these networks do not exist in the void but are coordinated through various formal and informal social institutions. We will therefore need explanatory social mechanisms relating human behaviours and social institutions to existing scaling phenomena in order to drive and govern cities towards sustainable pathways. At the same time some known scaling relationships in social networks already enable simulations in certain areas, e.g. direct rebound effects in traffic. If income is systematically integrated in future analyses we may also arrive at explaining indirect rebound-effects. However, income is not just a key variable for carbon emissions but also for the diffusion of technology, which needs to be taken into account. The article argued for a careful use of path dependency concepts when designing diffusion policies to drive the built environment towards its efficiency frontiers. It exemplified social mechanisms that facilitate urban transitions and encourage agency. Acknowledgements The author would like to thank Paul Jensen and Christopher Knight for their comments on earlier versions of this article. Appendix Scaling Most studies of complex networks have analysed the typology of networks by using graph theory (Caldarelli and Vespignani, 2007). A key criterion is the degree distribution which shows different properties, e.g. small-world networks are characterised by a Poisson- or Gauss-distribution. By contrast scale-free networks reveal logarithmic distributions that follow a power law. This means that an invariant relationship between a reference parameter and a second or more variables exists as in Bettencourt's theory of urban allometry. Scale-free networks include fractal structures of self-similarity as well as self-affinity. Whereas in self-similar networks the power law is the same in all directions in self-affine structures different directions obey different power laws. Scaling phenomena cannot only be found in the typology but also in the evolution of networks. For the analysis of empirical data two methods are applied: binning and cumulative distributions (Caldarelli, 2007, 72). Both curve fitting methods discriminating the statistical noise from the curve by assuming that the fluctuations bellow and above the average cancel out. The binning method divides the x-axis into intervals (bins) and averaging the data of each bin. For power laws Newman (2005) suggests logarithmic bins due to the strong noise in the long tail. Nevertheless, the bin size has to be determined by trial and error since the statistical noise cannot be entirely eliminated. By contrast, the method of cumulative distributions averages the fluctuations across the complete data set. It focuses on the probability P(x) that an outcome occurs with a frequency greater P > ðxÞ. This amounts to the probability distribution Z∞ Pðx0 Þdx0 . In cases where the function P(x) represents a P > ðxÞ ¼

P > ðxÞ ¼

Z∞

PðxÞ0 dx0 ¼

x

4279

Z∞

Ax0g dx0 ¼

x

A gþ1 x : g1

As a result the exponent is altered to: (gþ1), filtering most noise. Nevertheless, if the exponent is close to 1 the integral does not behave like a power law but like a logarithm. In addition to that the upper limit usually has a finite maximum value xmax which changes the curve: the smaller xmax the more does the curve depart from a power law; the more xmax approaches infinity the more the curve resembles a power law (Caldarelli, 2007). Integrating the above function gives the following equation: >

Zxmax

P ðxÞ ¼ x

Axg dx ¼

 A  g1 x ;  xg1 max g1

which allows to approximate a power law if xmax goes towards infinity. If, by contrast, xmax is too small the diviation from a line is difficult to assess and therefore the value for g hard to determine. Both methods have been criticised since they cannot establish with certainty whether the curve actually represents a power law. For this reason Clauset et al. (2007) combined the maximumlikelihood method with a statistical test (KolmogoroveSmirnov) and provided algorithms for a variety of statistical programmes.5 Batty et al. (2008) suggest computing the cumulative distribution function without binning from the raw data when examining densities. The density pi of an object i is ordered from the smallest to the largest densities resulting in the index changing from i to k. Hence, pk follows the order from the smallest to the largest. They compute the cumulative distribution function F(pK  pk) with PK equation k¼1 pk, which is equivalent to the integral of the continuous density. This is studied with the help of the counter or complementary-cumulative distribution function corresponding to the Zipf rank-size distribution. It is defined as:

r ¼ FðpK  pkÞ ¼ N  FðpK  pKÞ; where r is the rank in terms of the ordered sizes defined by k and N the number of occurrences. If a system is to be considered self-organising the entire distribution has to persistently follow a power law. This is also true for any sample thereof (cf. Cristelli et al., 2012). A surprisingly easy method for studying rank-size distributions that adheres to this requirement is the head-tail-breaks-method (Jiang, 2013). The method proceeds from the probability density and divides the data set at the arithmetic mean into two parts and repeats doing so by dividing the part above the arithmetic mean until no heavy tail distribution is left. In this way, the number of classes, as well as the intervals between these, is derived from the data themselves. As an additional advantage the classes can be reduces (e.g. for visualising) without additional computation since the hierarchy between the classes remains untouched. City Clustering Algorithm (CCA) In order to avoid the contingencies of administrative city boundaries Rozenfeld et al. (2008) have created the City Clustering Algorithm (CCA) analysing population clusters. These clusters are created from contiguous populated sites at defined length scales. The length scales are variable. The method scans the grid and applies the CCA to the coarse-grained dataset using

x

power law P(x) ¼ Axg where g stands for the scaling parameter integration results in the equation:

5

http://tuvalu.santafe.edu/~aaronc/powerlaws/.

4280

F. Schiller / Journal of Cleaner Production 112 (2016) 4273e4282

different length scales. In this way the algorithm defines “cities” as maximally connected clusters of populated sites at high resolution. References Anderson, Chris, 2008. The end of theory: will the data deluge make the scientific method obsolete? Wired Mag. http://archive.wired.com/science/discoveries/ magazine/16-07/pb_theory/. Andersson, Claes, Frenken, Koen, Hellervik, Alexander, 2006. A complex network approach to urban growth. Environ. Plan. A 38 (10), 1941e1964. http:// dx.doi.org/10.1068/a37418. €rnberg, Anton, To €rnberg, Petter, 2014. Societal systems e Andersson, Claes, To complex or worse? Futures 63, 145e157. http://dx.doi.org/10.1016/ j.futures.2014.07.003. s, van den Bergh, Jeroen C.J.M., 2014. Re-spending rebound: a macroAntal, Miklo level assessment for OECD countries and emerging economies. Energy Policy 68 (0), 585e590. http://dx.doi.org/10.1016/j.enpol.2013.11.016. Arcaute, Elsa, Hatna, Erez, Ferguson, Peter, Youn, Hyejin, Johansson, Anders, Batty, Michael, 2015. Constructing cities, deconstructing scaling laws. Interface 12 (iii), 20140745. http://dx.doi.org/10.1098/rsif.2014.0745. Arthur, W. Brian, 1994. Increasing Returns and Path Dependence in the Economy. University of Michigan Press, Ann Arbor. Axtell, Robert, Florida, Richard, 2001. Emergent Cities: a Microeconomic Explanation for Zipf's Law. In: Computing in Economics and Finance. Society for Computational Economics (April). Ayres, Robert U., 2001. The minimum complexity of endogenous growth models: the role of physical resource flows. Energy 26 (9), 817e838. http://dx.doi.org/ 10.1016/S0360-5442(01)00031-7.  , 1999. Emergence of scaling in random networks. Science Barab asi, Albert-L aszlo 286 (5439), 509e512. http://dx.doi.org/10.1126/science.286.5439.509. Batty, Michael, 2008. The size, scale, and shape of cities. Science 319 (5864), 769e771. http://dx.doi.org/10.1126/science.1151419. Batty, Michael, 2012. Building a science of cities. Cities 29, S9eS16. http:// dx.doi.org/10.1016/j.cities.2011.11.008 (March). Batty, Michael, Carvalho, R., Hudson-Smith, A., Milton, R., Smith, D., Steadman, P., 2008. Scaling and allometry in the building geometries of greater London. Eur. Phys. J. B 63 (3), 303e314. http://dx.doi.org/10.1140/epjb/e2008-00251-5. Baur, Albert Hans, Thess, Maximilian, Kleinschmit, Birgit, Creutzig, Felix, 2014. Urban climate change mitigation in europe: looking at and beyond the role of population density. J. Urban Plan. Dev. 140 (1), 04013003. http://dx.doi.org/ 10.1061/(ASCE)UP.1943-5444.0000165. Bettencourt, Luís M.A., 2013. The origins of scaling in cities. Science 340 (6139), 1438e1441. http://dx.doi.org/10.1126/science.1235823. Bettencourt, Luís M.A., West, Geoffrey B., 2010. A unified theory of urban living. Nature 467 (7318), 912e913. http://dx.doi.org/10.1038/467912a. , Helbing, Dirk, Kühnert, Christian, West, Geoffrey Bettencourt, Luís M.A., Lobo, Jose B., 2007. Growth, innovation, scaling, and the pace of life in cities. Proc. Natl. Acad. Sci. U. S. A. 104 (17), 7301e7306. http://dx.doi.org/10.1073/ pnas.0610172104. , Strumsky, Deborah, West, Geoffrey B., 2010. UrBettencourt, Luís M.A., Lobo, Jose ban scaling and its deviations: revealing the structure of wealth, innovation and crime across cities. PLoS ONE 5 (11), e13541. Public Library of Science. Bettencourt, Luís M.A., Samaniego, Horacio, Youn, Hyejin, 2014. Professional diversity and the productivity of cities. Sci. Rep. 4, 5393. http://dx.doi.org/ 10.1038/srep05393 (January). Biesbroek, G. Robbert, Termeer, Catrien J.A.M., Klostermann, Judith E.M., Kabat, Pavel, 2014. Rethinking barriers to adaptation: mechanism-based explanation of impasses in the governance of an innovative adaptation measure. Glob. Environ. Change 26, 108e118. http://dx.doi.org/10.1016/j.gloenvcha.2014.04.004 (May). Boschma, Ron A., Lambooy, Jan G., 1999. Evolutionary economics and economic geography. J. Evol. Econ. 9 (4), 411e429. http://dx.doi.org/10.1007/ s001910050089. Bourdic, Loeiz, Salat, Serge, Nowacki, Caroline, 2012. Assessing cities: a new system of cross-scale spatial indicators. Build. Res. Inf. 40 (5), 592e605. http:// dx.doi.org/10.1080/09613218.2012.703488. Brattebø, Helge, Bergsdal, Håvard, Sandberg, Nina Holck, Hammervold, Johanne, Müller, Daniel B., 2009. Exploring built environment stock metabolism and sustainability by systems analysis approaches. Build. Res. Inf. 37 (5e6), 569e582. http://dx.doi.org/10.1080/09613210903186901. Routledge. Bristow, David, Kennedy, Christopher, 2015. Why do cities grow? Insights from nonequilibrium thermodynamics at the urban and global scales. J. Ind. Ecol. 19 (2) http://dx.doi.org/10.1111/jiec.12239. Brockway, Paul E., Barrett, John R., Foxon, Timothy J., Steinberger, Julia K., 2014. Divergence of trends in US and UK aggregate exergy efficiencies 1960e2010. Environ. Sci. Technol. 48, 9874e9881. http://dx.doi.org/10.1021/es501217t. Broto, Vanesa Cast an, Allen, Adriana, Rapoport, Elizabeth, 2012. Interdisciplinary perspectives on urban metabolism. J. Ind. Ecol. 16 (6), 851e861. http:// dx.doi.org/10.1111/j.1530-9290.2012.00556.x. Buhaug, Halvard, Urdal, Henrik, 2013. An urbanization bomb? Population growth and social disorder in cities. Glob. Environ. Change 23 (1), 1e10. http:// dx.doi.org/10.1016/j.gloenvcha.2012.10.016.

Caldarelli, Guido, 2007. Scale-free Networks Complex Webs in Nature and Technology. Oxford University Press, Oxford. Caldarelli, Guido, Vespignani, A., 2007. Large Scale Structure and Dynamics of Complex Networks: from Information Technology to Finance and Natural Science. World Scientific Publishing Company. Clauset, Aaron, Shalizi, Cosma Rohilla, Newman, M.E.J., 2007. Power-law distributions in empirical data. SIAM Rev. 51 (3), 661e703. http://dx.doi.org/10.1137/ 070710111. Clifton, Kelly, Ewing, Reid, Knaap, Gerrit-Jan, Song, Yan, 2008. Quantitative analysis of urban form: a multidisciplinary review. J. Urban. Int. Res. Placemaking Urban Sustain. 1 (1), 17e45. http://dx.doi.org/10.1080/17549170801903496. Conte, R., Gilbert, Nigel, Bonelli, G., Cioffi-Revilla, C., Deffuant, G., Kertesz, J., Loreto, V., et al., 2012. Manifesto of computational social science. Eur. Phys. J. Special Top. 214 (1), 325e346. http://dx.doi.org/10.1140/epjst/e2012-01697-8. €mer, Julia, 2012. Decarbonizing urban transport Creutzig, Felix, Mühlhoff, Rainer, Ro in European cities: four cases show possibly high co-benefits. Environ. Res. Lett. 7 (4), 044042. http://dx.doi.org/10.1088/1748-9326/7/4/044042. Creutzig, Felix, Baiocchi, Giovanni, Bierkandt, Robert, Pichler, Peter-paul, Seto, Karen C., 2015a. Supporting information for ‘a global typology of urban energy use and potentials for an urbanization mitigation wedge’. PNAS S1eS27. Creutzig, Felix, Baiocchi, Giovanni, Bierkandt, Robert, Pichler, Peter-paul, Seto, Karen C., 2015b. Global typology of urban energy use and potentials for an urbanization mitigation wedge. PNAS 1e6. http://dx.doi.org/10.1073/ pnas.1315545112. Cristelli, Matthieu, Batty, Michael, Pietronero, Luciano, 2012. There is more than a power law in zipf. Sci. Rep. 2, 812. http://dx.doi.org/10.1038/srep00812 (January). Eames, Malcolm, Dixon, Tim J., May, Tim, Hunt, Miriam, 2013. City futures: exploring urban retrofit and sustainable transitions. Build. Res. Inf. 41 (5), 504e516. http://dx.doi.org/10.1080/09613218.2013.805063. Economist Intelligence Unit, 2009. European Green City Index. Siemens, Munich. Edenhofer, Ottmar, Kowarsch, Martin, 2015. Cartography of pathways: a new model for environmental policy assessments. Environ. Sci. Policy 51, 56e64. http:// dx.doi.org/10.1016/j.envsci.2015.03.017. Elsevier Ltd. Edenhofer, Ottmar, Pichs-Madruga, R., Sokona, Y., Farahani, E., Kadner, S., Seyboth, K., Adler, A., et al., 2014. Climate Change 2014: Mitigation of Climate Change: Contribution of Working Group III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, vol. III. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. Epstein, Joshua M., 1999. Agent-based computational models and generative social science. Complexity 4 (5), 41e60 doi:10.1002/(SICI)1099-0526(199905/06)4: 5<41::AID-CPLX9>3.0.CO;2-F. nkov Finka, Maros, Kluva a, Tatiana, 2015. Managing complexity of urban systems: a polycentric approach. Land Use Policy 42, 602e608. http://dx.doi.org/10.1016/ j.landusepol.2014.09.016. Flanagan, Kieron, Uyarra, Elvira, Laranja, Manuel, 2011. Reconceptualising the ‘policy mix’ for innovation. Res. Policy 40 (5), 702e713. http://dx.doi.org/ 10.1016/j.respol.2011.02.005. Foxon, Timothy J., 2011. A coevolutionary framework for analysing a transition to a sustainable low carbon economy. Ecol. Econ. Sustain. Res. Inst. Pap. 70 (12), 2258e2267. http://dx.doi.org/10.1016/j.ecolecon.2011.07.014. Leeds. Foxon, Timothy J., Pearson, Peter J.G., Arapostathis, Stathis, Carlsson-Hyslop, Anna, Thornton, Judith, 2013. Branching points for transition pathways: assessing responses of actors to challenges on pathways to a low carbon future. Energy Policy 52 (January), 146e158. http://dx.doi.org/10.1016/j.enpol.2012.04.030. , Strumsky, Deborah, Seto, Karen C., 2013. Does size Fragkias, Michail, Lobo, Jose matter? Scaling of CO2 emissions and U.S. Urban areas. PLoS ONE 8 (6), e64727. Public Library of Science. Galvin, Ray, Sunikka-Blank, Minna, 2014. The UK homeowner-retrofitter as an innovator in a socio-technical system. Energy Policy 74, 655e662. http:// dx.doi.org/10.1016/j.enpol.2014.08.013. Elsevier. Geels, Frank W., 2004. From sectoral systems of innovation to socio-technical systems. Insights about dynamics and change from sociology and institutional theory. Res. Policy 33 (6e7), 897e920. http://dx.doi.org/10.1016/ j.respol.2004.01.015. Geels, Frank W., 2011. The multi-level perspective on sustainability transitions: responses to seven criticisms. Environ. Innov. Soc. Transit. 1 (1), 24e40. http:// dx.doi.org/10.1016/j.eist.2011.02.002. Elsevier B.V. Geels, Frank W., Schot, Johan, 2010. The dynamics of transitions: a socio-technical perspective. In: Grin, J., Rotmans, J., Schot, J. (Eds.), Transitions to Sustainable Development. New Directions in the Study of Long Term Transformative Change. Routledge, New York, pp. 9e101. Grazi, Fabio, van den Bergh, Jeroen C.J.M., van Ommeren, Jos N., 2008. An empirical analysis of urban form, transport, and global warming. Energy J. 29 (4), 97e122. International Association for Energy Economics. Hassler, Uta, 2009. Long-term building stock survival and intergenerational management: the role of institutional regimes. Build. Res. Inf. 37 (5e6), 552e568. http://dx.doi.org/10.1080/09613210903189533. Hodson, Mike, Marvin, Simon, 2010. Can cities shape socio-technical transitions and how would we know if they were? Res. Policy 39 (4), 477e485. http:// dx.doi.org/10.1016/j.respol.2010.01.020. Hodson, Mike, Marvin, Simon, Robinson, Blake, Swilling, Mark, 2012. Reshaping urban infrastructure. J. Ind. Ecol. 16 (6), 789e800. http://dx.doi.org/10.1111/ j.1530-9290.2012.00559.x. International Energy Agency, 2013. Technology Roadmap. Energy Efficient Building Envelops, Paris.

F. Schiller / Journal of Cleaner Production 112 (2016) 4273e4282 Irwin, Elena G., Jayaprakash, Ciriyam, Munroe, Darla K., 2009. Towards a comprehensive framework for modeling urban spatial dynamics. Landsc. Ecol. 24 (9), 1223e1236. http://dx.doi.org/10.1007/s10980-009-9353-9. Jiang, Bin, 2013. Head/tail breaks: a new classification scheme for data with a heavy-tailed distribution. Prof. Geogr. 65 (3), 482e494. http://dx.doi.org/ 10.1080/00330124.2012.700499. Jiang, Bin, Claramunt, Christophe, 2004. Topological analysis of urban street networks. Environ. Plan. B Plan. Des. 31, 151e162. http://dx.doi.org/10.1068/b306. Jong, Martin de, Joss, Simon, Schraven, Daan, Zhan, Changjie, Weijnen, Margot, 2015. Sustainable-smart-resilient-low-carbon-eco-knowledge cities; making sense of a multitude of concepts promoting sustainable urbanization. J. Clean. Prod. 109, 25e38. http://dx.doi.org/10.1016/j.jclepro.2015.02.004. Kalmykova, Yuliya, Rosado, Leonardo, Patrício, Jo~ ao, 2015. Resource consumption drivers and pathways to reduction: economy, policy and lifestyle impact on material flows at the national and urban scale. J. Clean. Prod. 1e11. http:// dx.doi.org/10.1016/j.jclepro.2015.02.027. Kohler, Niklaus, Hassler, Uta, October 2012. Alternative scenarios for energy conservation in the building stock. Build. Res. Inf. 37e41. http://dx.doi.org/10.1080/ 09613218.2012.675714. Krugman, Paul R., 1991. Geography and Trade. The MIT Press, Cambridge, MA. Kubiszewski, Ida, Costanza, Robert, Franco, Carol, Lawn, Philip, Talberth, John, Jackson, Tim, Aylmer, Camille, 2013. Beyond GDP: measuring and achieving global genuine progress. Ecol. Econ. 93, 57e68. http://dx.doi.org/10.1016/ j.ecolecon.2013.04.019. Elsevier B.V. Kühnert, Christian, Helbing, Dirk, West, Geoffrey B., 2006. Scaling laws in urban supply networks. Phys. A Stat. Mech. Appl. 363 (1), 96e103. http://dx.doi.org/ 10.1016/j.physa.2006.01.058. €mmer, S., Gehlsen, B., Helbing, Dirk, 2006. Scaling laws in the spatial structure of La urban road networks. Phys. A Stat. Mech. Appl. €hmer, Stefan, Barthel, Claus, Merten, Frank, Schneider, Clemens, 2010. Lechtenbo Redesigning urban infrastructures for a low-emission future an overview of urban low-carbon technologies. S.A.P.I.EN.S 3 (October 2012): online. Lenzen, Manfred, Wood, R., Foran, B., 2008. Direct versus embodied energy. The need for urban lifestyle transitions. In: Droege, Peter (Ed.), Urban Energy Transition: from Fossil Fuels to Renewable Power. Elsevier, Amsterdam, pp. 91e120. , Bettencourt, Luís M.A., Strumsky, Deborah, West, Geoffrey B., 2013. UrLobo, Jose ban scaling and the production function for cities. PLoS ONE 8 (3), e58407. Public Library of Science. mi, Barthe  le my, Marc, 2014a. A typology of street patterns. J. R. Soc. InterLouf, Re face 11 (101). http://dx.doi.org/10.1098/rsif.2014.0924, 20140924e20140924. mi, Barthe  le my, Marc, 2014b. How congestion shapes cities: from mobility Louf, Re patterns to scaling. Sci. Rep. 4, 5561. http://dx.doi.org/10.1038/srep05561 (January). mi, Jensen, Pablo, Barthe  le my, Marc, 2013. Emergence of hierarchy in costLouf, Re driven growth of spatial networks. Proc. Natl. Acad. Sci. U. S. A. 110 (22), 8824e8829. http://dx.doi.org/10.1073/pnas.1222441110. mi, Roth, Camille, Barthe  le my, Marc, 2014. Scaling in transportation netLouf, Re works. PloS One 9 (7), e102007. http://dx.doi.org/10.1371/journal.pone.0102007. Lützkendorf, Thomas, Fan, Wei, Lorenz, David, 2011. Engaging financial stakeholders: opportunities for a sustainable built environment. Build. Res. Inf. 39 (5), 483e503. http://dx.doi.org/10.1080/09613218.2011.597206. Malpass, Peter, 2011. Path dependence and the measurement of change in housing policy. Hous. Theory Soc. 28 (4), 305e319. http://dx.doi.org/10.1080/ 14036096.2011.554852. Mandelbrot, Benoit B., 2009. ‘New methods of statistical economics,’ revisited: short versus long tails and Gaussian versus power-law distributions. Complexity 14 (3), 55e65. http://dx.doi.org/10.1002/cplx.20264. Mayntz, Renate, 2004. Mechanisms in the analysis of social macro-phenomena. Philos. Soc. Sci. 34 (2), 237e259. http://dx.doi.org/10.1177/0048393103262552. McCormick, K., Anderberg, Stefan, Coenen, L., Neij, L., 2013. Advancing sustainable urban transformation. J. Clean. Prod. 50, 1e11. Mitchell, Sandra D., 2002. Contingent generalizations: lessons from biology. In: €ln, Mayntz, Renate (Ed.), Akteure-mechanismen-modelle. Campus Verlag, Ko pp. 179e195. Mitchell, Sandra D., 2009. Unsimple Truths: Science, Complexity, and Policy, 1st ed. University of Chicago Press, Chicago. Müller, Daniel B., Liu, Gang, Løvik, Amund N., Modaresi, Roja, Pauliuk, Stefan, Steinhoff, Franciska S., Brattebø, Helge, 2013. Carbon emissions of infrastructure development. Environ. Sci. Technol. 47 (20), 11739e11746. http://dx.doi.org/ 10.1021/es402618m. Nelson, Richard R., Winter, Sidney G., 1982. An Evolutionary Theory of Economic Change. Belknap Press of Harvard University Press, Cambridge, MA. Newman, M.E.J., 2005. Power laws, Pareto distributions and Zipf's law. Contemp. Phys. 46 (5), 323e351. http://dx.doi.org/10.1080/00107510500052444. €lle, 2012. Improving the energy efficiency of buildings: the impact of Noailly, Joe environmental policy on technological innovation. Energy Econ. 34 (3), 795e806. Elsevier B.V. http://dx.doi.org/10.1016/j.eneco.2011.07.015. Noulas, Anastasios, Scellato, Salvatore, Lambiotte, Renaud, Pontil, Massimiliano, Mascolo, Cecilia, 2012. “A tale of many cities: universal patterns in human ur~ el. PloS ONE 7 (5), e37027. http://dx.doi.org/ ban mobility.” Edited by Juan A. An 10.1371/journal.pone.0037027. Nowotny, Helga, 2005. The increase of complexity and its reduction: emergent interfaces between the natural sciences, humanities and social sciences. Theory, Cult. Soc. 22 (5), 15e31. http://dx.doi.org/10.1177/0263276405057189.

4281

 S., Makse, Herna n a, 2014. Large cities are less Oliveira, Erneson a, Andrade, Jose green. Sci. Rep. 4, 4235. http://dx.doi.org/10.1038/srep04235. Pachauri, Shonali, Jiang, Leiwen, 2008. The household energy transition in India and China. Energy Policy 36 (11), 4022e4035. http://dx.doi.org/10.1016/ j.enpol.2008.06.016. Pauliuk, Stefan, Venkatesh, G., Brattebø, Helge, Müller, Daniel B., 2014. Exploring urban mines: pipe length and material stocks in urban water and wastewater networks. Urban Water J. 11 (4), 274e283. http://dx.doi.org/10.1080/ 1573062X.2013.795234. Pierson, Paul, 2004. Politics in Time. Princeton University Press, Princeton, Oxford. Powell, Walter, W., 1990. Neither market nor hierarchy: network forms of organization. Edited by Grahame Thompson Res. Organ. Behav. 12 (1), 295e336. Sage [u.a.]. Pumain, Denise, 2006. Alternative explanations of hierarchical differentiation in urban systems. Methodos Series. In: Pumain, Denise (Ed.), Hierarchy in Natural and Social Sciences, vol. 3. Springer Netherlands, Dordrecht, pp. 169e222. http://dx.doi.org/10.1007/1-4020-4127-6_8. line, Lobo, Jose , 2006. An Pumain, Denise, Paulus, Fabien, Vacchiani-Marcuzzo, Ce evolutionary theory for interpreting urban scaling laws. Cybergeo. http:// dx.doi.org/10.4000/cybergeo.2519 (July). Ragin, Charles C., 2008. Redesigning Social Inquiry: Fuzzy Sets and beyond. University of Chicago Press, Chicago, London. Redfearn, Christian L., 2009. Persistence in urban form: the long-run durability of employment centers in metropolitan areas. Reg. Sci. Urban Econ. 39 (2), 224e232. http://dx.doi.org/10.1016/j.regsciurbeco.2008.09.002. Reinders, A.H.M.E., Blok, Kornelis, Vringer, K., 2003. The direct and indirect energy requirement of households in the European Union. Energy Policy 31 (2), 139e153. http://dx.doi.org/10.1016/S0301-4215(02)00019-8. Rennings, Klaus, 2000. Redefining innovation d eco-innovation research and the contribution from ecological economics. Ecol. Econ. 32 (2), 319e332. http:// dx.doi.org/10.1016/S0921-8009(99)00112-3. Rickwood, Peter, Glazebrook, Garry, Searle, Glen, 2008. Urban structure and energyda review. Urban Policy Res. 26 (1), 57e81. http://dx.doi.org/10.1080/ 08111140701629886. Rodrik, Dani, Subramanian, Arvind, Trebbi, Francesco, 2002. Institutions rules: the primary of institutions over integration and geography in economic development. J. Econ. Growth 9 (2), 131e165. Roelich, Katy, Christof Knoeri, Steinberger, Julia K., Varga, Liz, Blythe, Philip T., Butler, David, Gupta, Rajat, Harrison, Gareth, Martin, Chris, Purnell, Phil, 2014. Towards resource-efficient and service-oriented integrated infrastructure operation. Technol. Forecast. Soc. Change 92, 40e52. http://dx.doi.org/10.1016/ j.techfore.2014.11.008. Elsevier B.V. le my, Marc, 2012. A longRoth, Camille, Moon Kang, Soong, Batty, Michael, Barthe time limit for world subway networks. J. R. Soc. Interface 9 (75), 2540e2550. http://dx.doi.org/10.1098/rsif.2012.0259. , Van Asselt, Marjolein, 2001. More evolution than revRotmans, Jan, Kemp, Rene olution: transition management in public policy. Foresight 3 (1), 15e31. http:// dx.doi.org/10.1108/14636680110803003.  S., Batty, Michael, Eugene Rozenfeld, Hern an D., Rybski, Diego, Andrade, Jose Stanley, H., Makse, Hern an A., 2008. Laws of population growth. Proc. Natl. Acad. Sci. U. S. A. 105 (48), 18702e18707. http://dx.doi.org/10.1073/ pnas.0807435105. Salat, Serge, 2009. Energy loads, CO2-emissions and building stocks: morphologies, typologies, energy systems and behaviour. Build. Res. Inf. http://dx.doi.org/ 10.1080/09613210903162126. Salat, Serge, Bourdic, Loeiz, 2011. Power laws for energy efficient and resilient cities. Procedia Eng. 21, 1193e1198. http://dx.doi.org/10.1016/j.proeng.2011.11.2130 (January). Samaniego, Horacio, Moses, Melanie E., 2008. Cities as organisms: allometric scaling of urban road networks. J. Transp. Land Use 1 (1), 29e39. Sartori, Igor, Bergsdal, Håvard, Müller, Daniel B., Brattebø, Helge, 2008. Towards modelling of construction, renovation and demolition activities: Norway's Dwelling stock, 1900e2100. Build. Res. Inf. 36 (5), 412e425. http://dx.doi.org/ 10.1080/09613210802184312. Scharpf, Fritz W., 1997. Games Real Actors Play: Actor-centered Institutionalism in Policy Research. In: Theoretical Lenses on Public Policy. Westview Press, Boulder. Scharpf, Fritz W., 2002. Kontingente Generalisierung in Der Politikforschung. In: €higkeit Mayntz, Renate (Ed.), Akteure e Mechanismen e Modelle: Zur Theoriefa € ln, pp. 213e235. Makro-sozialer Analysen. Campus Verlag, Ko Schipper, Lee, 2004. International comparisons of energy end use: benefits and risks. Encycl. Energy. Schwartz, Herman, 2010. Property bubbles, housing provision and welfare state reorganization in the United States and the European Union e policy choices and the constitution of the new welfare society. In: Welfare State Reform in the United States and the European Union e Policy Choices and the Constitution of the New Welfare Society. Oxford University Press, New York. Schwarz, Nina, 2010. Urban form revisiteddselecting indicators for characterising European cities. Landsc. Urban Plan. 96 (1), 29e47. http://dx.doi.org/10.1016/j. landurbplan.2010.01.007. Sorensen, Andre, 2015. Taking path dependence seriously: an historical institutionalist research agenda in planning history. Plan. Perspect. 30 (1), 17e38. http://dx.doi.org/10.1080/02665433.2013.874299. Sorrell, Steve, 2007. The Economics of energy service contracts. Energy Policy 35 (1), 507e521. http://dx.doi.org/10.1016/j.enpol.2005.12.009.

4282

F. Schiller / Journal of Cleaner Production 112 (2016) 4273e4282

Sorrell, Steven R., Dimitropoulos, John, 2008. The rebound effect: microeconomic definitions, limitations and extensions. Ecol. Econ. 65 (3), 636e649. http:// dx.doi.org/10.1016/j.ecolecon.2007.08.013. €th, Philipp, Rohracher, Harald, 2014. Conflicting strategies towards sustainable Spa heating at an urban junction of heat infrastructure and building standards. Energy Policy 78, 273e280. http://dx.doi.org/10.1016/j.enpol.2014.12.019. Elsevier. Steadman, P., Evans, S., Batty, Michael, 2009. Wall area, volume and plan depth in the building stock. Build. Res. Inf. 37 (5e6), 455e467. http://dx.doi.org/10.1080/ 09613210903152531. Steinberger, Julia K., Niel, Johan van, Bourg, Dominique, 2009. Profiting from Negawatts: reducing absolute consumption and emissions through a performance-based energy economy. Energy Policy 37 (1), 361e370. http://dx.doi.org/ 10.1016/j.enpol.2008.08.030. Stiglitz, Joseph E., Sen, Amartya K., Fitoussi, Jean-Paul, 2009. Report by the Commission on the Measurement of Economic Performance and Social Progress (Brussels). Tacchella, Andrea, Cristelli, Matthieu, Caldarelli, Guido, Gabrielli, A., Pietronero, L., 2013. Economic complexity: conceptual grounding of a new metrics for global competitiveness. J. Econ. Dyn. Control 37 (8), 1683e1691. http://dx.doi.org/ 10.1016/j.jedc.2013.04.006. Elsevier. Tiwari, Rajnish, Herstatt, Cornelius, 2012. Frugal innovation: a global networks' perspective. Die Unternehmung (Swiss J. Bus. Res. Pract. 66 (3), 245e274.

s, Steniec, Maja, 2009. Cutting Ürge-Vorsatz, Diana, Petrichenko, Ksenia, Antal, Miklo the energy use of buildings : how deep can the planet go ?. In: ECEE 2009 Summer Study Proceedings, pp. 1543e1553. Van den, Bergh, Jeroen, C.J.M., 2007. Evolutionary thinking in environmental economics. J. Evol. Econ. 17, 521e549.  ska, Karolina, 2010. Transitions: taking complexity Vasileiadou, Eleftheria, Safarzyn seriously. Futures 42 (10), 1176e1186. http://dx.doi.org/10.1016/j.futures.2010. 07.001. Wiedenhofer, Dominik, Lenzen, Manfred, Steinberger, Julia K., 2013. Energy requirements of consumption: urban form, climatic and socio-economic factors, rebounds and their policy implications. Energy Policy 63, 696e707. http:// dx.doi.org/10.1016/j.enpol.2013.07.035 (December). Elsevier. Wilson, Charlie, Grubler, Arnulf, Gallagher, Kelly S., Nemet, Gregory F., 2012. Marginalization of end-use technologies in energy innovation for climate protection. Nat. Clim. Change 2 (11), 780e788. http://dx.doi.org/10.1038/ nclimate1576. Wu, Fulong, 2014. Commodification and housing market cycles in Chinese cities. Int. J. Hous. Policy 1e21. http://dx.doi.org/10.1080/14616718.2014.925255 no. June 2015. € hmer, Stefan, Fischedick, Manfred, Gro €ne, MarieYetano Roche, María, Lechtenbo Christine, Xia, Chun, Dienst, Carmen, 2013. Concepts and methodologies for measuring the sustainability of cities. Annu. Rev. Environ. Resour. 39, 2.1e2.29. http://dx.doi.org/10.1146/annurev-environ-012913-101223.