WPS3712 Governance and the City: An Empirical Exploration into Global Determinants of Urban Performance Daniel Kaufmann, Frannie Léautier, and Massimo Mastruzzi 1 Abstract We contribute to the field of urban governance and globalization through an empirically-based exploration of determinants of performance of cities. We construct a preliminary worldwide database for cities, containing variables and indicators of globalization (at the country and city level), city governance, city performance (access and quality of infrastructure service delivery), as well as other relevant city characteristics. This city database, encompassing hundreds of cities worldwide, integrates existing data with new data gathered for this research project. We present a very simple conceptual framework and a set of hypotheses, and then test them econometrically. The findings suggest that good governance and globalization (at both the country as well as at the city level) do matter for city-level performance in terms of access and quality of delivery of infrastructure services. We also find that globalization and good city governance are significantly related with each other. There appear to be dynamic pressures from globalization and accountability that result in better performance at the city level. Furthermore, the evidence suggests that there are particular and complex interactions between technology choices, governance and city performance, as well as evidence of a non-linear (u-shaped) relationship between city size and performance, challenging the view that very large cities necessarily exhibit lower performance and pointing to potential agglomeration economies. Our framework also suggests a way of bridging two seemingly competing strands of the literature, namely viewing the city as a place or as an outcome. We conclude pointing to the need for expanding the database and the econometric framework, as well as to more general future research directions and policy implications emerging from this initial empirical investigation in the field of governance and the city. World Bank Policy Research Working Paper 3712, September 2005 The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent. Policy Research Working Papers are available online at http://econ.worldbank.org. 1The authors are at the World Bank Institute (WBI). This paper is a preliminary draft, for comment and discussion, and will undergo revisions. We benefited from the valuable assistance of Erin Hoffmann and Fatima Sheikh. The views, data, and research findings aim to further debate and analysis. The margins of error in any governance, institutional quality and urban dataset imply that interpretative caution is warranted in general, and in particular argue against inferring seemingly precise city or country rankings from the data. For details on data, visit: http://www.worldbank.org/wbi/governance/. 1 Introduction: The empirical challenge Does globalization at the city and country level matter for sound urban governance? Do sound urban governance and globalization affect city performance? While studies on the first question are particularly scarce, some insights into the second question can be found in the existing literature. In particular, there has been an explosion in the literature on globalization and urban management. A simple search on the internet reveals thousands of citations and references on the topic. These range from theoretical explorations such as the work of the "Globalization and World Cities Group" based in Loughborough University, England (GaWC) but also include newsworthy pieces that refer to specific decisions made by cities to position themselves in a global world. The work of Saskia Sassen dominates the citations on the subject of globalization and cities, and a number of sites dedicate themselves to organizing the literature and the data on globalization with some reference to cities, such as the work of the "Transnationalism Project" of the University of Chicago. The most used terminology to refer to globalization of cities is the term "networked", followed by the word "flows", both of which appear in the vast majority of the internet searches. However, the reference to globalization in the sense of world cities appears in only 3% of the references--and even much sparser are references to governance2. The virtual absence of data and empirically based treatment of the subject is particularly striking. A search of the empirical literature on globalization and urban governance suggests a significant gap, largely due to limited comparable cross-city data. This is particularly the case on governance and globalization issues, where an internet-based search elicits very few empirically-based entries. Researchers have in fact lamented the difficulty of getting reliable comparative data to analyze the issues related to globalization of cities (see Short et al (1996) and Knox P.L. (2002)). But does the data at our disposal permit even an initial exploration into those issues at the city level? If so, is there significant variation in city-level performance, governance and globalization--even within countries? Are differences across cities significant enough to warrant this line of empirical inquiry? And even if they are, do these differences matter? Within this initial empirical exploration, we try to address these issues. We find that the cross-city variance worldwide is rather substantial; and in fact even within countries such intra-city variance is not small--in general the extent of within-country cross-city variance is above one-half the cross-country variation. Such apparent rich diversity of experiences and performance across cities is worth reviewing, and thus exploring the potential determinants of the variation in performance is warranted. Consequently, in this paper we try to contribute to the field of urban governance and globalization through an empirically-based exploration of some key determinants of the performance of cities. This empirical inquiry is made possible through the construction of a worldwide database for cities that contains variables and indicators of globalization (at 2Due to the dynamic nature of data entry into the internet, these search percentages are valid at the time of the search, which was April, 2004. 2 the country and city level), of city governance, and of city performance (access and quality of service delivery). This city database integrates existing data with new data gathered specifically for this research project.3 The paper is organized as follows. Section 2 provides a discussion of conceptual and definitional issues on city globalization, providing a partial survey of the literature. In Section 3 we discuss the issue of urban governance in the context of globalization and review some of the relevant literature to capture the dynamics of city performance. Section 4 provides simplified hypotheses on the globalization and governance nexus in explaining city-level performance in terms of service delivery, includes a working definition of urban governance, and presents a reduced-form model specification to test key hypotheses. Section 4 also describes the extended city level dataset we have compiled, comprising data on 412 cities in 134 countries, from which we make use of over 35 variables.4 In this section we also present in brief the results, both in terms of simple summary charts synthesizing the uncontrolled correlations, as well as the econometric specifications and the synthesis of the econometric results of the determinants of city performance. Given the extensive set of variables in the dataset, and multiple econometric specifications, we present synthesis tables and graphs summarizing the main econometric results. In Section 5 we present the results of the preliminary empirical investigation of the potential determinants of city governance. Future research questions are dealt with in the concluding section, where we also present emerging policy implications from this analysis. 2 City Performance in the Context of Globalization There are two main threads in the literature that treat the subject of globalization of cities, which mostly refer to the term "world city" to describe the phenomenon. We take the theoretical underpinnings of this definition from the work of Doel and Hubbard (2002), who clarify key definitions and develop the conceptualization of globalization of cities. The two conceptualizations of world cities that we draw from include the following definitions: 3 See Annex D for the list of cities in the current version of the database. We use various sources of data for this analysis, detailed in the Annex. First, a comprehensive city level data set is taken from the UN Urban Observatory which has data on 232 cities in 114 countries for the year 1998. We refer to this database as the UN database. We also use data from the GaWC research which we call the Taylor database, which includes a roster of 265 cities constructed from data on the number of firms (advertising, accounting, banking, and law) and number of practitioners/offices in 265 cities for the year 2000. We construct a special database which we refer to as the KLM (Kaufmann, Léautier, Mastruzzi) database which includes data in 412 cities from 134 countries for the year 2003, based in large part from a worldwide enterprise survey (EOS). This database consists of 36 variables of governance as well as city and country economic data. 4The expanded city level dataset includes over a hundred variables. However, since part of this consolidated dataset is aggregated from different existing data sources comprising a different set of cities, the consolidated dataset contains many missing values. One of the objectives for the future is to build up this city level dataset and fill in these data gaps. 3 1. Globalized City as a place: a city is a location within the world, defined with precise city boundaries, but plugged into a global space of flows such as foreign direct investment that comes along with the decisions companies make to locate in particular cities, as well as the flow of people who come to work for these companies in terms of personnel in the country or key people visiting for business purposes, and in the flow of goods and services that are produced by these companies as they decide to locate in a particular city. In such a definition, a global city would get value out of its indigenous characteristics such as its past investments in critical infrastructure and the quality of its institutions. But in particular such a city would also get benefits from the decisions of companies to locate in that city and conduct economic activities such as banking, finance, accounting, legal services, or advertising. Urban politicians and city managers would be concerned about bringing local value to their citizens, reacting to events that impact the city on a global scale, and creating strategies to adapt the city to best advantage. They would therefore be seeking out to attract such companies or make sure they stay in the city, providing the key inputs and incentives such firms and their related personnel would need. 2. Globalized City as a sustained achievement of performances whether measured as the quality of services a city provides to its citizens (access to telephone services, water, sewerage, or electricity), the reliability of such services over time (such as measured by the quality of infrastructure services), and the degree to which a city involves its citizens in decision-making, is responsive to their demands, and the way it is governed in general. Such a city would have relationships with other cities and key players in the global economy such as people, firms, and organizations that are distributed, with varying degrees of clustering and dispersion across the world-city network. Urban politicians would strive to sustain good performance, and would interact with citizens and firms to make choices on investment at the city level as well as ensure the delivery of key services. How such politicians govern is as important as what services they deliver in such a definition. The first definition of a globalized city refers to cities which have a leadership role at the international scale, are externally oriented towards a global economy, have a high ranking in the world's urban hierarchy, or serve as a major gateway for migration. Such cities have existed as far back as the sixteenth century. Knox (2002) highlights the fact that there has been a shifting of the position that cities have had in the world, when you look at how they have organized trade or geopolitical strategies beyond their boundaries. Cities like London, Amsterdam, Antwerp, Genoa, Lisbon, and Venice were key global players in the seventeenth century. In the eighteenth century cities like Paris, Rome, and Vienna became main players in the world, while Antwerp and Genoa lost influence. In the nineteenth century Venice lost some of its influence while cities like Berlin, New York, and St. Petersburg joined the league of world cities. The shifting in and out of global influence, demonstrated by the example of a city such as Genoa, however, supports the second definition of a globalized city, as an outcome of decisions made by firms and individuals, 4 which we wish to explore. We would like to investigate empirically what determines the staying power of cities in their performance on a global scale, and whether governance has anything to do it. Empirical analyses that have been done using the first definition of a globalized city--a city as a place interacting with the global space--have included a number of indicators such as the seven defined by Friedman (1986): (a) the financial assets of a city such as its capital and equity markets; (b) a city's endowment of key infrastructure which firms and citizens would respond to in terms of their decisions to locate there, exit to alternative options, or push to get governance reforms that would allow them to receive improved services;5 (c) the size of the city in terms of its population with the assumption that benefits of agglomeration are important; (d) the availability and concentration of key business services and advanced producer services such as credit rating, risk management, and multi- jurisdictional law; (e) the importance of a city's manufacturing output; (f) how many Transnational Corporations have located their headquarters in a city; and (g) the presence of international institutions in the city. Other authors have used indicators such as the presence of internet domain names (Townsend, 2001); the extent of public-private partnerships (Kresl, 1995); measures of cultural vitality (Smith and Timberlake, 1995); the proportion of foreign residents in the city (Doel and Hubbard, 2002); and the numbers of service sector employees in the city (Doel and Hubbard, 2002). Very few authors have carried out empirical investigation using the second definition of globalized cities, which considers scale (size of a city, size of its economy including the size of private sector activity) or services (access to infrastructure and social services) as something produced rather than given, requiring cities to constantly perform to remain influential in a global world. New literature which focuses on connectivity of cities within a network includes that of Sassen, (2000), Castells, (2000), and Doel and Hubbard (2002). Uncovering the variables that drive the dynamics of city growth and performance interests us, and we particularly look at the role of governance, using a set of empirical tests. From the data we have been able to gather from various sources, we can look at the effect of both the scale issue as well as the quality of services and investigate the Friedman (1986) and Townsend (2001) hypotheses.6 5Kyu Sik Lee and Alex Anas in "The Benefits of Alternative Power Tariffs for Nigeria and Indonesia" (1996), demonstrate how companies locate in a city and if they do not find the services they need either invest in their own generation, thereby exiting the system, or put pressure to get better services, with potential improvements in local governance, or relocate to cities with better services. 6The Taylor database gives us data on financial/banking companies located in a city, as well as data on the availability of key companies in law, accounting, and advertising, all measures of indicators (a), (d), and (f) in the Friedman (1986) hypothesis. From the UN and KLM databases we get measures of key infrastructure endowments (indicator (b)). City size (indicator (c)) is obtained from the website: http://www.citypopulation.de. We use the Kearney composite measure of globalization for measure (e) and have collected web statistics that show whether a city has a website and if it posts budget information or business information there to capture the Townsend (2001) indicators. The only measure we do not use in our analysis is the presence of international institutions in the city. 5 In latter sections, we develop and implement a simple framework to study empirically the potential determinants of urban performance, within which globalization is hypothesized to be an important factor. From that perspective, we are to view the two definitions of a globalized city as complementary rather than as alternative hypotheses. Simply, the first definition above of a globalized city (particularly regarding `companies locate in their cities and conduct economic activities such as banking, finance, accounting, legal services, or advertising') is to be seen as a potential input to sound urban performance, while the second definition (`achievement of performances', including service delivery and quality of governance) is to be seen as the output or outcome of various inputs (including whether the city is global or not). In this respect, to avoid a quasi-tautology, we move away from equating the second view with a global city, instead regarding it simply as urban performance outcomes.7 We therefore posit the simple view that the two definitions of a global city are not necessarily substitutes or competing with each other, but potentially complementary, the first notion potentially being a determinant of city performance. Such a construction builds on the Tiebout (1956) hypothesis of "voting with your feet" and allows us to accept endogeneity. A city with good performance in terms of amenities it provides to its citizens (schools, health services, parks) attracts more migrant residents who "vote" to locate there. Such cities would therefore grow in size and if size of a city gives it advantage in a global world, the issue of "performance" may have an endogenous component, similarly with respect to location of firms. The dynamics of such a construct dictate that well performing cities would be well managed and hence even better performing. 3 Urban Governance in the Context of Globalization 3.1 Governance and Globalization at the Country level: basic definitions In the recent past, much work has been done on globalization and on governance at the country level. While this is not the main focus of the paper (and thus we do not provide a literature review), it is of relevance to briefly review the notions of governance and globalization at the country level for two reasons. First, as background to these same twin notions, but at the city level, which will be covered in more detail. And second, because country level variables may also obviously matter significantly as determinants of city- level performance. 7 In the `Popperian falsifiability' sense, taking the second definition of global city as is would not be very useful to test the hypothesis that global cities perform better. Thus, we obviate such a problem by interpreting the first definition as a potential input and the second definition as an outcome (to be empirically tested). 6 We have defined country-level governance as the exercise of authority through formal and informal traditions and institutions for the common good, thus encompassing: (1) the process of selecting, monitoring, and replacing governments; (2) the capacity to formulate and implement sound policies and deliver public services, and (3) the respect of citizens and the state for the institutions that govern economic and social interactions among them. The three dimensions in this definition of governance are then further unbundled to comprise two measurable concepts per dimension, for a total of six components: (1) voice and external accountability (i.e., the government's preparedness to be externally accountable through citizen feedback and democratic institutions, and a competitive press); and also (2) political stability and lack of violence, crime, and terrorism. Then (3) government effectiveness (including quality of policymaking, bureaucracy, and public service delivery); (4) lack of regulatory burden; and, finally, (5) rule of law (protection of property rights, judiciary independence, and so on); and (6) control of corruption. These components are subject to empirical measurement at the country level, for which a worldwide dataset has been constructed.8 3.2 Defining Urban Governance Despite the fewer numbers of citations that can be found in the literature search which give access to specific empirical data sets that are comparable, there is no lack of citations on the definition and use of the term "urban governance". In simple terms, UNESCO defines urban governance as the processes that steer and take into account the various links between stakeholders, local authorities and citizens. It involves bottom-up and top-down strategies to favor active participation of communities concerned, negotiation among actors, transparent decision-making mechanisms, and innovation in strategies of urban management policies.9 More comprehensively, one can find definitions of urban governance, as well as descriptions of potential indicators of urban governance in the work of Mehta (1998) and the UNCHS (1999). Mehta (1998) looks at urban governance through a set of attributes. He introduces the attribute of accountability which he suggests is derived from how cities manage their finances, communicate on use of funds and achievements to their citizens, and adhere to legal requirements and administrative policies. Embedded in Mehta's concept of accountability is the question of responsiveness, which includes the ability of a decentralized entity to determine and respond to the needs of its constituents. In doing so, city officials need to have processes of citizen participation, and a system for monitoring and evaluation, as well as reporting on results achieved. These three measures are critical if one is using the definition of a city as a place, where it is performing with respect to local issues, and managing the consequences of globalization locally. If we consider the other definition of a globalized city, as an outcome of globalization but also shaping its role within a globalized world then we need to look at dynamic indicators 8 See Kaufmann, Kraay, Mastruzzi, Governance Matters III: Governance Indicators for 1996­2002, World Bank Policy Research Working Paper 3106. 9http://www.unesco.org/most/most2.htm 7 of governance. In addition to the static attributes of accountability, responsiveness, and participation, Mehta also includes a dynamic concept of urban governance which he calls management innovation, which measures the degree to which urban managers have been able to successfully implement changes in their systems of administration to achieve superior performance. Mehta introduces three measures which relate to the ability of cities to transform global opportunities to local value. These include measures such as public- private partnerships, local government-citizen interaction, and networking. The ability of cities to engage in public-private partnerships is seen as a key capability for staying power in a global scene by a number of authors (see Doel and Hubbard (2002) and Knox (2002)). Mehta suggests measures such as presence of business sector initiatives to improve efficiency of local government functioning, as well as the degree to which implementation of policies and incentive schemes exist to encourage private sector participation in development. Linking back to the literature on global cities, such indicators would be relevant due to the consideration that strategic decisions by Transnational Corporations to operate in a given city are based on their determination of the advantages of locating there and would therefore depend on the quality of engagement with the local government in key areas as well as the local investment climate they would face. Having city governments interact with citizens and non-governmental organizations, opens up the space for introducing the global civil society in decision-making at the local level. Cities with a large diaspora that is actively involved in local decisions are one example but so are the networks of civil society that interact on a global scale with respect to local issues (see Sassen (2002)). Mehta's definition of urban governance also allows us to introduce the concept of networking of one city with other cities or with key actors such as firms, labor unions and business associations, but also other states. He suggests a number of indicators of networking such as the number of inter-city, regional, and international networks as well as the extent of technological interchange and collaboration. Other measures suggested are the exchange between cities of expertise and training. The recent example of the city of Rome sending experts to Kigali, indicates the extent to which this indicator is relevant. Urban governance as an outcome that is visible to a citizen is a key feature that allows empirical tests of the city as a place or as a sustained achievement of performances. There are very few existing indicators that can be used to capture dynamic changes in governance at the city level. We draw on measures defined by authors such as Mehta (1998) and the UNCHS (1999) which include: (1) consumer satisfaction (survey/complaints); (2) openness of procedures for contracts/tenders for municipal services; (3) percentage of population served by services; and (4) access of the public to stages of the policy cycle10. 10The UNCHS Expert Group Meeting on Urban Poverty and Governance Indicators, 29 April to 1 May 1999, Nairobi, Kenya highlights the top twelve indicators of urban governance which include: (1) consumer satisfaction; (2) openness of procedures for contracts/tenders for municipal services; (3) equity in the tax system; (4) sources of local government funding (taxes, user charges, borrowing, central government, international aid); (5) percentage of population served by services; (6) access of public to stages of policy 8 With these measures, we construct a vector of urban governance indicators which we use to test the impact of urban governance on city performance. The vector consists of measures of voice and participation and transparency and accountability which we take from the UN database. We also use measures like illegal financing, state capture, bribery in utility, and bribery in judiciary from the Kaufmann, Léautier, and Mastruzzi (KLM) database. Finally, we construct a dynamic indicator of city transparency which uses proxies like, whether a city has a web page, what information is included in the web page, whether the city budget is publicly available on the web or whether companies can register on the web. Note that this indicator captures similar aspects as the one described earlier in Section 2 and attributed to Townsend (2001). 3.3 Defining the Impact of Urban Governance on Globalized Cities A `globalized city' as an enduring performance is a concept worth testing empirically. We can ask "what aspects of governance give cities this capability to constantly translate the global complexities into an advantage to them which can be seen in the way they treat their citizens (access to services, quality of services); attract firms to invest (public-private partnerships, FDI, firm location); or measures such as maintaining economic growth over several years and increasing the average per capita income of a city's citizens. Doel and Hubbard (2002) suggest that a globalized city is always a work in progress being shaped and managed by firms, states, sectors, and cities. Thus a city which has good governance should be able to perform well and remain in the league of important global cities as well as deliver value to its citizens. The literature on globalization and urban governance has two schools of thought on the importance of globalization for the performance of cities. There are those who argue that cities can be well performing if they have sufficient spatial agglomeration of know-how and capacity in the city and that a city's international competitiveness is quite different from the concept of an international city (Kresl, 1995 p.54). Their superior performance in delivering local value to their citizens can come from their interaction with the rural hinterland, as well as with cities within the same country. Thus, well governed cities that are not highly globalized would fit in this category, where we would expect that such cities would be well performing because they are well governed rather than because they are globalized. Others argue that successful cities will be those that are considered world cities, that deliver local value through interpreting and tapping into global opportunities (Doel and Hubbard, 2002). Such cities would be attending to their own position in a global space of flows, and will be well governed not only in local terms but also in global terms. Their ability to remain well performing cities in a globalized world is determined by their constant cycle; (7) fairness in enforcing laws; (8) incorporation of excluded groups in the consultation process; (9) clarity of procedures and regulations and responsibilities; (10) existing participatory processes; (11) freedom of the media and existence of local media; and (12) autonomy of financial resources. 9 attention to good governance. De Long and Shleifer (1992) have used such a construct to look at the differential performance of cities that were ruled by princes (little voice and participation by citizens and merchants) and those ruled by merchants (more voice for citizens and firms). Because they have a long time series (from the 2nd to the 19th century, they are able to test the relationship between good governance (measured by voice) and city performance (measured by city population/size). 3.4 Capturing the Staying Power of Cities in a Globalized World Another important consideration is the one of scale. If scale is something produced rather than given (as pointed out by Doel and Hubbard (2002)), then the size of cities is an outcome of contingent encounters between cities, firms, individuals, and sectors (business associations, labor unions etc). This means that in due course (when sufficient time series data becomes available) one should also aim to gather "emergent governance indicators" which would allow us to look at the changes in governance over time as a result of globalization.11 We use a function that allows us to include variables at both city and country-level globalization as an input to delivering service performance within a city, as well as globalization as an outcome of good governance. In this manner, governance achievements provide the environment in which cities can perform better and hence tap into global opportunities. 3.5 Capturing the Dynamics of Globalization Measures that are important to capture both the dynamics of globalization as well as the staying power of cities include concepts of connectivity, flow and traffic which can be measured by indicators defined by Doel and Hubbard (2002) such as volume and direction of flow of people (via international migration); flow of capital (via international banking); flow of products (via import and export); flow of ideas (via broadcasting and the media); and flow of pollution (via dumping policies) or even crime (drugs, human traffic, money laundering). These measures would change over time and it is the direction of change as well as continuity of flows at given levels that define the staying power of cities. This definition posits the capability of cities to translate complex relationships at various levels into an advantage and this capacity to translate is rarer than the capacity to command or control. In its most concise definition, globalization "simply refers to the complex of forces that trend toward a single world society. Among these forces are mass communications, commerce, increased ease of travel, the internet, popular culture, and the increasingly 11This would permit, inter alia, to answer questions such as: do cities/countries that have an improving trend in governance also continue to perform better in terms of their degree of globalization and their ability to translate global opportunities into local value or both?, and would the scale of a city be a good measure of such a phenomenon? 10 widespread use of English as an international language."12 Globalization has most often been analyzed at the country level. Sources of data and analysis judging country level globalization as a dynamic concept are rare. Some authors use rankings of countries on a number of dimensions of globalization, such as the work of A.T. Kearney (Foreign Policy, 2002) who use a composite index of globalization. This index builds on indicators of trade, finance, personal contact, and information exchange. Using the Kearney data base, we investigate the issue of country level globalization and its impact on city performance, with cities judged on the ability to sustain a track record of services to their citizens. For globalization at the city level, we utilize data from the GaWC research (Taylor, Walker, 2000), which has city level data constructed from a superposition of networks of accountancy, advertising, banking/finance, insurance, law, and management consultancy firms--the business services indicators of Friedman (1986). This data base demonstrates the "unevenness of globalization" when one looks at the concentration of cities by region for example. We refer to this as the Taylor database. With such data, we can pursue the question of why some cities are left behind in globalization. 4 Determinants of City Performance: Key Hypothesis Tests and Simple Model Specification Because of the paucity of comparable data on cities worldwide, we need to be nimble on the model specification we use. We search for a specification that allows us to capture the dynamics of globalization, governance and performance of cities, as well as define specific hypotheses to be tested. 4.1 A Simple Framework We specifically consider the role that governance plays at different levels of globalization of cities. The schematic chart below allows us to pursue this analysis empirically. We distinguish between three dimensions; two of which are `inputs' (quality of governance and extent of city globalization, in the vertical and horizontal axis, respectively), and one an outcome (service delivery/city performance, in the `virtual' third dimension in the chart, illustrated in the chart by the 45 degree diagonal line). With such a schematic chart we define the key testable hypotheses on the effects of globalization and governance on city level performance. 12http://www.progressiveliving.org 11 Chart 1: City governance and extent of globalization as potential determinants of city performance High Performance Well Governed Well Governed Local City Global City I IV Governance City of Poorly Governed Poorly Governed Local City Global City Quality II III Low Performance Extent of City Globalization Hypotheses to be tested: Governance and globalization interact, and do matter for city performance. Hypothesis 1: Governance matters. We first ask: can a city be well-performing regardless of whether it is a `local' or `global' city, significantly driven by its good governance? And, more specifically, would a `local' city necessarily have to exhibit good governance to `compensate' and be able perform, i.e. could a city be located in quadrant II and still likely to be well performing?. We posit that, governance matters significantly for city performance, controlling for its level of globalization. While a `local' city could exceptionally exhibit good performance, this is highly unlikely unless the city exhibits relatively high levels of governance. Governance in this specification is an endowment that allows a city to perform well either locally or globally. In other words, we can test if a well performing city, whether local or global, has good governance, hence measuring the independent value added of governance to a local city or to a global city. The added value of governance for a local city is then the difference between the performance of a city in Quadrant I compared to one in Quadrant II, 12 while the added value of governance for a global city is the differential performance of a city in Quadrant IV compared to one in Quadrant III. Hypothesis 2: Globalization matters. We can then also test whether a global city is better performing than a local city, for a given quality of governance (whether for a given level of good governance, i.e. comparing quadrants I and IV, or poor governance, as in quadrants II vs. III, which would measure the independent effect of globalization on city performance). From the literature, it is expected that a city that is global is able to attain such global status because of specific actions taken by its political leadership or by its citizens, as well as the firms that locate within such a city. We would expect citizens of global cities to have better quality and access to services such as water, sewerage, electricity and telephones than local cities. Hypothesis 3: Globalization and governance interact positively. While in principle it is conceivable that a city could compensate for its `local' (vs. global) character by exhibiting good governance, in practice the dynamics of good governance may point to a linkage between globalization and governance: those cities that globalize, bring about the checks and balances of competition (say by the discipline imposed by FDI and credit markets), as well as the techniques, that are likely to positively impact on city governance. In turn, improved governance, involving transparency, control of corruption and property rights protection, may attract further demand and pressures for a local city to become more global. 13 Thus, a dynamic virtuous circle may be at play in which globalization and governance interactive positively to further improve performance. In particular, we will empirically investigate whether good governance is less likely in local cities, and thus whether becoming a global city, and a globalized country, helps improve city governance. Thus, in terms of schematic chart 1, we would expect that the worst performance would take place in cities in quadrant II (poorly governed local city). If it attempts to attain higher levels of globalization, that is unlikely to take place without pressures to take measures to also improve governance. Thus, a rapid and fully horizontal move out of quadrant II to quadrant IV, or a rapid and fully vertical move to quadrant I, are less likely than either the city staying within a low equilibrium trap, or moving out of such a low level quadrant II towards the good performance equilibrium in quadrant III via the diagonal route. 13 Taylor defines alpha cities as those with global significance in four key global city functions (international accountancy, advertising, banking, and legal services). A beta city is one that has global significance in three of the four key world-city (global city) functions. A gamma city is one that has global significance in two of the four key global city functions. Using the network aspects of the GaWC is a key advantage that we consider in looking at the issue of governance, asking questions like: "Is it that less globalized cities are relatively less well governed and that results in them trailing even further behind?" We group cities into two categories of "global" and "local" to present some of the results in charts that visually capture the main empirical findings. 13 4.2 Simple Model Specification. A very simple model specification to analyze what emerges from these hypotheses tests is advanced as follows: (1) Y = f(X, Z, c, C) + Y = city performance as measured by a vector of qualitative service variables which include access to water supply, sewerage, electricity, and telephones from the UN dataset, as well as the overall quality of infrastructure services and electricity supply, access to telephones and/or cell phones and access to internet in schools, using the EOS database; X = measure of globalization at the city level, which we specify using the Taylor dataset as the total number of offices of major international advertising, accounting, and financial firms in a city; Z = vector of governance indicators, which includes control of corruption, bribery in utility, and state capture, largely drawn from the EOS database; c = other city characteristics, such as city size, whether it is the capital city, and/or a port, as well as availability of websites and availability of information on city budget and business regulations in such websites; C=country characteristics, such as income level, size, extent of urbanization, and degree of globalization at the country level, and, = error term. 4.3 Data description and Mapping of Variables The KLM database, which integrates existing data with newly collected data and indicators, covering a total of 412 cities in 134 countries, was constructed by drawing from a variety of sources. Each variable we utilize in this paper is listed in Table 1 below. From the UN Observatory (1998), we downloaded 4 service access/performance variables, covering about 80 cities in 60 countries. From the enterprise-specific EOS survey database (2003), we construct city-specific averages for 12 governance indicators, as well as 5 service access/performance variables, covering 271 cities in 101 countries (see Table 1 below for details). For instance, two governance-related variables for which we constructed city-wide averages, were frequency of bribery in utility, and the extent of state capture (from firms' reports on the distortionary impact on competition of illegal payments made by certain firms) to influence government policies, laws or regulations. From the Taylor database, we downloaded the number of offices of major advertising, accounting and financial firms in each city and we constructed a "global city" variable by adding up the three variables, covering a total of 261 cities in 114 countries. 14 In addition, we also constructed indicators through internet searches, for particular city attributes, such as city population, existence of city websites and availability of data on budget and business regulations in such websites. We completed this list of city attributes by including two dummies for whether the city is the country's capital and/or a port. At the country level, we draw from the variable "Control of Corruption" from the Worldwide Aggregate Governance Indicators discussed above, where corruption (conventionally defined as the exercise of public power for private gain) is one of its components. The indicator is the aggregate of many individual sources, covering several aspects of corruption, ranging from administrative corruption to "grand corruption" in the political arena and "state capture". We also constructed an income per capita variable (PPP) by drawing from the Heston-Summers database and the CIA World Factbook (2001). For globalization at the country level, we draw from the A.T. Kearney/Foreign Policy Globalization Index. It tracks and assesses changes in four key components of global integration, incorporating such measures as trade and financial flows, movement of people across borders, international telephone traffic, Internet usage, and participation in international treaties and peacekeeping operations. It covers 62 countries, including industrialized and emerging economies.14 In Table 1 below we present the legend of all the variables we utilize in the empirical testing of these hypotheses, and their sources and characteristics. 14 Specifically, in this index, economic integration combines data on trade, foreign direct investment (FDI), and portfolio capital flows, as well as investment income payments and receipts. Personal contact tracks international travel and tourism, international telephone traffic, and cross-border remittances and personal transfers (including worker remittances, compensation to employees, and other person-to-person and nongovernmental transfers). Technological connectivity counts the number of Internet users, Internet hosts, and secure servers through which encrypted transactions are carried out. Finally, political engagement tracks each country's memberships in international organizations, personnel and financial contributions to U.N. Security Council missions, ratification of selected multilateral international treaties, and the amount of governmental transfer payments and receipts. 15 Table 1: Variable Legend and Data Sources Coverage Variable Description / Source Range / direction Country Cities City Variables Access to Water UN Observatory, 1998 % 61 83 Access to Sewerage UN Observatory, 1998 % 57 78 Access to Electricity UN Observatory, 1998 % 61 81 Access to Telephone Lines 1 UN Observatory, 1998 % 51 58 Access to Telephone Lines 2 EOS, 2003 1 (bad) - 7 (good) 101 271 Quality of Infrastructure EOS, 2003 1 (bad) - 7 (good) 101 271 Quality of Electricity EOS, 2003 1 (bad) - 7 (good) 101 271 Access to Cell Phones EOS, 2003 1 (bad) - 7 (good) 101 271 Access to Internet in schools EOS, 2003 1 (bad) - 7 (good) 101 271 Bribery in Utility EOS, 2003 1 (bad) - 7 (good) 101 271 State Capture EOS, 2003 1 (bad) - 7 (good) 101 271 Informal Money Laundering EOS, 2003 1 (bad) - 7 (good) 101 271 Street Crime EOS, 2003 1 (bad) - 7 (good) 101 271 Red Tape Cost of Imports EOS, 2003 1 (good) - 9 (bad) 101 271 Bribery to Affect Laws EOS, 2003 1 (bad) - 7 (good) 101 271 Diversion Public Funds EOS, 2003 1 (bad) - 7 (good) 101 271 Illegal Party Financing EOS, 2003 1 (bad) - 7 (good) 101 271 Bribery in Permits EOS, 2003 1 (bad) - 7 (good) 101 271 Bribery in Tax EOS, 2003 1 (bad) - 7 (good) 101 271 Soundness of Banks EOS, 2003 1 (bad) - 7 (good) 101 271 Trust in Politicians EOS, 2003 1 (bad) - 7 (good) 101 271 Organized Crime EOS, 2003 1 (bad) - 7 (good) 101 271 Quality of Postal System EOS, 2003 1 (bad) - 7 (good) 101 271 Health Access Gap EOS, 2003 1 (bad) - 7 (good) 101 271 Offices of major advertising, Global City financial & accounting firms Hundreds 114 261 (Taylor 2000) Source: www.citypopulation.de, City Population Logs 134 410 2001 Website Dummy City has a website (KLM 2003) 0 - 1 (yes) 133 398 City website has info on how to Business Dummy 0 - 1 (yes) 133 398 start a business (KLM 2003) City website has info on city Budget Dummy 0 - 1 (yes) 133 398 budget (KLM 2003) City has port facilities (KLM Port Dummy 0 - 1 (yes) 134 411 2003) Capital Dummy City is the capital (KLM 2003) 0 - 1 (yes) 134 411 Country Variables Globalization (Kearney) Journal of Foreign Policy, 2002 0 (close) - 1 (open) 59 254 Control of Corruption Governance Indicator, KK 2002 -2.5 (bad) / 2.5 (good) 134 411 Country Population WDI 2003 Logs 134 411 % people not in agriculture (WDI Urbanization % 100 366 2002) Heston-Summers and CIA World Income per Capita Logs 134 411 Factbook, 2001 16 4.4 Determinants of City Performance: Main Empirical Results First we performed a simple univariate, or at most, bivariate, testing of whether globalization and/or governance are highly and significantly related to city performance. In simplified chart form, we first present these results in Figures 1a through 1i, which summarize the initial empirical support for the main hypotheses we test. These figures present the results utilizing only the sample of emerging and transition economies (`non- OECD'), as a simple way of quasi-controlling for large income differentials (indeed, the differences are even larger with the full sample, not depicted in the figures, but shown in subsequent tables below). The figures below present the differential performance that a global (vs. local) city makes on a vector of performance in services such as access to telephones [1]15 (1a), sewerage (1b), water (1c), electricity (1d), overall quality of infrastructure (1e), access to telephones [2] (1f), access to cell phones (1g), access to internet in schools (1h) and quality of electricity supply (1i). Further, by the simple sample segmentation and calculation performed for each panel of Figure 1, we can also depict here the difference that good governance can make. This is done utilizing three measures of governance, of which one is a country average (control of corruption), and the other two are at the city level (bribery in utility, and state capture). We find that governance is significantly associated with performance, both for local and global cities. Well governed cities, whether local or global, perform better than poorly governed ones across all quality and access to service variables. The performance advantage of well governed cities holds across all three measures of governance and all service variables except access to the electricity grid in global cities. Globalization has a more nuanced impact on city performance. We find that on balance global cities do perform better than local cities. However, the differences are not that marked, and they vary depending on the type of service. In particular, the difference between local and global cities is clear for certain services, such as quality of infrastructure, yet non-existent for access to electricity, for instance. The rest of the services rate lies in between such extremes both in terms of magnitude of its impact (often modest) as well as varying levels of significance. Consistent with the depiction of some of the results in the various panels of Figure 1 above, we present the fuller set of average rating results for all relevant variables in the Table 2 below, for each one of the relevant samples (full sample, OECD, and non-OECD). From tables 2a, 2b, and 2c, we can clearly detect a strong positive correlation between city performance as measured by service access and the extent of globalization (both at city and country level), regardless of sample. Similarly, there appears to be a strong positive correlation between city performance/service access and city and country governance. In conclusion, more globalized cities are better performing, and better governed cities are well performing. Cities in well governed countries are also better performing. 15As per table 1, access to telephones [1] is drawn from UN Observatory, 1998 and measures the percentage of the population with access to telephones. Access to telephones [2] is drawn from EOS, 2003 and captures on a scale from 1-7 the availability of new telephone lines for business. 17 F igure 1a A ccess to T elephones (1), C ity G overnance and G lobalization ­ N on O E C D Sam ple 1 0 0 % P o o r G o v ernance G o o d G o v erna nce G lob al C ity ) %(s L ocal C ity ne phoeleTot 5 0 % sseccA 0 % C o ntro l o f B ribe ry in U tility State C o ntro l o f B ribe ry in State C apture C o rruptio n C apture C o rruptio n U tility S ou rces: E O S 2003, U N 1998, K K 2002 & K L M 2004 F ig u r e 1 b A c c e ss to S e w e r a g e , C ity G o v e r n a n c e a n d G lo b a liz a tio n ­ N o n O E C D S a m p le 1 0 0 P o o r G o v ern a n ce G lo b a l C ity G o o d G o v ern a n ce ) %(eg L o ca l C ity era ewSotss cceA 2 0 C o n tro l o f B rib e ry in U tility C o ntro l o f B ribe ry in U tility State C ap tu re C o rruptio n State C apture C o rruptio n S o u r c e s : E O S 2 0 0 3 , U N 1 9 9 8 , K K 2 0 0 2 & K L M 2 0 0 4 F ig u r e 1 c A c c e ss to W a te r , C ity G o v e r n a n c e a n d G lo b a liza tio n ­ N o n O E C D S a m p le P o o r G o v ern a n ce G lob al C ity G o o d G o v ern a n ce 100 L ocal C ity ) %(reta Wots Acces 20 C o ntro l o f B ribe ry in U tility State C apture C o ntro l o f B ribe ry in U tility State C apture C o rruptio n C o rruptio n S o u r ces: E O S 2 0 0 3 , U N 1 9 9 8 , K K 2 0 0 2 & K L M 2 0 0 4 18 F igu re 1d A ccess to E lectricity, C ity G overn ance an d G lob alization ­ N on O E C D Sa m ple Poor G overnance G lobal City G ood G overnance 100 Local City ) %(y citri ectlEots 70 Acces 40 C ontrol of C ontrol of B ribe ry in U tility State C apture C orruption B ribe ry in U tility State C apture C orruption S ou rces: E O S 2003, U N 1998, K K 2002 & K L M 2004 F ig u re 1e Q u ality o f In fra stru ctu re, C ity G o vern a n ce a n d G lo b a liza tio n ­ N o n O E C D S a m p le 6 Poor G overnance H ig h G ood G overnance G lobal City erutcurtsarf Local C ity In 3.5 ofyt aliuQ L ow 1 C ontrol of B ribe ry in U tility State C apture C ontrol of B ribe ry in U tility State C apture C orruption C orruption S o u rces: E O S 2 00 3 , K K 2 0 0 2 & K L M 2 0 04 F ig u r e 1 f A c c ess to T e le p h o n e s (2 ), C ity G o v er n a n c e a n d G lo b a liz a tio n ­ N o n O E C D S a m p le H ig h P o o r G o v ern a n ce 7 G o o d G o v ern a n ce G lo b a l C ity s L ocal C ity ne Li ne phoel 4 TeotsseccA L o w 1 C o ntro l o f B ribe ry in U tility State C apture C o ntro l o f B ribe ry in U tility State C apture C o rruptio n C o rru ptio n S o u r c e s : E O S 2 0 0 3 , K K 2 0 0 2 & K L M 2 0 0 4 19 F igu re 1g A ccess to C ell P h on es, C ity G overn an ce an d G lob alization ­ N on O E C D S a m p le H igh P oor G overnance 7 G ood G overnance G lob al C ity L ocal C ity senohPlelC 4 tos cescA L ow 1 C o ntro l o f B ribe ry in U tility State C apture C o ntro l o f B ribe ry in U tility State C apture C o rruptio n C o rruptio n S ources: E O S 2003, K K 2002 & K L M 2004 F ig u r e 1 h A c c e s s to I n te r n e t in S c h o o ls , C ity G o v e r n a n c e a n d G lo b a liz a tio n ­ N o n O E C D S a m p le H ig h P o o r G o v e r n a n c e 7 G o o d G o v e r n a n c e G lo b a l C ity L o c a l C ity oolshcS intenret 4 Inot sseccA L o w 1 C o n tro l o f B rib e ry in U tility S tate C ap tu re C o n tro l o f B rib e ry in U tility S tate C ap tu re C o rru p tio n C o rru p tio n S o u r c e s: E O S 2 0 0 3 , K K 2 0 0 2 & K L M 2 0 0 4 F igu re 1i Q uality of E lectricity, C ity G overn ance an d G lob alization ­ N on O E C D Sam ple H igh P oor G overnance 7 G ood G overnance G lobal C ity tyci Local C ity ectrilEfoytilauQ 4 L ow 1 C ontrol o f B ribe ry in U tility State C apture C ontrol o f B ribe ry in U tility State C apture C orruption C orruption Sources: E O S 2003, K K 2002 & K L M 2004 20 TABLE 2a: City Performance by Levels of Globalization and Governance ­ Full Sample (412 Cities) Average Ratings GLOBALIZATION GOVERNANCE CITY PERFORMANCE City Country Bribery in State Country Globalization, Globalization, Utility EOS Capture, EOS Control of Taylor 2002 Kearney 2003 2003 Corruption, KK02 Low High Low High Poor Good Poor Good Poor Good Access to Sewerage (%, UN) 62% 94% 75% 99% 55% 91% 60% 89% 62% 95% Access to Water (%, UN) 75% 95% 81% 100% 68% 96% 72% 94% 73% 98% Access to Electricity (%, UN) 90% 94% 90% 100% 82% 95% 83% 95% 85% 100% Access to Telephone lines 1 (%, UN) 46% 80% 60% 95% 37% 79% 41% 80% 40% 89% Access to telephone lines 2 (1-7, EOS) 4.8 6.0 5.4 6.4 4.4 6.2 4.6 6.0 4.6 6.2 Access to Cellular phones (1-7, EOS) 5.7 6.5 6.1 6.7 5.5 6.5 5.6 6.4 5.6 6.5 Access to Internet in schools (1-7, EOS) 2.9 4.7 3.7 5.3 2.9 4.8 3.0 4.7 2.9 5.0 Quality of Infrastructure (1-7, EOS) 3.1 5.0 3.5 5.5 2.9 4.7 2.9 4.7 2.9 5.0 Quality of Electricity (1-7, EOS) 3.9 5.6 4.4 6.2 3.6 5.7 3.8 5.4 3.7 5.8 Overall Number of Cities 132 129 131 123 137 134 136 135 211 201 Note: Each column variable was segmented into two equal samples, below and above the median value of the full sample. TABLE 2b: City Performance by Levels of Globalization and Governance ­ Non OECD Sample (274 Cities) Average Ratings GLOBALIZATION GOVERNANCE CITY PERFORMANCE City Country Bribery in State Country Globalization, Globalization, Utility EOS Capture, EOS Control of Taylor 2002 Kearney 2003 2003 Corruption, KK02 Low High Low High Poor Good Poor Good Poor Good Access to Sewerage (%, UN) 57% 72% 71% 79% 50% 87% 51% 82% 56% 72% Access to Water (%, UN) 74% 79% 78% 85% 64% 93% 66% 88% 70% 81% Access to Electricity (%, UN) 87% 90% 88% 92% 80% 93% 80% 92% 83% 90% Access to Telephone lines 1 (%, UN) 43% 57% 48% 76% 32% 74% 38% 64% 37% 59% Access to telephone lines 2 (1-7, EOS) 4.4 5.1 5.4 5.4 4.2 5.6 4.4 5.4 4.1 5.6 Access to Cellular phones (1-7, EOS) 5.4 6.1 6.2 6.1 5.4 6.2 5.5 6.0 5.4 6.1 Access to Internet in schools (1-7, EOS) 2.4 3.3 3.5 3.9 2.7 4.0 2.8 3.9 2.6 4.1 Quality of Infrastructure (1-7, EOS) 2.6 3.7 3.5 3.4 2.8 3.8 2.7 3.9 2.7 3.8 Quality of Electricity (1-7, EOS) 3.6 4.4 4.2 4.5 3.4 4.8 3.6 4.6 3.4 4.8 Overall Number of Cities 70 65 61 61 102 102 106 98 138 136 Note: Each column variable was segmented into two equal samples, below and above the median value of the full sample. 21 TABLE 2c: City Performance by Levels of Globalization and Governance ­ OECD Sample (138 Cities) Average Ratings GLOBALIZATION GOVERNANCE CITY PERFORMANCE Country City Globalization, Country Bribery in State Control of Taylor 2002 Globalization, Utility EOS Capture, Kearney 2003 EOS 2003 Corruption, KK02 Low High Low High Poor Good Poor Good Poor Good Access to Sewerage (%, UN) 98% 100% 99% 100% 98% 100% 98% 100% 99% 100% Access to Water (%, UN) 99% 100% 100% 100% 100% 100% 100% 100% 99% 100% Access to Electricity (%, UN) 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% Access to Telephone lines 1 (%, UN) 92% 96% 92% 99% 93% 100% 93% 99% 92% 99% Access to telephone lines 2 (1-7, EOS) 5.9 6.6 6.2 6.7 6.2 6.8 6.2 6.7 6.0 6.8 Access to Cellular phones (1-7, EOS) 6.5 6.7 6.5 6.8 6.5 6.8 6.5 6.8 6.4 6.8 Access to Internet in schools (1-7, EOS) 4.8 5.4 4.8 5.9 5.0 5.6 4.8 5.8 4.5 5.9 Quality of Infrastructure (1-7, EOS) 4.8 5.7 5.1 5.9 4.9 6.1 4.9 6.0 4.6 6.1 Quality of Electricity (1-7, EOS) 5.5 6.3 5.8 6.6 5.8 6.6 5.7 6.7 5.6 6.6 Overall Number of Cities 64 62 90 42 34 33 34 33 73 65 Note: Each column variable was segmented into two equal samples, below and above the median value of the full sample. The significance of the results of the link between governance and city performance is also summarized in Table 3 below, which presents both the magnitude (numerical) and significance (asterisks) of the coefficient of a regression of service performance on governance. As we see in the table, it does suggest the extent to which governance is important when utilizing the full sample, as well as when dissecting whether it matters in particular for local and global cities. We note that the results are large and significant for both local and global cities (with very rare exceptions), further supporting the chart results illustrated in Figures 1a through 1i and Tables 2a, 2b and 2c above.16 The results also suggest that these substantial additional benefits of improving governance generally do not differ significantly depending on whether the city is of a global or local nature (as indicated in the table by small magnitudes and lack of significance of the `difference' row for each service). The exception is the case of control of corruption and access to water and to telephones, which seem more important for local cities than global cities, and bribery in utility and quality of infrastructure and access to internet at schools, which is more important for global cities. These results may hint at the possibility of differential weights on the type of service that matters for local versus global cities. The results for sewerage are also quite high, and could be due to other efforts. More details on the technological and network aspects of services can be found in section 6. 16 Telephone access in Tables 2a and 2b seems to show overall a higher difference than other services. This may be because telecommunications, which is also a voice enhancing tool, is rather important--a more globalized city with better telecommunications services may offer better opportunities for voice, both local and international, and hence derive more transparency. Such a result has been envisaged by some researchers (see Sassen 2002). 22 TABLE 3: Governance Effect on Service Performance/Access for Global and Local Cities ACCESS TO WATER (%) SAMPLE Bribery in Utility State Capture Control of Corruption Full City Sample 0.11*** 0.10*** 0.11*** Global City 0.10* 0.04 0.06* Local City 0.09** 0.10** 0.15*** Differential 0.01 -0.06 -0.09* ACCESS TO SEWERAGE (%) Bribery in Utility State Capture Control of Corruption Full City Sample 0.15*** 0.13*** 0.16*** Global City 0.13 0.07 0.08* Local City 0.13** 0.11 0.21*** Differential 0.00 -0.04 -0.13 ACCESS TO ELECTRICITY (%) Bribery in Utility State Capture Control of Corruption Full City Sample 0.05** 0.05** 0.05*** Global City 0.05 0.06* 0.06** Local City 0.05* 0.07* 0.07** Differential 0.00 -0.01 -0.01 ACCESS TO TELEPHONES [1] (%) Bribery in Utility State Capture Control of Corruption Full City Sample 0.19*** 0.15*** 0.24*** Global City 0.17*** 0.13** 0.14*** Local City 0.16*** 0.09 0.32*** Differential 0.01 0.04 -0.18** QUALITY OF INFRASTRUCTURE (1-7) Bribery in Utility State Capture Control of Corruption Full City Sample 0.99*** 0.98*** 1.10*** Global City 0.96*** 0.86*** 0.99*** Local City 0.57*** 0.82*** 0.87*** Differential 0.39*** 0.04 0.12 QUALITY OF ELECTRICITY (1-7) Bribery in Utility State Capture Control of Corruption Full City Sample 1.06*** 0.92*** 1.10*** Global City 1.05*** 0.82*** 0.99*** Local City 0.80*** 0.79*** 1.10*** Differential 0.25 0.03 -0.11 ACCESS TO TELEPHONES [2] (1-7) Bribery in Utility State Capture Control of Corruption Full City Sample 0.87*** 0.85*** 0.82*** Global City 0.81*** 0.53*** 0.70*** Local City 0.92*** 0.78*** 1.10*** Differential -0.11 -0.25 -0.40 ACCESS TO CELL PHONES (1-7) Bribery in Utility State Capture Control of Corruption Full City Sample 0.43*** 0.35*** 0.44*** Global City 0.39*** 0.30*** 0.34*** Local City 0.35*** 0.21* 0.53*** Differential 0.04 0.09 -0.19 INTERNET ACCESS AT SCHOOL (1-7) Bribery in Utility State Capture Control of Corruption Full City Sample 0.99*** 0.93*** 1.12*** Global City 0.97*** 0.82*** 0.98*** Local City 0.64*** 0.78*** 1.21*** Differential 0.33* 0.04 -0.23 Note: Initial sample of 262 cities was divided into two equal samples according to whether the city is global or local. We then ran a set of univariate pooled regressions of service performance/access against selected governance variables(bribery in utility, state capture and control of corruption, as per columns). The results for the magnitude and significance of coefficients for global and local samples are reported in the global city and local city rows, respectively. In each econometric specification, we also included a dummy (for global city=1, local city=0) to identify the two samples as well as an interactive term (dummy * governance) to capture the differential impact of governance on city performance between global and local cities. Last row reports the differential of the coefficients (bolded, with significance levels) along with coefficients on governance variables for full sample and subsets. Refer also to Annex A for correlation matrix. 23 We broaden our presentation of the statistical results by summarizing the evidence obtained from the multi-variate econometric specifications, where a fuller set of controls in multiple specifications explaining service performance were included. As per description of the simple model and legend with the variables presented in the previous section, we performed econometric tests through OLS specifications with the various service performance outcomes as dependent variables, and with city characteristics, city and country governance and globalization indicators, as well as country characteristics, as independent variables. Table 4 below presents the combined synthesis of the results of all econometric specifications. This synthesis presents the magnitude values (+,0,- on the left) and significance (with asterisks, on the right), which we obtain by averaging the magnitude and significance of the regression results of 21 different OLS specifications, for each city performance indicator.17 From the implementation of the fuller econometric model, for which the results are summarized in Table 4 above, we obtain a number of salient results. There is an effect of City Globalization on performance using both the Taylor (Global City variable) and the UN (water, electricity, telephone, and sewerage access variables) databases. This effect is relatively modest in some cases yet present, though with varying levels of significance depending on the type of service, controls used, and model specifications. The variable that is most correlated to city globalization (global city (3)) is the quality of electricity. Other city globalization proxies, such as city web portal with information on steps to start a firm (business dummy), and whether the budget is transparently available on the web (budget dummy), show positive and significant results, but these results are not robust across all service variables. In particular, access to cell phones (as opposed to telephones), and the overall quality of infrastructure are significant, while the other service variables, such as access to water, electricity and sewerage, and to telephone lines show less significance. See Section 6 below for a more detailed treatment of some of the city characteristics and related results. Bribery in utility seems to also be important, showing significant results for all service variables in Table 4. Definition of city as a space is sensitive to the definitions of city boundaries, whether core city or broader agglomeration. We have corrected for thisby using city as an outcome for the analysis. Such a construct: (a) allows us to use "city as a place" as an input to the analysis and hence is less sensitive to the boundary definitions; and (b) the UN database 17 In particular, for each independent variable, for calculating the magnitude of the coefficient rating (left hand side), we computed the product of the average coefficient magnitude across all specifications times twice the standard deviation of the variable. Then we assigned magnitude values according to the following criteria: for positive coefficients, we applied a "++++" for any value above 1.2; "+++" for any value between 0.7 and 1.2; "++" for any value between 0.4 and 0.7; "+" for any value between 0.1 and 0.4; and finally a "0"'for any value below 0.1. Similarly, minus signs were applied to negative coefficients. For the rating on the significance of the coefficient in each cell (right hand side), we averaged the various significance levels across the 21 specifications, where "***" indicates significance at 1%, "**" indicates significance at 5%, "*" indicates significance at 10% and ` indicates significance at 15%. 24 has data on city size by core city and urban agglomeration, and we did not find any differences when either of the two measures were used. In any case, the data has measurement errors as any data base, and this is why we used similar indicators from different sources (e.g. access to telephones in UN databases and the same indicator in EOS database) to check the sensitivity of the results across databases. These results are robust and independent of the definition of city boundaries. We did control for various definitions of city boundaries by running regressions both on core city population and metropolitan population, with same results. TABLE 4: DETERMINANTS OF CITY PERFORMANCE ­ SYNTHESIS OF ECONOMETRIC REGRESSION RESULTS Source UN 1998 UN 1998 UN 1998 UN 1998 EOS 2003 EOS 2003 EOS 2003 EOS 2003 EOS 2003 Access to Access to Dependent Variables Access to Access to Access to Telephones Telephones Access to Internet Quality of Quality of Water Electricity Sewerage [1] [2] Cell Phones Access Infrastructure Electricity % % % % 1-7 1-7 1-7 1-7 1-7 City Variables City Population 0 / 0 0 / 0 0 / 0 0 / 0 + / 0 + / 0 0 / 0 + / * + / 0 Global City (3) 0 / 0 0 / 0 0 / 0 0 / 0 + / 0 0 / 0 0 / 0 + / 0 + / ** Bribery in Utility + / ' 0 / ' + / ' + / ** ++++ / *** ++ / *** - / 0 ++++ / *** ++++ / *** Business Dummy 0 / 0 0 / 0 0 / 0 + / * + / ' 0 / * 0 / 0 +++ / ** ++ / ** Website dummy 0 / 0 0 / ' 0 / 0 + / 0 ++ / ** + / ** 0 / 0 + / * + / ' Budget Dummy 0 / ' 0 / 0 0 / * 0 / 0 + / * 0 / 0 + / 0 + / 0 0 / 0 Port Dummy - / ** 0 / ' 0 / 0 - / ** + / 0 0 / 0 + / * 0 / ** - / * Capital Dummy 0 / 0 0 / 0 0 / 0 0 / 0 - / 0 - / 0 + / *** - / * 0 / 0 Country Variables Country Population 0 / 0 0 / 0 0 / 0 0 / 0 0 / 0 0 / 0 ++ / ** 0 / 0 - / * Urbanization (%) + / ' 0 / ' 0 / 0 + / ' + / 0 + / 0 0 / * 0 / ' + / ' Kearney Globalization Index 0 / ' 0 / 0 0 / 0 + / 0 + / * + / * + / * +++ / *** +++ / *** Control of Corruption + / ** 0 / * + / ' + / ** ++++ / *** ++ / ** + / 0 ++++ / *** ++++ / *** Income per capita + / *** + / ** ++ / *** ++ / *** ++++ / *** +++ / *** +++ / *** ++++ / *** ++++ / *** # of cities 63 61 59 42 194 194 194 194 194 Adjusted R-squared 0.28 0.11 0.29 0.37 0.39 0.28 0.28 0.45 0.50 Source: KLM city database, drawing from EOS-WEF 2003 as well as other sources as per below. Sample size: 412 cities in 134 countries. Values for magnitude (+,0,-) and significance (*,0,') in each cell were obtained by averaging results of 21 different OLS specifications. In particular, for each independent variable, for calculating the magnitude of the coefficient rating (left hand side), we computed the product of the average coefficient magnitude across all specifications times twice the standard deviation of the variable. Then we assigned magnitude values according to the following criteria: for positive coefficients, we applied a "++++" for any value above 1.2; "+++" for any value between 0.7 and 1.2; "++" for any value between 0.4 and 0.7 ; "+" for any value between 0.1 and 0.4; and finally a "0"'for any value below 0.1. Similarly, minus signs were applied to negative coefficients. For the rating on the significance of the coefficient in each cell (right hand side), we averaged the various significance levels across the 21 specifications, where *** indicates significance at 1%, **" indicates significance at 5%, * indicates significance at 10% and ` indicates significance at 15%. Values for number of cities and R-square were computed as simple averages across all specifications. Sources: All dependent variables drawn from UN 1998 and EOS 2003; globalization index downloaded from Foreign Policy Magazine, 2002 (Kearney); country population and urbanization drawn from WDI 2002; city population downloaded from website: www.citypopulation.de ; port, capital, budget and business dummies drawn from KLM 2003; Control of Corruption drawn from KK02; Global city (3) drawn from GAWC (Taylor, 2001); income per capita drawn from Herston-Summers and CIA World Factbook (2001); bribery in Utility drawn from EOS 2003. 25 These econometric results are consistent with and support what was described in the previous figures and tables. Further, it is important to observe differences across services in terms of the importance of globalization on performance. Indeed, the results suggest that there are differential impacts of globalization on city performance, depending on the options city residents have for exit or choice. In the services where they can exit, such as water and electricity, then globalization has limited impact on city performance. In cases where they cannot exit, there seems to be more impact. The role of voice and participation, when you cannot exit seems to be critical, and significant, as shown by the significance of the variables such as bribery in utility and control of corruption (see Table 4 above). Lee and Anas (1996) potently gave an illustration consistent with these results. They analyzed the decisions firms made to locate or not in cities in Nigeria and Indonesia, and also found that where public supply of electricity was unreliable, they could substitute by private generation capacity that they themselves installed. A sister paper (Kaufmann, Léautier, and Mastruzzi, 2004) provides a more detailed look at both the issue of technology and that of scale and city characteristics. Country globalization (Kearney Globalization Index) also matters for city level performance. It is significant for all service variables except access to electricity, sewerage and telephone lines. What is interesting in the results in Table 4 is that city performance for network infrastructure (electricity grid, sewerage, telephone lines) is impacted by two aspects of governance (bribery in utility at the city level, control of corruption at the country level) and per capita income. Rich countries can afford to provide network infrastructure in cities. And better control of city and country level corruption leads to better performance of cities in providing access to services. In summary, we find that there is an effect of city globalization on performance of cities using both the Taylor and the UN databases. While the strength of the effects appears to be somewhat more robust for governance variables than for globalization, it is present for the latter. 5 Determinants of City Governance 5.1 A simple Framework The dramatic increase in empirical work on governance over the past few years has spearheaded a more in-depth analysis of the manifestations, causes, and consequences of misgovernance and corruption. It has drawn from cross-country data as well as in-country diagnostic perspectives. Significant inroads have been made in unbundling governance to measure and analyze its detailed components, as well as in exploring empirically the governance performance of different institutions--such as parliament, police, and customs. We use three indicators of governance to investigate the relationship between globalization and governance. The first is an indicator of governance from the political perspective which we represent by an indicator of low illegal party financing. The second is a measure 26 of public sector governance for which we use an indicator on low diversion of public funds. Third, we use an indicator of the effectiveness of public services where we use a measure of trust and performance of the postal system. With these three indicators, we see that globalization is positively related to good governance (Figure 2a). For city governance we use a measure of whether a city is providing services to its citizens (low health access gap) and whether it has a well functioning public sector with low bribery in taxes or in the provision of utilities. The data shows that, the more urbanized a country is, the better its level of city governance (Figure 2b). This finding is of historical importance as most of what we now practice in the form of country governance was first developed in city states such as in Greece at the time of Plato in the 4th century B.C. or Jericho over 10,000 years ago. As illustrated in Figure 2c, the data suggest that neither city nor country size negatively affect governance at the city level. Furthermore, there is no evidence that a higher degree of urbanization within a country has a detrimental impact on the governance of cities; in fact, for some dimensions the extent of urbanization may have a positive effect. We find no support for the argument that capital cities are better or worse governed which somewhat contrasts with the empirical importance of the extent of globalization, if we assume that capital cities tend to be more globalized (Figure 2c). This early exploration suggests that some commonly held beliefs on city-level performance can be empirically challenged--such as the notions that in larger cities performance and governance will suffer due to diseconomies of agglomeration; that port cities, or the country's capital, will generally tend to be more corrupt than other cities; or that urbanization or globalization may bring about mis-governance. Complementing the stylized results shown in the various panels of Figure 2, we summarize in table 5 below the key average ratings for the governance variables we utilize, against the various potential determinants. 27 Figure 2a: City governance and globalization Good Governance 7 Low Illegal Party Financing Low Diversion of Public Funds Effectiveness/Trust of Postal System 4 Poor 1 Low Globalization Medium globalization High globalization Source: Executive Opinion Survey (2003). Figure 2b: City governance and urbanization City Governance 7 Good Low Health Access Gap Low Bribery in Taxes Low bribery in Utilities 4 Poor 1 Low Urbanization Medium Urbanization High Urbanization Source: Executive Opinion Survey (2003). Figure 2c: City governance and city characteristics 7 Low Bribery to affect laws, regulations Good Low Diversion of Public Funds Low Organized Crime City/country City Population Capital City Size Ratio 4 Poor 1 Low ratio High ratio Low High Not the Capital Population Population Capital City Source: Executive Opinion Survey (2003). 28 Next we review whether the results we have found are robust across databases and across different definitions of globalization and governance. The results are shown in Table 5 below. Overall, we observe higher levels of governance in global cities, although the magnitude of the differences between local and global cities varies depending on the type of governance variable used. These results are similar irrespective of whether one considers the full sample of 412 cities (shown in table 5 below), or when the samples are segmented into non-OECD and OECD cities (not shown). TABLE 5: Average governance ratings by multiple city/country factors Low Informal % Red Tape Low Low Bribery Absence of Low Bribery Low Illegal Money Cost of Diversion of in Utility street crime to affect laws Financing Laundering Imports Public Funds 1-7 1-7 1-7 % 1-7 1-7 1-7 Local 4.39 3.46 3.78 23.8 3.61 3.14 3.06 Global city (Taylor) Global 5.64 4.46 4.84 14.2 4.76 3.93 4.42 Low 4.95 3.72 4.27 17.4 3.99 3.33 3.44 Size ratio city/country High 5.05 4.00 4.32 18.2 4.24 3.53 3.76 Low 5.06 3.85 4.32 16.4 4.11 3.46 3.66 City population High 4.94 3.92 4.27 19.7 4.17 3.43 3.59 No 5.01 3.85 4.26 17.2 4.14 3.45 3.63 Capital City Yes 4.99 3.92 4.33 19.2 4.11 3.42 3.60 Website has information on No 4.90 3.76 4.16 18.5 4.04 3.37 3.50 how to start business Yes 6.19 5.19 5.42 10.7 5.18 4.26 5.02 Kearney Country Low 4.69 3.58 4.00 20.6 3.90 3.03 3.16 Globalization Index High 6.21 4.88 5.46 9.6 5.11 4.36 5.04 Low 4.40 3.58 3.93 21.5 3.65 3.12 3.10 Country Urbanization High 5.84 4.26 4.77 13.0 4.79 3.88 4.34 All variables range from 1 (bad) to 7 (good) with exception of cost of imports which is shown in percentage terms The synthesis results of the full set of econometric specifications is presented in table 6 below, and the full regression results appear in Table 4 of the paper. We find that globalization at both city and country level (Kearney Globalization Index) is positively correlated with all the measures of governance. Globalization at a country level, as measured by the Kearney Globalization Index is significant and positive against all the governance indicators, except cost of imports where it is significant and negative. Per capita income is also positively correlated with all dimensions of governance. The results shown in Table 6 are relatively robust and significant. With some specifications, country level globalization seems to dominate the city globalization variable. Partly this may be the fact that country level globalization in the more fundamental (macroeconomic and trade) sense vis-à-vis the rest of the world is even more important than city-specific measures or city specific attractiveness. Yet in part this result may also reflect the fact that the country globalization variable was constructed specifically for the purpose to have a globalization indicator, while the city globalization variables we are utilizing are indirect (and partial) proxies. Thus, the measurement errors may differ, with the city variables being more (statistically) `noisy'. 29 Two proxies for city level globalization which look at cities that have attracted business in finance, accounting, and advertising (global city 3) and those that have a transparent way of allowing firms to set up a business (business dummies) are better governed in all dimensions of governance. TABLE 6: DETERMINANTS OF CITY GOVERNANCE ­ SYNTHESIS OF ECONOMETRIC REGRESSION RESULTS Source EOS 2003 EOS 2003 EOS 2003 EOS 2003 EOS 2003 EOS 2003 EOS 2003 Low Bribery in Low Informal Money Red Tape Cost of Low Bribery to affect Low Diversion Low Illegal Party Dependent Variables Low Street Crime Utilities Laundering Imports laws public funds Financing 1-7 (good) 1-7 (good) 1-7 (good) 1-9 (bad) 1-7 (good) 1-7 (good) 1-7 (good) City Variables City Population + / ' + / ' 0 / 0 0 / 0 + / 0 + / ' 0 / 0 Global City (3) + / ' ++ / *** + / ' - / ** ++ / ** ++ / ** ++ / *** Country city size ratio + / ' + / ' + / ' 0 / 0 + / ' + / ' + / * Business Dummy ++ / *** +++ / *** ++ / ** - / ' ++ / ** +++ / *** ++ / ** Website dummy + / 0 + / ' + / 0 0 / 0 + / 0 + / ' 0 / 0 Budget Dummy 0 / 0 + / 0 + / 0 0 / 0 - / 0 0 / 0 0 / 0 Port Dummy 0 / 0 + / ' + / 0 0 / 0 0 / 0 + / ' + / ' Capital Dummy 0 / 0 - / ' 0 / 0 + / 0 - / 0 - / ' - / ' Country Variables Country Population --- / *** -- / ' - / 0 ++ / ' -- / * -- / * --- / ** Urbanization (%) ++ / ** 0 / 0 -- / ' 0 / * + / ' 0 / 0 0 / 0 Kearney Globalization Index ++++ / *** ++++ / *** ++++ / *** --- / *** +++ / *** ++++ / *** ++++ / *** Income per capita ++++ / *** +++ / *** ++++ / *** ---- / *** ++++ / *** ++++ / *** +++ / *** # of cities 193 193 193 193 193 193 193 Adjusted R-squared 0.30 0.21 0.19 0.24 0.21 0.29 0.17 Source: KLM city database, drawing from EOS-WEF 2003 as well as other sources as per below. All dependent variables range from bad to good with exception of costs of imports. Sample size: 412 cities in 134 countries. Values for magnitude (+,0,-) and significance (*,0,') in each cell were obtained by averaging results of 21 different OLS specifications. In particular, for each independent variable, for calculating the magnitude of the coefficient rating (left hand side), we computed the product of the average coefficient magnitude across all specifications times twice the standard deviation of the variable. Then we assigned magnitude values according to the following criteria: for positive coefficients, we applied a "++++" for any value above 1.2; "+++" for any value between 0.7 and 1.2; "++" for any value between 0.4 and 0.7 ; "+" for any value between 0.1 and 0.4; and finally a "0"'for any value below 0.1. Similarly, minus signs were applied to negative coefficients. For the rating on the significance of the coefficient in each cell (right hand side), we averaged the various significance levels across the 21 specifications, where *** indicates significance at 1%, **" indicates significance at 5%, * indicates significance at 10% and ` indicates significance at 15%. Values for number of cities and R-square were computed as simple averages across all specifications. Sources: All dependent variables drawn from UN 1998 and EOS 2003; globalization index downloaded from Foreign Policy Magazine, 2002 (Kearney); country population and urbanization drawn from WDI 2002; city population downloaded from website: www.citypopulation.de ; port, capital, size ratio, budget, website and business dummies drawn from KLM 2004; Global city (3) drawn from GAWC (Taylor); income per capita drawn from Heston-Summers and CIA World Factbook (2001). 30 We have summarized the results of the many econometric specifications to explain the potential determinants of city governance in the synthesis in Table 6 (the full set of regressions are available upon request)18. 7 Concluding: Future Research and Emerging Policy Implications 7.1 Brief synthesis In this paper we attempted to contribute to the field of urban governance and globalization through an empirically-based exploration of some key determinants of the performance of cities. This empirical inquiry was made possible through the construction of a worldwide database for cities that contains variables and indicators of globalization (at the country and city level), of city governance, and of city performance (access and quality of infrastructure service delivery). This city database integrates existing data with new data gathered specifically for this research project. We find that good governance and globalization matter for city-level performance in terms of access and quality of delivery of infrastructure services, and also that globalization and good city governance are related. There appear to be dynamic pressures from globalization and accountability that result in better performance at the city level. Furthermore, we find that there are complex interactions between technology, governance and city performance, as well as evidence of a non-linear (u-shaped) relationship between city size and performance, challenging the view that very large cities necessarily exhibit lower performance and pointing to potential agglomeration economies. Our framework also suggests a way of bridging two seemingly competing strands of the literature, namely viewing the city as a place or as an outcome. We find evidence that port cities seem to be in general more dependent on good governance for the city performance variables that matter for globalization (access to cell phone, internet access) and that capital cities tend more to serve local sewerage access better (water sewerage, electricity). A preliminary analysis of city level data from EOS 2004 corroborates the main findings in our paper (summarized in Annexes A and B). These results were corroborated when replacing 2003 data with 2004 data. Finally, 2004 data confirms the strong positive correlation between city performance and globalization as well as the strong positive correlation between city performance and city governance that we highlighted in Sections 4 and 5 . This is indicative of the robustness of our results. 18 We have presented synthesis results and have not shown the results of multiple correlations. We also have chosen to use data plots and not the actual econometric results. We thought the synthesis results were a better way to present the results. The actual regressions were tested for sign switches, and except for the access to electricity variable, the regressions did not show sign switches for the key variables. This is why we thought it best to show the range of coefficient estimates and the statistical significance. All the synthesis tables we have developed present controlled statistical results, as opposed to bivariate regressions, even if shown only in synthesis form. 31 7.2 Issues for Future Research There are a number of issues to consider for future research. First, with our dataset and empirical approach we need to keep in mind both a purely statistical point, as well as one of a more methodological/conceptual nature. The first refers to the limits we have after crossing variables from different datasets (Taylor/UN/EOS) in terms of the overlapping number of observations/cities (N), which for some indicators in Table 1 is very small. Thus, a major challenge in the next stage is to expand this database, and fill in the significant gaps within the existing set of over 400 cities. We expect to be able to collaborate with partners in this initiative. This expanded database would be made available on the web to all researchers and policy-makers (see Annex C for the current list of cities for which this dataset has information on at least some of the variables of interest in this research project). Second, we need to probe much deeper than was possible in this preliminary exploration on the issue of the dynamics between globalization and governance, and their interface with city performance. For instance, we found that globalization and governance are positively correlated with each other at the city level. Our simple conceptual and econometric model pointed to possible causality from globalization to good governance. Yet it cannot be ruled out obviously that in reality there is bi-directional causality, within an endogenous model where there is a possible virtuous circle between both variables. Within such a framework, an extension of the simple model presented in this paper would be warranted.19 This requires further research with additional data and a more complex econometric model with instruments, inter alia accounting for endogeneity issues. Furthermore, in this context the brief references to ancient times made earlier are suggestive of the need to probe further with an historical perspective. Given the constraints in data availability at the city and country level, in this initial research inquiry we did not include some other important institutional dimensions, such as decentralization and local finance. Obviously it cannot be ruled out that some of the significance of the governance variables for which we did have measures at the city and country level, such as the corruption-related ones, may be picking up in part some of the importance of other governance and institutional variables. Future research would need to 19 In essence we could conceptualize of another hypothesis, where the ability of cities to translate global opportunity into local value depends on governance, and that cities compete with each other on the basis of governance and not on the basis of initial endowments (infrastructure, culture etc). Cities with better governance remain globalized for longer periods of time (attract higher flows of capital, people, firms, finance, products) or attract such flows systemically over time (growth or change in volumes of these flows over time). This test would allow us to distinguish between those cities that are highly global and then drop in their rankings over time. For such a test, we would need at least two time periods. A possible model specification could include: Y = f (X1, X2, X3)+ where Y = change in the vector of governance indicators, to capture improvements or deterioration in governance over time which is a proxy for staying power in translating global opportunity to local value; X1 = a dummy for an alpha city from the Taylor database; X2 = dummy for a beta city, and X3 = dummy for a gamma city. With such a specification one could investigate whether alpha and beta type cities have better governance compared to the gamma type cities. 32 delve more in depth into these issues of potential omitted variable biases, and expand the database to account for additional institutional variables at the city level. Similarly, in future research it will be desirable to include proxies for economic output or productivity at the city level (there is a limited database for city product per capita by the UN, which suitably expanded and updated could be used for instance. Simple plotgrams with these limited dataset does suggest that city product does improve with globalization, for instance (see Léautier, Mehta, and Saito, 2004).20 7.3 Policy Implications The analysis we have done has provided empirical evidence on many observations and statements made in the past by experts in the field, prior to the gathering and analysis of a data. From this perspective, a number of policy conclusions emerge. First, our results are suggestive on the importance of focusing on country-wide governance, thereby tapping into economic benefits for the country at large but also providing a platform for cities to benefit from such environment-wide improvements. This suggests that reformist majors have an incentive to promote good governance at the national level (given the large externalities for the city). Further, International Financial Institutions (IFIs, such as the World Bank) and donors ought to continue to monitor and assess in some detail, how countries are faring with respect to governance at the national level. This makes sense not only from a country-wide performance standpoint, but also to point to ways of improving city-level performance. Second, our framework and evidence also points to the fact that improving governance at the city level allows cities to translate global opportunity into local value for their citizens. This implies that reformist city leaders do have important local policy and institutional levers at their disposal: governance at the local level matters significantly for urban performance. Similarly, this finding suggests that donor agencies could focus more on supporting improvements in city level governance. Finally, it is important to work with cities to support their globalization process and status. In this context, the interface between globalization, technology and governance deserves attention. In practical terms, web-related transparency and information outreach innovations are worthy illustrations, such as placing procurement and the budget on the city website, as well as a one-stop information and licensing `e-window' for business start-ups. In this context, it may also be important to have a set of incentives and strategies for cities to network with other cities to tap into global opportunities that will benefit a given city.21 20Other areas of investigation could include looking at the influence of diaspora networks on the governance of a city, as well as the influence of NGO movements for instance on human rights, environment, and corporate social responsibility issues within the context of the city, and its links to public governance. Further, expanding on the research on the interface between technology choices, governance and city performance also appear to be a promising line of inquiry from an empirical standpoint, which will be addressed in forthcoming research. 21 This would imply that IFIs such as the World Bank work with inter-city governance alliances and networks, such as the association of local governments and the city-to-city partnership, such as "Cities Alliance". 33 References Acemoglu, D., S. Johnson and J. Robinson, 2003. "Understanding Prosperity and Poverty: Geography, Institutions and the Reversal of Fortune.", NBER Working Paper #8460. Castells, M. 2000. "Materials for an exploratory theory of the network society." British Journal of Sociology 51: 5-24. Deas, I. and B. Giordano. 2001. "Conceptualizing and Measuring Urban Competitiveness in Major English Cities." Environment and Planning A 33: 1411-1429. De Long, J. Bradford and A. Shleifer, "Princes and Merchants: European City Growth Before the Industrial Revolution", NBER and Harvard University, December 1992. Doel, M.A. and P.J. Hubbard. (2002) "Taking World Cities Literally: Marketing the City in a Global Space of Flows." City 6(3). Foreign Policy. 2002. 2002 A.T. Kearney/Foreign Policy Magazine Globalization Index. Online. http://www.foreignpolicy.com/wwwboard/g-index2.php. Released as "Globalization's Last Hurrah." Foreign Policy (Jan/Feb): 38­51. Friedmann, J. 1986. "The World City Hypothesis", Development and Change 17, pp. 69- 83. Kaufmann, D. and A. Kraay. 2002. "Growth without Governance." Economia 3(1): 169­ 229. Online. http://www.worldbank.org/wbi/governance/pubs/growthgov.html. Kaufmann, D., A. Kraay. and M. Mastruzzi .2003. "Governance Matters III: Governance Indicators for 1996-2002." World Bank Policy Research Working Paper 3106. Kaufmann, D., M. Mastruzzi and D. Zavaleta. 2003. "Sustained Macroeconomic Reformd, Tepid Growth: A Governance Puzzle in Bolivia." Chapter 12 in In search of Prosperity: Analytic Narratives on Economic Growth (ed) D. Rodrik, Princeton University Press. Knack, S. and P. Keefer. 1997. "Does Social Capital Have an Economic Payoff? A Cross Country Investigation." Quarterly Journal of Economics 112(4): 1251­1288. Knox, P.L. 2002. "World Cities and the Organization of Global Space", in P.J. Taylor and M.J. Watts (eds), Geographies of Global Change, 2nd edition Oxford: Blackwell, 328-38. Kresl, P. 1995. "The Determinants of Urban Competitiveness: a Survey", in Kresl P. and Gappert, G. (eds) North American Cities and the Global Economy, Beverly Hills: Sage. Lee, K. S. and A. Anas.1996. "The Benefits of Alternative Power Tariffs for Nigeria and Indonesia." World Bank Policy Research Working Paper 1606. 34 Mehta, D. 1998. "Urban Governance: Lessons from Best Practices in Asia." UMP-Asia Occasional Paper No. 40. Sassen, S. 2000. "Locating Cities on Global Circuits." GaWC Research Bulletin No. 46. Online. http://www.lboro.ac.uk/gawc/rb/rb46.html. Sassen, S. 2002. "Global Cities and Diasporic Networks: Microsites in Global Civil Society." in Global Civil Society 2002, Chapter 9, The Centre for the Study of Global Governance. Short, J., Y.H. Kim, M. Kuus, and H. Wells. 1996. "The dirty little secret of World City research-data problems in comparative analysis." International Journal of Urban and Regional Research 20: 697-719. Smith, D.A. and M. Timberlake. 1995. "Conceptualizing and Mapping the Structure of the World Systems City System." Urban Studies 32: 287-302. Taylor, P.J, and D.R.F. Walker. "World City Network: Data Matrix Construction and Analysis." from the GaWC Research Group and Network, Data Set 6. Online. http://www.lboro.ac.uk/gawc/. Based on primary data collected by J.V. Beaverstock, R.G. Smith and P.J. Taylor (ESRC project "The Geographical Scope of London as a World City"). Tiebout, Charles M., (1956). .A Pure Theory of Local Expenditures,. Journal of Political Economy, 64, 416-424. Townsend, A. 2001. "The Internet and the Rise of the New Network Cities." Environment and Planning B: Planning and Design 28: 39-58. UNCHS. 1999. "UNCHS Expert Meeting on Urban Poverty and Governance Indicators." 29 April to May 1999, Nairobi Kenya. Online. http://www.gdrc.org/u-gov/indicators.html. Determinants of City Governance ­ Synthesis of Econometric Regression Results World Economic Forum. Various years. "Executive Opinion Survey." Prepared for The Global Competitiveness Report. 35 ANNEX A: Correlation Matrix of Selected Governance Variables and Service Access/ Performance FULL SAMPLE Bribery in Control of Utility State Capture Corruption Access to Telephones [1] 0.69*** 0.47*** 0.66*** Access to Sewerage 0.58*** 0.45*** 0.55*** Access to Water 0.57*** 0.45*** 0.52*** Access to Electricity 0.31** 0.27** 0.31*** Quality of Infrastructure 0.79*** 0.82*** 0.83*** Quality of Electricity 0.85*** 0.77*** 0.84*** Access to Telephones [2] 0.83*** 0.64*** 0.74*** Access to Cell Phones 0.67*** 0.57*** 0.64*** Access to Internet in school 0.82*** 0.80*** 0.88*** AVERAGE 0.68 0.58 0.66 GLOBAL CITY SAMPLE Bribery in Control of Utility State Capture Corruption Access to Telephones [1] 0.75*** 0.52** 0.69*** Access to Sewerage 0.57 0.38 0.52* Access to Water 0.58* 0.35 0.54* Access to Electricity 0.32 0.26* 0.37** Quality of Infrastructure 0.74*** 0.76*** 0.79*** Quality of Electricity 0.87*** 0.76*** 0.84*** Access to Telephones [2] 0.84*** 0.60*** 0.73*** Access to Cell Phones 0.67*** 0.57*** 0.59*** Access to Internet in school 0.81*** 0.77*** 0.84*** AVERAGE 0.68 0.55 0.66 LOCAL CITY SAMPLE Bribery in Control of Utility State Capture Corruption Access to Telephones [1] 0.58*** 0.25 0.60*** Access to Sewerage 0.44* 0.29 0.38*** Access to Water 0.41** 0.34** 0.39*** Access to Electricity 0.33* 0.32* 0.23** Quality of Infrastructure 0.67*** 0.70*** 0.66*** Quality of Electricity 0.66*** 0.51*** 0.63*** Access to Telephones [2] 0.71*** 0.49*** 0.60*** Access to Cell Phones 0.43*** 0.22* 0.47*** Access to Internet in school 0.63*** 0.58*** 0.81*** AVERAGE 0.54 0.41 0.53 Note: This Annex complements Table 3 in the paper. It shows pairwise correlations between selected governance variables and service access/performance variables for the full sample and for the two equal sub- samples of global and local cities. 36 ANNEX B: DETERMINANTS OF CITY GOVERNANCE ­ SYNTHESIS OF ECONOMETRIC REGRESSION RESULTS ANNEX C: DETERMINANTS OF CITY GOVERNANCE ­ SYNTHESIS OF ECONOMETRIC REGRESSION RESULTS Source EOS 2003 EOS 2003 EOS 2003 EOS 2003 EOS 2003 EOS 2003 EOS 2003 EOS 2003 EOS 2003 EOS 2003 EOS 2003 EOS 2003 EOS 2003 EOS 2003 Low Illegal Low Quality of Low Informal Red Tape Low Trust in Low Health Dependent Variables Low Bribery Money Low Street Cost of Low Bribery Low Diversion Party Bribery in Low Bribery Soundness of Organized Postal Politicians gap Financing crime System in Utilities Laundering Crime Imports to affect laws public funds Permits in Tax Banks 1-7 (good) 1-7 (good) 1-7 (good) 1-9 (bad) 1-7 (good) 1-7 (good) 1-7 (good) 1-7 (good) 1-7 (good) 1-7 (good) 1-7 (good) 1-7 (good) 1-7 (good) 1-7 (good) City Variables City Population + / ' + / ' 0 / 0 0 / 0 + / 0 + / ' 0 / 0 ++ / * 0 / 0 0 / 0 0 / 0 0 / 0 0 / 0 0 / 0 Global City (3) + / ' ++ / *** + / ' - / ** ++ / ** ++ / ** ++ / *** ++ / ** ++ / ** + / ** ++ / *** + / ** ++ / ** + / ' Country city size ratio + / ' + / ' + / ' 0 / 0 + / ' + / ' + / * + / * + / ' 0 / 0 + / * + / * + / 0 0 / 0 Business Dummy ++ / *** +++ / *** ++ / ** - / ' ++ / ** +++ / *** ++ / ** ++ / *** ++ / * ++ / ** ++ / ** ++ / ** ++ / ** +++ / *** Website dummy + / 0 + / ' + / 0 0 / 0 + / 0 + / ' 0 / 0 + / ' + / ' 0 / 0 + / ' + / * + / ' + / ' Budget Dummy 0 / 0 + / 0 + / 0 0 / 0 - / 0 0 / 0 0 / 0 0 / 0 0 / 0 + / 0 - / 0 0 / 0 + / ' + / 0 Port Dummy 0 / 0 + / ' + / 0 0 / 0 0 / 0 + / ' + / ' 0 / 0 - / ' + / 0 + / 0 + / 0 0 / 0 0 / 0 Capital Dummy 0 / 0 - / ' 0 / 0 + / 0 - / 0 - / ' - / ' - / * - / 0 - / ' - / ' 0 / 0 - / ' - / 0 Country Variables Country Population --- / *** -- / ' - / 0 ++ / ' -- / * -- / * --- / ** --- / *** --- / ** -- / ** -- / * -- / ' - / ' --- / *** Urbanization (%) ++ / ** 0 / 0 -- / ' 0 / * + / ' 0 / 0 0 / 0 + / * ++ / ' + / ' 0 / 0 - / 0 - / 0 - / 0 Kearney Globalization Index ++++ / *** ++++ / *** ++++ / *** --- / *** +++ / *** ++++ / *** ++++ / *** ++++ / *** ++++ / *** +++ / *** ++++ / *** ++++ / *** ++++ / *** ++++ / *** Income per capita ++++ / *** +++ / *** ++++ / *** ---- / *** ++++ / *** ++++ / *** +++ / *** ++++ / *** ++++ / *** +++ / *** ++++ / *** +++ / *** ++++ / *** ++++ / *** # of cities 193 193 193 193 193 193 193 193 193 193 193 193 193 193 Adjusted R-squared 0.30 0.21 0.19 0.24 0.21 0.29 0.17 0.25 0.28 0.15 0.21 0.18 0.29 0.31 Source: KLM city database, drawing from EOS-WEF 2003 as well as other sources as per below. All dependent variables range from bad to good with exception of costs of imports. Sample size: 412 cities in 134 countries. Values for magnitude (+,0,-) and significance (*,0,') in each cell were obtained by averaging results of 21 different OLS specifications. In particular, for each independent variable, for calculating the magnitude of the coefficient rating (left hand side), we computed the product of the average coefficient magnitude across all specifications times twice the standard deviation of the variable. Then we assigned magnitude values according to the following criteria: for positive coefficients, we applied a "++++" for any value above 1.2; "+++" for any value between 0.7 and 1.2; "++" for any value between 0.4 and 0.7 ; "+" for any value between 0.1 and 0.4; and finally a "0"'for any value below 0.1. Similarly, minus signs were applied to negative coefficients. For the rating on the significance of the coefficient in each cell (right hand side), we averaged the various significance levels across the 21 specifications, where *** indicates significance at 1%, **" indicates significance at 5%, * indicates significance at 10% and ` indicates significance at 15%. Values for number of cities and R-square were computed as simple averages across all specifications. Sources: All dependent variables drawn from UN 1998 and EOS 2003; globalization index downloaded from Foreign Policy Magazine, 2002 (Kearney); country population and urbanization drawn from WDI 2002; city population downloaded from website: www.citypopulation.de ; port, capital, size ratio, budget, website and business dummies drawn from KLM 2004; Global city (3) drawn from GAWC (Taylor); income per capita drawn from Heston-Summers and CIA World Factbook (2001). ANNEX C: LIST OF CITIES IN THE KLM DATABASE CITIES COUNTRY CITIES COUNTRY CITIES COUNTRY Kabul Afghanistan Xiamen China Munich Germany Tirana Albania BARRANQUILLA Colombia Stuttgart Germany Algiers Algeria Bogota Colombia Accra Ghana CONSTANTINE Algeria Manizales Colombia TEMA Ghana ORAN Algeria MEDELLIN Colombia Athens Greece REGHATA Algeria PEREIRA Colombia Guatemala City Guatemala Luanda Angola Kinshasa Congo, dem. Rep. Conakry Guinea Buenos Aires Argentina ALAJUELA Costa Rica Georgetown Guyana SAN FERNANDO Argentina CARTAGO Costa Rica PORT-AU-PRINCE Haiti Yerevan Armenia HEREDIA Costa Rica LA CEIBA Honduras Adelaide Australia CAKOVEC Croatia SAN PEDRO SULA Honduras Brisbane Australia KARLOVAC Croatia TEGUCIGALPA Honduras Canberra Australia OSIJEK Croatia VILLANUEVA Honduras Hobart Australia PULA Croatia Hong Kong Hong Kong Melbourne Australia RIJEKA Croatia Budapest Hungary Perth Australia SPLIT Croatia GYONGYOS Hungary Sydney Australia VARAZDIN Croatia KAPOSVAR Hungary Wellington Australia ZADAR Croatia TATABANTA Hungary Linz Austria Zagreb Croatia REYKJAVIK Iceland Vienna Austria Havana Cuba Ahmadabad India Baku Azerbaijan Limassol Cyprus Bangalore India Manama Bahrain Nicosia Cyprus Calcutta India Dhaka Bangladesh Prague Czech Republic Chennai India Minsk Belarus Aarhus Denmark DELHI India Antwerp Belgium Copenhagen Denmark Hyderabad India Brussels Belgium Santo Domingo Dominican Republic Jaipur India WATERLOO Belgium CUENCA Ecuador Lucknow India La Paz Bolivia Guayaquil Ecuador Mumbai India SANTA CRUZ Bolivia IBARRA Ecuador New Delhi India GABORONE Botswana Quito Ecuador PUNE India Belo Horizonte Brazil 10TH OF RAMADAN Egypt Bandung Indonesia Brazilia Brazil 6TH OF OCTOBER Egypt Batam Indonesia Curitiba Brazil Cairo Egypt Jakarta Indonesia Porto Alegre Brazil GIZA Egypt Medan Indonesia Recife Brazil ANTIGUO CUSCATLAN El Salvador Tehran Iran Rio de Janeiro Brazil San Salvador El Salvador Baghdad Iraq Salvadore Brazil TALLINN Estonia CORK Ireland Sao Paulo Brazil TARTU Estonia Dublin Ireland Sofia Bulgaria Addis Ababa Ethiopia Jerusalem Israel VARNA Bulgaria ESPOO Finland Tel Aviv Israel Douala Cameroon Helsinki Finland BELLUNO Italy Yaounde Cameroon Lille France Bologna Italy Calgary Canada Lyon France Genoa Italy Edmonton Canada Marseilles France Milan Italy Montreal Canada Paris France Naples Italy Ottawa Canada Strasbourg France PORDENONE Italy Quebec Canada BACAU Gambia Rome Italy Toronto Canada BANJU Gambia Turin Italy Vancouver Canada KANIFING Gambia Abidjan Ivory Coast Winnipeg Canada SEREKUNDA Gambia Kingston Jamaica N'DJAMENA Chad Berlin Germany Kyoto Japan Santiago Chile Bonn Germany Nagoya Japan TALCAHUANO Chile Dortmund Germany Osaka Japan Beijing China Dresden Germany Tokyo Japan Dalian China Dusseldorf Germany Yokohama Japan Guangzhou China Essen Germany Amman Jordan Nanjing China Frankfurt Germany Almaty kazakhstan Shanghai China Hamburg Germany Mombasa Kenya Shenzhen China KOELN Germany Nairobi Kenya Tianjin China Leipzig Germany Pyongyang Korea, north ANNEX C: LIST OF CITIES IN THE KLM DATABASE (contd.) CITIES COUNTRY CITIES COUNTRY CITIES COUNTRY Pusan Korea, South Casablanca Morocco PREVIDZA Slovak Republic Seoul Korea, South Rabat Morocco CELJE Slovenia Kuwait Kuwait MAPUTO Mozambique KRANJ Slovenia AIZKRAUKLES Latvia MATOLA Mozambique Ljubljana Slovenia DAUGAVPILS Latvia Yangon Myanmar MARIBOR Slovenia KULDIGA Latvia Windhoek Namibia MURSKA SOBOTA Slovenia LIEPAJA Latvia Amsterdam Netherlands NOVO MESTO Slovenia OGRE Latvia ARNHEM Netherlands SLOVENJ GRADEC Slovenia RIGA Latvia Rotterdam Netherlands Cape Town South Africa VENTSPILS Latvia The Hague Netherlands Durban South Africa Beirut Lebanon Utrecht Netherlands Johannesburg South Africa Monrovia Liberia Auckland New Zealand MIDRAND South Africa Tripoli libya Christchurch New Zealand Pretoria South Africa ALYTUS Lithuania Managua Nicaragua Barcelona Spain KAUNAS Lithuania IKEJA Nigeria Bilbao Spain KLAIPEDA Lithuania Lagos Nigeria Madrid Spain MARIJAMPOLE Lithuania Bergen Norway Seville Spain PANEVEZYS Lithuania Oslo Norway Valencia Spain SIAULIAI Lithuania STAVANGER Norway Colombo Sri Lanka SILUTE Lithuania Ruwi Oman MOUNT LAVINIA Sri Lanka UKMERGE Lithuania Islamabad Pakistan Khartoum Sudan VILNIUS Lithuania Karachi Pakistan Gothenburg Sweden Luxembourg Luxembourg Lahore Pakistan Stockholm Sweden STRASSEN Luxembourg Rawalpindi Pakistan Basel Switzerland Macau Macao Panama City Panama Geneva Switzerland BITOLA Macedonia Asuncion Paraguay ST-GALLEN Switzerland KUMAHOVO Macedonia CIUDAD DEL ESTE Paraguay Zurich Switzerland OHRID Macedonia FDO.DE LA MORA Paraguay Damascus Syria PRILEP Macedonia SANTA RITA Paraguay KAOHSIUNG Taiwan SKOPJE Macedonia CALLAO Peru TAICHUNG Taiwan STIP Macedonia Lima Peru Taipei Taiwan STRUMICA Macedonia HO-PASIG Philippines ARUSHA Tanzania VELES Macedonia MAKATI Philippines DAR-ES-SALAAM Tanzania ANTANANARIVO Madagascar Manila Philippines MOROGORO Tanzania TOAMASINA Madagascar PARANAQUE Philippines Bangkok Thailand BLANTYRE Malawi Warsaw Poland Lome Togo LILONGWE Malawi Lisbon Portugal ARIMA Trinidad and Tobago Kuala Lumpur Malaysia Doha Qatar COUVA Trinidad and Tobago Labuan Malaysia ALBA IULIA Romania Port of Spain Trinidad and Tobago Malacca Malaysia Bucharest Romania SFAX Tunisia Penang Malaysia CONSTANTA Romania SOUSSE Tunisia PETALING JAYA Malaysia CRAIOVA Romania Tunis Tunisia SHAH ALAM Malaysia IASI Romania Ankara Turkey BAMAKO Mali ORADEA Romania BALIKESIR Turkey SEGOU Mali PIATRA NEAMT Romania Istanbul Turkey FLORIANA Malta PLOIESTI Romania Kampala Uganda GZIRA Malta RAMNICO VALCEA Romania Kueb Ukraine MARSA Malta SLATINA Romania Abu Dhabi United Arab Emirates SAN GWANN Malta TARGOVISTE Romania Dubai United Arab Emirates SLIEMA Malta CAMAPA Russia Birmingham United Kingdom VALLETTA Malta Moscow Russia Bristol United Kingdom ZEJTUN Malta St Petersburg Russia Cardiff United Kingdom PORT LOUIS Mauritius Jeddah Saudi Arabia Edinburgh United Kingdom GUADALAJARA Mexico Riyadh Saudi Arabia Glasgow United Kingdom Mexico City Mexico Dakar Senegal Leeds United Kingdom Monterrey Mexico Dar es Salaam Senegal Liverpool United Kingdom SAN PEDRO GARZA GCIA Mexico Freetown Sierra Leone London United Kingdom Tijuana Mexico Singapore Singapore Manchester United Kingdom TORREON Mexico Bratislava Slovak Republic Newcastle United Kingdom XALAPA Mexico PRESOV Slovak Republic Nottingham United Kingdom 39 ANNEX C: LIST OF CITIES IN THE KLM DATABASE (contd.) CITIES COUNTRY Alexandria United States Atlanta United States Baltimore United States Boston United States Buffalo United States Charlotte United States Chicago United States Cincinnati United States Cleveland United States Columbus United States Dallas United States Denver United States Detroit United States Houston United States Kansas City United States Los Angeles United States Miami United States Minneapolis United States New Orleans United States New York United States Philadelphia United States Phoenix United States Pittsburg United States Portland United States Richmond United States San Diego United States San Francisco United States San Jose United States Seattle United States St Louis United States Washington United States Wilmington United States Montevideo Uruguay Tashkent Uzbekistan Caracas Venezuela BIENHOA Vietnam Hanoi Vietnam HO CHI MINH Vietnam Sana'a Yemen BABEBO Yugoslavia Belgrade Yugoslavia KRALJEVO Yugoslavia NIS Yugoslavia NOVI SAD Yugoslavia SR.MITROVICA Yugoslavia SUBOTICA Yugoslavia UZICE Yugoslavia KITWE Zambia Lusaka Zambia NDOLA Zambia Bulawayo Zimbabwe Harare Zimbabwe 40