Measuring Networks across Countries: an Empirical Exploration
|June 6, 2012||Posted by Administrator under Ekonomske teorije|
Axel Marx and Jadir Soares
“A world in which horizontal and vertical government networks comprise different types of government institutions (regulatory, judicial, legislative), perform different functions (information exchange, enforcement cooperation, technical assistance and training), have different members, have different degrees of formality and coexists in different ways with international organizations is a messy world indeed. It may seem impossibly complex.” Anne-Marie Slaughteriii
Notwithstanding the importance of networks in the academic literature, little solid empirical data is available to measure networks on the country level and the effect on relevant outcome parameters such as government effectiveness, industrial development and/or GDP/per capita on country level. There are many excellent network studies available which quantitatively analyse network effects on organizational level, and case studies which describe the importance of networks. However, few studies are available which aim to capture the degree to which a country is ‘networked’ or connected taking into account that networks develop and are influential on distinct levels (intra-organizational networks, inter-organizational networks and international networks) (for an exception see Maoz, 2010). No overall network index currently exists which enables a comparison between countries and which substantiates the importance of networks for relevant outcomes. So, is it “impossibly complex” to measure these networks?
For the purpose of this report, we explore the possibilities for constructing such an index. Given the limited scope of the study, in terms of both duration and budget, we were able to collect little new data on the basis of surveys, expert interviews, data-mining in raw datasets of existing databases, etc. Figure 2.1, based on the work of Adcock and Collier (2001) presents the ideal-typical process of constructing and validating new concepts and indicators. Starting from the distinction between intra-organizational networks, inter-organizational networks and international networks, we therefore used an
inductive approach to construct an index of connectedness and hence to work our way from level 4 to level 1 in the concept development process. As a result, more than 70 databases (see Annex 1) containing country level data for a significant number of countries were screened for indicators which can be related to international, inter-organizational and intra-organizational networks. In total more than 7000 existing indicators were considered (Annex 2 contains more information on the variable selection
process). Some indicators were identified as being potentially relevant, i.e. proxies for indicators on the levels identified in the report, intra-organizational, inter organizational and international.
This approach has several disadvantages. First, we have to work with available data. The report develops new indicators on the basis of existing data which is not the same as gathering new data on the basis of the concept development framework outlined in figure 2.1. Chapter 1 stressed the importance of institutionalized/embedded ties for knowledge management and knowledge creation. Such a finegrained assessment is not possible if data is not specifically collected from that theoretical perspective. Working with existing data makes it difficult to differentiate between arm-length and embedded networks. Chapter 1 also argued that the ‘ecosystem’ of private sector development consists of many different actors, potentially creating a wide diversity of networks which in all likelihood is not captured in existing datasets. Secondly, while many existing variables were screened there are likely to be many more relevant databases. Future work could focus on identifying these. Nevertheless, the present report does succeed in developing an initial indicator for connectedness. Its use will shed some light on the fruitfulness of continuing the effort to develop more fine-grained measures of connectedness. In subsequent chapters, several proposals for the development of new indicators are made.
What is the result of screening more than 7000 variables with the purpose of identifying network indicators? Surprisingly few indicators are available. Figure 2.2 presents the seven variables which were selected for the purpose of the connectedness index. For international networks we aimed to identify indicators that capture the flows of information and policy diffusion between public authorities, as well as the information flows between economic actors (Slaughter, 2004; Martínez-Diaz &Woods, 2009).
Two indicators were selected to capture this degree of international connectedness, namely the KOF (Swiss Economic Institute) political globalization indicator and the KOF economic networks indicator. The political globalization index captures inter alia the membership in international inter-governmental organizations and the number of international treaties which are signed and ratified by a country.
The economic networks indicator measures the actual economic and financial flows between countries (trade, FDI, portfolio investments). Several other economic indicators capture economic flows, but the KOF is the most comprehensive and suitable one for the purpose of this report.
Three variables were selected to capture the degree of inter-organizational interconnectedness within a country, namely university-industry collaboration, networks and supporting industries and the degree to which individuals are members of professional organizations which are often established for networking purposes. The first two indicators are drawn from the Global Competitiveness Report.
University industry collaboration measures the extent to which business and research collaborate on research and development. It captures the networks between business and universities, when working together pursuing innovations. Networks and supporting industries captures the number and quality of local suppliers and the extent of their interaction (i.e. clusters, or the concentration of interconnected businesses). Both are in the literature on inter-organizational networks and economic geography recognized as important indicators to capture the degree of connectedness between these
organizations. (Podolny & Page, 1998; Powell &Smith-Doerr, 1994; Saxenian et al. 2001; European Commission, 2008) The third indicator is drawn from the World Values Survey and aims to capture networks of professionals that collaborate each other for specific purposes. Networking in the context of professional association can be regarded as a relevant networking strategy in the context of information exchange (see Burt, 1995; Baker, 2000; Putnam, 2000 for a more general argument on the importance of association).
Intra-organizational networks are hard to capture. To measure intra-organizational networks we identified two proxies based on the degree to which firms offer training (Cross & Parker, 2004). The idea is that training enhances internal networks and learning resulting from increased interaction between people within an organization. One measure comes from the World Bank Enterprise Surveys and measures the percentages of firms offering formal training. A second measure is based on the Global
Competitiveness report and focuses on-the-job training which is in turn based on the local availability of specialized research and training services in a country and the extent to which companies in a country invest in training and employee development.
The indicators will be discussed more extensively in the following sections. Figure 2.2 presents the different components of the connectedness index.
Economic Globalization (KOF)
Political Globalization (KOF)
University Industry Collaboration (GCR)
Networks and Supporting Industries (GCR)
Professional Association (WVS)
Firms Offering Training (WB-ES)
On the Job Training (GCR)
To analyse the relationship with relevant outcome variables, the report focuses on four variables, namely two policy-related variables (government effectiveness and regulatory quality) and two economy-related variables (industrial development and GDP per capita). Government effectiveness and regulatory quality are chosen since networks are assumed to contribute to better policy formulation and implementation (see discussion in Part 1).
Government effectiveness and regulatory quality in turn are important for better private sector development and economic development, the ultimate parameters in which we are interested (see also Altenburg (2011, pp. 35-36)). Government effectiveness, from the World Bank governance indicators series, captures different aspects of policymaking and implementation, including the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government’s commitment to such
policies. The link with private sector development is specifically made in the concept of regulatory quality, from the World Bank governance indicators series, which refers to the ability of governments to formulate and implement sound policies and regulations that permit and promote private sector development (Kaufman et al. 2009). The UNIDO Competitive Industrial Performance (CIP) Index benchmarks competitive industrial activity at the country level and is an indicator for industrial development. GDP per capita from the World Development Indicators is included as a second general measure for economic development.
The analysis will focus on the one hand on analysing the variation in the connectedness index and its subindices, and on the other hand on the relationship with other relevant parameters such as policy effectiveness, industrial development and economic development, without implying any causal relationship. Data on the latter indicators is drawn from the World Bank governance indicators, the UNIDO Competitive Industrial Performance Index and World Bank development indicators on GDP per capita PPP. Table 2.1 presents the variables which are used to compose the connectedness index, the sources from which they are drawn and the name of the variable in the source database.
|Table 2.1: Variables and Sources of the Connextedness Index|
|Political Networks||KOF Index of Globalization||Political Globalization|
|Economic Networks||KOF Index of Globalization||Actual flows in economic terms|
|University-Firm Networks||Global Competitiveness Report||University-industry collaboration in R&D|
|Inter-firm Networks||Global Competitiveness Report||Networks and supporting industries|
|Personal Networks||World Values Survey||A072: Member of professional associations or|
|A104: Active/inactive membership of professional organization|
|Formal Training||Enterprise Surveys||L.10: Over fiscal year … [last complete fiscal year], did this establishment have formal training programs for its permanent, full-time employees?|
|On-the-job Training||Global Competitiveness Report||On-the-job training|
|Government Effectiveness||Worldwide Governance Indicators||Government effectiveness|
|Regulatory Quality||Worldwide Governance Indicators||Regulatory quality|
|Competitive Industrial Performance (CIP)||Industrial Development Report||Competitive industrial performance|
|GDP per capita||World Development Indicators||GDP per capita, PPP (current international $) (NY.GDP.PCAP.PP.CD)|
The discussion of the available indicators makes clear that several potential relevant networks are currently not captured in the datasets which were screened in the context of this report. The private sector development ecosystem is such that many types of actors can form relevant knowledge networks. Most importantly no indicators are available, to our knowledge, which capture the degree to which governmental structures are connected to the ‘private’ sector in a country, neither on the level of interportfolio organizational networks nor on the intraorganizational level (for example number of bureaucrats with significant private sector experience).
The following sections discuss the different subindices, the connectedness index and the relationship with relevant other variables, government effectiveness, CIP and GDP/per capita.
2.2. The international networks sub-index
The International Networks sub-index is based on two indicators from the KOF Index of Globalization, political and economic globalization. Political globalization is a proxy for the degree to which states are networked on an international level.
This indicator is based on the number of embassies in a country, the number of international organizations of which the country is a member, the number of UN peace missions in which a country participated, and the number of international treaties a country signed (Dreher, 2006). The proxy for economic globalization (networks) is based on the flows of goods and services (KOF actual flows). This indicator takes into
account the exports and imports of goods and services, foreign direct investments (FDI stocks), the interportfolio of investments of a country, and the income payments to foreign nationals.
After the selection of the indicators, the International Networks Sub-index was created based on the arithmetic mean of political and economic networks, transformed on a scale from 0-1.The sub-index of International Networks covering 121 countries is presented in table 2.2.
|Table 2.2: International Networks Sub-index|
|ISO code||Country||International Network Index||International Network Rank||ISO code||Country||International Network Index||International Network Rank|
|POL||Poland||0.730||33||PNG||Papua New Guinea||0.520||79|
|ZAF||South Africa||0.730||34||BIH||and Herzegovina||0.519||80|
|BRA||Brazil 0.583||60||Sri Lanka||0.408||104|
|EGY||Egypt||0.563||65||Bolivarian Republic of|
|LVA||Latvia||0.538||74||SYR||Syrian Arab Republic 0.260||117|
|TTO||Trinidad||0.538||75||BFA||Burkina Faso 0.255||118|
The international sub-index shows significant variation in the degree to which countries are linked to each other on the international level, both politically as well as economically. The comparison with the median indicates that a significant proportion (more than 50 per cent) of the countries achieve relatively high scores on the sub index. However, several countries also receive lower scores and are outside the international dynamics between countries. It should be noted that a score of zero does not imply that a country is totally unconnected, but that – taken the variation between countries into account and due to the re-scaling of the variables, which is necessary for index-creation (see annex 2) – a country has a score of zero, indicating that in comparison to other countries the international connectedness is very low.
2.3 The interorganizational networks sub-index
The Inter-organizational Networks Sub-index was created based on following three indicators. First, the indicator on networks and supporting industries is taken from the Global Competitiveness Report 2008.
This indicator is based on an Executive Opinion Survey, and takes into account the quality and quantity of local suppliers, and the state of cluster development. The University-Industry Collaboration indicator is also taken from the Global Competitiveness Report that measures to what extent business and universities collaborate on research and development (R&D) in a country. Finally, the professional association indicator, which captures the degree to which individuals are involved in professional associations, was taken from the World Values Survey.
The Inter-organizational Networks Sub-index was created by the arithmetic mean the three indicators, transformed on a scale from 0-1. The Interorganizational Networks sub-index, covering 81 countries, is presented in table 2.3.
|Table 2.3: Inter-organizational Networks Index|
|ISO code||Country||International Network Index||International Network Rank||ISO code||Country||International Network Index||International Network Rank|
|TWN||Taiwan, Province of China||0.817||7||KOR MYS||Korea, Republic of Malaysia||0.730 0.688||15 16|
|JPN||Japan||0.807||8||HKG Hong Kong SAR, China 0.658||17|
|BIH||Bosnia and Herz||0.062||79|
The inter-organizational sub-index also varies significantly between countries. Inter-firm networks (clusters), firm-university networks and personal networks are very highly developed in some countries but underdeveloped in a large number of countries. The median indicates that overall the degree of interorganizational interconnectedness is below 0.5, indicating that a significant number of countries have less developed inter-organizational networks as operationalized in the inter-organizational network sub-index. In our sample, it is partly a consequence of the low level of personal networks measures by the professional association indicator. It should be stressed that this is only a very partial operationalization on the basis of available data, which does not take into account several other elements which could be important in terms of interorganizational networks, most importantly the links between other actors of the private sector development eco-system which are not included in the sub-index. Again, the zero score does not indicate a complete absence of inter-organizational networks, but is a result of the re-scaling method, indicating a comparatively low level of inter-organizational connectedness.
2.4 The intraorganizational network sub-index
The Intra-organizational Networks Sub-index was created based on two indicators. The Percentage of Firms Offering Formal Training comes from the World Bank Enterprise Surveys, most specifically from the question L10 which assessed whether an establishment offered formal training programs for its permanent, full-time employees.
The On-the-job Training indicator from the Global Competitiveness Report 2008-2009 is based on the local availability of specialized research and training services and the extent to which companies invest in training and employee development.
The Intra-organizational Networks sub-index was created by the arithmetic mean of the two training indicators. The index, covering 163 countries, is presented in table 2.4.
|Table 2.4: Intra-organizational Networks Index|
|ISO code||Country||International Network Index||International Network Rank|
|CHE Switzerland 1.000||1|
|DNK Denmark 0.975||2|
|USA United States 0.972||3|
|SWE Sweden 0.940||4|
|NLD Netherlands 0.908||5|
|SGP Singapore 0.893||6|
|WSM Samoa 0.890||7|
|FIN Finland 0.886||8|
|JPN Japan 0.880||9|
|BEL Belgium 0.833||10|
|CAN Canada 0.817||11|
|GBR United Kingdom 0.817||12|
|FRA France 0.804||13|
|NOR Norway 0.801||14|
|ISL Iceland 0.789||15|
|AUS Australia 0.766||16|
|IRL Ireland 0.759||17|
|AUT Austria 0.757||18|
|HKG||Hong Kong SAR, China||0.646||30|
|KOR||Korea, Republic of||0.573||37|
|ARE||United Arab Emirates 0.542||40|
|DOM Dominican Republic 0.433||59|
|VNM Viet Nam||0.393||68|
|LKA||Sri Lanka 0.388||69|
|COD||Congo, Democratic 0.362||77|
|RUS||Russian Federation 0.340||84|
|Bolivarian Republic of|
|Plurinational State of|
|and Gaza Strip|
|SYR||Syrian Arab Republic 0.182||140|
The intra-organizational sub-index varies significantly between countries. The low median score indicates that these instruments to strengthen internal networks are less widespread among countries. A limited number of countries achieve high scores, while a large group of countries receive lower scores, as is indicated by the median. Again, the zero score does not indicate a complete absence of intraorganizational networks, but is a result of the rescaling method, indicating a comparatively low level of intra-organizational connectedness.
2.5 The Connectedness Index
The Connectedness Index is the average of three sub-indices (International, Inter-organizational, and Intra-organizational Networks). It is presented in table 2.5.
|Table 2.5: Connectedness Index|
|Network Index||Network Index||Network Index||Index||Rank|
|SGP NOR CAN||SingaporeNorwayCanada||0.8490.8310.787||0.7600.7980.823||0.8930.8010.817||0.834 0.810 0.809||6 7 8|
|DEU GBR FRA||GermanyUnited KingdomFrance||0.837 0.6960.840||0.8650.7810.616||0.6080.8170.804||0.770 0.7650.754||9 10 11|
|JPN MYS CZE NZL||JapanMalaysiaCzech RepublicNew Zealand||0.8980.844 0.852 0.726||0.8070.6880.5930.629||0.880 0.6080.6620.678||0.728 0.713 0.7020.678||13 14 15 16|
|KOR THA||Korea, Republic of Thailand||0.6390.719||0.730 0.577||0.573 0.640||0.648 0.646||19 20|
|ESP CHL||SpainChile||0.829 0.833||0.494 0.500||0.5060.485||0.6100.606||2425|
|HUN CYP ITA BRA||HungaryCyprus ItalyBrazil||0.940 0.837 0.787 0.583||0.4640.4520.5340.508||0.362 0.4510.3940.572||0.5890.5800.5720.554||27 282930|
|ARG BGR COL||Argentina BulgariaColombia||0.602 0.820 0.554||0.3350.2410.413||0.4500.2910.364||0.4630.451 0.444||383940|
|PHL NGA ROU||PhilippinesNigeriaRomania||0.580 0.714 0.647||0.344 0.313 0.279||0.4070.2920.364||0.4440.440 0.430||414243|
|DOM ZMB||DominicanRepublicZambia||0.4200.687||0.408 0.356||0.4330.215||0.420 0.419||44 45|
|ETH TTO GTM||EthiopiaTrinidad and Tobago
|TUR MEX||TurkeyMexico||0.5140.487||0.381 0.396||0.2860.282||0.3940.388||5354|
|URY ARM||UruguayArmenia||0.654 0.281||0.2210.567||0.2410.224||0.3720.357||5758|
|KGZ BIH VEN||KyrgyzstanBosnia and Herzegovina Venezuela,Bolivarian Republic of||0.506 0.519 0.377||0.0690.062 0.152||0.2930.2800.325||0.2890.2870.285||6566 67|
|DZA MDA||AlgeriaMoldova||0.5390.472||0.0750.000||0.093 0.208||0.236 0.227||70 71|
|GEOTZA||GeorgiaUnited Republic of Tanzania||0.4430.091||0.0640.273||0.142 0.275||0.216 0.213||7374|
The connectedness index clearly shows the overall variation in the degree to which countries are networked, both internally as well as internationally (for a discussion on using the median for comparison purposes see annex 1). Some countries obtain consistently high scores across the various network indicators and hence on the connectedness index, whereas other receive consistently lower scores. Also, it is interesting to note that similar connectedness scores were reached following very distinct paths. For example, Hungary (0.589) and Brazil (0.554) occupy the 27th and 30th ranking positions, respectively.
However, while Brazil is very consistent in the three components of connectedness (0.583 for International Networks, 0.508 for Inter-organizational Networks, and 0.572 for Intra-organizational Networks), the scores of Hungary vary significantly: a very high score is achieved (0.940) in the International Networks Subindex, a mean score in the case of the Inter-organi -zational Networks Sub-index, and a low score (0.362) in the Intra-organizational Networks Sub-index. The similar result in the Connectedness index is, in part, a consequence of our choice of the aggre gation procedure (equal weighting) that uses a full compensability system, i.e., a low score in one indicator is equally compensated by a high score in other.
More generally, the differences on country level between indices are interesting. Some Asian countries, such as Japan and China, score below median on international networks but (very) highly on interorganizational and intra-organizational networks. Others, including some European countries such as Poland and Hungary, score highly on international networks but show only median scores on interorganizational and intra-organizational networks. Still others, such as India, score very highly on one indicator, in casu inter-organizational networks, but below median on the other two indices. This variation, both across countries and within countries, and across types of networks, reveals that very different dynamics are unfolding with regard to the development of networks.
Graphs 2.1-2.3 present the scatter plots between the three sub-indices: international, inter-organization and intra-organization networks. The X and Y-axis present the median scores. The graphs help us to visualize the different scores of countries and between countries on the different network subindices. For example, on the top left of graph 2.2 one can observe that Bulgaria scores very highly in the international sub-index but below the median in the intraorganizational networks sub-index. Another example of the disparity between the sub-indices is the case of India (top of graph 2.3), whose score is very high on inter-organizational networks, but only median on intra-organizational networks.
2.6 The relationship between connectedness and government, industrial and economic performance
In order to analyse the relationship between connectedness and government effectiveness, regulatory quality, competitive industrial performance, and GDP per capita PPP a correlation matrix was constructed. The graphs clearly show a strong positive linear relationship between on the one hand connectedness and on the other hand different performance indicators.
Given the linear relationship between the variables (see graphs 2.4-2.7) the Pearson Product-Moment Correlation Coefficient was used to calculate the correlation between the different indicators (see annex 2).
The correlations are presented in table 2.6.
The analysis clearly shows the strong relationship between connectedness and government effectiveness, regulatory quality, industrial competitiveness and economic development. This is further supported by the high correlations which are all highly significant (see table 2.6). Both the overall connectedness index as well as the subindices on international networks, inter-organizational networks and intra-organizational networks are highly and significantly correlated to the performance indicators.
There are two interesting exceptions. First, regulatory quality, an indicator which is directly related to private sector development, is still highly and statistically significantly correlated with connectedness but the correlation is much less strong than in case of government effectiveness (also compare graphs 2.4 and 2.5). This is an interesting finding which needs to be analysed more in depth, especially since regulatory quality and government effectiveness are highly correlated. An in-depth comparison of countries which score very differently on regulatory quality and government effectiveness in its relationship with connectedness should be further pursued. Secondly, on the level of personal networks measured as membership in professional associations, the table indicates that this network measure is not significantly correlated to any of the performance measures. This can be the result of different methodological and substantial reasons which need to be further explored.
On the one hand, this high correlation is of course an interesting and relevant finding. No correlation would indicate that networks are ‘much ado about nothing’ and that we would not be able to find empirical evidence to support the increased attention for networks. This is clearly not the case. Networks do play an important role. However, the high
correlations also show that much more work is necessary to further understand the concept of networks and assess the impact of networks. The correlations are simply too high to draw many definite conclusions. Several methodological and substantial points are at stake.
First of all, we have to ask ourselves whether the results are spurious, i.e. whether there are any latent variables that drive connectedness and/or its subindices as well as the other variables. With regard to connectedness (especially intra-organizational networks as measured by training) and economic development, it might for example be the case that both are influenced by the development of human capital. Other theoretical reasons might probably be identified which could hypothesize why high correlations occur. Further theoretical development is necessary in this respect.
Secondly, the results might indicate that several of the indicators used are correlated proxies for the same phenomenon and that they are influenced by a same underlying dynamic. The latter can be explored a bit further by a closer inspection of the ranking of the connectedness index. The top 30 consists mostly of OECD Member States with the exception of Brazil, China, Cyprus, Malaysia, Singapore, South Africa and Thailand. Some of these exceptions score (very) highly on indicators such as the Human Development Index (UNDP) or economic development indicators. Hence, connectedness is very high in highly developed or rapidly developing countries. This indicates that networks are highly correlated to the development level. Whether they are a cause, consequence or both cannot be disentangled on the basis of the present analysis.
The latter is related to a third and obvious point that correlation is not causation since we do not know the direction of the cause; a third variable might be involved which is responsible for the covariance between X and Y. Hence the correlations and identified relationships should definitely not be considered causally relevant. International political networks for example can be a consequence of economic development as highly developed economies are more likely to have more embassies because they can afford it. The presence of such a large and highly educated diplomatic corps is also likely to affect the number of agreements a country can initiate, which is another element in the international political networks indicator. Similarly, the degree of university-industry interactions is affected by the presence of an elaborated tertiary educational tier, which in turn is partially a result of the development level of a country.
Although these arguments might reverse causality it should also be noted that the analysis on the subindex level shows that there are several cases where the level of economic development (as measured by CIP or GDP per capita) or policy effectiveness (as measured by government effectiveness and regulatory quality) is the same but the variation in networks very substantial (see graphs in annex 3), indicating that if a reverse causal argument would hold other factors contribute to network development. Taking it a step further, it might be the case that network dynamics emerge which further in time have an effect on the other variables. Much more theoretically informed empirical research is needed to figure out how networks causally play out in the dynamics of increased policy effectiveness, private sector development and economic development. In addition, we need more refined data and time-series to get grip on the issue of causality.
This chapter explored the possibility of constructing an index to capture the degree to which a country is networked on different levels.
The exploration was carried out on the basis of an inductive, data-searching approach. Many datasets and variables were screened. Very few contain data on networks. In addition, the data displays limitations:
Insufficient time series are available for a better causal analysis.
The connectedness index could only be calculated for 75 countries because data is lacking.
The data only very partially captures the idea of networks, both in terms of their structures (the many potential networks which might arise out of the eco-system of private sector development) and of their nature (embedded versus arm-length networks).
The remaining indicators which are included in the index and which are considered as a proxy for networks, such as the intra-organizational ones on training, also capture other aspects such as human capabilities development.
So far, general indicators capturing network effects were considered. One good way forward to capture more precise networks and network effects, especially on the international level, would begin by making use of social network analysis tools and develop indicators on the basis of dyadic relations between countries. Zeev Maoz
(2010) in a very recent publication explored this further and makes convincing arguments for a better exploitation of network tools in the context of international relations and international political economy research.
Finally, the available data only allows for an indirect link to the nexus of networks and knowledge management. Data on knowledge networks is limited and more conceptualization is needed to guide empirical research in this area.
Notwithstanding the limitations of the data, especially from a theoretical and conceptual perspective, it was possible to create a connectedness index to further substantiate the relevance of examining networks. The results show that there is significant variation in networks across countries and also within countries across levels of networks. This is an interesting finding which triggers many questions on how to explain this variation. The variation correlates highly with other outcome variables such as government effectiveness, industrial development and economic development. As such this finding is highly interesting, but not definite causal arguments can be drawn from this link at this stage. Networks are probably cause and consequence and influence other parameters in causality loops. In general, concept development with the aim of developing indicators which capture the ‘network effect’ would best follow the process outlined in figure 2.1. More conceptual and empirical refinement is required. Given the rise of the importance of networks this might be further explored by bringing together experts on international relations, on economic clusters and inter-organizational networks, intra-organizational networks, international and national datasets and social network analysis in order to explore further existing datasets, identify opportunities to create more data and further conceptualize the concept of connectedness as a measurable indicator to capture the degree of network formation.