Printer Friendly

International ICT Trade Dynamics 2004-2014: an explorative network analysis.

Abstract: To assess the impact of information and communication technologies (ICT) in respect to established trade relationships, constructive insights are derived from empirical observations on international bilateral trade of ICT goods and services. Since ICT are often considered a determinant boost for economic growth, the current study investigates countries' interconnections in terms of trade of ICT products and services worldwide. Based on a newly formed dataset, a directed, weighted, multi-layered and temporal network is introduced. This unbiased approach allows the development of an initial exploratory network analysis. This assessment reveals, over the period 2004-2014, that three main central nodes are China, the EU28, and the US. Furthermore, the notable position of EU28 is unveiled in product's exports, in spite of China's prevailing role.

Key words: Network analysis, ICT international bilateral trade, European Union, US, China

The positive effects of information and communications technologies (henceforward, ICT), are assessed by several studies. Following the evolution of endogenous macroeconomic models (ROMER, 1990; AGHION & HOWITT, 1998), the impact of ICT on international trade (CLARKE & WALLSTEN, 2006; FREUND & WEINHOLD, 2002; FREUND & WEINHOLD, 2004; CHOI, 2010; VEMURI & SIDDIQI, 2009), and on economic growth is shown (HARDY, 1980; NORTON, 1992; GREENSTEIN & SPILLER, 1996; ROLLER & WAVERMAN, 2001; OLINER et al., 2007; OLINER & SICHEL, 2000). The decrease in trade costs as a result of ICT use is particularly studied through the implementation of gravity models (FREUND & WEINHOLD, 2002; FREUND & WEINHOLD, 2004; VEMURI & SIDDIQI, 2009; CHOI, 2010; YUSHKOVA, 2014; XING, 2017), which establish the relationship between the dissemination of ICT technologies and greater economic growth rates. However, the ICT role from a structural perspective is still not inquired, allowing the identification of key actors in international trade, and of specific relationships and hierarchies that are present in the sub-domains of international trade of ICT goods and services.

In order to develop an approach based on a systemic perspective, academic research employed methods for complex network analysis. Recent focus is set particularly on international trade networks (imports/exports), observing scale-free degree distribution properties (1) of real networks, thus denoting essential non-random relations (SERRANO & BOGUNA, 2003; LI et al., 2003) within these 'robust and yet fragile' networks (2) (LI et al., 2003). Topological statistics of different orders appear also in relevant literature to apprehend a dissortative tendency (3) in trade networks, and to investigate its evolution over time (FAGIOLO et al., 2009; GARLASCHELLI & LOFFREDO, 2005; De BENEDICTIS & TAJOLA 2011). Lastly, recent analyses have shifted their aim to multi-layered structure (BARIGOZZI et al., 2010), and to the emergence of some relevant regularities of the trade exchange relationships (ZHOU et al., 2016).

In the current study a specific segment of the international trade network is analysed; the international bilateral trade of ICT goods and services. Since the ICT are often interpreted as a determinant variable positively related to economic growth (MINOLA & GIORGINO, 2008; OECD, 2015; World Bank, 2016), the objectives are set on the investigation of countries' interconnections and their hierarchical structure, in terms of international trade in ICT, and on the evolution of this specific structure of relationships over time, for a time period of eleven years (2004-2014). The role of certain countries in the network of ICT trade, and in the sub-networks determined by the exchange of specific categories of products and services, is evaluated through network statistics. As trade exchanges address also the diffusion of innovative information and competences, through the use of products and the integration of services in the economic processes of the receiving countries, thence a central role in the network implies a key position in the dissemination of innovative competences.

In this context, this work complements previous efforts to analyse ICT from different perspectives, such as the performance of the ICT industry of the most important global players using the PREDICT Dataset (DESRUELLE & STANCIK, 2014), the increasingly internationalised environment of ICT R&D by exploring ICT intellectual property and internationalisation dynamics through patent data (De PRATO & NEPELSKI, 2014) and company level data (De PRATO et al., 2011). Selected emerging countries have been the focus of other studies, such as the BRICs countries (SIMON, 2011), China (SIMON, 2012) or Asian countries (De PRATO et al., 2017). In line with previous PREDICT reports on R&D in ICT in the European Union (STANCIK & DESRUELLE, 2012; MAS & FERNANDEZ De GUEVARA RADOSELOVICS, 2013, 2014, 2015), the latest edition (MAS et al. 2017) confirms the progress of several Asian countries (China, Taiwan, South Korea, and India), in terms of macro-economic indicators, in the ICT sector worldwide.

The study is structured as follows; in Section 1 the process of data collection and of data management is described. Subsequently, in Section 2 the definition and construction of the network are presented, and relevant network statistics are introduced. In the last section, an integrated interpretation of the results and the perspectives of the current approach conclude the analysis.

* Data Description

In the framework of the PREDICT project, a dataset regarding international bilateral trade of ICT goods and services is formed. Import, export, re-import and re-export (4) data are collected for the period 2000-2014 for ICT goods and 2004-2014 for ICT services, by country of origin and destination, and by type of goods and services exchanged.

The dataset consists of trade flows (5) between reporting countries and partner countries. Both imports and exports are expressed in Free On Board value. In the case of imports, the reporting country is the country that imports goods and services, namely the country of destination of the trade flow; while in the case of exports, the reporting country is the country of origin of the goods and services. There are no thresholds used to sample the most relevant trade flows, hence all trade flows are considered.

Countries

The coverage of reporting and partner countries is not symmetric in the dataset. The geographical areas covered by the dataset are the EU 28 Member States (6), the EU28 aggregate (7), and other 12 ICT leading countries (8) as both reporting and partner countries. In addition to these 41 geographical entities, the five continents (America, Africa, Asia, Europe and Oceania) and the total (World) are also included as partners, thus as place of origin for imports, and as place of destination for exports. This allows the computation of partners as residual units, by aggregating the non-participant countries in other relation, e.g. Rest of America is computed as the aggregation of all American countries different than the US, Brazil and Canada. Consequently, the coverage of the entire world as reporter or partner (9) is attained, even with a mix of aggregation levels. Ultimately, the units considered are the following 45: Austria, Belgium, Bulgaria, Croatia, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Poland, Portugal, Romania, Slovak Republic, Slovenia, Spain, Sweden, United Kingdom, Norway, Switzerland, Australia, Brazil, Canada, China, India, Japan, Korea, Russia, Taiwan, United States, Africa (10), rest of American countries, rest of Asian countries, rest of European countries and rest of Oceanian countries. The time period that is eventually studied for each of the countries, with data of ICT import and export of goods and services, is the decade 2004-2014, as the unavailability of data for services trade flows before 2004 for certain countries, renders impossible the exhaustive comparative analysis between 2000 and 2004.

Goods and Services Classification

Concerning the disaggregation by typology of goods, the OECD Guide to measuring the Information Society (OECD, 2011) is followed. The ICT goods are defined at 6-digit level using the Harmonised System (henceforward, HS) classification (11), aggregated into 70 blocks of items and organised in five product categories: (i) B1 - computers and peripheral equipment, (ii) B2 -communication equipment, (iii) B3 - consumer electronic equipment, (iv) B4 - electronic components, and (v) B5 - miscellaneous. To provide an analysis with substantial discernible products, ultimately the five categories are considered. For ICT services, the data has been organised according to the extended balance-of-payments categories (EBOPS) for each country. The categories used for the analysis are: (i) C1 - telecommunications services, and (ii) C2 - computer services.

Data Sources and Processing

The main sources for the assembly of the dataset on bilateral trade in ICT goods are OECD's International Trade by Commodity Statistics (ITCS) database, and the Commodity Trade Statistics Database (UN-COMTRADE) developed by the United Nations Statistical Division (UNSD). The data are initially collected in national currency, and ultimately converted into euros using the exchange rates provided by Eurostat. To ensure trade data consistency with National Accounts, total exports (or imports) of goods from ITCS or UN-COMTRADE are rescaled to total exports (or imports) of goods reported in official National Accounts (12). This adjustment creates a homogeneous dataset, which is compatible with the concerned variables included in the PREDICT database, which rely on National Accounts as the main source. Moreover, in order to have constant prices that allow comparability over time, the series are deflated by means of price indices with 2010 as base year (implicit deflators). To compute constant prices, the price index of the country of origin of the trade flow is considered.

Mirror values

Since available data exist from exports and imports, the value for the same trade flow may be reported by the importer and the exporter, hence producing "mirror values" (13). Due to numerous reasons, such as omitting transactions, misidentifying trade partners or product misclassification among others, mirror values differences are observed. Therefore, reconciliation between the mirror flows is required to reduce discrepancies. Several reconciliation approaches are established (GAULIER & ZIGNAGO, 2010; GEHLHAR, 1996), with the most significant referring to the modelling of the bias or variance of the error, or difference between the two reported values. In GAULIER & ZIGNAGO (2010), a weighted average is computed, where the weight increases for countries with low error variances. The inverse relation is based on the rationale that countries producing more reliable data present lower error variances. For this analysis, the reconciliated flow is computed as the arithmetic mean of the mirror flows, following FAGIOLO et al. (2009), thus considering data from both countries (importer and exporter) as equally reliable. In the cases with one value reported for a specific trade flow (14), this single value is directly used.

* Network and Methodology

In this Section, the process of network assembly and of the respective statistics for the analysis are concisely presented.

Network development

For the analysis of ICT international trade, a network approach is followed, and a single network consisted of 18 nodes is formed. From an initial set of 45 geographical areas, in order to perform an initial exploratory analysis intended to investigate world patterns, the countries belonging to the European Union are considered as a single aggregate of 28 countries, the EU28. The network is constructed to provide a multi-graph, thus between two nodes there can be more than one connection. The connections, which if undirected are named edges, or arcs if direction information is provided, represent trade flows between countries, and are characterized by two attributes: (i) the year of the transaction, and (ii) the typology of good or service exchanged.

The direction of the arcs is set to represent the flow of goods or services, signifying that the country exporting the good, or providing the service, is intended as the origin of the arc. The weight of each trade flow is set as the value of the trade in constant euros. The result represents a directed (according to the origin and destination of the trade flow), weighted (according to trade value in euros), multi-layered (according to the typology of goods and services) and temporal network. The computational part for the ensued single network is accomplished through an open source function developed recently (CSARDI & NEPUSZ, 2006 - R package igraph). This structure is adopted in order to extract and analyse from the occurred single network, sub-networks that allow forthcoming analyses on layers and dynamics. Figure 1 shows a graphical representation of the network of all ICT trade flows in year 2014, without any distinction concerning the typology of goods and services.

Network statistics

The analysis begins with the nodes degree, defined as the number of connections that each node comprises. This measure is important for the expression of connectivity, in terms of number of connections (15), of each node of the network. Nevertheless, this statistic is computed acknowledging exclusively the presence of connections, without any information regarding their importance. Thus, to develop observations considering the intensity of the relations that each node attains with the entirety of its neighbours, a second statistic is required, namely the node strength, which is defined as the sum of the weights of the connections of a node. These two statistics are computed considering in-connections and out-connections (i.e., imports and exports). Thence, the computed statistics are the in-degree (D.IN), out-degree (D.OUT), in-strength (S.IN), and out-strength (S.OUT).

As information on topological properties of nodes is essential, a third statistic is computed. As a result of the computation of nodes' weighted betweenness centrality (henceforth, WBC), considerations concerning the position and the relevance of each node, with respect to the entire network structure, are developed. The betweenness centrality is defined as the number of all shortest paths between any two nodes that pass through a given node (FREEMAN, 1977; FREEMAN 1979). Based on the provided information on the connections' weight (i.e., the values of the trade flows), WBC is computed by the definition of NEWMAN (2001) and BRANDES (2001). The reciprocals of the weight of the connections are considered (16) for the calculation of the shortest paths. Betweenness measures are computed with and without addressing the directedness of the trade flows, so as to examine the network as undirected and directed, respectively. On these accounts, the computed statistics are the weighted betweenness centrality in the undirected network (WBC.UND), and weighted betweenness centrality in the directed network (WBC.DIR).

All the statistics are calculated for the time period 2004-2014, for the whole network and for the seven sub-networks (or layers) determined by the seven considered categories of products and services (five categories of goods and two categories of services).

* Explorative statistical analysis

The computation of network statistics for the trade network, namely the network including the aforementioned categories of goods and services, reveals that over the considered period, the countries with the highest number of degrees (both in and out degrees) are the EU28, Russia and the United States. Japan, Australia, China and Canada have a constant number of connections over time, although at levels lower than those of the country with the highest degrees (Table 1). The number of degrees overall remains approximately constant for each country (Figure 2).

The interpretation of the nodes' strength overrules these observations. The total ICT imports (in-strength) indicate that EU28 and United States remain among the top countries (Figure 3, left). China and the rest of the Asian countries (17) show levels considerably higher than those of other countries, while Russia, which presented a significant, high number of degrees for the whole studied period, does not maintain its relevancy on account of the strength of connections. In view of the total exports (out-strength), China noticeably emerges as the only country showing a strong increasing pattern, apart from the period between 2007-2009 (Figure 3, right). Moreover, the relevance of China is also confirmed by the observation of its trade balance (Table 2); for all the period the difference between exports and imports is positive and, with the exception of a single year (2009), it has positive yearly increases.

The result of the nodes' strength computation is established by the observation of the weighted betweenness centrality of the undirected network (WBC.UND) (Figure 4, left panel). Since this statistic measure is computed considering the values of the trade as weights, and without considering the directedness of the trade flows, so without the distinction between imports and exports, China emerges as more central node in ICT trade than EU28 and US, with an increasing trend since 2008 (illustrated by a red dashed line in left panel of Figure 4). WBC.UND underlines this deduction for this country on the importance of its node, in terms also of overall magnitude in ICT trade flows.

Nevertheless, it is similarly crucial to consider a centrality measure that takes into account the direction of the observed trade flows, so distinguishing between imports and exports. An additional pattern emerges from the computation of the weighted betweenness centrality of the directed network (WBC.DIR) (Figure 4, right panel). China remains the country with the sharpest increase and the node with the highest centrality at the end of the period, however the distance with other major countries is not as substantial. At the beginning of the studied time period, EU28 and US apprehend the most central positions in the network and, in spite of China's reinforcement in terms of centrality since 2006, they maintain their key role. Moreover, it is important to mention that after 2012 (red dashed line in right panel of Figure 4), while the WBC.DIR of US and of the rest of Asian countries decreases, EU28 and China both ameliorate their positions.

The centrality comparison for undirected and directed network, promotes the understanding that from the point of view of total trade, regardless of the direction, China is the most central node in the ICT trade network, and other countries follow with significant difference. Nonetheless, the network directedness enables the perception that, for the EU28 and the US, the more balanced proportion between imports and exports, in conjunction with the solid structure of their relationships, support their central role, even if regarding the trade balance appear to be distantiated from China. From an economic point of view, WBC.DIR in a trade network is explicitly relevant to products and services exchanges that are strongly connected to a part of the most innovative economic processes, and consequently to the most innovative information and competences.

In order to further inquire into this observation, the ICT trade network is divided into seven sub-networks, which include exclusively the trade flows of each of the seven specific typologies of exchanged goods and services (Figure 5). Following the previous reasoning, there are not notable findings on the time evolution of nodes' degree, as both in and out degree statistics remain essentially constant since the beginning of the studied period (Figure 5, D.IN and D.OUT, B1 to B5). However, regarding the degrees of the EU28 and China in goods and services, while for goods both have the relatively highest number of degrees for the entire period (Figure 5, D.IN and D.OUT, B1 to B5, masked behind the representation for United States illustrated in red continuous line), in the services sub-networks the number of China's trade partners is considerably lower than that of the EU28 (Figure 5, D.IN and D.OUT, C1 and C2).

Two important remarks arise from the statistics for the in and out strength of goods and services sub-networks. The strength of China in exports is sustained only as a result of its strength in the trade of goods, as in services it does not appear as a relevant node. The EU28 presents high levels of imports of ICT goods that are not always aligned with the values of the corresponding categories of exports (Figure 5, S.IN and S.OUT, B1 to B5). Regarding services, the node representing EU28 indicates that it maintains both the highest imports and exports over the entire studied period (Figure 5, S.IN and S.OUT, C1 and C2).

The previous findings assent to the observation that, in terms of WBC.UND, China evidently emerges as the most important presence among the sub-networks of ICT goods (Figure 5, WBC.UND, B1 to B5), while the EU28 is the most central node among the sub-networks of ICT services (Figure 5),

WBC.UND, C1 and C2). Nevertheless, the resulting statistics from computation of the weighted betweenness centrality of nodes based on the directedness of the network (WBC.DIR), reveal an essential contradiction that indicate the emergence of the EU28 as a key node also in the subnetworks of ICT goods (Figure 5, WBC.DIR B1 to B5). In particular, according to the statistics for sub-networks of ICT goods categories 1, 2 and 3 (B1 - computers and peripheral equipment, B2 - communication equipment, and B3 - consumer electronic equipment, respectively), although the highest WBC.UND occurs for China, and there is a considerable distance between China and the other studied countries, which reflects its total strength in the specific sub-networks, the WBC.DIR denotes that the centrality of EU28 is closer to that of China.

Following the network statistics, the European Union, the United States and China emerge as prevailing actors in the ICT trade flows. Hence, focusing on further assessing their substantial ensued trends and patterns in terms of aggregated ICT goods and services (categories B and C), the latent generating mechanism of the ICT trade data over time is addressed, namely the time series of bilateral trade flows between the three aforementioned parties. For the detailed trend of exports between these actors, the normalised constant values per goods and services categories are presented in Figure I of Appendix I. However, since it is more beneficial to interpret the reaction on trade from previous observations and/or shocks, the variances of normalised constant values of trade flows between EU28, US and China are subsequently analysed. In Figure 6a-c, ICT aggregated goods are depicted in green, and the aggregated services in blue.

In the exporting relationship between the EU28 and the US regarding both goods and services, a discernible influence before and after 2009 is ascertained, while the crisis' onset in 2009 may be suggested since 2006, as it is inferred by the decreasing trend of the variance between 2006 and 2009 (Fig. 6a). After 2010, a steadily increasing variance in the services (Fig. 6a, 6b), implies that the exporting levels of the EU28 towards the US are recovered. A decrease in the imports of goods of the EU28 from the US is also implied, as the decreased trend of the variance for normalised constant values of import of EU28 from the US is illustrated in Fig. 6b (for the disaggregated products time series see Annex I, Fig Ib). Furthermore, a noteworthy increase in the EU28 exporting services to the US is realised after 2010, and a reinforced exporting pattern is also captured towards China after the crisis. Harmonising to the crisis impact, China exporting goods decreased during the period between 2007 and 2009, with one year hysteresis, that is time-invariant as the time series analysis denoted. It is also of paramount importance to highlight that during and after the crisis, the EU28 is consistently exporting services to China, which from an economic point of view reveals the EU28's centrality in a sector where China's presence is limited, as the variance for China's exporting services (Fig. 6c) is substantially zero for the entire considered time period.

* Conclusions

The impact of ICT on international bilateral trade relationships is explored in this study through statistical analyses in networks and time series. Over the period 2004-2014, the considered approach uncovers patterns that establish the European Union, the United States and China as prevailing actors in the ICT trade flows.

The EU28 arises as the most central node in the sub-networks of ICT services, while China appears as the most important in terms of overall magnitude in ICT trade flows, and as the most noticeable country in the subnetworks of ICT goods. During and after the crisis, EU28 has been consistently exporting services to China and the US, which from an economic point of view confirms the EU's high centrality in services subnetworks, where China's presence is acutely limited. It should be also stressed that EU's high levels of imports of ICT goods do not consistently match to the magnitude of export values in the corresponding categories. At the beginning of the period, EU and US led the trade relationships, holding the most central positions. Following China's sharp increase that granted its prominent position since 2011, US and EU maintained a key role, with the latter following China closely. The acknowledgment of the trade flows direction, namely discerning imports from exports, ascertains the centrality of the EU28 comparable to that of China. Therefore, the more balanced proportion between imports and exports of the EU28 and the US, in addition to the findings of strong structure of their relationships, grant them a central role. This observation is masked when considering total trade, namely the sum of imports and exports. The absence of direction information enables China as the most distinct central node, particularly in goods categories B1, B2 and B3 (computers and peripheral equipment, communication equipment, consumer electronic equipment, respectively), with the sharpest increase at the end of the period, and a gap with other major countries.

The network analysis indicated an essentially constant number of trade partners among the countries studied. Russia is among the countries with the highest number of connections, in conjunction with the EU28 and the United States for the whole period, although it is not relevant when taking into account the total value of trade flows. Considering the economic value of total imports and exports, China appeared as expected, as the main exporting country with a markedly positive trade balance, thus justifying its high centrality in the undirected network.

The integrated analysis of ICT trade provides essential insights of the ICT sector, which is relevant to economic growth. As the ICT sector has a mutual relation with innovation and technological development, and since the flows of goods and services are coupled with the associated flows of information and competences - which are required for the appropriate integration of goods and services in the importing economies - the centrality in such a network signifies the prevalence in the management, circulation and diffusion of crucial knowledge.

Further perspectives of this study encompass the implementation of an improved reconciliation method for mirror flows, and the establishment of weights according to the GDP of importing and exporting countries, in order to ensure a common scale for the unbiased impact assessment of trade flows on the involved economies. Moreover, a multivariate stochastic analysis of trade flows trends in subsequent time instances is envisaged, employing network statistics and economic indicators as variables.

Appendix I

References

BARIGOZZI, M., FAGIOLO, S. & GARLASCHELLI, D. (2010): "Multinetwork of international trade: A commodity-specific analysis", Physical Review E., 81.

BRANDES, U. (2001): "A faster algorithm for betweenness centrality", Journal of mathematical sociology, 25 (2), p. 163-177.

CHOI, C. (2010): "The effect of the Internet on service trade", Economics Letters, Volume 109, Issue 2, p. 102-104.

CLARKE, G. & WALLSTEN, S. J. (2006): "Has the Internet Increased Trade? Developed and Developing Country Evidence", Economic Inquiry, 44, p. 465-484.

CSARDI, G. & NEPUSZ, T. (2006): "The igraph software package for complex network research", InterJournal, Complex Systems.

De BENEDICTIS, L. & TAJOLI, L. (2011): The World Trade Network, The World Economy.

De PRATO G. & NEPELSKI, D. (2014): International Protection of ICT Intellectual Property and the Internationalization of ICT R&D, JRC Scientific and Policy Reports, Report EUR 26650 EN.

De PRATO, G., NEPELSKI, D. & STANCIK, J. (2011): Internationalisation of ICT R&D, JRC Scientific and Policy Reports, Report EUR 24888 EN.

De PRATO, G., LOPEZ-COBO, M. & SIMON, J. P. (2017): Dynamics of ICTs: assessing investments in R&D. A global cross-comparison. JRC Technical Report, July, Forthcoming.

DESRUELLE, P. & STANCIK, J. (2014): "Characterizing and comparing the evolution of the major global economies in information and communication technologies", Telecommunications Policy, 38, p. 812-826.

FAGIOLO, G., REYES, J. & SCHIAVO, S. (2009): "The evolution of the world trade web: a weighted-network analysis", Journal of Evolutionary Economics, 20, p.479-514.

FREEMAN, L.:

- (1977): "A Set of Measures of Centrality Based on Betweenness", Sociometry, 40, p. 35-41.

- (1979): "Centrality in Social Networks: Conceptual Clarification", Social Networks, 1, p. 215-239.

FREUND, C. & WEINHOLD, D.:

- (2004): "The effect of the Internet on international trade", Journal of International Economics, 62, Issue 1, p. 171-189.

- (2002): "The Internet and International Trade in Services", American Economic Review, 92, Issue 2, p. 236-240.

FULLER, W. A. (1976): Introduction to Statistical Time Series, New York: John Wiley and Sons, ISBN 0-471-28715-6.

GARLASCHELLI, D. & LOFFREDO, M. I. (2005): Structure and evolution of the world trade network, Physica A, p. 138-144.

GAULIER, G. & ZIGNAGO, S. (2010): BACI: International Trade Database at the Product-level, CEPII Document de travail, No. 23.

GEHLHAR, M. (1996): "Reconciling Bilateral Trade Data for Use in GTAP", GTAP Technical Papers, 11.

GREENSTEIN, S. & SPILLER, P. (1996): "Estimating the Welfare Effects of Digital Infrastructure", No 5770, NBER Working Papers, National Bureau of Economic Research.

HARDY, A. P. (1980): "The role of the telephone in economic development", Telecommunications Policy, 4, Issue 4, p. 278-286.

HOWITT, P. & AGHION, P. (1998): "Capital Accumulation and Innovation as Complementary Factors in Long-Run Growth", Journal of Economic Growth, 3, Issue 2, p. 111-30.

MAS, M. & FERNANDEZ De GUEVARA RADOSELOVICS, J.:

- (2013): The 2013 PREDICT Key Facts Report, An Analysis of ICT R&D in the EU and Beyond.

- (2014): The 2014 PREDICT Key Facts Report, An Analysis of ICT R&D in the EU and Beyond.

- (2015): The 2015 PREDICT Key Facts Report, An Analysis of ICT R&D in the EU and Beyond.

MAS, M., FERNANDEZ De GUEVARA RADOSELOVICS, J., ROBLEDO, J. C., & LOPEZ-COBO, M. (2017): The 2017 PREDICT Key Facts Report, An Analysis of ICT R&D in the EU and Beyond, EUR 28594 EN.

MILLS, T. C. (1991): Time series techniques for economists, Cambridge University Press.

MIN ZHOU, G. & WU, H. (2016): "Structure and formation of top networks in international trade 2001-2010", Social Networks, 44, p. 9-21.

MINOLA, T. & GIORGINO, M. A. (2008): Who's going to provide the funding for high tech start-ups? A model for the analysis of determinants with a fuzzy approach, R&D Management, vol. 38, No. 3, p. 335-351.

NEWMAN, M. E. J. (2001): Scientific collaboration networks. II, Shortest paths, weighted networks, and centrality, Physical Review E, 64.

NORTON, S. (1992): "Transaction Costs, Telecommunications, and the Microeconomics of Macroeconomic Growth", Economic Development and Cultural Change, 41, Issue 1, p. 175-96.

OECD:

- (2015): OECD Digital Economy Outlook 2015.

- (2011): Guide to Measuring the Information Society.

OLINER, S. D., & SICHEL, D. E. (2000): "The Resurgence of Growth in the Late 1990s: is Information Technology the Story?", Journal of Economic Perspectives, Vol. 14, No. 4, p. 3-22.

OLINER, S. D., SICHEL, D. E. & STIROH, K. J. (2007): "Explaining a Productive Decade", Brookings Papers on Economic Activity, No. 2.

ROLLER, L.-H. & WAVERMAN, L. (2001): "Telecommunications Infrastructure and Economic Development: A Simultaneous Approach", American Economic Review, 91, Issue 4, p. 909-923.

ROMER, P. (1990): "Endogenous Technological Change", Journal of Political Economy, 98, Issue 5, p. S71-102.

SERRANO, M. A. & BOGUNA, M. (2003): "Topology of the World Trade Web", Physical Review E, 68.

SIMON, J. P.:

- (2017): "Dynamics of ICTs: Industries, R&D, Markets", DigiWorld Economic Journal, no 105, 1st quarter 2017, p. 121-138.

- (2012): "The ICT Landscape in BRICS Countries", Digiworld Economic Journal, no 85, p. 191-202.

- (2011): "The ICT Landscape in BRICS Countries: Brazil, India, China", JRC Scientific and Technical Reports.

STANCIK, J. & DESRUELLE, P. (2012): The 2012 PREDICT Key Facts Report, An Analysis of ICT R&D in the EU and Beyond.

The World Bank, (2016): World development report 2016: Digital dividends.

United Nations, (2017): International Trade Statistics Knowledgebase.

VEMURI, V. K. & SIDDIQI, S. (2009): "Impact of Commercialization of the Internet on International Trade: A Panel Study Using the Extended Gravity Model", The International Trade Journal, 23:4, p. 458-484.

XIANG, L., YU, Y. J. & GUANRONG, C. (2003): "Complexity and synchronization of the World trade Web", Physica A, 328, p. 287-296.

XING, Z. (2017): The impacts of Information and Communications Technology (ICT) and E-commerce on bilateral trade flows, International Economics and Economic Policy.

YUSHKOVA, E. (2014): Impact of ICT on trade in different technology groups: analysis

Riccardo RIGHI, Sofia SAMOILI, Montserrat LOPEZ-COBO, & Giuditta De PRATO

European Commission, Joint Research Centre (JRC), Growth and Innovation, Digital Economy Unit, Seville, Spain

(*) Disclaimer: The views expressed in the present work are purely those of the authors and may not under any circumstances be regarded as stating an official position of the European Commission. Neither the European Commission nor any person acting on behalf of the Commission is responsible for the use which might be made of this work.

Box 1 - The PREDICT research project

PREDICT, "Prospective insights on R&D in ICT", has been producing statistics and analyses on ICT industries and their R&D in Europe since 2006. The project covers major world competitors including 40 advanced and emerging countries--the EU28 plus Norway, Russia and Switzerland in Europe, Canada, the United States and Brazil in the Americas, China, India, Japan, South Korea and Taiwan in Asia, and Australia. It also covers a growing array of indicators related to the ICT content of economic activities.

PREDICT is a collaboration between the JRC and the European Commission Communications Networks, Content and Technology (CNECT) Directorate General. Since 2013, data collection and analysis has been carried out jointly by the JRC and the Valencian Institute of Economic Research (Instituto Valenciano de Investigaciones Economicas - Ivie). PREDICT relies on the latest available official statistics delivered by countries, Eurostat and the OECD. The project has been designed to help policy makers to understand dynamics in the ICT sector and to foster its growth.

More information, data and reports produced by the PREDICT project are available online at the project website: https://ec.europa.eu/jrc/en/predict.

For a summary of the main results of the latest PREDICT Dataset see De PRATO et al. 2017, MAS et al. 2017 and SIMON 2017.

Box 2 - Univariate time series analysis

To evaluate in an unbiased approach the latent mechanism that generates international bilateral trade of ICT goods and services over time, and any external implied impact or shock factors, the behaviour of the respective time series and the stationarity aspect is assayed. The latter is essential to the time series analysis, in order to provide robust conclusions and anticipate patterns, based on previous observations, random shocks, or both, as in non-stationary time series two behavioural effects mainly occur. The persistence and the magnitude of a behavioural effect could be misinterpreted due to its time-independence, and moreover, time series with similar trend could appear to be correlated, without substantial mutual economic causality basis. The existence and underlying structure of potential correlations among observations of the ICT trade flows time series is identified through stationarity assessment, so as to form suitable trading policy frameworks disengaged from ensued distortions from unidentified correlations leading to bias.

The stochastic process is assumed stationary when its variables are time-invariant, hence given the studied ICT trade data, if the studied historical trade values are denoted as the univariate random variable Y, then accordingly, the joint distributions of the conditional random variable [Y.sub.t] |[F.sub.t-1], namely [D.sub.Yt], Yt-1, ..., [D.sub.Yt], Yt-p for a finite number of time instances p, of the cumulative distribution function [F.sub.t-1]-={[Y.sub.t-1], ..., [Y.sub.t-p]}, are equal for each time instance over the specified time period. Assuming normality, from the joint distributions, the two measures that are studied expecting to be constant over time, are the conditional mean E([Y.sub.t]|[Y.sub.t-1],...,[Y.sub.t-p]) and conditional variance Var([Y.sub.t]|[Y.sub.t-1],...,[Y.sub.t-p]) of each t, while the covariance is explained as a function of temporal distance between two observations, and not by the time instance (MILLS, 1991). Otherwise, in a regression analysis the time-related variables of a non-stationary time series, although unrelated to each other, would lead to: i) false significant models, where the developed model would be falsely fitted to data, and to ii) unreliable assumptions for asymptotic analyses and tests (t-/F-distribution tests will not follow the corresponding distributions) that will undermine the hypothesis tests for the parameters. To avoid a spurious regression, diagnostic stationarity tests can be performed, so as to estimate correlations and autocorrelations between time-dependent variables of a stochastic process, which yield meaningful observations.

The Augmented Dickey-Fuller (ADF) autoregressive process is employed in this study to test the null hypothesis (H0) of the presence of a unit root in a time series, with alternative hypothesis its stationarity, on the model of Equation 1:

[mathematical expression not reproducible] (1)

If a unit root occurs from the equation, [alpha]=1, or [alpha]-1=0 or [beta]=0, then the series is not stationary. Equation (1) is computed for a first-order autoregressive coefficient, and the regression resumes when the upper bound of the rate at which the number of lags used to increase the sample size of a general autoregressive model, is reached. Based on the critical values of the Dickey-Fuller tables (FULLER, 1976) and the interpolated p-values of BANERJEE et al. (1993), the null hypothesis is rejected for the studied trade flows, so the examined ICT trade series are assumed stationary.

(1) A scale free network is a network where the degree follows a power law distribution, thence few nodes have a very large number of connections, and the majority of the nodes have very few connections.

(2) Scale free networks are considered as 'robust and yet fragile', because although the random removal of some nodes induces insignificant alterations in their structure, if certain nodes with many connections are removed, the probability that the network dismantles is significantly increasing. The fact that most real networks are scale free renders this typology of networks greatly relevant in the study of complex networks.

(3) The tendency of actors to establish connections with actors with different features. One example is a network in which nodes with few connections tend to establish connections only with nodes that have a large number of connections, and vice-versa.

(4) Re-exports are exports of foreign goods in the same state as previously imported. Re-imports are goods imported in the same state as previously exported. There is multiple causality for the return of an exported good to the country of origin: i) the exported good might be defective, ii) the importer might have defaulted on payments or cancelled the order, iii) the authorities might have imposed an import barrier, or iv) demand or prices in the country of origin might have rendered more advantageous the return of the good to the origin (UN, 2017).

(5) The trade flows are corrected deducting re-imports from total imports and re-exports from total exports. This correction produced negative values in a very small proportion of cases (less than 0.3%), which were excluded from the analysis.

(6) Austria, Belgium, Bulgaria, Croatia, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Poland, Portugal, Romania, Slovak Republic, Slovenia, Spain, Sweden, United Kingdom.

(7) The EU28 aggregate was constructed as the sum of data from the 28 countries that conformed the European Union at the end of the period analysed, irrespective their adhesion date. Therefore, throughout the analysis, the EU28 entity refers to the aggregate of these 28 countries, although some of them did not belong to the European Union in part of the period considered.

(8) Norway, Switzerland, Australia, Brazil, Canada, China, India, Japan, Korea, Russia, Taiwan and United States.

(9) The only missing trade flows in the dataset are the bilateral flows between non-reporting countries, that is, when considering continents and residual areas (e.g. Asia to America or Asia to Rest of Europe).

(10) Since no African country is considered as reporting country, there is not residual unit for this continent.

(11) The HS has been revised three times since the first revision in 1996 (in 2002 [Rev. 2], 2007 [Rev. 3] and 2012 [Rev. 4]). To address the gap between the HS2002 and the HS2007, as it is impossible to have one-to-one correspondences between these classifications, "blocks of items" are used following OECD (2011).

(12) In foreign trade statistics, exports of goods are reported FOB (free on board), whereas imports are reported CIF (cost, insurance and freight).Exports and imports in National Accounts are both expressed in FOB terms. Thus, by rescaling the data to total imports of goods in National Accounts, the CIF term is assumed to be homogeneously distributed among the different sectors, and the series previously expressed in CIF are converted into FOB terms.

(13) In this dataset there are mirror values for bilateral trade flows between the 40 reporting countries.

(14) This occurred for trade flows between one of the 40 reporting countries and a continent or a residual geographical unit.

(15) The degree of a vertex is calculated as the number of adjacent edges/arcs. Since the network is created as a multigraph, two countries may establish more than one connection per year. Thus, in the computation of the countries degrees, the maximum value for each country (in a single year) is given by the product of the number of countries minus one, and the number of products/services categories (i.e., 17*7=119).

(16) This implies that relative high weights of the connection between two nodes, shorten the length of the path.
Table 1 - In-degrees and out-degrees of the seven countries with the
highest number of overall degrees, in 2004, 2009 and 2014

           In       In       In       Out      Out      Out
           degrees  degrees  degrees  degrees  degrees  degrees
           in 2004  in 2009  in 2014  in 2004  in 2009  in 2014

EU28       115      115      115      117      116      116
Russia     113      117      118      112      116      117
US         107      117      117      108      118      118
Japan       97       98      101       98       98      100
Australia   99       98       97       98       99       99
China       90       91       91       90       91       91
Canada      90       91       91       90       91       91

Table 2 - Trade balance (total exports minus total imports) of China,
EU28, US and the rest of Asian countries in 2004, 2009 and 2014

Balance of ICT trade normalised   2004     2009    2014
values of in and out strength

China                             0.189    0.398   0.622
EU28                             -0.202   -0.127  -0.108
Rest of Asian countries           0.073   -0.113  -0.300

Values are calculated as the difference between the scaled values (to
the maximum recorded all over the period) of total exports and total
imports (i.e. out-strength minus in-strength).
COPYRIGHT 2017 IDATE
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2017 Gale, Cengage Learning. All rights reserved.

 
Article Details
Printer friendly Cite/link Email Feedback
Author:Righi, Riccardo; Samoili, Sofia; Lopez-Cobo, Montserrat; De Prato, Giuditta
Publication:DigiWorld Economic Journal Communications & Strategies
Geographic Code:9CHIN
Date:Jul 1, 2017
Words:7033
Previous Article:Cross-Border Data Flows and Growth in Europe.
Next Article:Implications of the GATS for Digital Trade and Digital Trade Barriers in Services.

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters