The Architecture of Globalization:
A Network Approach to International Economic Integration.
Raja Kali and Javier Reyes1
Department of Economics
Sam M. Walton College of Business
University of Arkansas
Fayetteville, AR 72701
Abstract
We combine data on international trade linkages with network methods to examine the global
trading system as an interdependent complex network. We map the topology of the international
trade network and suggest new network based measures of international economic integration, at
both a global system-wide level and a local country-level. We develop network based measures that
incorporate not only the volume of trade but also the influence that a country has on the international
trading system. These measures incorporate the structure and function of the network and may
provide a more meaningful approach to globalization than current measures based on trade volumes.
We find that in terms of participation and influence in the network, global trade is hierarchical with a
core-periphery structure at higher levels of trade, though integration of smaller countries into the
network increased considerably over the 1990’s. The network is strongly “balkanized” according to
geography of trading partners but not as strongly by income or legal origin. Using these new
measures we find that a country’s position in the network has substantial implications for economic
growth. We therefore suggest that a network approach to international economic integration has
potential for useful applications in international business, finance and development.
Keywords: globalization, economic integration, networks, international trade
1
Email: rkali@walton.uark.edu, jreyes@walton.uark.edu
We are grateful to Jon Johnson, Fabio Mendez and Anand Swamy for helpful discussions. We thank seminar participants
at Harvard Business School, the University of Arkansas, Williams College, IGIDR-Mumbai and MWIEG Fall 2005 for
their comments. Viktoria Riiman provided outstanding reseach assistance.
,I. Introduction.
While popular usage of the term “globalization” provokes strong and polarizing opinions across the
world, such sentiments are usually associated with the effects, real or perceived, of what economists refer to
as international economic integration. The increase in international economic integration that has
characterized the last half-century has been associated with the spectacular economic performance and move
out of poverty for large parts of the world (Sachs and Warner, 1995), but also with the increase in the
volatility of country-level performance, reflected in several recent episodes of economic and financial “crises”
(Forbes 2001). There is also a growing perception that the process of globalization has accelerated over the
last decade and that the benefits and costs of increasing economic integration have not been evenly distributed
across the world (Stiglitz, 2002; Bhagwati, 2004).
Despite a sharp increase in interest on these issues, discussions are often handicapped by the dearth of
appropriate measures of international economic integration. Most studies of international economic
integration or globalization in the economics literature focus on the volume of trade (exports and/or imports
as a fraction of total trade) between countries, or define “trade integration” as the sum of exports and imports
divided by GDP (see for example Rodrik, 2000, IMF World Economic Outlook, 2002). While these
indicators2 have been useful, the literature recognizes their shortcomings (which we describe in more detail
below). Nevertheless, they are still widely used for studying international economic integration, primarily for
lack of better alternatives.
Recent advances in the study of networks (Albert and Barabasi, 2002; Newman, 2003) have placed
elegant and powerful tools at our disposal, enabling us to suggest alternative measures of international
economic integration (henceforth IEI) that turn from a sole focus on individual country trade levels to a
consideration of the pattern of linkages that tie together countries around the world as a whole. In this paper
we combine a network approach with data on international trade linkages in order to examine the global
trading system as an interdependent complex network3. A network approach enables us to derive statistics
that describe the structure and evolution of global trade in ways that existing measures do not capture, such as
the number of actual and potential trading partners, the structure of regional trading and the influence of
individual countries and groups of countries for the whole network and for specific regions. We use this
change in perspective toward IEI to suggest new measures of integration that provide insights into global
trade that have been overlooked by the literature.
2 Other measures based on volumes such as gross private flows to GDP, and total trade to merchandise value added also
fall into this category.
3
Complex networks are large scale graphs that are composed of so many nodes and links that they cannot be
meaningfully visualized and analyzed using standard graph theory. Recent advances in network research now enable us
to analyze such graphs in terms of their statistical properties. Albert and Barabasi (2002) and Newman (2003) are
excellent surveys of these methods.
2
, With this objective, we first map the topology of the international trade network with a view to
understanding its structure and properties. Armed with such an understanding, we then suggest new measures
of IEI, at both a “local”, country-level, and a “global”, system-wide level, that incorporate the structure and
function of the network. We use these measures to parse IEI along a number of different lines: geography,
income and legal origin. This enables an examination of whether global trade has become more integrated or
“balkanized” along these dimensions. We suggest network-based measures that capture not only the volume
of trade but also the “influence” that a country may have on the international trading system. We have data
on the network of international trade linkages at two points in time, 1992 and 1998, and are able to construct
these measures for both years and examine how the network and thus “globalization” has evolved over the
1990’s. Since trade levels vary considerably from country to country and there could be some debate over
what constitutes “consequential” levels of trade, we construct the network for different trade level thresholds4.
We find that at low levels of trade, the global trading network has become much more integrated, while at
higher levels of trade it has not changed much. At low levels of trade, the global trade network is quite
decentralized and homogenous but at higher levels of trade the network looks much more hierarchical and
heterogeneous, with a core-periphery structure. We also find that there is a high level of multilateralism in
global trade and this has not changed much between 1992 and 19985.
As an application, and to demonstrate the potential of the network approach to IEI, we use our
measures of network importance in a cross-country growth regression and find they are all statistically and
economically significant, have the expected signs and raise the explanatory power of the regression above that
obtained using only volume based measures current in the literature. Using one of our measures of local
integration, degree centrality, a measure of how centrally located a country is in the network6, we find that an
improvement in the centrality ranking by 10 units at the two percent trade-link threshold increases the average
growth rate of per capita GDP by 1.1 percentage points.7 A country’s position in the network can thus have
substantial implications for development outcomes.
The paper is organized as follows. Section II describes the data and definitions that we use to
organize the trade-link data. Section III applies concepts from network analysis to understand properties of
4 We describe this procedure in more detail in section II.
5
While we believe this is the first exercise to explicitly chart the topology of the international trade network and suggest
the use of this topology for the understanding of economic integration, we are by no means the first to use network ideas
in international business and economics. An excellent introduction to this literature is Rauch and Casella (2001) and the
critique by Zuckerman (2003). Systems- or network-based measures of globalization have, to the best of our
knowledge, not been used in economics before, but there is antecedent in the sociology literature. A paper by Smith and
White (1992) uses international trade flow data to consider the change in the structure of the international division of
labor with the goal of understanding patterns and cycles of hegemony in the world-system. The focus of this work is
thus quite different from ours.
6 We describe various network measures in more detail below.
7 This is judged to be a substantial effect by the standards of the literature. For example, Yanikkaya (2003) finds that an
increase of 10% in the total trade to GDP ratio would increase the average growth rate of per capita GDP by 0.18%.
3
, the network. We first provide an overview of the topology of the network and then delve deeper into the data
and propose measures of local and global economic integration. Section IV is our application to economic
growth. Section V summarizes our findings and suggests further applications of these measures.
II. Definitions and Data.
The first step in our approach is to identify the fundamental building blocks of the network and their
specific properties. A network is a set of points, called nodes or vertices, with connections between them,
called links or edges. In our context, each country is considered to be a node of the network. Since
international trade is usually measured using the monetary value of exports and imports between countries,
trading relationships are analogous to valued links in a network, and these vary from country to country. In
order to chart the structure of the network we wish to take into account the magnitude of these relationships
but not specifically their exact value.
We do this by considering a network link between two countries to be present if the trade level
between them is above a certain threshold. Specifically, we define a trade-link between country i and country
j to be present if the value of exports from country i to country j as a proportion of country i’s total exports is
greater than or equal to a given magnitude. Since exports of country i to country j are in effect imports of j
from i we are able to construct both export and import networks in order to understand IEI from both sides.
Moreover, since trade levels vary considerably from country to country and there could be some debate over
what constitutes “consequential” levels of trade, we construct the network for different trade level thresholds,
which we explain below. Examining how the structure of the network changes as the trade threshold used to
define the presence of links varies also enables us to understand the sensitivity of various topological
characteristics of the network to differing trade magnitudes. Constructing the network for different thresholds
enables us to incorporate both magnitudes and network features into our analysis. Using thresholds enables
us to avoid working directly with valued-directed links even though implicitly these thresholds embody the
values of the trade links in our data.
The data used for our international trade network was extracted from the COMTRADE Database of
the United Nations8. We use the US dollar value of exports and imports of all commodities between 189
countries for 1992 and between 192 countries in 19989. Countries are the nodes of the network and a link
8
United Nations database STIC 1.
9
A list of countries is included in Table 1A of the Data Appendix. It should be noted that even though our trade network
is extensive, it is not all-inclusive. The United Nations database includes more than 230 countries/areas, plus some NES
(not elsewhere specified) areas. Although we compute the total exports and total imports from the all-inclusive raw
database, in our trade network analysis we only include countries. In other words we ignore regions and NES figures.
Additionally some countries, like Guadeloupe, Martinique, Reunion, and others, are excluded from our analysis because
there are some inconsistencies in the data reported by these countries.
4
