INA - Papers








INA Working Paper Series

Who Calls Whom? Global Hierarchies and Telephone Networks

Hugh Louch, Eszter Hargittai, and Miguel Angel Centeno

Princeton University

(The most recent version of this paper appears in The Washington Quarterly.The correct citation for that paper is:
Louch, Hugh, Eszter Hargittai and Miguel Angel Centeno. 1999. Phone Calls and Fax Machines: The Limits to Globalization.
The Washington Quarterly. 22:2 83-100)

Globalization is everywhere. States, economies, and societies are increasingly integrated. Flows of goods, capital, humans, and cultural objects now link all of us in a global integrated web (Mittelman 1997, Hirst and Thompson, 1996). The development of international trade has had the most immediate (or most visible) consequence, but money, in and of itself, has come to play an arguably even larger role than does the transfer of material goods. Labor, while still subject to much greater control than capital, moves transnationally while tourism now involves an estimated 600 million international travelers a year. Finally, the ubiquity of CNN is already a cliché, and entertainment industry budgets now make calculations on the basis of a global market.

There is little doubt that we are undergoing a process of compression of international time and space; no country, economy, or society can expect to remain an island. The separation of production and consumption that is the heart of the market economy appears to have reached its zenith. Political structures are not immune to apparently overwhelming forces of globalization; even the most powerful states tremble before the vagaries of the international market. Globalization is not just another "buzz-word" (globaloney?), but very much a real and significant phenomenon.

But, what does it mean? What does a globalized world look like? Much of the writing on globalization takes one of two equally simplistic positions (Epstein, Crotty, and Kelly, 1996, Rodrik 1997b). The optimists view globalization as a practically unmixed blessing destined to increase economic production, reduce political tensions, and create new forms of communication over and above national loyalties and customs. Pessimists see it as increasing the already wide gulf between haves and have-nots. With few exceptions (Barnett and Salisbury 1996, Snyder and Kick 1979, Smith and White 1992), however, analyses of globalization have been largely lacking in any systematic empirical basis and much of it has been descriptive. Anecdotal evidence of Rambo screenings in Borneo, Chunnels, and 24 hour stock trading, can only give a flavor of the changes taking place; they do not make for a clear picture of what, if any, structural transformations are occurring.

We use changes in international telecommunications to explore the process of globalization. Telephone communication nicely captures all the various elements involved in globalization. It reflects contacts on a variety of social levels; from the frantic calls of high finance to the family conversations of illegal workers. Unlike most measures, telephone contact includes both the social and economic dimensions involved in globalization. Telecommunications transmit information on the movement of goods, capital, and human beings (Kellerman 1985, Dökmecí and Lale Berköz 1996). Analyses of individual country telephone patterns indicate that telephone traffic is correlated with both economic relations (e.g. trade) and social ties (e.g. tourism and migration). Telephone networks, therefore, are not only a part of the increasing interconnectivity, but they also serve as a wonderful reflection of the myriad of links.

We ask a set of simple questions: How much has telecommunications grown? How has that growth been distributed? How has the shape of the international network of countries changed over the past 15 years? In short, who calls whom? Readers should note that we are not making causal claims regarding how or why these trends developed. Such analysis will require parallel data on trade, migration, etc. (we are currently in the process of gathering this information). For now, however, we are limited to describing what we see rather than explaining it. Given the central importance of this phenomenon and the relative scarcity of solid information regarding it, we consider even such a limited goal worthwhile.

Models of the Global Web

There are currently three dominant interpretations of the process of global interdependence and we have used them to structure our discussion of the data.

Interdependent Globalization is defined by a generic and system wide increase in reciprocal ties between countries. From this perspective, the web of international contact is expanding in fairly uniform shape. This has been the largely accepted assumption behind much of the discussion of increasing inter-connection. A universal globalization process would be characterized by an exponential increase in absolute telephone traffic and a homogenous rate of growth of contact between definable clusters of countries.

Civilizations and Empires is a much less optimistic view defined by increasing concentration of communication within clusters of countries united by a common cultural heritage or congruent to ex-imperial countries, historical flows of trade and contemporary financial flows. According to the first variant, associated with the recent work of Samuel Huntington (1993, 1996) we have and will continue to see increasing divisions across cultural lines. For our purposes, this would be characterized by a presence of distinct sub-networks congruent with language or culture and an increase in contact within these sub-networks and relative decline in contact between these clusters. The Empires perspective serves as a materialist parallel to the Huntingtonian clash. In this case, we would expect clustering around the three major powers of US, European Union (Germany), and Japan. This pattern would be characterized by a presence of distinct sub-networks following economic, ex-colonial links, or regional links with asymmetrical positions of states within these sub-networks.

Finally, we define Hegemonic Globalization by the consistent or increasing centrality of a small core of rich countries and perhaps the domination by a single power. This view sees globalization as merely an acceleration of the concentration of resources and influence in European and North American cluster with some limited East Asian additions. In our data, this would be reflected in an increasing asymmetrical network centrality of the United States and an increasing centrality of core countries.

These three generic perspectives obviously do not exhaust the possible models. They do serve as examples, however, of the three basic network patterns. The general globalization model assumes a world in which each unit is equally connected to each other unit as in the first diagram below. The second perceives the world as a series of closed cliques with little communication between each other, and intensive interaction within the receptive groups as in the middle diagram. The third imagines the world system as the spokes on a wheel with little or no contact between the various points on the circumference and each devoting most of their attention to contact with the center. It should be obvious that globalization will have very different consequences depending on which model it follows. Each of the following sections discusses how the data fit these different perspectives.

How Big a Web?

While there are more international phone calls going on every day, there is no evidence of a dramatic growth over the past decade. The rate of annual change has been quite consistent. When we standardize the data to take into account population and economic growth (using the number of total telephones), the propensity to call has remained remarkably stable after a sharp increase in the late 1980s (Figure I). Thus, globalization (as measured by telephone contact) does not seem to be accelerating. Individual countries have experienced this growth in quite different ways. While the average amount of contact increases over time, there is a parallel increase in variance. That is, in later years, there is a larger gulf between countries that make and receive a lot of calls and those who make few (Figure I). Figure II illustrates the basic inequality in the distribution of calls. It separates the 20 most frequently called countries from the bottom 50. Obviously the more "popular" are getting ever more so.

To what extent are countries speaking with the same partners of a decade ago? We defined telephone cliques or groups of countries that called each other. Almost half of all these cliques defined by levels of mutual contact from 1983 still exist in their exact form in 1995 (54 of 109). For 63% of 1983 cliques there is a clique in 1995 which matches 90% of the actors. Finally, for each clique in 1983 excepting one, there is a clique in 1995 where at least 60% of the members match.

This evidence makes us question the supposed universality of globalization. Rather, it appears that certain countries have increased the amount of contact with each other, but that this has been a relatively steady process over the past two decades. No communication revolution has taken place, nor has the "social hierarchy" of telephone contact been remarkably changed. This is not to deny the technological transformation that has occurred, but to question the extent to which this has transformed the underlying structure of international relations.

If it is fairly clear that the international telephone network of countries is highly stratified, the next step is to determine its underlying structure.

Clashing Cliques?

With one exception, there is no evidence of increasing integration along cultural lines or homogenization of phone contacts. If we analyze the Islamic world, the "civilization" that most concerns Huntington, we find that the proclivity to contact other countries within this supposed block has remained consistent around 38-39% since 1983. This pattern remains even when we standardize the data to take into account development and population growth. Countries in the "European category" (Western Europe and the Mediterranean sans Iberia, and the United States and Canada) are actually the ones most likely to call fellow cultural partners, but even here, there has been a decline in the relative importance of intra-category communication. The one exception where we do note an increase in "auto-communication" is what we have called the Buddhist/Confucian "civilization" (Figure IIIa & IIIb). This is an obvious instance where comparisons with other data are critical. Given its exceptionality and the changes that occurred in this region during the past decade, we suspect that this calling effect has little to do with a common culture and is more a result of economic integration.

If we look at selected countries we note the same pattern: no sign of an increase in the clustering of "Islamic" countries, but some sign of a similar process among "Confucian/Buddhist" ones which could be driven by non-cultural reasons. Egypt’s contact with other Islamic countries has remained relatively constant as a proportion of its total telephone traffic since 1983 (64% in and 41% out to 61% in and 43% out). The clique analysis does not support an increasing integration of Egypt within an Islamic world. It has only added one dominantly Islamic clique in the past decade (which also includes the United States) as well as one featuring Israel. Taiwan and Singapore’s contact with "Confucian" countries has increased marginally (20% in/28% out to 30% in/30% out and 40% in/55% out to 45% in 56% out, respectively). While Taiwan has increased its membership in cliques dominated by other countries in the region (see below), it would be difficult to see a dominant cultural pattern without verging on a racist judgment that "Asians are all alike".

The notion of "empire", must also be treated with some caution (Figure IV a-e). Obviously, language, personal contacts, and economic legacies will continue to shape the pattern of international relationships for some time to come. The telephone data do reflect the influence of France in West Africa, that of Britain in parts of the Middle East and East Asia, and the dominant position of the United States in Latin America. But, historical legacies may not be as strong as is often assumed in classic dependency literature (Chase-Dunn and Grimes 1995, Gereffi and Fonda 1992). Despite its efforts to maintain a francophone zone, for example, France’s centrality within its old Empire indicates declining importance for the old metropol. The United States' relative importance appears also to have declined in terms of Latin American international communication (reflecting the growth of regional ties such as Mercosur). On the other hand, the pattern for the ex-British Empire would indicate increasing importance of the United Kingdom. Again, the point is not to deny the important roles contacts with the imperial power play in the ex-colonies, but to analyze this trend within a pattern of globalization. With regards possible "new" empires we note that Germany does not appear to have established a dominant position vis-à-vis Eastern Europe, despite all the rhetoric about Mitteleuropa. Following or findings on East Asia, Japan does appear to play a much more central role in the telecommunications of that region.

The Rich Get Richer

The most salient finding from the ITU data is the continuing and, in some cases increasing, concentration of calls to rich countries. Even when we standardized the data (thereby controlling for the greater number of phones in the developed world) the pattern continues. Separate analyses exploring the number of calls among the other categories indicate little contact between the varying levels of peripheries. There appears to be an inverse relationship between income level and likelihood to concentrate calls on the rich core. (The exception is the ex-socialist countries and again this shift is partly accounted for by the increase in number of countries and thus the shift in calls from national to international). Overall, the lesser-developed countries have little independent contact between each other and concentrate their international inter-action with the dominant powers (Table I).

The "rich" also tend to pay more attention to each other. Looking at clique structure, we note that both the G7 and other high-income countries have grown increasingly integrated and now tend to share a larger number of common cliques (the mean number of rich members in any clique increased from 5.5 in 1983 to 6.3 in 1995; that of the G7 increased from 3.1 to 3.5). These same countries are also in everyone’s cliques. There were only 2 cliques without "rich" countries in 1995! The developed countries also dominate the top cliques in terms of proportion of "inter-clique" calls sent or received.

On the other side of the global hierarchy stands an apparently permanent "underclass" of countries. The number of countries that do not qualify to belong to a single clique (they do not speak with any two other countries that also call each other) has remained remarkably stable (64 in 1983, 63 in 1995) and largely concentrated in Africa. The only sign of progress here is the decline in the percentage of cliques that do not posses a single "poor" member (from 89% in 1983 to 76% in 1995). Even when these countries do belong to cliques, they are often in only one (40 countries) and are the only member from their region or income category. In short, the poor tend to be either non-members of the global community or isolated from each other.

Analyzing the position and composition of what may be called the semi-periphery or the "newly industrializing countries" is more difficult. We analyzed data for 6 countries that at least anecdotally would represent that global class: Brazil, Egypt, Nigeria, Poland, Singapore and Taiwan. We only see signs of significant regional integration in the case of Taiwan (25% in/35% out to 40% in/55% out). Taiwan and Singapore have essentially developed two sets of cliques--- those they share with dominant countries and a largely autonomous group of regional cliques. Brazil’s links with South America have declined (14% in/23% out to 9% in/20% out). In terms of clique membership the isolation of Africa in a newly "globalized" world is indicated by the performance of Nigeria, one of the leading powers in the region. In 1983, 1989, and 1995, it only belonged to a single clique made up of rich countries. Brazil’ s numbers of clique memberships have remained remarkably consistent while Poland, Taiwan and Singapore’s have skyrocketed. The latter two have arguably graduated into "core" status as they have expanded their set of contacts to beyond serving as a sub-node for European and North American economies. Poland has arguably done the same, but it remains something of an outlier in a series of cliques dominated by much richer countries. It does not, however, seem to fulfill the classic role of semi-periphery as a bridge to poorer economies. Brazil’s clique memberships, however, are a classic example of such a role. First, the majority of the members of its cliques are wealthier countries. Second it rarely shares a clique with a large group of South American countries; rather, Brazil and a single other continental neighbor accompany the rich.

Analysis of regional blocks highlights the tenuous position of even the middle range countries. The NAFTA community gives a good indication of why universal statements regarding the nature of globalization are insufficient and why the information on structural position provided by network analysis will provide a much better picture of whatever "new world order" emerges from the 20th Century. While intra-NAFTA traffic has increased quite dramatically in absolute terms, the relative importance or network centrality of the community for the U.S. and for Canada has actually declined. For Mexico, however, the NAFTA countries were and remain the only game. We suspect that these findings indicate a wider pattern. The international connections of developed countries have become more dispersed and heterogeneous. Those of less developed countries, however, may simply have intensified, but have not changed in any structural form (Figure V).

If there is a center to the international telephone network, however, it may not necessarily be the generic rich, but a single power. The data reveal the importance of distinguishing a core or center from the rise of a global hegemon. If we look at the importance of the G7 (a reasonable starting definition of a global center), we find an apparent decline in overall centrality of those other than the United States. This would suggest the increasing mono-polarization of power with the US as the sole hegemon and the possible appearance of significant semi-peripheral centers. The evidence certainly supports the idea of the United States as the major player in the new world order. The American share of international telephone traffic has been slowly but consistently increasing and now accounts for nearly 40%. This parallels American domination in other areas such as popular culture (Rothkopf 1997) (Figure VI).

Looking at clique membership, the sub-hierarchy among the dominant players becomes even clearer. The most outstanding characteristics is the overpowering presence of the United States, which is included in close to 90% of all cliques. Germany is a distant second while France and Great Britain appear to have switched positions in terms of global network centrality. The most significant contrast is between Japan and the United States. Not even during the glory years of the late 1980s did Japan belong to more than a quarter of global telephone cliques. No matter its powerful role in the world economy and within its own region, it did not have a global reach. The rich may be getting richer, but at least as measured in telecommunications, there is only one center.


We can therefore make several observations about the process of "globalization" as measured by international telephone traffic:

  • While the amount of international communication did increase dramatically from 1983 to 1995, we do not observe a significant acceleration of this process. Moreover, we find remarkable stability in terms of who calls whom with the most popular countries remaining so.
  • There is little evidence of an increase in intra-cultural contact. The one exception is East Asia where the likely cause has more to do with economic development than with increasing cultural chauvinism. Similarly, the evidence indicates the decline of "imperial" zones with the possible exception of Japan and East Asia.
  • There is a clear hierarchy of telephone contact with most of it concentrated on and in the wealthiest countries. Less developed countries are either marginal to international communications or linked via asymmetrical relationships to a group of richer nations.
  • The "center" is increasingly monopolized by the United States with the relative position of Europe weakening.
At least as measured by telecommunications, the global web may be expanding, but so is the importance and significance of the country at its heart, the United States. We are only beginning to explore what accounts for this pattern. Certainly differences in tariff structures explain a great deal. Patterns of migration may also account for much as would trade and political influence. The point, however, is that in our conversations about globalization we should not forget its basic shape, i.e. increasingly unipolar.

Methodological Note

For each year, we have created a matrix which simultaneously represents all of the contact (and the direction thereof) between each country and every other. Thus, the ijth entry in any year’s matrix represents the volume of telephone contact from country i to country j; the jith entry would record the reciprocal contact. Each individual matrix represents a particular year, and the matrices are stacked on top of one another (providing a three dimensional matrix) to analyze the data over time.

Several points need to be clarified. First, our data may suffer from the distortion brought about through "call-back" systems that allow for calls to be channeled through countries with the lowest rates. Thus a Peruvian might use a U.S. call back system to call Venezuela -- such a call would register as a contact between Peru and the U.S. (direction would depend on the type of call back system used) and one from U.S. to Venezuela. We do not believe, however, that these constitute a major distortion. We have also been unable to ascertain the possible effect of cellular communications and how these may be accounted for in the data. Depending on the years and the country pairs, there was a significant amount of missing information. Given the sensitivity of network data to this problem we decided to extrapolate or interpolate data for missing years. When possible we interpolated assuming a linear progression between the two years for which we had data. In the absence of adequate starting or end points with which to obtain a trend line, we used the global mean of growth in telephone traffic.

We also faced the difficult situation of the changing composition of the global state system. This was particularly troublesome for post-1989. Our default definition for all years was based on the countries existing in 1995. If a country did not exist in a previous year contacts to and from it were input as 0. In the case of the USSR, we have it listed as the only participant from 1983 to 1992. Then, beginning in 1993, USSR receives and sends 0 calls and we listed its component states. In the case of Yugoslavia, we retain that entry despite its changing composition (so by 1995 it only includes calls to and from Montenegro-Serbia). Other components of ex-Yugoslavia are added as we have data and listed as having no phone contact prior to their creation. (This is important to keep in mind when we look at aggregate data on countries with no contacts in different years.) Because of reporting problems, Czechoslovakia includes only data from the Czech Republic after 1992.

A major concern involves the unit of analysis. As it stands, we are inferring attributes of actions made by individuals to countries. The real network we are describing is one between 5 billion individuals, not 200 countries. But, given the modes of reporting telephone data, there is no way to avoid the artificial categorization by nation states. The important question must be the extent to which the individuals who dominate the network enjoy their privileged position because of their ascribed characteristic of citizenship. In the case of the United States, we would argue that the answer is yes. What we cannot address, obviously, is the fact that only certain members of each of these societies are making the phone calls and that beneath the global stratification that we find lies an equally great domestic division.

Of equal concern is that in using network analysis we have treated the countries as equivalent units. While not so relevant in our discussion of volume, such an assumed equivalence may be problematic for cliques. In these cases, however, the difference in volume (a function of size and development) would help to ameliorate this shortcoming. Conversely, the differences in size and development level may also overwhelm indications of strength of ties. In the analysis that follows we report two different sets of data. The first, to which we most often refer, is the absolute number of minutes of telephone traffic between any pair of countries or country-groups. The second is a standardized measure through which we sought to reduce the effects of size and development level. For the latter, we divided each data cell by the product of the number of phones in each of the countries in the dyad in any particular year. This measure thus represents the amount of contact given the total potential number of telephone links between two countries.

The main network procedures useful for analyzing this type of data are centrality and clique analysis and structural equivalence. Centrality and clique-based measures examine actual contact between countries. Centrality can measure the country most phoned, controlling for all calls made by other countries, and the country which phones the most (again with appropriate controls). Clique analysis is useful because it allows us to separate regions of the matrix (groups of countries) which have more relative contact with each other than with other countries. Cliques are formed, for this sort of data, by determining an acceptable minimum level of phone calls that act as a cut-off point for determining if there exists a strong, a weak, or no connection between the countries. Combinations of clique and centrality measures allow us to determine local or regional centrality. We can also define partial cliques and those countries that serve as bridges between different cliques or regions. Structural equivalence methods maximize the similarity of in and out ties of any two countries, whether or not they have ties to one another. If countries A and B make calls to countries C and D, but not to country E, then they are structurally equivalent. We would expect that structurally equivalent countries should play similar roles in international relations.


Barnett, George and Joseph G.T. Salisbury. 1996. "Communication and Globalization: A Longitudinal Analysis of the International Telecommunications Network". Journal of World System Research, 2, 16. (http://csf.Colorado.EDU/wsystems/jwsr/vol2/v2_ng.htm)

Baud, Marie-France. 1996. "Market Globalization". UNESCO Courier, Nov. 1996, pp. 33-36.

Castells, Manuel. 1996. The Rise of the Network Society. Cambridge, MA : Blackwell Publishers.

Chase-Dunn, Christopher and Peter Grimes. 1995. "World Systems Analysis". Annual Review of Sociology, 21, pp. 387-417.

Dökmecí, V. and Lal Berköz, 1996. "International Telecommunications in Turkey". Telecommunications Policy, 20, 2, pp. 125-130.

Epstein, Gloria, James Crotty and Patricia Kelly. 1996. "Winners and Losers in the Global Economics Game". Current History, 95, 604, pp. 377-381.

Falk, Richard. 1997. "Will Globalization Win Out". International Affairs, 73, 1, pp. 123-136.

Gereffi, Gary and Miguel Korzeniewicz, eds. 1994. Commodity chains and global capitalism. Westport, Conn. : Greenwood Press.

Gereffi, Gary and Stephanie Fonda. 1992. "Regional Paths of Development". Annual Review of Sociology, 18, pp. 419-48.

Giussani, Paolo. 1996. "Empirical Evidence for Trends Toward Globalization". International Journal of Political Economy, 26,3, Fall, pp. 15-38.

Hirst, Paul and Grahame Thompson. 1996. Globalization in Question. London: Polity Press.

Hudson, Heather E. 1997. Global Connections: International Telecommunications Infrastructure and Policy. New York: Van Nostrand Reinhold.

Huntington, Samuel 1993. "The Clash of Civilizations?". Foreign Affairs, 72,3, pp. 22-49.

Huntington, Samuel. 1996. The Clash of Civilizations and the Remaking of the World Order. New York: Simon and Shuster.

Kellerman, A. 1985. "Telecommunications and the Geography of Metropolitan Areas". Progress in Human Geography, 8, pp. 222-246.

Kellerman, A. 1992. "US International Telecommunications, 1961-88: An International Movement Model". Telecommunications Policy, 16, 3, pp. 156-166.

Kellerman, A. and A. Cohen. 1992. "An International Communications as Internal Movement: the Case of Israel, 1951-1988. Telecommunications Policy, 6,2, pp. 156-166.

Mittelman, James H. 1996. "How Does Globalization Really Work". In James H. Mittelman, ed., Globalization: Critical Reflections. Boulder: Lynne Rienner Publishers.

Powell, Walter W. 1990. "Neither Market Nor Hierarchy." In Barry Staw and L.L. Cummings, ed., Research in Organizational Behavior. v. 12. Greenwich, CT: JAI Press.

Rodrik, Dani. 1997a. Has Globalization Gone Too Far? Washington, D.C.: Institute for International Economics.

Rodrik, Dani. 1997b. "Sense and Nonsense in the Globalization Debate". Foreign Policy, Summer, 107, pp. 19-36.

Rothkopf, David. 1997. "In Praise of Cultural Imperialism". Foreign Policy, Summer, 107, pp. 38-53.

Smith, D.A. and D.R. White. 1992. "Structure and Dynamic of the Global Economy: Network Analysis of international Trade 1965-1980." Social Forces, 70, pp. 857-893.

Snyder, D. and E. Kick. 1979. "Structural Position in the World System and Economic Growth, 1955-1970. American Journal of Sociology, 84, pp. 1096-1126.

Valenzuela, J. Samuel and Arturo Valenzuela. 1978. "Modernization and Dependency: Alternative Perspectives in the Study of Latin American Underdevelopment". Comparative Politics, 10, pp. 535-557.

Various. 1996. "Global Circus: Narratives of Globalization". International Journal of Political Economy, 26,3, Fall.