The cohesion effect of structural equivalence on global bilateral trade,1948

Abstract

Spanish

French

This article bridges two important approaches in the study of global trade – social network analysis and the gravity model – and examines how countries’ structural locations in the global trade network influence bilateral trade from 1948 through 2000. The authors identify a cohesion effect of structural equivalence (the degree to which two nodes have similar ties with other nodes in the network) in global trade: two structurally equivalent countries develop more bilateral trade even after controlling for conventional dyadic factors. This is because common trading ties with other countries promote similar sociocultural values, information flows, and converging institutions, thereby boosting bilateral trade. The authors further demonstrate that the cohesion-generating role of structural equivalence has become more salient over time in the increasingly complex global trade network. Overall, this study shows that bilateral trade is embedded within the structural context of the overall global trade network, and this structural effect is historically contingent on the evolving nature of global economic activities.

Keywords

Bilateral trade economic sociology globalization quantitative analysis

Introduction

The evolution of global trade networks has been a key part in the overall economic globalization. Overall global trade has been expanding rapidly since the end of the Second World War (see Figure 1). Within the web of trade relations, individual countries face both opportunities and constraints in developing trade with other countries. Studies on global trade demonstrate that countries do not trade freely with all other countries, but increasingly trade within blocks of countries (Clark and Beckfield, 2009; Frankel, 1997, 1998; Kim and Shin, 2002; Zhou, 2010). Such fragmentation of global trade suggests that substantial structural constraints on trade relations between countries exist. This increasingly networked but fragmented global trade raises important questions: How does a country choose its trading partners? Does its structural location matter for its bilateral trade with others and, if so, which dimension of network location matters?

Figure 1.

Annual world trade (exports in US constant million dollars).

These questions can be best answered through analyzing the impact of two countries’ structural location within the overall global trade network on the volume of their bilateral trade. Bilateral trade captures a country’s tendency to trade with particular countries. Scholars have long been interested in finding out what factors attract two countries together or push them away from each other. These pull and push forces are the micro foundation for the macro global trade network.

The economists’ conventional way of studying bilateral trade is the gravity model, in which bilateral trade volume is explained by two countries’ economic development level and geographical distance between them (Feenstra et al., 2001; Frankel, 1997, 1998; Rose, 2004). Despite the popularity of the gravity model among trade economists, there is a critical gap in this approach. It focuses exclusively on dyadic characteristics of two countries and neglects how bilateral trade can be affected by their structural location within the overall global trade network. Economic sociology has long been criticizing the assumption of structureless markets in economics. Instead, economic sociologists view markets as social structures in which actors exchange goods or services based on shared understanding and expectations. To cope with uncertainties and risks inherent in economic exchange, market actors frequently seek to develop stable social relationships with their exchange partners. This social embeddedness of markets indicates the relevance of the actor’s structural location within the market to various market outcomes (Burt, 1992; Uzzi, 1996, 1999). However, so far very few empirical studies have actually shown this structural effect on global bilateral trade.¹

In this article we offer an important sociological extension to the gravity model by examining the effect of structural equivalence, a key measure of two countries’ relative location in the overall global trade network, on bilateral trade. Structural equivalence refers to the degree to which two actors have similar ties with other actors in a network. In the case of global trade, structurally equivalent countries trade with each other. We argue that structurally equivalent countries develop more bilateral trade due to the cohesion-generating effect of structural equivalence. Specifically, structural equivalence promotes bilateral trade through three channels – it facilitates common sociocultural values, information flows, and institutional convergence. Moreover, we realize that the global trade network is not static, but has constantly been evolving over time in terms of structural complexity. Therefore, we also investigate the historical contingency of the effect of structural equivalence on bilateral trade. To this end, we analyze a longitudinal dataset on global bilateral trade from 1948 through 2000. Overall, this article argues that bilateral trade is embedded within the overall structural context of the global trade network, and that this structural effect is historically contingent on the evolving nature of global economic activities.

Existing empirical literature on bilateral trade

The gravity model

Empirical studies on bilateral trade have been greatly facilitated by the use of the gravity model, which is claimed to be ‘one of the great success stories in empirical economics’ (Feenstra et al., 2001: 431). The gravity model relates trade between two countries positively to both of their economic size (often using GDP as a proxy), and negatively to the distance between them, usually with a functional form that is similar to the law of gravity in physics. This model was sometimes criticized for its lack of theoretical foundations, but recently economists have found solid theoretical support (Feenstra et al., 2001).

Using this gravity model scholars have identified many factors that potentially influence a country’s trading ties with other countries, including its economic size, geographic location, cultural characteristics, political regime, and global institutional connections (e.g., Frankel, 1997, 1998; Ingram et al., 2005; Zhou, 2010, 2011). Bilateral trade is substantially larger between two economically large countries than between small ones. The growing network of intergovernmental organizations (IGOs) boosts bilateral trade by tearing down trade barriers, providing regulations for global markets, and reducing uncertainties inherent in global trade (Ingram et al., 2005). Moreover, inter-country geographic distance, differences in national cultures and political systems between countries are all found to be significant influences on bilateral trade (Zhou, 2010, 2011). These empirical studies based on the gravity model have revealed the multidimensional nature of bilateral trade and improved our knowledge about the forces shaping bilateral trade.

However, the existing literature exclusively focuses on the effect of dyadic factors on bilateral trade, thereby ignoring how countries’ structural locations within the broader global network affect their trade with each other. Although global trade has long been seen as a huge global network (Breiger, 1981; Kim and Shin, 2002; Smith and White, 1992; Snyder and Kick, 1979), scholars so far have made very few attempts to incorporate two countries’ relative locations in the global trade network into the study of bilateral trade. This negligence is largely due to a lack of dialogue between social network studies and the gravity model. To fill this gap in the existing literature, we bridge the two approaches in this study – first using social network analysis to generate a globe-level variable indicating two countries’ similarity in global trade profiles, structural equivalence, and then incorporating this variable into the gravity model.

The social network approach and world system theory

The other independently developed empirical literature on global trade is from social network studies inspired by world system theory. World system theory has developed a macro-historical description of the world economy over a long time period. According to world system theory, structural locations of countries within the three-tiered global capitalist system affect economic development through unequal economic exchanges between core and peripheral countries (Wallerstein, 1974). Using social network analysis, a large number of empirical studies have explored structural partitioning of the global trade network (Breiger, 1981; Clark and Beckfield, 2009; Kim and Shin, 2002; Mahutga, 2006; Sacks et al., 2001; Smith and White, 1992; Snyder and Kick, 1979). The key motivation of these network studies is to empirically classify countries into three structural positions, namely core, semi-periphery, and periphery, as predicted by world system theory.

To generate countries’ structural positions, these studies need to examine how similar or different countries are from each other in terms of trade relations with other countries. Early studies (e.g., Snyder and Kick, 1979) employed the concept of structural equivalence – two countries would be structurally equivalent if both countries trade with the same other countries (Boorman and White, 1976; Lorrain and White, 1971), while Smith and White (1992) applied role equivalence where two countries would be role equivalent if they trade with similar types of countries. Despite different measures in these studies, they all applied some form of blockmodeling to a distance matrix of countries, which classified individual nodes into groups with maximum within-group density and minimum between-group density.

These studies contribute greatly to our understanding of the overall structure of the global trade network. We further extend the utility of social network analysis in the study of global trade. Instead of searching for a simple macro-level partitioning, we seek to explicitly examine how bilateral trade volume is affected by countries’ relative network locations in the global trade system.

Structural equivalence in the global trade network

In addition to the conventional factors in the gravity model, we argue that structural equivalence in the global trade network influences the level of trade between two countries. Structural equivalence is a network concept that measures the level of structural similarities between individual actors in a network.² Two actors are structurally equivalent if they have the same relations with the same actors (Lorrain and White, 1971). Two perfectly structurally equivalent actors have exactly the same relations with the exact same others. Of course, perfect structural equivalence rarely exists in real networks. Thus, a measure has been developed to indicate the degree of structural equivalence between actors. It is important to recognize that the concept of structural equivalence does not presuppose a direct tie between actors. Structural equivalence between two actors is based on the comparison of their respective relations with others, not the direct relation between themselves. In other words, the measure of structural equivalence is independent of direct relations between the focal dyad. It is a useful tool to approach the embeddedness of the focal dyad within a higher-level social structure.

The social network literature often finds social similarities or homogeneity between structurally equivalent actors (Burt, 1987; Friedkin, 1984; Mizruchi, 1990, 1993). Due to those social similarities, it is inferred, sometimes assumed, that those actors are likely to form a direct tie, but this effect of structural equivalence on tie formation has not been tested empirically for global trade. We extend the scholarship of structural equivalence to dynamic formation of network ties in the field of global trade.

The trade-promoting effect of structural equivalence

As a key dimension of two actors’ relative structural location in the overall network, some social network studies have examined the effects of structural equivalence in market settings. These studies suggest a positive influence of structural equivalence on tie formation between actors (Burt, 1987, 1988; Friedkin, 1984; Mizruchi, 1990, 1993), or the so-called cohesion effect of structural equivalence. Borrowing this important insight from the social network literature, we argue that there are three major mechanisms or channels through which the cohesion effect of structural equivalence is at work in global trade.

First, structural equivalence in global trade promotes common sociocultural values and market preferences. Network ties are an important source of social diffusion (Burt, 1987; Coleman et al., 1957; Mizruchi, 1990). A key finding of economic sociology is that economic exchanges are not only embedded in social relations, but also generate new social relations (Barber, 1995; Emirbayer and Goodwin, 1994; Zelizer, 1988).³ When two countries trade, they also exchange social values and cultural preferences. In the case of structural equivalence, trading with the same partners exposes two countries to the same social values and market preferences, which help the two countries to develop common sociocultural values.

Second, trading with the same countries provides the two countries with the same source of information and the opportunities to know about each other. Information, especially market-related information, flows directly through trade relations. Structurally equivalent actors send and receive information through shared ties (Burt, 1987; Coleman et al., 1957; Mizruchi, 1990). Common trading partners are ideal locales where two countries meet and discover business opportunities in each other’s markets (Kim and Skvoretz, 2010).

Third, trade with the same countries exposes the two countries to common formal institutions such as legal regulations, product or service-related standards, labor standards, and common informal institutions such as manners of doing business. When countries trade with the same countries, they cumulate common experience of dealing with the same set of institutions. This shared experience often generates convergence of institutions between the two countries.

Taken together, structural equivalence diffuses similar values and social trust, facilitates information flows, and fosters converging institutions. They all in turn lead to increased bilateral trade. First, bilateral trade, as a form of economic exchange, is imbued with meaning and ideas and is shaped by the cultural conceptions of trading partners (DiMaggio, 1994; Zelizer, 2003). Bandelj (2002, 2009) demonstrates that countries choose their investment destinations based on shared sociocultural understanding because common values and understanding promote trust, reduce risks, and ease communications during economic transactions. Following the same logic, shared sociocultural values and ideas facilitate bilateral trade.

Second, the information generated by common business networks includes market opportunities, product standards, prices, and localized knowledge in each other’s markets, which can be easily transformed into direct bilateral trade (White, 2002). The shared information also makes it easier to understand and predict the behavior of others, and promotes the development of mutual trust (Guiso et al., 2004).

Third, similar institutions provide critical underpinnings of the market between the two countries, smooth transactions across borders, and reduce risks and uncertainties in bilateral trade (Fligstein, 2001; Mansfield and Reinhardt, 2008). As a result, the convergence of institutions facilitated by sharing the same trading partners generates more bilateral trade.

In a nutshell, common social values, intensified information flows, and converging institutions facilitated by structural equivalence all promote bilateral trade. The three mechanisms underlying the cohesion effect of structural equivalence is summarized in Figure 2.

Figure 2.

Mechanisms underlying the cohesion effect of structural equivalence.

Hence, the cohesion argument suggests that structurally equivalent countries in the global trade network are likely to develop more bilateral trade.⁴ Accordingly, we propose our first hypothesis, as follows:

Hypothesis 1: Structural equivalence promotes bilateral trade.

It is worth mentioning that another relevant argument, the competition argument, implies that competition induced by structural equivalence impedes bilateral trade. Structurally equivalent actors often experience a high level of market competition with each other (Burt, 1992). In global trade, trading with the same other countries means that the two countries often have to compete in order to succeed in their common markets. Facing direct market competition, actors often emulate the successful competitor to avoid being substituted by the other in global markets (Meyer and Rowan, 1977; Mizruchi, 1993). This can lead to convergence in the industrial structure of the two countries and a wide range of overlapping goods. Consequently, structurally equivalent countries have no need to trade much with each other. In contrast with the cohesion argument, the competition argument implies a negative effect of structural equivalence on bilateral trade. Although this argument is not as well developed as the cohesion one in the existing literature, we still present it here as a possible alternative hypothesis.

Historical contingency of the effect of structural equivalence

We further argue that increased complexity in the global trade network makes the trade-promoting effect of structural equivalence more salient over time. In addition to the explosive expansion of global trade flows (see Figure 1), the increasing number of countries participating in global trade can be another source of increased complexity of the global trade network (see Figure 3). The growing pool of potential trading partners opens up more trade opportunities, but also increases the difficulty in choosing trading partners.

Figure 3.

Change in the number of countries in global trade over time.

Another change is even more profound. Recent studies on economic globalization demonstrate that the production process within the same industry is often divided across a number of countries (e.g., product design in country A, manufacturing in country B, and assembly in country C) (Arndt and Kierzkowski, 2001; Coe et al., 2004; Gereffi et al., 2005).⁵ Such fragmentation is made possible by technological advancement, which allows ‘modularization’ of production processes across borders to maximize efficiency and to cope with increasing differentiation of consumer demands (Sturgeon, 2002). This disintegration of production increases the importance of trade in components and parts (often called ‘intermediate inputs’) relative to raw materials and finished products. It is found that global trade in intermediate inputs has grown faster than trade in other products, highlighting the growing internationalization of production operations (Yeats, 1998). It was around the late 1970s and early 1980s when this trend increased substantially, and has continued up to now (Lall et al., 2004).

The rising popularity of these internationalized productive activities, on the one hand, brings in more global trade, as intermediate inputs cross borders several times during the production process. On the other hand, it also brings in more uncertainties and complexities in global trade because any delays or missing links in trade hamper the whole production process (Gereffi and Korzeniewicz, 1994; Hummels, 2001). Structural equivalence can play an important role in mitigating the increased uncertainties and facilitating coordination in global trade, according to the cohesion argument. Structural equivalence fosters common social values, facilitates information flows, and promotes converging institutions, which all have positive influences on internationalized productive activities. Thus, it is likely that structurally equivalent countries are located in different stages of the same production process, and trade increasingly more with each other in intermediate inputs.

As a result of the increasing network complexity and internationalized productive activities, the role of structural equivalence should have become more salient in promoting bilateral trade. Accordingly, we arrive at our second hypothesis:

Hypothesis 2: Structural equivalence increasingly promotes bilateral trade over time.

Data and method

The data used in this article are pooled time-series cross-sectional data and span 53 years, consisting of annual data from 1948 through 2000 for more than 150 countries. They were compiled by Rose (2004) from the IMF’s Direction of Trade Statistics.⁶ The unit of observation is a ‘dyad-year’, that is, a given pair of countries in a given year. The dependent variable is the volume of bilateral trade between a pair of countries recorded in constant US dollars.

The independent variable of main interest is structural equivalence. We use the UCINET software to calculate structural equivalence, which is the Pearson correlation coefficient of the profile vectors of Country 1 and Country 2 in their relationships with all countries in the network (Breiger, 1981; Burt, 1987). It measures the degree of similarity between two countries in terms of their profiles of trading partners. The value of structural equivalence may change over time.⁷

Other independent variables serve as controls in the analysis. Time-varying variables include GDP, population, and several binary variables indicating whether the two countries are in any regional trade arrangements (RTA), whether they are both in the World Trade Organization (WTO), whether they both participate in the Generalized System of Preferences⁸ (GSP), whether they use the same currency, respectively. Time-constant variables include distance between the two countries, and two binary variables indicating whether the two countries share the same language, and whether the two countries have ever been in a colonial relationship, respectively. These variables are all potentially associated with the volume of bilateral trade, and are commonly controlled in empirical studies.

We incorporate structural equivalence into the gravity model and make it one of the variables that explain bilateral trade flows. The general setup of the gravity model is as follows:

\ln ({Trade}_{ijt}) = β_{0} + β_{1} {SE}_{ijt} + β_{2} \ln {({GDP}_{i} {GDP}_{j})}_{t} + β_{3} \ln {({Pop}_{i} {Pop}_{j})}_{t} + β_{4} \ln ({Dist}_{ij}) + β_{5} {WTO}_{ijt} + β_{6} {Lang}_{ij} + β_{7} {Colo}_{ij} + β_{8} {Rta}_{ijt} + β_{9} {GSP}_{ijt} + ε_{ijt}

where i and j denote the countries in a dyad, t represents the time point, and the variables are:

Trade_ijt is the real value of bilateral trade between i and j in year t;

SE_ijt is the degree of structural equivalence between i and j in year t, ranging between −1 and 1. A larger value indicates a higher degree of structural equivalence;

GDP_i is the real value of GDP;

Pop_i is the population of country i;

Dist_ij is the distance between two countries;

WTO_ijt is a binary dummy for GATT/WTO membership, and is 1 if both countries are WTO members;

Lang_ij is a binary dummy for common language;

Colo_ij denotes colonizer–colony relations, and is a binary variable which is unity if country i ever colonized j or vice versa;

Rta_ijt is a binary variable that is unity if i and j belong to the same regional trade arrangement at time t;

GSP_ijt is a binary variable that is unity if both countries are in GSP;

ϵ_ijt is the model disturbance for a dyad (i and j) in year t, representing the omitted other influences on bilateral trade.

The gravity model is usually assumed to be dyadic independent. The conventional method is fixed country pair-specific (dyadic) fixed effects (Anderson and Van Wincoop, 2003; Rose, 2004). The analysis here employs dyadic fixed effects models, which should account for any non-independence of dyadic observations in the data. The greatest advantage of the fixed effects model (FEM) is that it allows explanatory variables to be correlated with unit-specific unobserved variables. In panel data analysis, disturbances often display some degree of stability over time, in which case the covariance structure of ϵ_ijt often follows the first-order autoregressive scheme, AR(1) (Finkel, 1995; Tuma and Hannan, 1984). To account for this possible autoregressive structure in disturbances, we employ the FEM with AR(1) disturbances, for which the Stata software (Release 10) has a built-in ‘xtregar’ fixed effects estimator. To control unobserved heterogeneity for each year, we also include a series of year dummy variables, one for each year. Since we use the fixed effects model, time-invariant variables such as geographical distance are not estimated. However, as part of the robustness checks, we include such time-constant variables in the random effects model.

Results

Table 1 shows some descriptive statistics about structural equivalence. Panel A presents the distribution of the structural equivalence score in the pooled data. While in theory structural equivalence can range from −1 to 1, the actual range is from –.067 to .999. Panel B lists the country pairs that receive the highest and lowest scores of structural equivalence in the year 2000. Panel C of Table 1 shows the correlations among the variables used in the analysis. There is no obvious concern for multicollinearity as all the correlations are moderate.

Table 1.

Descriptive statistics about structural equivalence.

Table 2 shows the results from the gravity model with structural equivalence and other dyadic predictors. In Model 1, we only control for basic gravity model variables, GDP and population. Structural equivalence has a statistically significant and positive effect on bilateral trade.

Table 2.

Unstandardized coefficients from the gravity model of bilateral trade.

	Model 1	Model 2	Model 3	Model 4
Structural equivalence	.380^*** (8.47)	.374^*** (8.34)	.420^*** (9.36)	.772^*** (17.64)
GDP	.651^*** (50.14)	.644^*** (49.56)	.619^*** (47.76)	.803^*** (43.92)
Population	−.110^*** (−9.01)	−.104^*** (−8.51)	−.087^*** (−7.15)	−.368^*** (−17.05)
Regional trade arrangement		.873^*** (9.54)	.852^*** (9.36)	.678^*** (7.76)
Currency union		.044 (.37)	.110 (.93)	.448^*** (3.94)
WTO			.088^*** (4.43)	.069^*** (4.46)
GSP			.625^*** (24.78)	.182^*** (6.71)
Constant	5.122^*** (237.95)	5.118^*** (238.59)	5.001^*** (231.58)	4.410^*** (185.49)
Year fixed effects				Controlled
ρ in AR(1)	.666	.666	.662	.647
R ²	.094	.095	.099	.159

Notes: (1) N = 241,766. (2) Numbers in parentheses are t-values. (3) From two-tailed tests, * p < .05; ** p < .01; *** p < .001. (4) The year fixed effects (52 year dummies) are controlled in Model 4 (not shown to save space).

In Model 2, we expand the basic gravity model by including two variables that measure different types of trade-related arrangements between countries. We control the two variables to make sure that the positive effect of structural equivalence is not caused by bilateral trade policies. The result does not differ from what we see in Model 1, and structural equivalence still has a significant and positive effect on bilateral trade.

In Model 3 we add multilateral trade-related arrangements, and in Model 4 we further control for the year fixed effects by adding a dummy variable for each year. In both models the significantly positive effect of structural equivalence still holds. The models above indicate that structurally equivalent countries, or those countries that trade with a similar list of countries, are more likely to trade with each other. This finding supports Hypothesis 1.

Next we examine how the effect of structural equivalence has changed over time. To test Hypothesis 2, we create an interaction term between year and structural equivalence. The base model is the year 1948, so now the coefficient of structural equivalence measures its effect on bilateral trade in 1948. The coefficient of the interaction term measures the change in this effect over the years. The results are shown in Table 3. According to Model 1, in early years structural equivalence actually had a significantly negative effect (−2.136 in 1948), but over time it gradually turned positive (increasing by .076 per year). Around the mid-1970s, structural equivalence began to have a positive effect on bilateral trade. This finding from Model 1 still holds even when we control for both bilateral and multilateral trade-related arrangements (see Model 2 and Model 3).

Table 3.

Unstandardized coefficients from the gravity model of bilateral trade: Change over time.

	Model 1	Model 2	Model 3
Structural equivalence	−2.136^*** (−19.92)	−2.114^*** (−19.65)	−2.124^*** (−19.86)
Structural equivalence *Year	.076^*** (27.55)	.075^*** (27.15)	.076^*** (27.44)
Year	−.016^*** (−8.93)	−.016^*** (−8.85)	−.020^*** (−10.75)
GDP	.637^*** (49.02)	.635^*** (48.75)	.628^*** (48.39)
Population	−.077^*** (−6.26)	−.075^*** (−6.08)	−.058^*** (−4.75)
Regional trade arrangement		.355^*** (3.88)	.397^*** (4.35)
Currency union		.282^* (2.38)	.318^** (2.69)
WTO			−.067^** (–3.18)
GSP			.458^*** (17.44)
Constant	5.252^*** (242.42)	5.247^*** (242.10)	5.134^*** (234.19)
ρ in AR(1)	.661	.661	.658
R ²	.103	.103	.105

Notes: (1) N = 241,766. (2) Numbers in parentheses are t-values. (3) From two-tailed tests, * p < .05; ** p < .01; *** p < .001.

Figure 4 visualizes the upward trend of the effect of structural equivalence on global bilateral trade over time. It is likely that the positive effect of structural equivalence in the post-1975 period overtakes its negative effect in the pre-1975 period, so the overall effect of structural equivalence for the whole period turns out to be positive in Table 2. Although structural equivalence displays a positive effect in the pooled data from 1948 through 2000, its effect has actually experienced a dramatic shift over the years.

Figure 4.

Change over time in the effect of structural equivalence on bilateral trade.

A changing network effect over time usually reflects profound changes in market environments (Gulati, 1995; Mizruchi et al., 2006). For instance, as the organizational field for strategic alliance becomes more complex with heightened uncertainties for future alliance, organizations tend to rely more on preexisting alliance network ties for new alliance formation (Gulati, 1995). We argue that the same logic is applicable here. Increased complexity in global trade has increased the importance of preexisting trade networks in developing bilateral trade. The increase in the number of countries involved in global trade, the explosive expansion of global trade flows, and especially, the increasingly globalized productive activities all contribute to the increased complexity. Since this globalization of production was not well developed before the 1970s, two countries trading with the same countries were more likely to be direct competitors in global markets. However, as globalized production has become increasingly popular since the 1970s, countries trading with the same countries often participate in different stages of the production process, which boosts trade in intermediate inputs between them and thus shifts the effect of structural equivalence from negative competition to positive cohesion.

Robustness checks

To test the robustness of the results produced by the fixed effects models above, we also try several other estimators. The results are shown in Table 4. First, instead of using structural equivalence of the same year as bilateral trade, we create structural equivalence with a one-year lag to better establish causal direction. As Model 1 in Table 4 shows, with this lagged structural equivalence, the fixed effects model generates substantively similar results. We also try two- and three-year lagged structural equivalence, which do not change the results either.

Table 4.

Unstandardized coefficients from the gravity model of bilateral trade: Robustness checks.

(1) Overall effect			(2) Change over time
	Model 1: Lagged structural equivalence	Model 2: Random effects model		Model 1: Lagged structural equivalence	Model 2: Random effects model
Structural equivalence		1.162^*** (28.72)	Structural equivalence		−1.850^*** (−20.16)
1-year lagged structural equivalence	.208^*** (4.97)		Structural equivalence * Year		.077^*** (32.38)
GDP	.575^*** (47.64)	1.137^*** (115.64)	1-year lagged structural equivalence	−1.626^*** (−17.34)
Population	−.050^*** (−4.41)	.781^*** (107.79)	1-year lagged structural equivalence * Year	.058^*** (24.18)
Regional trade arrangement	1.092^*** (14.11)	.584^*** (6.84)	Year	.004^** (2.70)	−.092^*** (−64.50)
Currency union	.106 (1.00)	.566^*** (5.62)	GDP	.540^*** (44.58)	1.434^*** (136.34)
WTO	.269^*** (14.83)	−.375^*** (−20.41)	Population	−.058^*** (−5.11)	.915^*** (128.08)
GSP	.696^*** (31.79)	.519^*** (21.61)	Regional trade arrangement	.481^*** (6.19)	.733^*** (8.66)
Distance		−1.346^*** (−53.05)	Currency union	.397^*** (3.78)	.569^*** (5.77)
Common language		.668^*** (12.71)	WTO	−.032 (−1.67)	−.053^** (−2.82)
Common colonial history		2.633^*** (14.36)	GSP	.363^*** (15.75)	.762^*** (31.93)
Year fixed effects	Controlled	Controlled	Distance		−1.393^*** (−58.58)
Constant	3.587^*** (151.40)	−9.629^*** (–32.84)	Common language		.739^*** (15.06)
			Common colonial history		1.758^*** (10.29)
			Constant	4.051^*** (167.81)	−13.193^*** (−45.34)
ρ in AR(1)	.608	.662	ρ in AR(1)	.605	.658
R ²	.158	.593	R ²	.169	.646

Notes: (1) N = 241,766. (2) Numbers in parentheses are t-values. (3) From two-tailed tests, * p < .05; ** p < .01; *** p < .001.

Then we apply the random effects model and report its results in Model 2. Unlike the fixed effects model, the random effects model also takes between-dyad difference into consideration, and it allows for inclusion of time-invariant variables. Thus, we further bring into the modeling such variables as distance (in logarithmic form), common language, and common colonial history between countries. The results from the random effects model also support our findings.

Since the same country belongs to a number of different dyads, dyadic dependence in the gravity model can be a potential concern. The main concern is network autocorrelation, which causes problems for standard statistical tests. One way to address this problem is the Multivariate Regression Quadratic Assignment Procedure (MRQAP) (Dekker et al., 2003; Gulati, 1995; Krackhardt, 1988; Mizruchi, 1993). We use the ‘double Dekker semi-partialling method’ available in the UCINET software. Structural equivalence still shows a significantly positive effect on bilateral trade.⁹

In the above analysis, by using the interaction between year and structural equivalence we assumed a linear yearly change over time in the effect of structural equivalence (see Figure 4). Here we relax this assumption and fit the fixed effects model to bilateral trade data in each decade separately. As a result, six regressions are estimated, each for one decade. The effects of structural equivalence are statistically significant in all six models. The effect of structural equivalence in the 1940s is −2.854. It increases to –.583 in the 1950s, –.214 in the 1960s, .045 in the 1970s, .432 in the 1980s, and 1.270 in the 1990s. The result is largely consistent with the linear pattern seen in the earlier analysis and indicates a continuous increase in the effect of structural equivalence over time.

Finally, the data used in the previous analyses are unbalanced, as not all countries are in the data throughout the whole period. Some countries (e.g., Armenia) were established after 1948 while some (e.g., Yugoslavia) ceased to exist before 2000. We apply the models to the balanced data and analyze only the countries that existed over the whole period. All results are substantively the same.

Further exploration

Besides the robustness checks we further explore the effect of structural equivalence through three additional models. First, the analysis above has shown that structural equivalence promotes the expansion of bilateral trade flows. Does it also contribute to the initiation of bilateral trade ties? Model 1 in Table 5 uses a binary variable of bilateral trade, instead of the trade volume, as the dependent variable; 0 indicates no bilateral trade ties, while 1 represents existence of ties. We use the random effects logit regression to estimate this model.¹⁰ The effect of structural equivalence is significantly positive, so structural equivalence also promotes initiation of bilateral trade ties.

Table 5.

Unstandardized coefficients from the gravity model of bilateral trade: Further exploration.

(1) Formation of trade ties		(2) Influence from role equivalence		(3) Influence from world trade expansion
	Model 1		Model 2		Model 3
Structural equivalence	3.400^*** (24.87)	Structural equivalence	.487^*** (14.29)	Structural equivalence	.463^*** (14.20)
GDP	.470^*** (42.44)	Role equivalence	.082 (1.13)	Overall world trade flows	.054^*** (26.63)
Population	.282^*** (14.90)	GDP	.753^*** (36.51)	GDP	.408^*** (42.25)
Regional trade arrangement	.541^* (2.21)	Population	−.148^*** (−10.84)	Population	−.110^*** (−12.11)
Currency union	−.144 (−1.02)	Regional trade arrangement	.403^*** (4.11)	Regional trade arrangement	.492^*** (7.57)
WTO	−1.060 (−1.73)	Currency union	.289^*** (4.22)	Currency union	.472^*** (5.61)
GSP	3.668^*** (19.28)	WTO	.071 (1.93)	WTO	.007 (.44)
Distance	−.468^*** (−20.81)	GSP	.417^*** (7.03)	GSP	.150^*** (8.04)
Common language	.220^*** (5.65)	Year fixed effects	Controlled	Year fixed effects	Not controlled
Common colonial history	3.465^** (3.46)	Constant	4.845^*** (66.46)	Constant	3.592^*** (111.98)
Year fixed effects	Controlled	ρ in AR(1)	.657	ρ in AR(1)	.651
Constant	−5.962^*** (−23.00)
R ²	.242	R ²	.173	R ²	.152

Notes: (1) N = 241,766. (2) Numbers in parentheses are z-values in Model 1 and t-values in Models 2 and 3. (3) From two-tailed tests, * p < .05; ** p < .01; *** p < .001. (4) The dependent variable in Model 1 is binary, and Model 1 uses the random effects logit regression. (5) Models 2 and 3 use the fixed effects regression with AR(1).

Second, both structural equivalence and role equivalence can be employed to measure structural positions in the global trade network. The cohesion effect argued in this study is based on the same trading partners, not the same generalized type of partners, so it can only be captured by structural equivalence, not role equivalence. To empirically test this choice we estimate a model which includes both structural equivalence and role equivalence.¹¹ The result is presented in Model 2 of Table 5. Structural equivalence shows a significant effect, whereas role equivalence does not. This empirical evidence lends further support to our argument. It is trading with the same partners, not trading with the same type of partners, that boosts bilateral trade.

Third, the effect of structural equivalence turns positive almost exactly when the growth of global trade starts accelerating around the mid-1970s (see Figures 1 and 4). Is the rise in the trade-promoting effect of structural equivalence simply a byproduct of the accelerated expansion of global trade flows? As we have explained above, the transformation of global trade markets is much more profound than global trade expansion. Model 3 in Table 5 confirms our expectation. After controlling for the world level of trade flows, the effect of structural equivalence is still significant. Mere expansion of world trade does not explain the whole effect of structural equivalence.

Conclusion and discussion

In this article we sought to apply the embeddedness insight from economic sociology to global trade, and to bridge two important approaches – the gravity model and social network analysis. In doing so, we went beyond those conventional dyadic factors used in the gravity model, and argued that the structure of the global trade network itself could be an important influence on bilateral trade. Bilateral trade is not simply a business of the two countries, but has to be studied in the structural context of the overall global trade network.

Specifically, in addition to those dyadic characteristics of the two trading countries indentified by the gravity model, their relative locations, which are determined by their network ties with other countries in the global trade network, also affect bilateral trade flows between them. To capture the relative network locations of the two countries, we brought in the concept of structural equivalence from the social network literature.

We proposed a cohesion argument that predicts a positive effect of structural equivalence on bilateral trade. The analysis of global bilateral trade data from 1948 through 2000 showed that, controlling for all conventional factors used in the gravity model, structural equivalence has a significantly positive effect on bilateral trade. We proposed three mechanisms to explain this finding. Common trading ties with other countries facilitate bilateral trade through cultivating similar sociocultural values, facilitating information flows, and fostering convergence of institutions. Moreover, rather than simply observing an overall positive effect of structural equivalence, we paid additional attention to the trend in its change and predicted the cohesion effect to become even more salient over time. This rising importance of structural equivalence can be attributed to increasing complexities in the global economy, especially increasingly globalized productive activities. Echoing a few social network studies (Gulati, 1995; Gulati and Gargiulo, 1999; Mizruchi et al., 2006), we argued that the effect of network structure is historically contingent. The more complex and intertwined the world economy becomes, the more likely actors rely on existing network ties in conducting economic exchanges.

This finding here has important implications for regionalization in the world economy. Regionalization is a prominent feature in global trade markets over the recent several decades, and has been attributed to such factors as transportation costs, regional institutional arrangements, and geocultural affinity (e.g., Frankel, 1997, 1998; Kim and Shin, 2002; Zhou, 2010, 2011). The cohesion effect of structural equivalence is another potential force driving the ongoing regionalization. Based on the UN definition of macro-geographic regions, on average within-region country pairs have a much higher degree of structural equivalence (.643) than cross-region pairs (.317). This difference is statistically significant at the .05 level. Thanks to the increasing importance of structural equivalence in stimulating bilateral trade, countries within the same regions have an extra edge in strengthening bilateral trade ties. Hence, the cohesion effect of structural equivalence contributes to trade regionalization, and its rising salience is likely to further accelerate the regionalization process.

Nevertheless, this study also has its limitations and generates research questions for future studies. First, although we explained the positive effect of structural equivalence on bilateral trade through the three mechanisms derived from the cohesion argument, we were not able to directly assess or distinguish the three mechanisms in the analysis. Furthermore, we did not examine how these mechanisms actually operate in reality. In future studies, we will collect firm-level trade data and conduct case studies to see how those benefits of structural equivalence materialize in practice, thereby conferring more concrete form on the theory and findings here.

Second, as the trend of economic globalization continues, we expect an increasing degree of structural complexity in the global trade network. As a market becomes more complex structurally, an actor’s structural location becomes more important in determining its ties with others. In this study, although we illustrated this increasing importance of structural equivalence over time, in the future we plan to develop a good indicator for the structural complexity in global trade, and to establish a better quantified link between structural complexity and the effect of structural equivalence.

Footnotes

Acknowledgements

The authors contributed equally to this article. We thank Jason Beckfield, the editor of International Sociology, and two anonymous reviewers for their helpful comments. Early versions of this article were presented at the 2010 Annual Meeting of the Eastern Sociological Society in Boston and the 2010 Annual Meeting of the American Sociological Association in Atlanta.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Notes

Author biographies

Min Zhou is an Assistant Professor of Sociology at the University of Victoria, Canada. He received his PhD in Sociology from Harvard University. His research interests include institutionalization of global governance, global market networks, and China–Japan relations.

Chan-ung Park is an Associate Professor of Sociology at Yonsei University, South Korea. His research focuses on social policies and economic sociology. He is currently working on social enterprises, welfare–work nexus in Asia, FTA network and its effect on inequality.

References

Anderson

Van Wincoop

(2003) Gravity with gravitas: A solution to the border puzzle. American Economic Review 93: 170–192.

Arndt

Kierzkowski

(2001) Fragmentation: New Production Patterns in the World Economy. Oxford: Oxford University Press.

Bandelj

(2002) Embedded economies: Social relations as determinants of foreign direct investment in Central and Eastern Europe. Social Forces 81(2): 411–444.

Bandelj

(2009) The global economy as instituted process: The case of Central and Eastern Europe. American Sociological Review 74: 128–149.

Barber

(1995) All economies are ‘embedded’: The career of a concept, and beyond. Social Research 62: 387–413.

Boorman

White

(1976) Social structure from multiple networks II: Role structures. American Journal of Sociology 81: 1384–446.

Breiger

(1981) Structures of economic interdependence among nations. In: Blau

Merton

(eds) Continuities in Structural Inquiry. Beverly Hills, CA: Sage, pp. 353–380.

Burt

(1987) Social contagion and innovation: Cohesion versus structural equivalence. American Journal of Sociology 92: 1287–1335.

Burt

(1988) Some properties of structural equivalence measures derived from sociometric choice data. Social Networks 10: 1–28.

10.

Burt

(1992) Structural Holes: The Social Structure of Competition. Cambridge, MA: Harvard University Press.

11.

Clark

Beckfield

(2009) A new trichotomous measure of world-system position using the international trade network. International Journal of Comparative Sociology 50: 5–38.

12.

Coe

Hess

Yeung

HW-C

. (2004) ‘Globalizing’ regional development: A global production networks perspective. Transactions of the Institute of British Geographers 29: 468–484.

13.

Coleman

Katz

Menzel

(1957) The diffusion of an innovation among physicians. Sociometry 20: 253–270.

14.

Dekker

Franses

Krackhardt

(2003) An equilibrium-correction model for dynamic network data. Journal of Mathematical Sociology 27: 193–215.

15.

DiMaggio

(1994) Culture and economy. In: Smelser

Swedberg

(eds) The Handbook of Economic Sociology. Princeton, NJ: Princeton University Press, pp. 27–57.

16.

Emirbayer

Goodwin

(1994) Network analysis, culture, and the problem of agency. American Journal of Sociology 99: 1411–1454.

17.

Feenstra

(1998) Integration of trade and disintegration of production in the global economy. Journal of Economic Perspective 12: 31–50.

18.

Feenstra

Markusen

Rose

(2001) Using the gravity equation to differentiate among alternative theories of trade. Canadian Journal of Economics 34: 430–447.

19.

Finkel

(1995) Causal Analysis with Panel Data. Thousand Oaks, CA: Sage.

20.

Fligstein

(2001) The Architecture of Market. Princeton, NJ: Princeton University Press.

21.

Frankel

(1997) Regional Trading Blocs in the World Economic System. Washington, DC: Institute for International Economics.

22.

Frankel

(ed.) (1998) The Regionalization of the World Economy. Chicago: The University of Chicago Press.

23.

Friedkin

(1984) Structural cohesion and equivalence explanations of social homogeneity. Sociological Methods and Research 12: 235–261.

24.

Friedkin

(1998) A Structural Theory of Social Influence, Vol. 13. New York: Cambridge University Press.

25.

Gereffi

Korzeniewicz

(eds) (1994) Commodity Chains and Global Capitalism. Westport, CT: Praeger.

26.

Gereffi

Humphrey

Sturgeon

(2005) The governance of global value chains. Review of International Political Economy 12: 78–104.

27.

Granovetter

(1985) Economic action and social structure: The problem of embeddedness. American Journal of Sociology 91: 481–510.

28.

Guiso

Sapienza

Zingales

(2004) Cultural biases in economic exchange. Kellogg School of Management of Northwestern University Working Paper.

29.

Gulati

(1995) Social structure and alliance formation patterns: A longitudinal analysis. Administrative Science Quarterly 40: 619–652.

30.

Gulati

Gargiulo

(1999) Where do interorganizational networks come from? American Journal of Sociology 104: 1439–1493.

31.

Hummels

(2001) Time as a trade barrier. Purdue University Center for Global Trade Analysis Working Paper No. 1152.

32.

Ingram

Robinson

Busch

(2005) The intergovernmental network of world trade: IGO connectedness, governance, and embeddedness. American Journal of Sociology 111: 824–858.

33.

Kim

Shin

E-H

(2002) A longitudinal analysis of globalization and regionalization in international trade: A social network approach. Social Forces 81: 445–468.

34.

Kim

Skvoretz

(2010) Embedded trade: A third party effect. Social Science Quarterly 91(4): 964–983.

35.

Krackhardt

(1988) Predicting with networks: Nonparametric multiple regression analysis of dyadic data. Social Networks 10: 359–381.

36.

Krugman

(1995) Growing world trade: Causes and consequences. Brooking Papers on Economic Activity 1: 327–362.

37.

Lall

Albaladejo

Jinkang

(2004) Mapping fragmentation: Electronics and automobiles in East Asia and Latin America. Oxford Development Studies 32: 407–432.

38.

Lorrain

White

(1971) Structural equivalence of individuals in social networks. Journal of Mathematical Sociology 1: 49–80.

39.

Mahutga

(2006) The persistence of structural inequality? A network analysis of international trade, 1965–2000. Social Forces 84: 1863–1889.

40.

Mansfield

Reinhardt

(2008) International institutions and the volatility of international trade. International Organization 62: 621–652.

41.

Meyer

Rowan

(1977) Institutionalized organizations: Formal structure as myth and ceremony. American Journal of Sociology 83: 340–363.

42.

Mizruchi

(1990) Cohesion, structural equivalence, and similarity of behavior: An approach to the study of corporate political power. Sociological Theory 8: 16–32.

43.

Mizruchi

(1993) Cohesion, equivalence, and similarity of behavior: A theoretical and empirical assessment. Social Networks 15: 275–307.

44.

Mizruchi

Stearns

Marquis

(2006) The conditional nature of embeddedness: A study of borrowing by large U.S. firms, 1973–1994. American Sociological Review 71: 310–333.

45.

Rose

(2004) Do we really know that the WTO increases trade? American Economic Review 94: 98–114.

46.

Sacks

Ventresca

Uzzi

(2001) Global institutions and networks: Contingent change in the structure of world trade advantage, 1965–1980. American Behavioral Scientist 44: 1579–1601.

47.

Smith

White

(1992) Structure and dynamics of the global economy: Network analysis of international trade 1965–1980. Social Forces 70: 857–893.

48.

Snyder

Kick

(1979) Structural position in the world system and economic growth, 1955–1970: A multiple-network analysis of transnational interactions. American Journal of Sociology 84: 1096–1126.

49.

Sturgeon

(2002) Modular production networks: A new American model of industrial organization. Industrial and Corporate Change 11: 451–496.

50.

Tuma

Hannan

(1984) Social Dynamics. New York: Academic Press.

51.

Uzzi

(1996) The sources and consequences of embeddedness for the economic performance of organizations: The network effect. American Sociological Review 61: 674–698.

52.

Uzzi

(1999) Embeddedness in the making of financial capital: How social relations and networks benefit firms seeking financing. American Sociological Review 64: 481–505.

53.

Wallerstein

(1974) The Modern World-System. New York: Academic Press.

54.

White

(2002) Markets from Networks: Socioeconomic Models of Production. Princeton, NJ: Princeton University Press.

55.

Whitford

Potter

(2007) Regional economics, open networks and spatial fragmentation of production. Socio-Economic Review 5: 497–526.

56.

Yeats

(1998) Just how big is global production sharing? World Bank Policy Research Working Paper No. 1871.

57.

Zelizer

(1988) Beyond the polemics on the market: Establishing a theoretical and empirical agenda. Sociological Forum 3: 614–634.

58.

Zelizer

(2003) Enter culture. In: Guillen

Collins

England

Meyer

(eds) The New Economic Sociology: Developments in an Emerging Field. New York: Russell Sage Foundation, pp. 101–128.

59.

Zhou

(2010) Multidimensionality and gravity in global trade, 1950–2000. Social Forces 88(4): 1619–1643.

60.

Zhou

(2011) Intensification of geo-cultural homophily in global trade: Evidence from the gravity model. Social Science Research 40(1): 193–209.

The cohesion effect of structural equivalence on global bilateral trade,1948–2000

Abstract

Keywords

Introduction

Existing empirical literature on bilateral trade

The gravity model

The social network approach and world system theory

Structural equivalence in the global trade network

The trade-promoting effect of structural equivalence

Historical contingency of the effect of structural equivalence

Data and method

Results

Robustness checks

Further exploration

Conclusion and discussion

Footnotes

Acknowledgements

Funding

Notes

Author biographies

References