Trade networks and colonial trade spillovers

• DOI: http://doi.org/10.1111/roie.12288
• Date: 01 September 2017
• Published date: 01 September 2017

ORIGINAL ARTICLE

Trade networks and colonial trade spillovers

Antoine Berthou

1

|

H

elène Ehrhart

2

1

Banque de France and Centre d’Etudes

Prospectives et d’Informations

Internationales (CEPII), Paris,France

2

Agence Française de D

eveloppement,

Paris, France

Correspondence

Antoine Berthou, Banque de France,

31 rue Croix des Petits Champs,

75001 Paris, France.

Email: antoine.berthou@banque-france.fr

JEL Classification: F14, F15

Abstract

This paper provides new empirical evidence regarding the

formation of international trade networks. Established trade

relations may open the gate to new trade opportunities, as

they allow meeting new trade partners over time. We test

this prediction and its implications for aggregate trade pat-

terns by using the experience of ancient trade linkages

between former colonies and their former colonizers (colo-

nial trade linkages). We first show, using aggregate trade

data, that former colonies have more trade with former colo-

nizer’s neighbors (colonial trade spillovers). We then show

that the past export and import experience of former colonies

with the colonizer have an impact on the propensity to trade

similar products with third countries. In particular, the trade

spillover effect is negatively related to geographical distance

between third countries and the colonizer, and positively

affected by their degree of economic integration.

KEYWORDS

internationaltrade dynamics, networks formation, colonies

1

|

INTRODUCTION

Business networks are probably one of the most prominent forces shaping international trade flows. They

contribute to explain the geographical distribution of trade across countries (Chaney, 2013, 2014), as well

as the transmission of shocks across sectors and borders (Acemoglu, Carvalho, Ozdaglar, & Tahbaz-

Salehi, 2012; Di Giovanni & Levchenko, 2010; Johnson, 2014). Understanding better how these trade net-

works are formed, and their functioning, is therefore of a great interest for researchers and policy makers.

Our objective is to provide new empirical evidence regarding the formation of a trade network.

Our strategy is guided by the theoretical work by Chaney (2014), which predicts that trade networks

expand over time to more distant locations through the existing networks of trade partners. We test in

particular whether trade experience in a given country can generate new trade opportunities with other

countries, and investigate the role played by geographical and political factors. On the empirical side,

the challenge is to deal with the endogeneity of the trade links. Suppose the simple example of a three

countries’network A, B, and C. The trade experience of country A with country B could generate new

trade opportunities with country C, but the A–B trade relation could itself result from the A–C linkage.

This problem is similar to the “reflection problem”in the social networks’literature (Manski, 1993;

Rev IntEcon. 2017;25:891–923. wileyonlinelibrary.com/journal/roie V

C2017 JohnWiley & Sons Ltd

|

891

Received: 9 April 2015

|

Revised: 4 January 2017

|

Accepted: 16 January 2017

DOI: 10.1111/roie.12288

Bramoull

e, Djebbari, & Fortin, 2009; Topa & Zenou, 2015). The empirical analysis of the trade net-

work formation therefore requires starting from a pair of countries having a well established trade rela-

tion, which is not itself endogenously determined by other trade relations.

We make use of the experience of the preferential trade relations between former colonizers

(France and the United Kingdom) and their former colonies (colonial trade linkage), which has been

well established in the trade literature (Head, Mayer, & Ries, 2010). Former colonies trade more with

their former colonizer than with other countries, controlling for other gravity forces and multilateral

resistance terms. This empirical pattern has been attributed, beyond preferential trade policy or mone-

tary agreements, to the existence of informal institutions taking the form of rules and practices facilitat-

ing business, some of them having survived consecutive to the decolonization process (Rauch, 2001;

Head et al., 2010). Given the strength and the stability of the colonial trade linkage in empirical trade

studies, it is unlikely to result from other trade relations. We therefore use this linkage in order to

explore the formation of trade networks consecutive to trade experience.

Our empirical strategy focuses on French and British colonial empires in 1945, and the implica-

tions of this colonial history for former colonies’trade relations during the two decades preceding the

Great Recession.

1

In doing so, we are particularly interested in a network of three countries being

potentially linked through international trade relations: the former colonizer, its former colony, and a

third country. We investigate whether colonial trade linkages fostered the creation of new trade rela-

tions with third countries (colonial trade spillovers), i.e., had an impact on trade relations of former

British and French colonies with other countries, and contributed to shape the geographical distribution

of their aggregate exports and imports.

We first provide empirical evidence, using aggregate bilateral exports’data for the period 1995 to

2007 for 198 exporting and importing countries (Base pour l’Analyse du Commerce International

[BACI] dataset), that former colonies of British and French colonial empires have more trade—exports

or imports—with countries that are geographically closer to the former colonizer than other countries.

This result is obtained from the estimation of a gravity equation, controlling in particular for different

factors such as bilateral distance, trade policy, language or migrations that would directly affect the

value of trade between country pairs, as well as for multilateral resistance terms that would influence

former colonies’trade propensity with the rest of the world. In quantitative terms, we find that E.U.

countries trade about 50 percent more with former French and British colonies than what would be pre-

dicted by a gravity framework. Geographical distance to the colonizer is found to be a strong determi-

nant of former colonies’exports and imports with third countries, affecting the probability that a trade

relationship exists and the intensity of this relation especially in terms of the number of products

exported and imported. Finally, we find that the negative effect of distance to colonizer on bilateral

trade flows is more robust in differentiated goods sectors, consistent with the idea that trade networks

tend to foster the intensity of trade relations with former colonizers’neighbors.

We then test for the presence of trade spillovers in relation to the colonial trade linkage. We are

particularly interested in exploring how trade networks expand over time. Following the predictions

from the model of endogenous trade networks by Chaney (2014), we expect that the development of

trade networks is biased towards existing trade partners’neighbors. The empirical strategy relies on the

estimation of a linear probability model, and uses information about bilateral exports and imports val-

ues at the product level. The key variable in this model is the export or import experience for a certain

product in the colonizer’s market. This variable is then interacted with various factors related to geog-

raphy or economic integration, which link the trade partner with the former colonizer. The results indi-

cate that trading a certain product with the former colonizer increases the probability of exporting or

importing the same product with the former colonizer’s neighbors in subsequent years, or countries

characterized by a high degree of economic integration with the former colonizer (sharing the same

892

|

BERTHOU AND EHRHART

currency or having signed a free trade agreement [FTA] with that country). These results are robust to

the choice of the set of products that former colonies may possibly export or import.

Our findings are in line with the empirical predictions from the model of trade networks by Chaney

(2014). In this model, firms can expand in different markets through “direct search”or “remotesearch”in

the presence of search frictions. With the latter, firms can search for new trade partners using their net-

work of clients and suppliers, which allows them to expand their trade relations with more distant coun-

tries over time.

2

These results are also consistent with trade theories that emphasize the role played by

learning or matching through trade experience in a foreign market (Albornoz, Calvo Pardo, Corcos, &

Ornelas, 2012; Defever, Heid, & Larch, 2015; Nguyen, 2012; Timoshenko, 2015a,b). Both trade models

with learning into export markets and the model of trade networks by Chaney (2014) predict that trade

experience can generate a geographical spread of imports and exports. Also, our result showing that the

value of experience tends to decay over time is in line with Berman, Rebeyrol, and Vicard (2015).

Empirical evidence of such mechanisms has been produced mostly using firm-level data for single

countries. Chaney (2014) shows using French firm-level data that where a firm exports in the current

year is partly determined by where it was exporting previously, and that geography plays a key role in

the path of the expansion of exports of the firm. Albornoz et al. (2012) demonstrate that Argentinean

exporters sequentially export into foreign markets, starting from neighboring countries.

3

Defever et al.

(2015) use the end of multi-fibre agreements to predict Chinese firms’entry into European Union, U.

S. and Canadian markets. They show that after entry, Chinese firms tend to expand towards markets

that were geographically and culturally proximate from previous destinations. Morales, Sheu, and Zah-

ler (2011) use a moments’inequalities approach to predict the path of expansion of exports of firms

into foreign markets by Chilean exporters. Their empirical structural model emphasizes the role played

by gravity and “extended”gravity forces, the latter depending on the similarity between the destination,

and countries where the firm exported previously.

4

In our approach, we use exports data at the product

level for several countries and consider all potential destinations in this data. We use the experience of

the colonial trade linkage to explore how trade networks expand over time and space. We provide

empirical evidence that the micro channels that govern the formation of trade networks can explain

some empirical patterns in the geographical distribution of trade flows by countries, such as the impor-

tance of third E.U. countries in the total trade of former British and French former colonies.

Finally, our results are also consistent with the geographical concentration of production networks

emphasized in recent works on global value chains (Johnson and Noguera, 2012a,b), as trading with

companies located in the former colonizer’s market may generate new trading opportunities with other

members of the same production network.

The paper is organized as follows. In Section 2, we develop a brief history of colonization and dis-

cuss its implications for our empirical exercise. Section 3 presents the empirical methodology and the

data used in our estimations. Evidence of a colonial trade spillover based on aggregate trade data is

also presented in this section. In Section 4, we use the product detail of the data on bilateral exports

and provide evidence that the microeconomic dynamics of trade contribute to explain the existence of

a colonial trade spillover. We also discuss the relevance of these results with respect to theoretical pre-

dictions. The last section concludes.

2

|

COLONIAL TRADE LINKAGES: HISTORICAL

BACKGROUND AND MEASUREMENT

2.1

|

Colonization and the colonial trade linkage

The European explorations of the late 15th century, which opened new navigation routes to America

and Asia, initiated a process of territorial expansions leading to the constitution of the European

BERTHOU AND EHRHART

|

893