Export diversification and economic development: A dynamic spatial data analysis

• DOI: http://doi.org/10.1111/roie.12316
• Author: Rosanna Pittiglio,Roberto Basile,Aleksandra Parteka
• Date: 01 August 2018
• Published date: 01 August 2018

SPECIAL ISSUE PAPER

Export diversification and economic development:

A dynamic spatial data analysis

Roberto Basile

1

|

Aleksandra Parteka

2

|

Rosanna Pittiglio

1

1

Universit

a degli Studi della Campania

“Luigi Vanvitelli”, Caserta, Italy

2

Gdansk University of Technology,

Gdansk, Poland

Correspondence

Aleksandra Parteka, Gdansk University

of Technology, Faculty of Management

and Economics, Gdansk, Poland.

Email: aparteka@zie.pg.gda.pl

Funding Information

EU RTD Framework Programme: The

Action IS1104 “The EU in the new com-

plex geography of economic systems:

models, tools and policy evaluation”

(Gecomplexity); National Science Centre

(Poland), Decision No. DEC-2015/19/B/

HS4/02884 (Aleksandra Parteka)

Abstract

This paper contributes to the empirical literature on the rela-

tionship between “export variety”(export diversification) and

economic development by relaxing the assumption of cross-

country independence and allowing for spatial diffusion of

shocks in observed and unobserved factors. Export variety is

measured for a balanced panel of 114 countries (1992–2012)

using very detailed information on their exports (HS 6-digit

product level). The estimation results of a dynamic spatial

panel data model confirm the relevance of spatial network

effects in export diversification: indirect effects (spatial spill-

overs) strongly reinforce direct effects, while spatial proximity

to large countries accelerates the diversification process. In

about 10 years the whole space–time diffusion of the diversi-

fication shock is widely completed. We reveal that the long-

run spillover impact from European countries is much higher

than from other countries such as the United States, Japan, or

the BRICS (Brazil, Russia, India, China, and South Africa).

1

|

INTRODUCTION

Diversification paths during the process of economic development is a topic that has attracted the atten-

tion of many economists (Imbs & Wacziarg, 2003; Koren & Tenreyro, 2007; Klinger & Lederman,

2006; Cadot, Carrère, & Strauss-Kahn, 2011, 2013; Minondo, 2011; Parteka & Tamberi, 2013a,b;

Mau, 2016). Diversifying exports is one of the main strategies that a country may follow to reduce

uncertainty (Di Giovanni & Levchenko, 2011; Koren & Tenreyro, 2007, 2013). This ability is espe-

cially crucial in the case of developing countries, which are typically characterized by low diversifica-

tion of their economic structure (Amurgo-Pacheco & Pierola, 2008; Carrère & Strauss-Kahn, 2014).

From a theoretical point of view, increasing the variety of goods produced is expected to exert a posi-

tive impact on productivity and economic growth as shown, for instance, in models of “expanding

product variety”(Barro & Sala-i-Martin, 2004, pp. 285–315; Grossman & Helpman, 1991a, pp. 43–83,

1991b). Consequently, it is not surprising that the topic of evolving diversification along the path of

growth has been widely explored, mainly empirically.

1

634

|

V

C2017 JohnWiley & Sons Ltd wileyonlinelibrary.com/journal/roie Rev IntEcon. 2018;2 6:634–650.

DOI: 10.1111/roie.12316

Discussion so far has mainly regarded the relationship between GDP per capita and levels of diver-

sification of economic activity. Some authors (Imbs & Wacziarg, 2003; Koren & Tenreyro, 2007;

Klinger & Lederman, 2006; Cadot et al., 2011) argued that in the first stage, at low levels of income,

growth goes in line with an increase in the level of diversification; however, once countries reach a

certain level of income, further growth is accompanied by re-concentration.

2

Several scholars

(De Benedictis, Gallegati, & Tamberi, 2008, 2009; Parteka, 2010; Parteka & Tamberi, 2013a,b),

however, show skepticism about the robustness of these patterns, correcting conventional, absolute

measures of product diversity and find a nonlinear but monotonically decreasing trend reflecting

progressive relative de-specialization along the path of economic growth. More recently, Mau (2016)

has stressed that the above-cited nonmonotonic hump-shaped pattern is mainly due to an omitted log-

transformation of the income variable, as well as sample selection bias and lack of control variables.

By focusing on measurement issues (absolute vs. relative measures of export diversification),

3

the

functional form of the model (linear vs. quadratic) and other model specification issues (log-transforma-

tion, dynamic specification, and so on), the empirical literature has totally neglected another important

source of bias, namely the existence of cross-country (or spatial) dependence in the data-generating

process. Indeed, all the aforementioned studies analyze the relationship between trade diversification

and economic development under the (implicit) assumption of spatial independence.

4

In other words,

they do not consider any kind of spatial co ntagion among countries in the specialization process. This

is quite surprising, given the strong links between countries involved in the global trade network

(De Benedictis & Tajoli, 2011; Chaney, 2014) and the network structure of economic output

(Hausmann & Hidalgo, 2011).

Several terms are used in the literature to describe the phenomenon of the interaction between

agents (e.g., countries) being shaped by geography: spatial diffusion, spatial contagion, spatial spillover

effects, and network effects. Leaving aside other disciplines (such as sociology or urban studies), the

main areas of application of these concepts in economics include: economic geography and agglomera-

tion economics (Krugman, 1991; Fujita, Krugman, & Venables, 1999; Fujita & Thisse, 2002; Duranton

& Puga, 2004; Glaeser, 2008), the spatial diffusion of knowledge, technology and innovat ion (Comin,

Dmitriev, & Rossi-Hansberg, 2012; Ertur & Koch, 2007, 2011) and mechanisms of contagion in finan-

cial markets (Allen & Gale, 2000).

What kind of channels can lead to similar patterns of export structure (in particular, the level of

export diversification) among countries close to each other in geographical and/or economic terms?

The first obvious channel is trade itself.

5

Whatever its driving force (differences in endowments in the

Heckscher–Ohlin framework,

6

differences in productivity in the Ricardian framework, or others), inter-

national trade inevitably leads to the creation of ties among countries and to cross-country interdepend-

ence. Useful insights into possible transmission channels are also provided by endogenous growth

models with international R&D spillovers, imitation of innovation, and technology diffusion (Aghion

& Howitt, 1997; Howitt, 2000; Grossman & Helpman, 1991a; Coe, Helpman, & Hoffmaister, 1997)

especially in a Schumpeterian multi-country setting (Ertur & Koch, 2011).

In particular, an important reference point for the study of diversification dynamics in a spatial

dependence setting is still the New Trade Theory, NTT (Krugman, 1995; Neary, 2009), which explains

why similar countries trade intensively, exploiting economies of scale and drawing utility gains from

access to a wider variety of goods (“love of variety”). Product differentiation through “love of variety”

is also a key element of New Economic Geography (NEG) models (surveyed in Fujita et al., 1999;

Brakman, Garretsen, & Van Marrewijk, 2009) where the tension between agglomeration and disper-

sion forces determines the spatial distribution of economic activity. NEG endogenizes location in an

international trade model (Brakman, Garretsen, & Van Marrewijk, 2014), so in the context of our

study, one may think of the following theoretical explanation for the importance of spatial patterns and

BASILE ET AL.

|

635