Regional variations in exporters’ productivity premium: Theory and evidence

• DOI: http://doi.org/10.1111/roie.12398
• Author: Eiichi Tomiura,Toshihiro Okubo
• Date: 01 August 2019
• Published date: 01 August 2019

Rev Int Econ. 2019;27:803–821. wileyonlinelibrary.com/journal/roie

|

803

© 2019 John Wiley & Sons Ltd

1

|

INTRODUCTION

Exporters are, on average, more productive than nonexporters—this is a stylized fact that has been

firmly established in the literature referencing a wide range of data sources since Bernard and Jensen

(1995).1

Various related issues such as the self‐selection of productive firms in exports have been em-

pirically examined (e.g., Bernard & Jensen, 1995; Bustos, 2011; De Loecker, 2007; Roberts & Tybout,

1997). The economic geography literature, on the other hand, confirms that the average productivity

of firms or plants is higher in core regions than in peripheral areas within the same country (e.g.,

Rosenthal & Strange, 2004).2

These two lines of research, however, are yet to be integrated. Using

Japanese plant‐level longitudinal data, this study examines if the productivity premium of exporters

relative to nonexporters differs between the core and periphery regions within a country and accord-

ingly, discusses the theoretical interpretations of plant‐level empirical regularities.

Received: 13 March 2018

|

Revised: 15 December 2018

|

Accepted: 4 January 2019

DOI: 10.1111/roie.12398

ORIGINAL ARTICLE

Regional variations in exporters’ productivity

premium: Theory and evidence

ToshihiroOkubo1

|

EiichiTomiura2

1Faculty of Economics,Keio University,

Tokyo, Japan

2Graduate School of Economics,

Hitotsubashi University, Tokyo, Japan

Correspondence

Toshihiro Okubo, Faculty of Economics,

Keio University, 2‐15‐45 Mita, Minato‐ku,

Tokyo 108‐8345, Japan.

Email: okubo@econ.keio.ac.jp

Funding information

This research is financially supported by

the Research Institute of Economy, Trade

and Industry (RIETI) and Grant‐in‐Aid for

Scientific Research (Grant Nos. 17K03677,

16K03652, 25780169, 26220503,

18H03637). Access to plant‐level data is

arranged by RIETI.

Abstract

The international trade literature confirms that the average

productivity of exporters is higher than that of nonexporters,

while economic geography studies establish that urban

firms tend to be more productive than rural ones. By intro-

ducing region‐specific transportation costs in a Melitz‐type

heterogeneous‐firm trade model, the theory predicts that the

minimum threshold productivity level for export is higher

but that for survival by serving the local market is lower in

the periphery region than in the core. Using Japanese plant‐

level panel data, we find evidence supporting the theoretical

prediction that exporters in the peripheral regions, espe-

cially those distant from the core, have large productivity

premiums.

JEL CLASSIFICATION

F1, R12, L25

804

|

OKUBO and TOMIURa

A productivity advantage that exporters have over nonexporting domestic firms is the central element

of heterogeneous‐firm trade (HFT) models pioneered by Melitz (2003). Despite observed differences in

transportation costs across countries and regions, most theoretical models concentrate on the relation-

ship between inter‐firm productivity variations and the firms’ export decision. To consider such cross‐

regional differences, we introduce transportation costs that vary by region in a Melitz‐type HFT model.

Applying a Melitz‐type HFT model to the core–periphery structure within a country trading with

another, we propose the following two hypotheses. The productivity premium of exporters over non-

exporters is larger in the periphery (a region with limited access to the foreign country), because it is

more difficult for local firms to export to the foreign market but easier to survive in the local market.

This core–periphery gap in exporters’ productivity premium widens with an increase in the distance

between the periphery and core regions.

This study tests these predictions by examining variations in exporters’ productivity premium

across regions in Japan. To preview our principal findings, exporters tend to be more productive than

nonexporters particularly in the periphery region. That is, exporters in areas more distant from the

core (Tokyo or Osaka) have a larger productivity advantage. These findings are consistent with our

theoretical prediction and with the fact that most Japanese exports are through ports or airports located

at the core. The core–periphery contrast is observed not only in terms of the average productivity level

but also in the shape of productivity distribution.

While this study focuses on the core–periphery structure within a country, international compar-

ison is also an important research topic in the case of spatial variations in exporter premiums. The

International Study Group on Exports and Productivity (ISGEP, 2008) compares the productivity

premium across 14 countries.3

While they compare countries by constructing proxies for regulatory

qualities, the investigation of Japan, which is not a federated country, is suitable for the purpose of our

study because the central government holds strong authority in imposing regulations common to all

its regions.4

Further, even though we focus on within‐country differentials, regions vary sufficiently

in terms of market size within Japan. For instance, Tokyo and Osaka as prefectures report larger GDPs

than those of many countries in the ISGEP sample (2008).5

Bellone, Kiyota, Matsuura, Musso, and Nesta (2014), who also conduct a cross‐country compar-

ison, report that the productivity premium of exporters over nonexporters differs between Japan and

France. While they mainly discuss cross‐sectoral variations, the authors interpret their finding as an

indication that the cutoff threshold for export entry varies by country. Unlike Bellone et al.’s compar-

ison of Japan with France, this study focuses on regional variations in Japan.

Although productivity is not discussed, several studies analyze the impact of intra‐national geogra-

phy on the international trade of firms located across regions. Anderson and Van Wincoop (2004) em-

phasize that internal geography is a critical element in understanding the trade costs of international

transactions. Using French firm‐level data, Crozet and Koenig (2010) detect a significant negative

effect of interior distance on the number of exporters and average export values. Albarran et al. (2013)

show that longer domestic travel time reduces the probability of exporting among small‐ and medium‐

sized firms in Spain. Martincus and Blyde (2013) identify a causal effect of within‐country travel

distance on firms’ export values in Chile using an earthquake as a natural experiment. These findings

suggest that entry thresholds for exports should differ by region within a country.6

Accordingly, we

examine productivity premiums required for plants to export from different regions in Japan with

varying distance to the core, where major international ports and airports as well as various trade‐fa-

cilitating service functions are concentrated.

The remainder of this paper is organized as follows. Section 2 formalizes our theoretical predic-

tions. Section 3 describes our plant‐level data. Section 4 reports the productivity premium compari-

sons across prefectures. Section 5 directly analyzes plant‐level regularities. Section 6 concludes.