The gravity model and trade in intermediate inputs

• Date: 01 August 2020
• Published date: 01 August 2020
• DOI: http://doi.org/10.1111/twec.12947
• Author: Kozo Kiyota,Theresa M. Greaney

2034

|

wileyonlinelibrary.com/journal/twec World Econ. 2020;43:2034–2049.

© 2020 John Wiley & Sons Ltd

Received: 1 April 2019

|

Revised: 16 December 2019

|

Accepted: 21 January 2020

DOI: 10.1111/twec.12947

ORIGINAL ARTICLE

The gravity model and trade in intermediate inputs*

Theresa M.Greaney1

|

KozoKiyota1,2,3

1Department of Economics, University of Hawai‘i, Honolulu, Hawai‘i

2Keio Economic Observatory, Keio University, Tokyo, Japan

3Research Institute of Economy, Trade and Industry (RIETI), Tokyo, Japan

Funding information

Zengin Foundation for Studies on Economics and Finance; Japan Society for the Promotion of Science, Grant/Award

Number: JP18KK0348 and JP19H00598

KEYWORDS

Factory Asia, intermediate inputs, intranational trade, structural gravity model

1

|

INTRODUCTION

The gravity model has been used to analyse the determinants of bilateral trade for over 50years, since

Tinbergen (1962). As Yotov, Piermartini, Monteiro, and Larch (2016, p. 5) state "Often referred to

as the workhorse in international trade, the gravity model is one of the most popular and successful

frameworks in economics." Despite this high praise and voluminous literature using the gravity model

to estimate bilateral trade relationships, one may question whether the gravity model is equally appli-

cable to trade in intermediate inputs as it is to trade in final products. While consumers in many coun-

tries may share similar demands for apparel and autos, for example, the demands for textiles and auto

parts presumably differ widely across countries depending on whether a country hosts apparel and

auto manufacturing plants or not. Do the same determinants apply equally well to trade in consumer

(i.e., final) goods as to trade in producer (i.e., intermediate) goods?

This is not a trivial question since international trade in intermediate inputs accounts for the vast

majority of world trade flows in recent years thanks to production fragmentation and the development

of global value chains. The share of intermediate inputs in total imports increased from 56.9% in 1995

to 63.4% in 2011 for all industries, and a similar pattern is confirmed when we focus on the imports

of tradable industries and those of manufacturing industries, as shown in Figure1.1 Given the large

and increasing role of intermediate inputs in world trade, we set out to examine whether the gravity

1Based on the OECD Trade in Value-Added dataset covering 63 economies, which is described further in our data section.

*The authors have benefited greatly from conversations on this topic with Jota Ishikawa. We also thank an anonymous

referee and seminar participants at the Japan Society of International Economics (especially Kazunobu Hayakawa), RIETI

and the University of Hawai‘i. The usual disclaimers apply.

Kozo Kiyota was a Visiting Colleague at the University of Hawai'i when this research was undertaken and would like to thank

Keio University for providing financial support. Kiyota also acknowledges the financial support received from the JSPS

Grant-in-Aid (JP18KK0348, JP19H00598) and a grant-in-aid from Zengin Foundation for Studies on Economics and Finance.

|

2035

GREANEY ANd KIYOTA

model performs as well for describing trade in intermediate inputs as it does for trade in final con-

sumer products.

If the pattern of trade in intermediate inputs is largely proportional to trade in final goods, then our

inquiry may be short-lived. However, we find that the correlation between bilateral final goods trade

and bilateral intermediate goods trade has declined noticeably in recent years, as shown in Figure1.

Although the correlation is strongly positive, it declined over the period from 0.959 in 1995 to 0.861

in 2011. This result suggests that the pattern of trade in intermediate inputs is diverging somewhat

from that of trade in final goods. Baldwin and Taglioni (2014), hereafter referred to as "BT," find that

the gravity model performs well for bilateral trade in all goods, final goods and intermediate inputs

when the analysis covers a wide range of countries but it performs poorly when applied to a particular

subset of countries (i.e., "Factory Asia") in which intermediate inputs account for a rapidly growing

share of trade.

We pursue a similar line of inquiry while making two contributions to the literature by applying

a structural gravity equation inclusive of intranational trade and by explicitly taking into account the

dual nature of goods that can be used as either final consumer goods or intermediate inputs. Our major

findings are twofold. We find that the structural gravity equation performs extremely well for describ-

ing bilateral trade in final goods and in intermediate inputs. In fact, the gravity model performs well

even when we focus on a subset of countries in which intermediate inputs trade accounts for a large

and growing share of trade, namely "Factory Asia." This result is in contrast to the BT finding that the

standard gravity model performs poorly when focused on Factory Asia countries.

Second, we find that the performance of the gravity model declines if the model excludes intra-

national trade and/or if it focuses on a very small subset of countries, even after the model accounts

for the dual nature of products. Our results suggest that the poor performance of the gravity model

FIGURE 1 Share of intermediate inputs to total imports and correlation between the imports of intermediate

inputs and final goods, 1995–2011

Notes: Figure indicates the import share of intermediate inputs to total imports and the correlation between the

imports of intermediate inputs and those of final goods. For sources, see main text.

0.86 0.88 0.9 0.92 0.94 0.96

Correlation

0.55 0.6 0.65 0.7

Share of intermediate inputs

1995 1997 1999 2001 2003 2005 2007 2009 2011

All industry (left) Tradables (left)

Manufacturing (left) Correlation (right)