Trade balance dynamics and exchange rates: In search of the J‐curve using a structural gravity approach

• Published date: 01 September 2019
• DOI: http://doi.org/10.1111/roie.12426
• Date: 01 September 2019
• Author: Harald Badinger,Aurélien Fichet de Clairfontaine

1268

|

Rev Int Econ. 2019;27:1268–1293.

wileyonlinelibrary.com/journal/roie

1

|

INTRODUCTION

The prevalence of large and persistent global imbalances is seen as a major threat to the stability of

the world economic system. Hence, identifying and quantifying the effects of the main determinants

of the current (and financial) account is an issue that is repeatedly raised to the fore in both academic

and public debates. The exchange rate, as the most important single price of an economy and crucial

determinant of relative prices between domestic and foreign goods, is one key factor influencing

DOI: 10.1111/roie.12426

ORIGINAL ARTICLE

Trade balance dynamics and exchange rates: In

search of the J‐curve using a structural gravity

approach

HaraldBadinger1,2,3

|

AurélienFichet de Clairfontaine1

This is an open access article under the terms of the Creat ive Commo ns Attri bution License, which permits use, distribution and reproduction

in any medium, provided the original work is properly cited.

© 2019 The Authors. Review of International Economics Published by John Wiley & Sons Ltd

1Vienna University of Economics and

Business, Vienna, Austria

2Austrian Institute of Economic Research

(WIFO), Vienna, Austria

3CESifo, Munich, Germany

Correspondence

Harald Badinger, WU Vienna, Department

of Economics, Welthandelsplatz 1, A‐1020

Vienna, Austria.

Email: harald.badinger@wu.ac.at

Abstract

This paper uses a structural gravity approach, specifying

currency movements as trade cost component to derive an

empirical trade balance model, which incorporates multi-

lateral resistance terms and accounts for the cross‐country

variation in the exchange rate pass‐through into import

and export prices. The model is estimated using quarterly

bilateral trade flows between 47 countries over the period

2010Q1 to 2017Q2, disaggregated into 97 commodity

groups. Our results support the existence of an “aggregate”

J‐curve, pooled over commodity groups; at the same time

they point to considerable heterogeneity in the trade balance

dynamics across industries below the surface of aggregate

data.

JEL CLASSIFICATION

F12; F31; F32

|

1269

BADINGER AND FICHET DE CLAIRFONTAINE

global imbalances. In policy discussions of bilateral imbalances, the allegation of exchange rate ma-

nipulation and demands for realignments can be observed quite frequently.

From a theoretical perspective, the standard Marshall–Lerner condition specifies when a depreci-

ation leads to an improvement of the trade balance, assuming perfect competition, rigid prices, com-

plete exchange rate pass‐through and infinite export supply elasticities. It reveals that a depreciation

has three effects: a price effect, since imports become more expensive, and quantity responses of

exports and imports owing to changes in their relative prices. This basic insight also holds true under

more general assumptions.

The price effect typically materializes more quickly than the quantity effects. As a consequence,

a depreciation may lead to an incipient deterioration of the trade balance, which subsequently turns

into a positive effect after the quantity effects have worked themselves out. This gives rise to a J‐curve

effect of a depreciation on the trade balance (or an inverted J‐curve effect of an appreciation on the

trade balance).

The J‐curve phenomenon and the “sluggishness of quantity” was first considered in detail by Magee

(1973). Up to the late 1980s, the J‐curve hypothesis has then been repeatedly tested using aggregate

trade data, investigating the link between a country’s real effective exchange rate and its trade balance

vis‐à‐vis its most important trading partners using time‐series techniques (e.g., Bahmani‐Oskooee,

1985; Himarios, 1985). These types of studies, which show mixed results on the presence of J‐curves,

were criticized for being potentially subject to an aggregation bias that conceals effects taking place at

the bilateral level (Bahmani‐Oskooee & Brooks, 1999).

Rose and Yellen (1989) were the first to use bilateral trade data and test the J‐curve hypothesis for

country pairs, utilizing cointegration techniques proposed by Engle and Granger (1987), but they find

no support for the presence of a J‐curve. More recent studies make use of an error‐correction version

of an autoregressive distributed lag (ARDL) model, suggested by Pesaran, Shin, and Smith (2001).

Overall, as suggested by the comprehensive survey by Bahmani‐Oskooee and Ratha (2004), the em-

pirical evidence on the existence of a J‐curve is rather mixed.

The most widely used models for the analysis of trade balance dynamics strongly resemble early

empirical gravity equations by relating the export–import ratio to relative economic size (proxied by

GDP) and the (real) exchange rate. Additional (ad‐hoc) variables included in previous studies are

GDP growth, government consumption or the level of high‐powered money (see Bahmani‐Oskooee

& Ratha, 2004).

A shortcoming even of recent studies on trade balance dynamics is that they do not reflect the

considerable progress that has been made in the gravity literature, which emphasizes the importance

of multilateral resistance terms (Anderson & Van Wincoop, 2003) and incorporates the exchange

rate (and its pass‐through) as trade cost component (Anderson, Vesselovsky, & Yotov, 2016). This

widespread lack of a rigorous theoretical foundation may be an explanation for the mixed or negative

results about the presence of a J‐curve in the vast majority of previous studies.

The present paper addresses these shortcomings by setting up a trade balance model that builds

on a structural gravity model, shifting the focus from a bilateral to a multilateral analysis, account-

ing for third‐country effects and incorporating cross‐country differences in the exchange rate pass‐

through. The empirical model is tested for a comprehensive and recent dataset over the period 2010

to 2017, including quarterly observations on bilateral trade flows between 47 (mainly OECD) coun-

tries, disaggregated into 97 commodity groups, with a total of up to 64,860 observations per com-

modity group.1

We find that, when pooling across commodity groups, the trade balance deteriorates over the

first two quarters following a depreciation. This effect persists for four quarters and is then fol-

lowed by a trade balance improvement in the long run, thus providing evidence for an “aggregate”