Impact of commodity prices on exchange rates in commodity‐exporting countries
| Author | Amalia Morales‐Zumaquero,Rebeca Jiménez‐Rodríguez |
| Published date | 01 July 2020 |
| Date | 01 July 2020 |
| DOI | http://doi.org/10.1111/twec.12952 |
1868
|
wileyonlinelibrary.com/journal/twec World Econ. 2020;43:1868–1906.
© 2020 John Wiley & Sons Ltd
Received: 30 April 2019
|
Revised: 14 January 2020
|
Accepted: 25 February 2020
DOI: 10.1111/twec.12952
ORIGINAL ARTICLE
Impact of commodity prices on exchange rates in
commodity-exporting countries
RebecaJiménez-Rodríguez1
|
AmaliaMorales-Zumaquero2
1Department of Economics, Campus Miguel de Unamuno, University of Salamanca, Salamanca, Spain
2Department of Economics, Campus El Ejido, University of Málaga, Málaga, Spain
Funding information
Consejería de Educación, Junta de Castilla y León, Grant/Award Number: SA049G19; Spanish Ministry of Science,
Innovation and Universities, Grant/Award Number: PRX18/00501 Salvador de Madariaga grant
KEYWORDS
commodity prices, real effective exchange rate, structural breaks
1
|
INTRODUCTION
Commodity prices have shown a strong upward swing in the early 2000s. Such upward movements
have raised the academic interest in the long-run behaviour of these prices, especially in those coun-
tries whose exports mainly depend on commodities. In the late 1990s, the literature on the determi-
nants of the long-run exchange rate concluded that the key determinants of the long-run real exchange
rate were productivity differentials, cumulated current account imbalances, government spending and
interest rate differentials, among others (see Froot & Rogoff, 1995, and Rogoff, 1996, for nice re-
views). More recently, Frankel (2010) has shown how a considerable hike in the export commodity
prices can cause an appreciation of the currency, among other macroeconomic effects.1 However,
there have been very few authors who have focused on studying how real commodity prices can affect
real exchange rate in commodity-exporting countries (Bodart, Candelon, & Carpantier, 2012; Cashin,
Céspedes, & Sahay, 2004; Chen & Rogoff, 2003), and they all have concluded that an increase in
commodity prices appreciates the real exchange rate in most of countries analysed.
From a theoretical perspective, the positive relationship between commodity prices and real ex-
change rate, previously mentioned, has given rise to two explanations on the basis of what variable
anticipates the movements of the other. The first explanation is based on the so-called "commodity
currencies", which establishes that changes in commodity prices lead to changes in exchange rate
(Bodart et al., 2012; Cashin et al., 2004; Chen & Rogoff, 2003; Coudert, Couharde, & Mignon, 2011;
1Another strand of the literature has studied the key factors that explain the recent commodity boom (Baffes and Haniotis,
2010) and the common factors of commodity prices (see, for instance, Delle Chiaie et al., 2017).
|
1869
JIMÉNEZ-RODRÍGUEZ aND MORaLES-ZUMaQUERO
Dauvin, 2014). The second explanation suggests that movements in exchange rate lead to movements
in commodity prices (Engel & West, 2005; Meese & Rogoff, 1983; Rossi, 2013).
From an empirical perspective, Chen and Rogoff (2003) analyse three developed countries
(Australia, Canada and New Zealand) by using cointegration techniques and find that commodity
prices have a strong and positive influence on the real effective exchange rate in Australia and New
Zealand, but a significant negative impact on the Canadian dollar.2 Cashin et al. (2004) examine
whether real exchange rates and commodity prices move together over time by considering 58 individ-
ual commodity-exporting countries with data from 1980 to 2002. They show evidence of a significant
long-run relationship between commodity prices and real exchange rates for about one-third of their
sample of countries. Bodart et al. (2012) use a panel data model with 68 developing and emerging
countries and 42 commodities for the period 1988–2008 and conclude that the price of the dominant
commodity has a significant long-run impact on real exchange rate. Dauvin (2014) uses panel cointe-
grating techniques with 10 energy-exporting and 23 non-fuel commodity-exporting countries for the
period 1980–2011 and provides evidence of the existence of "energy currencies".3 Coudert, Couharde,
and Mignon (2015) consider a panel of 68 commodity exporters, distinguishing between advanced,
intermediate and low-income countries, and obtain that only advanced exporter's currencies are sensi-
tive in the short-run to changes in terms of trade. More recently, Zhang, Dufour, and Galbraith (2016)
analyse the link between exchange rates and commodity prices for three commodities (crude oil, gold
and copper) and four countries (Canada, Australia, Norway and Chile) with data from 1986 to 2015.
Their results suggest that the causality between exchange rates and commodity prices is in both direc-
tions and is stronger at short-run horizons.
This paper is motivated by the sharp upward movements in commodity prices in the early 2000s
and their influence on real exchange rate for commodity-exporting countries, as well as the lack of
recent studies that performed a country-specific analysis of such an influence. Therefore, we are in-
terested in analysing how commodity prices affect real exchange rate behaviour and, consequently, we
focus on the so-called "commodity currencies" explanation.
Most of recent studies existing in the related literature have been carried out by using panel data
techniques, which does not make possible to carefully analyse each country individually, and have
obtained general conclusions for the group of countries considered. Therefore, it is our aim to study
the effects of commodity prices on real exchange rates in those economies that highly depend on com-
modities by considering countries individually and taking into account the possibility of structural
breaks.4,5 In particular, we analyse how real exchange rate reacts when an increase in the price of the
2This result is consistent with the findings of Amano and van Norden (1995) and Issa et al. (2006).
3Additional evidence in favour of "oil currencies" can be seen in Habib and Kalamova (2007), Korhonen and Juurikkala
(2007) and Coudert et al. (2011), among others.
4Considering structural breaks is important not only in testing unit roots of time series (the presence of breaks in stationary
time series may induce apparent unit roots in those series), but also in cointegration testing (the failure to find cointegration
may reflect in fact a cointegrating relationship with a structural break). Moreover, structural breaks are also relevant to model
the relationship between cointegrated/non-cointegrated variables.
5To the best of our knowledge, there are few studies that consider the possibility of structural breaks in the relationship
between real exchange rates and commodity prices. The main exception is Cashin et al. (2004) who examine cointegration
allowing a structural shift by considering the Gregory and Hansen (1996) test and obtain evidence of a stable cointegration
relation between real exchange rates and commodity prices for their group of countries. It is worth noting that these authors
also consider a single structural change when analyzing the order of integration by using Zivot and Andrews (1992) unit root
test. Notice that our approach allows us not to impose the existence of a single structural break, but we allow the existence of
up to five shifts, which is a considerable gain.
1870
|
JIMÉNEZ-RODRÍGUEZ aND MORaLES-ZUMaQUERO
main commodity (i.e., the one whose share is at least 20% of total commodity export) occurs allowing
for structural breaks in the relationship.
As a first step, we analyse the properties of the individual time series considered (i.e., commod-
ity prices and real exchange rates). In particular, we study the order of integration of each time series
considered, although we previously take into consideration the possibility of structural changes in
each individual time series (given that at first glance there seems to be structural breaks). Thus, we
first test for the existence of multiple structural breaks in each individual time series by following
the Bai and Perron (1998, 2003) methodology, which is specified in order to analyse endogenous
changes in a univariate time series. Having established the presence of structural breaks in each time
series,6 we second study the order of integration of each time series allowing for multiple structural
breaks by applying the Carrion-i-Silvestre, Kim, and Perron (2009) test. Once the order of integra-
tion for each variable is established and in the case of I(1) variables, as a second step, we analyse
the existence of cointegration between commodity prices and real exchange rates for each particular
country by considering the methodology developed by Maki (2012) that allows us to test for cointe-
gration with multiple structural breaks. As a final step, two different empirical models are specified
on the basis of the existence or non-existence of cointegration between commodity prices and real
exchange rates.
The rest of the paper is organised as follows. Section2 presents data and methodology. Section3
reports the main empirical results. Section4 draws some concluding remarks.
2
|
DATA AND METHODOLOGY
2.1
|
Data
Original commodity prices come from International Monetary Fund (IMF) Primary Commodity
Prices database (version 6 March 2019).7 This database contains spot prices of 68 commodities that
are internationally traded, five groups (food, beverages, agricultural raw materials, metals and petro-
leum), three sub-blocks (edibles, industrial inputs and fertilisers) and two blocks (non-fuel and en-
ergy), as well as a global price index of all commodities. Table1 contains the structure of commodity
price database, including the weights.8 It is worth noting that crude oil (POILAPSP) and natural gas
(PNGAS) prices are price indices with base year 2016 (2016=100). Thus, we have transformed all
market prices into price indices considering 2016 as the base year in order to homogenise our com-
modity price database. Moreover, we obtain real commodity prices by deflating by US CPI.9
We consider the same 68 countries as Coudert et al. (2015). Following these authors, we categorise
them on the basis of the development level (advanced economies, intermediate economies—emerging
or developing countries—and low-income economies).
6In this study, there are no time series in which no structural breaks appear.
7See technical documentation of IMF Primary Commodity Prices database, http://www.imf.org/exter nal/np/res/commo d/
index.aspx
8Tables A1 and A2 describe each of the composite price indices and each individual commodity price index, respectively.
9It has been recently common to obtain real commodity prices by deflating nominal ones according to US CPI (see, for
instance, IMF database, https://www.imf.org/~/media /Files /Resea rch/Commo dityP rices /Month ly/Charts.ashx) given that the
IMF's index of the unit value of manufactured exports (MUV) is no longer available. We would like to thank Vincent Bodart
for providing us the MUV data up to 2009M12. This has allowed us to verify that there are no significant differences between
deflating by US CPI and deflating by MUV with data up to 2009M12.
To continue reading
Request your trial