Forecasting exchange rates with commodity prices—a global country analysis

• Author: Jens Klose,Martin Baumgärtner
• Published date: 01 September 2019
• Date: 01 September 2019
• DOI: http://doi.org/10.1111/twec.12802

2546

|

wileyonlinelibrary.com/journal/twec World Econ. 2019;42:2546–2565.

© 2019 John Wiley & Sons Ltd

Received: 29 March 2019

|

Accepted: 5 April 2019

DOI: 10.1111/twec.12802

ORIGINAL ARTICLE

Forecasting exchange rates with commodity

prices—a global country analysis

MartinBaumgärtner

|

JensKlose

Fachbereich Wirtschaft,Technische Hochschule Mittelhessen, Gießen, Germany

KEYWORDS

commodity prices, exchange rate, forecast, panel analysis

JEL CLASSIFICATION

F17; F31; F47; C23

1

|

INTRODUCTION

In economic forecasts, commodity prices serve as early indicators for future developments. The prices

for oil or copper are used to determine how the supply and demand of the industry will develop. They

are also of interest for monetary policy, as commodity prices are frequently used to forecast the ex-

change rates of various countries.1

Meese and Rogoff (1983) were the first to study the forecasting

properties of commodities on exchange rates. However, they were unable to beat the random walk,

which is why they concluded that commodities are not suitable for forecasting exchange rates.

Subsequent studies found the opposite effect (Bouoiyour, Selmi, Tiwari, & Shahbaz, 2015; Chen &

Chen, 2007; Ferraro, Rogoff, & Rossi, 2015; Mark, 1995). Thus, there is still no consensus in litera-

ture upon this issue. The results vary heavily depending on the period under investigation and the

countries used in the sample.

The paper extends the existing literature by various aspects. Based on the methodology of Chen

and Rogoff (2003) and Kohlscheen, Avalos, and Schrimpf (2017), this work provides a comprehensive

picture of the predictive properties of commodity prices for a large number of 126 countries over a

period from 1993 to 2016. To the best of our knowledge, this is the first paper that uses such a broad

country sample. While many studies concentrate only on a relatively limited sample of countries, for

example countries mostly exporting commodities, every country with a sufficient database is consid-

ered here. This knowledge is not limited to countries that export many raw commodities, but can be

extended to a broad country sample. In addition to the usual export price index, this is the first attempt

to investigate the significance of import prices and to better understand the interplay between these

two variables. Therefore, an import price index is also calculated in our work, which allows good

forecasts to be made even for countries without strong commodity exports. It has the same sign as the

1 See Bruno and Shin (2015), Gali and Monacelli (2005), Ghosh, Ostry, and Chamon (2016) or Ibhagui (2018) on this issue.

|

2547

BAUMGÄRTNER ANd KLOSE

export price index, which contradicts the theory but is in line with other empirical results. By dividing

the sample into two periods, we gain insight into whether the prediction power changes over time; that

is, our latter sample measures possible changes stemming from the global financial crisis.

The remainder of the paper proceeds as follows: Section 2 describes the economic channels how

commodity prices influence the exchange rate. It gives an overview of the main theoretical con-

cepts and impact channels. Section 3 gives a literature review. This also includes an overview of the

partly contradictory findings in the literature, which shows that the scientific debate is still ongoing.

Section 4 describes the data used in this study, while Section 5 presents our estimation and forecast-

ing approach. The results are shown in Section 6. We provide some robustness analysis in Section 7,

including different base currencies and expanding the model by other explanatory variables. Section

8 finally concludes.

2

|

ECONOMIC CHANNELS

According to economic theory, exchange rates and commodity prices are interrelated. We identify

four channels in which both variables are connected with each other. Most studies investigate the

effect of individual commodity prices, such as oil prices, on the exchange rate. The transfer of these

results to other commodities should, however, be possible without major restrictions because all of

them are traded in US dollars nowadays (Habib, Bützer, & Stracca, 2016).

The first channel is the expectations channel (Chen, Rogoff, & Rossi, 2010). While transaction

costs can cause arbitrage transactions not leading to a complete price compensation, investors should

rationally try to anticipate future market developments and adapt their behaviour accordingly. This has

an effect of commodity prices on the exchange rates. If an investor expects rising commodity prices,

he will invest more in commodity producing countries which export commodities, as interest rates

tend to rise in these countries. The induced capital inflows lead to an appreciation of the exchange rate

due to higher demand for domestic currency.

Second, the terms of trade channel describes the effect of a price increase of a traded good on

the economy in general and the exchange in specific. The channel works as follows: a rise in world

commodity prices increases the production costs of companies, which rely on imports of these com-

modities. This price increase leads to higher inflation rates as the producer passes it through into the

consumer prices. In this case, the general price level in a country which is dependent on commodities

rises more strongly than in a more self‐sufficient country whose industry is not affected by this price

increase. As a result, inflation rates in countries are different and the real exchange rate changes

(Beckmann, Czudaj, & Arora, 2017). On the one hand, companies depending on commodities lose

competitiveness. On the other hand, companies independent of commodities become more competi-

tive in the world market. Thus, in theory a price shock for commodities leads to a real devaluation of

the exchange rate for net importers.

The third important channel is the portfolio and wealth channel developed by Krugman (1980).

The portfolio channel has a short‐term effect. Starting point is again a rise in commodity prices. Thus,

the import costs for commodity scarce countries will initially increase, which deteriorates their current

account balance. At the same time, the investments from commodity‐exporting countries are increas-

ing due to higher profits. This increases the capital account balances of industrialised countries. The

degree of exchange rate variability thus depends on the investment behaviour of the commodity‐ex-

porting countries (Lane & Shambaugh, 2010). The dampening portfolio effect is usually not fully

converted into the domestic currency, but are mostly invested abroad by state‐controlled funds. For