EXCHANGE RATE PASS‐THROUGH INTO RETAIL PRICES

• Date: 01 November 2016
• Published date: 01 November 2016
• Author: Nicola Zaniboni,Alexis Antoniades
• DOI: http://doi.org/10.1111/iere.12203

INTERNATIONAL ECONOMIC REVIEW

Vol. 57, No. 4, November 2016

EXCHANGE RATE PASS-THROUGH INTO RETAIL PRICES∗

BYALEXIS ANTONIADES AND NICOLA ZANIBONI1

Georgetown University, U.S.A.; Transparent Value Advisors, U.S.A.

We study exchange rate pass-through and its determinants using scanner data on fast moving consumer goods sold

by 1,041 outlets in the United Arab Emirates between 2006 and 2010. The data are augmented with country of origin

information. Our main finding is that exchange rate pass-through varies more across retailers within regions than

across regions, and in particular that pass-through increases with retailer market share. We also find that exchange rate

pass-through is negatively correlated with both product quality and the elasticity of substitution of the product category

and positively correlated with the frequency of price adjustment.

1. INTRODUCTION

The extent to which prices respond to exchange rate movements and the underlying de-

terminants of those responses are key issues in international macroeconomics. This is not

only because of their direct implications for international transmission of shocks, but also for

the more general insights they can provide in the study of price-setting behavior (see, e.g.,

Gopinath and Itskhoki, 2011). In recent years, the increased availability of very detailed data

at the micro level has spurred important new work in the (already substantial) exchange rate

pass-through literature. Indeed, with the analysis of micro-level data, researchers have been

able to provide more reliable measurements of exchange rate pass-through into prices and

offer an improved understanding of its microeconomic determinants.2In particular, recent

work has decomposed pass-through along the supply chain (e.g., Hellerstein, 2008; Naka-

mura and Zerom, 2010) and has linked pricing to market to firm-level characteristics of ex-

porters (e.g., Berman et al., 2012; Chatterjee et al., 2013) to study the relevance of factors

such as variable markups, local costs, and heterogeneous quality in explaining incomplete

pass-through.

Variation in price setting behavior at the retailer’s level has however received very little atten-

tion.3Because retailers play a key role in the supply chain, exploring whether they significantly

contribute to variation in consumer prices can provide important insights on the transmission

mechanism of cost shocks.

Our work fills this gap by linking pass-through to retailer characteristics. Specifically,

we study exchange rate pass-through into retail prices using micro data on prices and

quantities for about 85% of the fast moving consumer goods (FMCGs) sold across hun-

∗Manuscript received July 2014; revised June 2015.

1This research was made possible by support of an NPRP grant from the Qatar National Research Fund. We thank

Robert Feenstra, John Romalis, Daniel Westbrook, Charles Engel, Kenneth West, Alan Blinder, Oleg Itskhoki, Ariel

Burstein, and seminars participants at Princeton University, Columbia University, Georgetown University, George

Washington University, University of Wisconsin–Madison, NYU-Abu Dhabi, the University of Cyprus, Bank of

Canada, and the University of Bern for helpful comments. We would also like to thank the editor and two anonymous

referees. All remaining errors are our own. The views expressed here are those of the author and not necessarily

those of Guggenheim Partners or its subsidiaries. Please address correspondence to: Alexis Antoniades, Georgetown

University, OD56, Education City, Doha, 23689, State of Qatar. E-mail: aa658@georgetown.edu.

2For a recent and thorough review of these developments in the literature, see Burstein and Gopinath (2014).

3An important exception is Nakamura (2008). The author considers a variance decomposition on prices of

100 products (UPCs) sold across 7,000 groceries in 50 U.S. states and finds that about 2/3 of the variation in prices is

driven by retailers’ chain-level effects.

1425

C

(2016) by the Economics Department of the University of Pennsylvania and the Osaka University Institute of Social

and Economic Research Association

1426 ANTONIADES AND ZANIBONI

dreds of grocery stores in the United Arab Emirates (UAE) between January of 2006 and

December of 2010. For each outlet and each period, we have access to transaction data at

the barcode level for 30 product categories, along with information on the outlet type, the

region it belongs to, and whether the outlet is part of a chain.4Because of this availability of

outlet-level information in our data, we are able to analyze variation in pass-through across

retailers.

Although this is a dimension of heterogeneity that is typically neglected in the literature,

our results suggest it matters: We find that exchange rate pass-through varies more across

retailers within regions than across regions, and in particular that pass-through increases with

retailer market share. We first estimate aggregate pass-through to be around 20% after one

year and find it to vary little across regions in the UAE.5We then estimate pass-through across

retailers within regions and find it is highest for supermarkets and lowest for mini-markets,

with groceries in between. To control for any differences in business models across types of

outlets, we repeat the analysis restricting the sample to supermarkets only. We estimate that

exchange rate pass-through for goods sold in supermarkets with high market share in a region

is in most instances significantly larger than for supermarkets with low market share within

that same region. The data suggest it is the size of the retailer itself that matters the most

for pass-through: We perform a similar exercise for market shares by country of origin and

brand and find there is less clear-cut evidence of a systematic relationship with exchange rate

pass-through.

Our data also allow us to analyze how pass-through varies with other characteristics of

goods. We look at its relationship with the elasticity of substitution within product categories,

which we estimate using the methodology in Broda and Weinstein (2006). We then study how

pass-through relates to product quality and follow Auer and Chaney (2009) using variation in

price within specific product category-weight-package type triplet as evidence of variation in

quality. Finally, we provide evidence of a relationship between pass-through and the frequency

of price adjustment, along the lines of Gopinath and Itskhoki (2010). These dimensions can

help shed light on the determinants of pass-through. Whereas product quality can be viewed

as a proxy for markups, which can be used to shield consumers from fluctuations in exchange

rates or shifts in costs, the elasticity is a proxy that captures the degree of product heterogeneity

of the available varieties within a product group. Product quality varies across products in a

specific group, but the elasticity of substitution is constant within a group. Finally, Gopinath and

Itskhoki (2010) argue that the observed variation of pass-through with the frequency of price

adjustment provides guidance on the construction of models of price setting behavior that are

more consistent with the data.

We show that exchange rate pass-through is negatively correlated with product quality and

elasticity of substitution. This is consistent with the findings in recent work, such as Atkeson

and Burstein (2008), Berman et al. (2012), and Auer and Chaney (2009), which focuses on the

behavior of exporters. We find that pass-through is positively related to products’ frequency of

price adjustment, as in Gopinath and Itskhoki (2010).

We view the evidence on the relationship between pass-through and retailer size as the main,

and novel, contribution of our article. We interpret this to strongly suggest that retailers exhibit

heterogeneity in price-setting behavior. We argue that such heterogeneity can be accounted for

in a framework where retailers differ in the size of their markups and/or the share of total costs

that they incur locally. In particular, if large retailers charge lower markups and/or have lower

local costs, then exchange rate pass-through will be positively related to retailer market size.

This is because the lower the local costs (Corsetti and Dedola, 2005) or markups (Hellerstein,

4Transactions data refer to the total units of a particular item (barcode) sold within a particular period in a specific

outlet and the price of the item in the particular store at the time (day) of the audit within that period. Outlet types

offer a characterization of outlets based on size and function, such as supermarkets, groceries, mini-markets, cafeterias,

and pharmacies.

5We discuss later in the article how our aggregate findings are in line with existing estimates in the literature.