Consumer arbitrage in cross‐border e‐commerce

• DOI: http://doi.org/10.1111/roie.12424
• Author: José Anson,Matthias Helble,Mauro Boffa
• Published date: 01 September 2019
• Date: 01 September 2019

1234

|

wileyonlinelibrary.com/journal/roie Rev Int Econ. 2019;27:1234–1251.

© 2019 John Wiley & Sons Ltd

1

|

INTRODUCTION

The breakthrough of the Internet and the propagation of new means of communication have substan-

tially altered international trade patterns. Today’s private consumers can interact directly and in real

time with e‐retailers around the world. In the quest for the best deal, they arbiter between the prices of-

fered by businesses locally and those offered by online shops worldwide. As pointed out by Hortaçsu,

Martínez‐Jerez, and Douglas (2009), the emergence of e‐commerce has lowered search costs for con-

sumers. If information on prices and products is getting easier to access, we need to ask whether con-

sumers are taking advantage of international arbitrage opportunities and if yes to what extent.

The key determinant of price differences between domestic and foreign markets is the exchange

rate. The paper investigates how favorable nominal exchange rate changes stimulate e‐commerce ex-

ports to a given destination, in the short run. The hypothesis is put under scrutiny with the help

of a dynamic specification of a demand equation using Pesaran, Shin, and Smith’s (1999) vector

Received: 3 September 2018

|

Revised: 10 February 2019

|

Accepted: 23 April 2019

DOI: 10.1111/roie.12424

ORIGINAL ARTICLE

Consumer arbitrage in cross‐border e‐commerce

JoséAnson1

|

MauroBoffa1

|

MatthiasHelble2

1Research and Strategy Programme,UPU,

Universal Postal Union, Bern, Switzerland

2Economic Research and Regional

Cooperation Department,ADBI, Asian

Development Bank Institute, Tokyo, Japan

Correspondence

Mauro Boffa, UPU, Universal Postal

Union, Weltpoststrasse 4, 3014 Bern,

Switzerland.

Email: mauro_boffa@outlook.com

Abstract

In today’s internet markets consumers can search for, find

and compare prices worldwide. Online, information circu-

lates faster than offline and arbitrage opportunities such as

the ones arising from currency shocks are easily unveiled.

In this paper, we estimate for the first‐time exchange rate

elasticities for cross‐border e‐commerce transactions.

Exploiting a new high‐frequency database on international

transactions of parcels, we find that a 1% appreciation of the

domestic currency increases e‐commerce imports by 0.7%.

Comparing the result with traditional estimates in offline

markets, this implies a 50% exchange rate pass‐through

online.

JEL CLASSIFICATION

F 14; F 31

|

1235

ANSON et Al.

error‐correction (VEC) model for panel data. To abstract from supply shocks, we exploit a new high‐

frequency data on international postal parcels collected by the Universal Postal Union (UPU) and

combine it with weekly exchange rate data. While international parcel postal flows are highly cor-

related to traditional trade flows, they originate mainly from e‐commerce transactions. Therefore, they

constitute the perfect measurement unit for cross‐border e‐commerce trade.1

Our analysis unveils two important dynamics in online markets. First, arbitrage is indeed happen-

ing. However, online exchange rate elasticities appear to be lower than the ones of traditional offline

estimates. While the average exchange rate elasticity (Head & Mayer, 2014) is about 1.4%, we find an

expenditure switching effect that is half as important (0.7%). Second, and as a direct consequence, the

average bilateral exchange rate pass‐through into import prices is about 50%. The finding is similar to

the one of Gorodnichenko and Talavera (2017) using price data scrapped from online retailers. They

show that the online exchange rate pass‐through between Canada and the United States oscillates be-

tween 60% and 75%. The paper confirms that phenomenon but instead of using listing prices it focuses

on actual shipped quantities between several trading partners.

The debate on exchange rate pass‐through has mostly focused on the prevalence of producer cur-

rency pricing (PCP) vs. local currency pricing (LCP) (among others; Campa & Goldberg, 2005;

Burstein & Gopinath, 2014; Devereux, Dong, & Tomlin, 2017). Low exchange rate pass‐through has

often been attributed to the prevalence of LCP. This stands in contrast with our second finding; low

pass‐through may be observed even when there is little LCP. A result that is intrinsic to the nature

of our database primarily dealing with flows deriving from international e‐commerce transactions,

where PCP is more prevalent. Thus, the overall contribution of the paper is twofold. We test the hy-

pothesis of international consumer arbitrage from demand data at an unprecedented high frequency

and we provide evidence of low exchange rate pass‐through even in the absence of LCP.2

We face two main challenges while testing the hypothesis of international arbitrage. First, we need

to obtain reliable high‐frequency data on e‐commerce trade flows. As with arbitrage in financial mar-

kets, arbitrage in the goods market requires observing how consumers react to frequent price changes.

We solve the issue by exploiting a new database on international parcel exchanges collected by the

UPU. Second, monetary policy may well be endogenous to trade flows. To deal with that, the empir-

ical analysis focuses only on highly integrated markets where exchange rate policy can be considered

exogenous to trade.3

The findings in the paper are important for at least three reasons. First, exchange rate movements

have regained substantive attention during and in the aftermath of the Great Recession. As several

countries have applied expansive monetary policies, macroeconomists have been repeatedly con-

fronted with questions related to the role of monetary policy in restoring economic competitiveness,

often questioning the impact of alleged currency manipulations. An important challenge for empirical

research is the observation of short‐run trade aggregates. In the paper, we present a new database that

observes trade at an unprecedented high frequency and allows for tracking those changes in almost

real time.

Second, with online markets growing more and more important it is necessary to assess if exchange

rate changes have the same impact as they would have in the offline world. The two key elements af-

fecting exchange rate elasticities offline are price stickiness and pricing to market (Corsetti & Dedola,

2005).4

One may argue that tracking IP addresses may make discrimination easier in the online world.

Ellison and Ellison (2009) report remarkably high elasticities for some goods, especially whenever

a price search engine is present. The assumption of price stickiness in the context of exchange rate

movements has been tested empirically in several studies. For example Álvarez et al. (2006) mainly

focus on internet prices and show, for a subset of products in Germany, Italy, the United Kingdom,

France, and the United States that the median average price change spans from 25 to 68days. The