Network effect and international currency

• Published date: 01 July 2024
• Author: Tao Liu,Xiaosong Wang,Wing Thye Woo
• Date: 01 July 2024
• DOI: http://doi.org/10.1111/sjoe.12554

Scand. J. of Economics 126(3), 600–626, 2024

DOI: 10.1111/sjoe.12554

Network effect and international currency

Tao Liu

Central University of Finance and Economics, Beijing 102206, China

taoliu_cufe@outlook.com

Xiaosong Wang

Renmin University of China, Beijing 100872, China

wangxiaosong@ruc.edu.cn

Wing Thye Woo

University of Malaya, Kuala Lumpur, 50603, Malaysia

wtwoo@ucdavis.edu

Abstract

The network effect is an important factor for the inertia and path dependence in the international

monetary system. In this paper, we explore its inf‌luence, both theoretically and empirically:

(a) by extending the standard monetary search model to a three-country three-currency model

with multiple transitions to clarify the network effect on currency choices; (b) by showing

empirically that the network effect is positive, statistically signif‌icant, and quantitatively

comparable with other factors, such as trade share. The empirical results hold both for major

international currencies and emerging currencies, such as RMB, and for various estimation

specif‌ications and subsamples.

Keywords: Core–periphery structure; monetary search model; network effect

JEL classif‌ication:F33; G15; G36

1. Introduction

The evolution of the international monetary system (IMS) displays strong

inertia and path dependence. When settling cross-border transactions, people

have strong incentives to use the same currency as others to reduce searching

friction and transaction costs (Krugman, 1980). In history, although the US

GDP surpassed that of the UK in around the 1870s, the international role of the

US dollar (USD) was quite limited and the pound sterling (GBP) enjoyed an

incumbency advantage until the 1920s (Eichengreen, 2011). More recently,

China has become the second largest economy and the biggest exporting

country in 2010, but the USD’s global status remains dominant in international

trade (Gopinath et al., 2020) and foreign exchange transactions.1These facts

1According to the BIS Triennial Central bank survey of foreign exchange (FX) and

over-the-counter (OTC) derivatives markets in 2019, 88.3 percent of foreign exchange

c

2023 F¨

oreningen f¨

or utgivande av the SJE.

T. Liu, X. Wang, and W. T. Woo 601

conf‌irm the prevalence of inertia in the IMS, and one main reason for the

inertia is the network effect (Eichengreen and Flandreau, 2012; Eichengreen

et al., 2016).

The network effect in the IMS refers to the phenomenon that a currency’s

value increases in the number of its users (Chit¸u et al., 2014).2In this paper,

we focus on the medium-of-exchange function of an international currency,3

where the adoption of a common currency helps to improve matching

probability and to reduce transaction costs (e.g., Matsuyama et al., 1993).

Accordingly, people stick to the USD because they are more likely to f‌ind

a trading partner who is also using the USD in cross-border transactions.

There is no shortage of theoretical and empirical studies that explore the

network effect and its inf‌luence on currency choices (e.g., Flandreau and

Jobst, 2005; Devereux and Shi, 2013; Eichengreen et al., 2016).

However, constrained by modeling techniques and data availability, most

of these works fail to measure explicitly the network effect, and thus are

unable to address a series of important issues. Specif‌ically, how can the

network effect be evaluated in the global f‌inancial market? What is the

structure of the IMS from the perspective of network analysis? What is the

network effect’s magnitude, especially in comparison with other determinants

of currency choices? Does the network effect exist only in major international

currencies, or does it also exist in emerging currencies such as the renminbi

(RMB)?

In this paper, we explore these issues by expanding the theoretical

framework, and by using detailed currency information in cross-border

transactions. Our three main f‌indings are the following.

(i) The extension of a two-country two-currency monetary search model to

a three-country three-currency model (with multiple transitions) shows

that agents are willing to hold more of a certain currency if that brings

a higher level of matching probability (which is proportional to the

pagerank centrality in network analysis).

transactions are related to the USD. Because bilateral FX transactions involve two currencies,

the share sums up to 200 percent.

2In this paper, we use “network effect”and “network externality” interchangeably, although there

are considerable differences between them. Strictly speaking, network effect is not equivalent

to network externality unless the participants in the market fail to internalize these effects

(Liebowitz and Margolis, 1998).

3Admittedly, there are many channels that could lead to the network effect, and the search/match

approach is only one of them. The existence of the network effect could be due to transaction

costs (Rey, 2001; Devereux and Shi, 2013), to convention and custom, to the coalescing effect

in pricing export goods (Goldberg and Tille, 2008), or to the positive feedback between various

functions of international currencies (Gopinath and Stein, 2021; Chahrour and Valchev, 2022).

c

2023 F¨

oreningen f¨

or utgivande av the SJE.