Trade Network Centrality and Currency Risk Premia

• Date: 01 June 2019
• Author: ROBERT J. RICHMOND
• DOI: http://doi.org/10.1111/jofi.12755
• Published date: 01 June 2019

THE JOURNAL OF FINANCE •VOL. LXXIV, NO. 3 •JUNE 2019

Trade Network Centrality and Currency Risk

Premia

ROBERT J. RICHMOND∗

ABSTRACT

I uncover an economic source of exposure to global risk that drives international asset

prices. Countries that are more central in the global trade network have lower interest

rates and currency risk premia. To explain these findings, I present a general equi-

librium model in which central countries’ consumption growth is more exposed to

global consumption growth shocks. This causes the currencies of central countries to

appreciate in bad times, resulting in lower interest rates and currency risk premia.

Empirically, central countries’ consumption growth covaries more with world con-

sumption growth, further validating the proposed mechanism.

BETWEEN 2001 AND 2016, CARRY TRADE INVESTORS that went long in currencies

with high average interest rates, by borrowing in currencies with low average

interest rates, obtained an annualized Sharpe ratio of 0.43. This Sharpe ratio

is similar to those found in U.S. equity markets and is surprising given the

strategy’s simple, unconditional nature. Although the returns to carry trade

strategies are well studied, less is known about their economic origins. In this

paper, I show that differences in interest rates that drive currency returns

are explained by countries’ trade network centrality, that is, their position in

the global trade network. By connecting returns to economic quantities, I shed

light on the fundamental origins of exposure to risk that drives international

asset prices.

To make the connection between returns and quantities, I begin with the

simple observation that countries share and are exposed to risk through trade

links. These trade links form a global trade network. Figure 1depicts the

global trade network in 2013. Each circle represents a country and each line

represents a trade link. The size of each circle corresponds to a country’s share of

∗Robert J. Richmond is with NYU Stern School of Business. The author has no relevant or mate-

rial financial interests that relate to the research described in this paper. I am incredibly grateful

to Hanno Lustig as well as Mike Chernov, Andrea Eisfeldt, Barney Hartman-Glaser, and Andy

Atkeson for invaluable feedback on this research. For helpful comments, I thank Daniel Andrei,

Ariel Burstein, Bruce Carlin, Ricardo Colacito, Xavier Gabaix, Lars Hansen, Tarek Hassan, Mah-

yar Kargar, Ralph Koijen, Rob Ready, Nick Roussanov,Jung Sakong, Yinan Su, Adrien Verdelhan,

Brian Waters, and seminar participants at Boston College, Carnegie Mellon, Chicago Booth, Uni-

versity of Colorado, Imperial College London, London Business School, MIT Sloan, Northwestern

Kellogg, NYU Stern, Stanford SITE, the Annual Conference on International Finance, UT Austin,

UCLA Anderson, University of Utah, and WFA.

DOI: 10.1111/jofi.12755

1315

1316 The Journal of Finance R

Figure 1. World trade network in 2013. This figure depicts country links as measured by

bilateral trade intensity—pairwise total trade normalized by pairwise total GDP.Links are drawn

if bilateral trade intensity is greater than the cross-sectional median. Circle position corresponds to

trade network centrality and circle size corresponds to GDP.Trade data are from the IMF Direction

of Trade Statistics and GDP data are from the World Bank, both in dollars. (Color figure can be

viewed at wileyonlinelibrary.com)

global gross domestic product (GDP). Trade links are measured using pairwise

total trade normalized by pairwise total GDP, and thus only the strongest half of

trade links are displayed. The position of each circle corresponds to a country’s

position in the trade network, or its trade network centrality. Countries are

central if they have many strong links to countries that are important for the

global output of tradable goods. Due to these trade linkages, central countries

turn out to be more exposed to global shocks than peripheral countries. Central

countries are not necessarily large. For example, global trade hubs such as

Singapore and Hong Kong are central but are not nearly as large as their key

Trade Network Centrality and Currency Risk Premia 1317

Figure 2. Risk premia and interest rate differentials versus centrality. This figure depicts

decade-long averages of annualized risk premia rx and annualized one-month interest rate dif-

ferences (measured using covered interest rate parity with forward spreads f−s)versustrade

network centrality for 39 countries. For real values, inflation expectations are lagged year-over-

year inflation as in Atkeson and Ohanian (2001). For each country, monthly observations are

averaged into three blocks (1988 to 1992, 1993 to 2002, and 2003 to 2016). Centrality is the export-

share weighted average of countries’ bilateral trade intensities—pairwise total trade divided by

pairwise total GDP. Trade data are from the IMF Direction of Trade Statistics and annual GDP

data are from the World Bank, both in dollars. For the euro area, I construct an aggregate with

all countries that adopted the euro, beginning in 1999. Foreign exchange data are monthly from

Barclays and Reuters.

trading partners such as the United States. In contrast, countries that trade

only a small amount with a few partners, such as New Zealand, are peripheral.

These cross-sectional differences in trade network centrality turn out to be a

significant determinant of countries’ unconditional interest rates and currency

risk premia.

Figure 2illustrates the relations between centrality vis-´

a-vis interest rates

and currency risk premia. To focus on unconditional variation, I plot 10-year

averages of interest rate differentials and risk premia for a U.S. investor ver-

sus 10-year averages of trade network centrality. Central countries, such as

Singapore, have low average interest rates and currency risk premia. In con-

strast, peripheral countries, such as New Zealand, have high average interest

rates and currency risk premia. In general, interest rates and currency risk

premia are decreasing in trade network centrality. These patterns hold for