Country Size, Currency Unions, and International Asset Returns

• Date: 01 December 2013
• Author: TAREK A. HASSAN
• DOI: http://doi.org/10.1111/jofi.12081
• Published date: 01 December 2013

THE JOURNAL OF FINANCE •VOL. LXVIII, NO. 6 •DECEMBER 2013

Country Size, Currency Unions, and

International Asset Returns

TAREK A. HASSAN∗

ABSTRACT

Differences in real interest rates across developed economies are puzzlingly large

and persistent. I propose a simple explanation: bonds issued in the currencies of

larger economies are expensive because they insure against shocks that affect a larger

fraction of the world economy.I show that, indeed, differences in the size of economies

explain a large fraction of the cross-sectional variation in currency returns. The data

also support additional implications of the model: the introduction of a currency union

lowers interest rates in participating countries, and stocks in the nontraded sector of

larger economies pay lower expected returns.

INTERNATIONAL INVESTORS EARN HIGH expected returns by borrowing in low in-

terest rate currencies and lending in high interest rate currencies. I propose

a simple explanation for this fact: bonds issued in the currencies of economies

that account for a larger share of world wealth are a better hedge against

consumption risk. As a consequence, larger economies have permanently lower

real and nominal interest rates than smaller economies, resulting in persistent

violations of uncovered interest parity (UIP).

Consider a world in which there are two countries, call them the United

States and Denmark (where the United States accounts for most of the world’s

wealth). Households in both countries consume a mixture of traded and non-

traded goods. In this world, bonds are country-specific consumption insurance:

the U.S. risk-free bond promises to deliver one unit of the U.S. consumption

bundle while the Danish risk-free bond promises to deliver one unit of the

∗Tarek A. Hassan is with University of Chicago, NBER, and CEPR. I thank Philippe

Aghion, John Y. Campbell, John Cochrane, Nicolas Coeurdacier, Emmanuel Farhi, Nicola Fuchs-

Sch¨

undeln, Pierre-Olivier Gourinchas, St´

ephane Guibaud, Elhanan Helpman, Thomas Mertens,

YvesNosbusch, Nathan Nunn, Helene Rey, Kenneth Rogoff, and Adrien Verdelhanfor helpful com-

ments. I also thank the workshop participants at Harvard Business School, Harvard University,

Massachusetts Institute of Technology Sloan, Berkeley Haas, Kellogg Graduate School of Manage-

ment, University of Chicago Booth, New York University Stern, Columbia Business School, Duke,

Northwestern, London Business School, London School of Economics, Brown, Boston University,

the World Bank, INSEAD, Institute for International Economic Studies Stockholm, the Federal

Reserve Bank of Boston, the Austrian Central Bank, Universitat Pompeu Fabra/Centre de Recerca

en Economia Internacional, University of Zurich/Institut fur Empirische Wirtschaftsforschung, the

Society for Economic Dynamics annual meeting, and the CEPR European Summer Symposia in

Financial Markets for valuable discussions. Special thanks also go to Doroth´

ee Rouzet and Simon

Rees.

DOI: 10.1111/jofi.12081

2269

2270 The Journal of Finance R

Danish consumption bundle. Intuitively, it is cheaper to provide this insurance

to Danish consumers. It takes a tiny fraction of the world’s supply of traded

goods to make up for any shortage of Danish nontraded goods. In contrast, a

relative shortage of U.S. nontraded goods automatically creates a worldwide

shortage of traded goods. The U.S. risk-free bond must therefore be more ex-

pensive than the Danish risk-free bond and the United States must have a

lower risk-free interest rate in equilibrium.

I show that differences in the size of economies do indeed explain a large frac-

tion of the cross-sectional variation in currency returns. Moreover,the data sup-

port a number of additional implications of the model: stocks in the nontraded

sector of larger economies pay lower expected returns and the introduction of

the euro lowered default-free interest rates and stock returns in the nontraded

sector of participating countries. Bonds denominated in currencies of larger

economies also appear to be a better hedge against long-run consumption risk

of U.S. investors.

Section Idevelops an international asset pricing model in which the correla-

tion structure of exchange rates and asset prices arises endogenously in general

equilibrium. The key insight from this model is that less of the country-specific

risk affecting large economies can be diversified internationally and that, as

a consequence, investors based in larger countries must have a more volatile

marginal utility of consumption than investors based in smaller countries.

I first show the main implications of this insight in a simplified model in

which asset markets are complete and the world is populated by a contin-

uum of households that receive stochastic endowments of both a traded and

a nontraded consumption good. All households within a given country receive

the same per capita endowments, where a large country accounts for a larger

share of world wealth (it has more households or richer households). House-

holds share all the risk from their traded endowment and some of the risk from

their nontraded endowment by shipping traded goods across countries. Assets

are priced with the marginal utility from traded goods, which is the same for

all households in equilibrium. Households demand a lower expected return on

assets that pay off well when this marginal utility is high, that is, when the

world endowments of traded and nontraded goods are low.

When a country has a relatively low per capita endowment in the nontraded

sector, its nontraded good becomes relatively more expensive and its real ex-

change rate appreciates. However, a low endowment in the nontraded sector

of a large country simultaneously triggers a large rise in the marginal utility

from traded goods, while a low endowment in the nontraded sector of a small

country does not. As a consequence, a larger country’s consumption bundle

tends to appreciate when marginal utility from traded goods is high. Any asset

that makes a payment that is partially fixed in terms of a larger country’s

consumption bundle is therefore a better hedge against consumption risk and

must pay lower expected returns. One such asset is a risk-free bond (think of an

inflation-indexed bond) that promises to pay one unit of a country’s consump-

tion bundle. Another is the ownership claim (stock) to the endowment in the

nontraded sector, which also increases in value when a country’s consumption

Country Size, Currency Unions, and International Asset Returns 2271

bundle appreciates. Larger countries should thus have lower risk-free interest

rates, and stocks in their nontraded sector should pay lower expected returns.1

I next analyze the full model in which markets are incomplete and the econ-

omy is affected by monetary shocks in addition to endowment shocks. Introduc-

ing money is important for two reasons. First, it provides us with a meaningful

way of thinking about currency areas that are distinct from countries. Second,

it breaks the tight link between exchange rates and aggregate consumption

growth, which is a counterfactual prediction of real models of exchange rate

determination.

In the full model, a subset of households in each country are precluded from

trading in financial markets and can only hold nominal bonds that make fixed

payments in terms of the currency used in the country in which they reside.

When inflation rises, these “inactive” households pay an inflation tax that

goes to the benefit of “active” households that consume proportionately more

whenever inflation is high.

Strikingly, this process again induces a positive correlation between the real

exchange rate of larger currency areas and the marginal utility from traded

goods of active households. It follows that all assets that make payments that

are partially fixed in terms of a larger currency area’s currency are better

hedges against consumption risk. These include risk-free and nominal bonds

as well as stocks in the nontraded sector. The introduction of a currency union

should thus lower risk-free and nominal interest rates as well as expected

returns on stocks in the nontraded sector within participating countries.

The theoretical part of the paper thus yields four testable predictions that

link differences in expected returns to differences in the size of economies and

currency areas: (1) real and nominal bonds issued in the currencies of larger

countries pay lower expected returns, (2) the introduction of a currency union

lowers expected returns on bonds within the union, (3) stocks in the nontraded

sector of larger countries pay lower expected returns than those of smaller

countries, and (4) the introduction of a currency union lowers expected returns

on stocks in the nontraded sector of participating countries.

Sections III and IV test these four predictions using a panel data set of devel-

oped economies over the 1984 to 2007 period. Figure 1shows the raw data. The

left-hand side of the figure plots the average forward premium of the curren-

cies of 27 Organisation for Economic Co-operation and Development (OECD)

countries against the U.S. dollar over a simple measure of the relative size of

their economies—the log of the share that each country contributes to total

OECD output. Each observation in the graph is an average over the years 1984

to 1990, 1990 to 1998, and 1998 to 2007. We see an economically large negative

correlation between this simple measure of country size and forward premia,

which implies that on average a country that contributes 10% of OECD output

(such as Germany) tends to have a 1.40 percentage points lower (annualized)

interest rate than a country that contributes only a negligible share of OECD

1These statements are subject to a slight restriction on the parameter space and to the assump-

tion that differences in the variances of countries’ endowments adhere to a regularity condition.