The Home Bias in Equities and Distribution Costs

• Date: 01 July 2015
• Published date: 01 July 2015
• DOI: http://doi.org/10.1111/sjoe.12105
• Author: Mathias Hoffmann,Christina Ortseifer,Philipp Harms

Scand. J. of Economics 117(3), 983–1018, 2015

DOI: 10.1111/sjoe.12105

The Home Bias in Equities and

Distribution Costs∗

Philipp Harms†

Johannes Gutenberg University Mainz, DE-55128 Mainz, Germany

philipp.harms@uni-mainz.de

Mathias Hoffmann

Deutsche Bundesbank, DE-60431 Frankfurt am Main, Germany

mathias.hoffmann@bundesbank.de

Christina Ortseifer

Johannes Gutenberg University Mainz, DE-55128 Mainz, Germany

christina.ortseifer@uni-mainz.de

Abstract

We show that incorporating distribution costs into a general equilibrium model of interna-

tional portfolio choice helps to explain the home bias in international equity investment.

Our model is able to replicate observed investment positions for a wide range of parameter

values, even if agents have an incentive to hedge labor income risk by purchasing foreign

equity. This is because the existence of a retail sector affects both the correlation of domestic

returns with the domestic price level and the correlation between financial and non-financial

income.

Keywords: Home bias; international financial market integration; international risk sharing;

JEL classification:F41; G11; G15

I. Introduction

Since Obstfeld and Rogoff (2001) established their claim that trade costs

help to explain the “six major puzzles in open economy macroeconomics”,

a growing body of literature has scrutinized the relationship between fric-

tions in goods trade and the observed home bias in equity holdings.1In the

presence of such frictions, technology shocks lead to stronger fluctuations

in the national price level, and the home bias might reflect the desire of

agents to insure against “real exchange rate risk”: if the profits of domestic

†Also affiliated to the Study Center Gerzensee.

∗The views expressed in this paper are those of the authors and do not necessarily reflect the

views of the Bundesbank. We are indebted to three anonymous referees, whose comments

have helped to improve the paper significantly.

1Surveys of both the empirical and theoretical literature on the “home bias puzzle” are

provided by Lewis (1997), Obstfeld (2007), and Coeurdacier and Rey (2013). The latter

show that while the equity home bias in industrialized economies has decreased substantially

during the past two decades, agents in these countries still allocate an excessive share of

their wealth to domestic equity.

CThe editors of The Scandinavian Journal of Economics 2015.

984 The home bias in equities and distribution costs

firms are high whenever the domestic price level is high (i.e., the real ex-

change rate appreciates), then this is a compelling reason to increase the

share of domestic equities relative to the frictionless benchmark. Variations

in the real exchange rate might be due to a relatively large weight of do-

mestic goods in the consumption baskets of agents; that is, a home bias in

consumption (Kollmann, 2006), the presence of goods that are non-tradable

(Stockman and Dellas, 1989; Collard et al., 2009), or the existence of trad-

ing costs that drive a wedge between the domestic and foreign prices of

a good. In Obstfeld and Rogoff (2001), Obstfeld (2007), and Coeurdacier

(2009), these costs take the form of “iceberg costs” (i.e., a leak in the

bucket that transports goods from one country to another).

Both Coeurdacier and Obstfeld point towards two important challenges

that this literature faces. First, they illustrate that trade costs in goods mar-

kets cannot generate a home bias in equities under standard preference

assumptions. Secondly, the equilibrium equity portfolios they derive are

extremely sensitive to variations in the model parameters. In this paper,

we propose distribution costs as an alternative source of international price

differences and real exchange rate fluctuations. We argue that including

these costs in an otherwise standard model of international portfolio choice

goes a long way towards explaining observed equity investment patterns,

even if consumers have no particular preference for domestic traded goods,

and even if there is an incentive to hedge labor income risk by purchasing

foreign equities (Baxter and Jermann, 1997; Baxter et al., 1998). Hence,

although we place ourselves in an unfavorable starting position relative to

endowment-economy models, our framework is able to generate a diversi-

fication pattern that is close to real-world portfolio shares. Moreover, we

demonstrate that, in such a set-up, agents’ portfolio choices do not exhibit

the dramatic sensitivity to small parameter variations that characterizes

many other models of international risk sharing.

Our results are derived within a stochastic two-country general equilib-

rium model in which agents purchase shares of domestic and foreign firms

in order to maximize their expected utility. Agents consume domestic and

foreign traded goods as well as non-traded goods. We follow Burstein et al.

(2003) in assuming that every traded good has to be augmented by a certain

amount of (non-traded) retail services. While it has long been recognized

that such a constellation results in international price differences at the

retail and wholesale levels (Corsetti et al., 2008a) and that the existence of

distribution services raises the effective share of non-traded goods in GDP

(Collard et al., 2009), we argue that accounting for distribution costs adds

several features that are important for agents’ portfolio choices. First, the

existence of retail services increases the demand for non-traded goods at

given prices and reduces the effective elasticity of the demand for traded

goods. Both aspects enhance the importance of profits in f inancing agents’

CThe editors of The Scandinavian Journal of Economics 2015.

P. Harms, M. Hoffmann, and C. Ortseifer 985

consumption and thus magnify any shock that causes a variation in profits.

Moreover, the existence of distribution costs limits consumers’ scope for

expenditure switching and thus reinforces the positive relationship between

non-traded goods firms’ prices and their prof its. Conversely, it cushions the

impact of terms-of-trade fluctuations on traded goods firms’ prof its. Finally,

distribution costs reinforce the influence of non-traded goods prices on the

real exchange rate.

These features affect agents’ demand for domestic and foreign assets by

influencing the second moments that are crucial for their portfolio choices:

domestic equities are a good hedge against real exchange rate risk if the

covariance between domestic profits and the real exchange rate is negative.2

If the profits of domestic f irms are high whenever the domestic price level

is high, agents find it advantageous to invest a large share of their wealth

in domestic equity, because the loss in purchasing power is offset by an

increase in income. Conversely, agents have an incentive to hedge their non-

diversifiable labor income by purchasing foreign equities if domestic profits

are positively correlated with labor income. The existence of distribution

costs influences these covariances: we find that, compared to a model

without a retail sector, the (negative) covariance between domestic profits

and the real exchange rate (divided by the variance of domestic profit

incomes) is enhanced. Conversely, the (positive) covariance between profit

and labor income (divided by the variance of domestic profit incomes)

is reduced for a wide range of parameter values. Thus, distribution costs

affect agents’ portfolio choices, both by weakening the incentive to hedge

labor income risk and by reinforcing the incentive to hedge real exchange

rate risk.

As we will show, our model generates realistic portfolio choices for

a wide range of parameter values. Hence, even if we depart from our

benchmark parametrization, the resulting portfolio shares do not assume

implausibly high or low values. This surprising stability – which stands in

contrast to the findings of Obstfeld (2007) and Coeurdacier (2009) – is

due to the fact that distribution costs cushion the influence of parameter

variations on the correlation of prices and incomes. Thus, in contrast to

Coeurdacier and Gourinchas (2011), who demonstrate that an asset structure

consisting not only of equities but also of internationally traded bonds can

help to overcome the problems mentioned above, we show that even if we

stick to the traditional assumption that agents only trade equity, distribution

costs go a long way towards explaining the home bias in international equity

investment, and help to resolve the high sensitivity of equity portfolios to

preference parameters.

2We are using the price notation. An increasing real exchange rate thus corresponds to a

real depreciation.

CThe editors of The Scandinavian Journal of Economics 2015.