Home bias in open economy financial macroeconomics

TL;DR: In this paper, the authors focus on three broad classes of explanations: hedging motives in frictionless financial markets, asset trade costs in international financial markets (such as transaction costs, differences in tax treatments between national and foreign assets), and informational frictions and behavioural biases.

Abstract: Standard theory would predict that investors hold a well diversified portfolio of equities across the world but despite the process of ‘financial globalization’, investors still hold a disproportionate share of local equities in their portfolio: the ‘equity home bias’ (French and Poterba (1991)). We review the various explanations of this puzzling phenomenon in the context of recent developments in macroeconomic modelling that have allowed to incorporate sophisticated international portfolio choices in standard two-country general equilibrium models. We refer to this literature as Open Economy Financial Macroeconomics. We focus on three broad classes of explanations : (i) hedging motives in frictionless financial markets (real exchange rate and non-tradable income risk), (ii) asset trade costs in international financial markets (such as transaction costs, differences in tax treatments between national and foreign assets...), (iii) informational frictions and behavioural biases. These recent developments raise the need for new portfolio facts beyond the equity home bias and we will present some new evidence on cross-border asset holdings across different types of assets: equities, bonds and bank lending. We also present some new micro data on institutional holdings of equity at the fund level. These data should inform macroeconomic modelling of the open economy and a growing literature of models of delegated investment.

Summary

1 Introduction

• Standard finance theory predicts that investors hold a diversified portfolio of equities across the world if capital is fully mobile across borders.
• The process of 'financial globalization' fostered by capital account liberalizations, electronic trading, increasing exchanges of information across borders and falling transaction costs has certainly led to a large increase in cross-border asset trade (Lane and Milesi-Feretti (2003) ).
• Explaining equity home bias has been one of the main motivations for this literature, which has first focused on models with equities only.
• The authors use standard dynamic models of the Open Economy Financial Macroeconomics literature.

2.1 Definition

• French and Poterba (1991) were the first to their knowledge to document domestic ownership shares across countries.
• Using data for the US, Japan, UK, France and Germany, they show that investors hold a disproportionate share of domestic assets in their equity portfolios.
• They label this lack of cross border diversification the equity home bias.
• This is a well-known puzzle in international finance: in a world with frictionless financial markets, the most basic International CAPM model with homogenous investors across the world would predict that the representative investor of a given country should hold the world market portfolio.

2.2 Evidence across time and across countries

• While one could argue that, at the end of the eighties, international capital markets were far from being frictionless and this could contribute to rationalize home bias, this line of explanation seems more doubtful today.
• In figure (1) , the authors show the evolution of home bias measures in developed countries across regions of the world: 4 it has decreased over the last twenty years with the process of 'financial globalization' but remains high in most countries (see also table (1) for a recent snapshot of home bias measures for selected countries).
• For the developed world, this means that the share of foreign equities in investors portfolios is roughly a 1/3 of what it should be if the benchmark is the basic International CAPM.
• This robust stylized fact has received considerable attention from both the finance literature and the macroeconomics literature.
• Papers differ in the extent to which they use real or nominal returns, how they estimate expected returns and how they deal with structural breaks and nonstationarity.

3 Open Economy Financial Macroeconomics

• The theoretical macroeconomic literature points towards potential gains from international diversification to hedge national production risk.
• In the presence of imperfectly correlated productivity shocks or output shocks across countries, owning foreign equity could help to smooth consumption.
• In reality, heterogeneity across investors from different countries leads to departure from the world market portfolio and potentially a bias towards national assets.
• Various sources of heterogeneity leading to equity home bias have been explored in the macro literature.
• They fall in two broad classes of explanations : (i) hedging motives (real exchange rate risk and non-tradable income risk), (ii) asset trade costs in international financial markets (such as transaction costs, differences in tax treatments, in legal framework and other policy induced barriers to foreign investment).

3.1 Methodological Breakthrough

• Until recently, most macroeconomic models of the international economy relied mostly on the following asset structures: either one non-contingent bond traded internationally or complete asset markets through Arrow-Debreu securities.
• None of these models could say anything about gross foreign asset holdings and the extent to which tradable assets could be used to share risks internationally.
• Secondly, they show that in order for the steady state portfolio to be well defined, a second order approximation of the portfolio equations (Euler equations) needs to be considered, while only the first order dynamics of the other equations of the model are required to pin down steady state portfolios (also called zero order portfolio).
• In addition to these conceptual insights, Devereux and Sutherland (2008a) also provides a formula which can be used to compute portfolios analytically in a fairly general class of models.
• In simultaneous work, Tille and van Wincoop (2008) develop a solution technique that is analogous to the one presented in Devereux and Sutherland (2008a) .

3.2 Shortcomings and extensions

• The main advantage of the perturbation methods developed by Devereux and Sutherland (2008a) and Barro (2006) ) or when the problem is non-stationary.
• Since the methods are mainly based on first and second order approximations, they may also be inaccurate in models that exhibit strong non-linearities, such as models with borrowing constraints.
• Lastly, the approximation of the decision rules in these methods is made around the deterministic steady-state.
• The deterministic steady-state might not be the stationary steady-state of the model in presence of risk.
• Exceptions of two-country/two goods models where solutions can be found without approximations are models with log-linear preferences as in Pavlova and Rigobon (2007, 2010a,b) .

4 Hedging motives in Open Economy Financial Macroeconomics

• Hedging motives lead to departure from the benchmark model of Lucas (1982) where homogeneous investors across the world hold identical portfolios.
• Investors receive a part of their income (wages in particular) that cannot be traded in financial markets, also known as -Non-tradable income risk.
• As Pavlova and Rigobon (2007, 2010ab), Devereux and Saito (2005) use a continuous time framework which allows some analytical solutions to be derived, but it can only be applied to a restricted class of models.
• In order to analyze how these hedging motives affect equity portfolios, the authors present a benchmark twocountry/two good model where the only traded assets are equities of both countries.
• The authors show how loglinearization techniques can be used to derive (zero-order) steady-state portfolios.

4.1.1 Set-up and First Order Conditions

• There are two symmetric countries, Home (H) and Foreign (F ), each with a representative household.
• The authors assume that capital is fixed for now and will allow for endogenous capital accumulation in the subsequent section (4.2).

Preferences

• In the deterministic steady state, a is the share of consumption spending devoted to the local good.
• The welfare based consumer price index that corresponds to these preferences is: EQUATION where p i,t is the price of good i.

Technologies and firms' decisions

• There is a firm in country i that hires local labor and produces output, using technology (5) .
• Thus, the country i wage incomes are: EQUATION where w i,t is the country i wage rate.

Household decisions and market clearing conditions

• Each household selects portfolios, consumptions and labor supplies that maximize her life-time utility (2) subject to her budget constraint (8) for t ≥ 0.
• The following equations are first-order conditions of that decision problem: EQUATION (9) represents the optimal allocation of consumption spending across goods, and the labor supply decision.
• (10) shows the Euler equations with respect to the two stocks.
• The authors show that the asset structure (two assets with two exogenous shocks) is "locally-complete" in the sense that up to a first order linear approximation, the consumption allocation is efficient (in other words there is perfect risk sharing up to a first-order approximation of the model).
• The method the authors use to solve for portfolios is then slightly different from Devereux and Sutherland (2008a) as it does not require a second-order expansion of the Euler equations (equation (( 10)).

Log-linearization of the model

• In an equilibrium with "locally-complete" markets, the ratio of Home and Foreign marginal utilities of aggregate consumption is proportional to the consumption-based real exchange rate (Backus and Smith (1993) , Kollmann (1995) ).
• Linearization of this risk sharing condition gives: EQUATION.
• Thus Home terms of trade worsen when the relative supply of Home goods increases as Foreign goods are scarcer.
• The authors will show that there exists a unique portfolio S, which, for consumptions consistent with the linearized risk sharing condition (14) , satisfies the following 'static' budget constraint: EQUATION.
• This expression shows the changes in country H income (relative to the income of F ) necessary to finance the changes in consumption consistent with efficient risk-sharing (up to first order).

Partial equilibrium zero-order portfolios

• The 'static' budget constraint is useful to derive the equilibrium portfolio as a function of variance/covariance ratios.
• This expression holds in many classes of models (with equity only) as the authors only need the budget constraints and generic first order conditions to derive it.
• The portfolio departs from the fully diversified one with weights 1/2 in both equities (as in Lucas (1982) ) in presence of labor income risk and/or real exchange rate risk.
• If local equities have higher returns (than abroad) when local returns to non-tradable wealth are lower (than abroad), households will bias their portfolio towards local equities.
• The optimal hedging of real exchange rate risk depends on two forces going in opposite direction: when local goods are more expensive, consumers need to generate more income in order to stabilize their purchasing power.

General equilibrium zero-order portfolios

• In general equilibrium macro models, the above variance/covariance terms can be expressed as a function of the underlying parameters of the model.
• The authors derive the Lucas portfolio when α → 1 (no human capital risk) and when a = 1/2 (no real exchange rate risk).
• Home investors exhibit Home equity bias as Home equity have higher returns when the Home real exchange rate appreciates.
• In the mean time, dividends net of investment spending are falling.
• The equity portfolio is the same as in Heathcote and Perri (2008) 29 but holds for all values of the preference parameters.

Hedging real exchange rate risk

• As appears clearly in the previous model, optimal portfolios are structured to hedge the risk arising from real exchange rate fluctuations.
• The above model (in line with Coeurdacier (2009)) 15 shows the opposite result for most parameter values (in particular for φ and σ above unity).
• With utility separable in tradable and nontradable consumption, optimal portfolios imply that domestic agents hold all of the equity of domestic nontradable firms.
• Thus, this model generates home bias in equity positions, and the home bias increases in the share of nontradable consumption in total output.
• The hedging of real exchange rate risk leads to equity home bias if local equities have higher returns (than abroad) when local prices are higher (than abroad).

Hedging non-tradable income risk

• In their model (see equation ( 18)), hedging non-tradable income risk implies picking stocks which have higher payoffs when labour income is low.
• From an empirical perspective, a series of papers have provided estimates of the covariance between relative equity returns and relative returns to human wealth which is the key empirical counterpart of portfolio biases in this class of models.
• (ii) an increase in Home investment (more than abroad) as Home investment uses more intensively cheaper Home goods (due to home bias in investment spending).
• The first generation of papers presented above focus on equity positions to rationalize home bias.
• Equities are used to hedge sources of risks that cannot be hedged through the bond positions, in particular the part of non-tradable income risk that is orthogonal to bond returns.

4.2.2 Set-up of the Model

• The authors use a similar set up as in section (4.1) but they add two important ingredients to formalize their above discussion on hedging motives: endogenous capital accumulation and trade in real bonds.
• They allow us to overcome the limitations of the model presented in (4.1): first, endogenous investment in a two-good model breaks the perfect link between returns on physical capital and returns on human capital; second, bond trading modifies the hedging properties of equities.
• Bonds will be used to hedge fluctuations in the real exchange rate.
• This model is similar Coeurdacier, Kollmann and Martin (2010) which extends Heathcote and Perri (2008) to multiple asset classes (bonds and equities).
• In presence of productivity shocks only, the authors would face a portfolio indeterminacy in a first-order approximation of the non-portfolio equations since the number of available assets (bonds and equities in each country) would exceed the exogenous sources of uncertainty.

Technologies and capital accumulation

• As before, production in each country uses capital and labor with a Cobb-Douglas production function: EQUATION (2008), Devereux and Sutherland (2007 and 2008b) .
• 24 Obviously, a different shock will have different business cycles implications but the authors are presently not interested in those.
• The associated investment price index is the same as for consumption P i,t (see equation ( 4)).

Firms' decisions

• This implies the following first-order condition: 26 EQUATION.
• This leads to the following intratemporal allocation for investment goods: EQUATION.

Financial markets and instantaneous budget constraint:

• There is now international trade in stocks and (real) bonds.
• There is a bond denominated in the Home good, and a bond denominated in the Foreign good.
• At date t, the country i household now faces the following budget constraint: EQUATION.

4.2.4 Empirical evidence on the hedging of non-tradable income risk

• In order to show the relevance of conditioning for bond returns, the authors now present some empirical evidence on the hedging of non-tradable risk (see Coeurdacier, Kollmann and Martin (2010) for similar evidence 30 ).
• The authors compute bond payments differentials q using the (trade-weighted) real exchange rate of one country with respect to the other six.
• Conditionally (lines (2) and ( 4), wages and dividends comove negatively for all countries, which lead to a home equity bias in their equity-bond model.
• Hence, returns on domestic nominal bonds and local taxes comove positively and the household prefers to hold local nominal bonds.

CA

• In the mean time, as government expenditures are biased towards local goods, the relative price of locally produced goods increase and so does the pay-off of the claim to local output (local equity).
• Hence, returns on local equity and taxes comove positively and households will optimally bias their portfolio towards local equity.

4.3.1 Other assets?

• The recent literature has shown the importance of extending existing models to a larger menu of assets.
• There is no simple answer to this question: potentially any asset that is traded publicly could affect the equity portfolio if it has some hedging properties in addition to what bonds and equities can achieve.
• The authors have ignored debt as a way to raise capital for firms and have focused on firms that are fully financed through equity.
• They show that in an environment where the financial structure is irrelevant for the value of the firm (when the Modigliani-Miller theorem applies), the presence of corporate debt has no impact on investment decisions and the equilibrium consumption allocation (up to the first order) since markets are complete.
• When firms are partially financed through debt, holding the same fraction of debt and equity guarantees that investors have their optimal exposure to the value of the firm.

4.3.2 Exchange Rates and Asset Prices

• While the models described above can claim some success in replicating some features of aggregate portfolio data, they cannot replicate realistic moments of asset prices and exchange rates (see Lewis (2011) for a recent survey on international asset pricing puzzles).
• This does not mean that the mechanisms highlighted would not survive in more general environments able to generate realistic asset prices but this remains to be seen.
• First attempts in that direction include Stathopoulos (2008 Stathopoulos ( ,2009) ) who introduces habit formation and Benigno and Nistico (2009) who model ambiguity aversion.
• The authors believe that explaining asset prices moments should help disentangling across the different potential channels generating portfolio biases.
• With standard CRRA preferences, this leads to a perfect correlation between real exchange rate changes and relative consumption growth (Home relative consumption falls when Home relative prices are higher; see equation ( 14) in the previous models).

4.3.3 Dispersion of home bias across time, countries and assets

• Data on home bias exhibit substantial variations across time and across countries and (2) and Table (1 in line with model predictions, they match the degree of equity bias to the degree of trade openness of countries).
• Since more data on aggregate foreign asset holdings are now available, both in the time series and in the cross-section, it seems natural to extend theories to heterogeneous countries.
• This would provide more accurate tests of the different theories available.
• Exploiting the bilateral dimension of the data using a multi-country framework could also help in that matter.
• While, on average, the advanced countries are (in net terms) borrowing in foreign-currency, some major countries have very large negative domestic-currency debt positions (most notably the US).

5 Asset trade costs in international financial markets

• Another strand of the macroeconomics literature considers frictions in financial markets as the main source of heterogeneity across investors.
• Portfolio home bias is the natural outcome of these frictions.

5.1 Transaction costs would need to be very large to explain equity home bias...

• French and Poterba (1991) initially argue in a mean-variance framework that these costs must be much larger than the one typically observed if one want to rationalize equity home bias.
• Using stock returns data from 1975 to 1989 for the US, Japan, UK, France, Germany, Canada, the authors use estimates of a covariance matrix of returns together with an optimal portfolio rule that is implied by constant relative risk aversion in order to back out the differences in expected returns needed to explain actual portfolio shares for these countries.
• The implicit excess return on domestic equity implied by observed portfolio holdings is then interpreted as a measure of the cost of international asset trading needed to generate the observed home bias: they find an order of magnitude for these costs of several hundred basis points, too big to be true!.
• Numerous subsequent studies have provided such indirect estimates of the costs.
• These costs are very large ranging from Another piece of evidence pointing that transaction costs cannot rationalize portfolio holdings is Tesar and Werner (1995): transaction costs based explanation of the equity home bias should in general imply that turnover should be lower for foreign equity holdings than for domestic ones (unless they apply only to dividend repatriation).

5.2 ...Unless diversification benefits are very small

• As stated above, transaction costs are often assumed to be small although direct measures of these costs do not often exist (see below).
• As shown by Martin and Rey (2004) and Coeurdacier and Guibaud (2009) , even small transaction costs may lead to sizable home bias when Home and Foreign stocks are close substitutes: any small transaction cost is amplified if the benefits of diversification provided by foreign assets are small.
• Cole and Obstfeld (1991) show that the equivalence between portfolio autarky and complete markets also obtains in a setting with investment under the following assumptions: i) unitary elasticity of substitution between the two goods, ii) unitary elasticity of intertemporal substitution, iii) full depreciation.
• They find small welfare gains for a broad range of values of the elasticity of substitution between the two goods.
• Since the seminal paper of Cole and Obstfeld (1991) , a large number of papers using consumption data have computed welfare gains from international risk sharing with quite a lot of variation across studies.

5.3 Direct measures of the costs?

• It is important to note that most existing work provides indirect measures of transaction costs using stock returns data and observed portfolio allocation (as in French and Poterba (1991) ).
• When considering the rules of international taxation, one could expect some large impact on home portfolio biases.
• First, dividends when repatriated are subject to nonnegligible withholding taxes (of a magnitude of roughly 10% in developed countries).
• To avoid this double-taxation, shareholders receive a tax rebate.

6.1.1 Exogenous information sets

• The impact of informational asymmetries on portfolio decisions has been first studied in the finance literature.
• Wincoop (2009) applies the noisy rational expectations framework from the finance literature to a standard two country/one good DSGE model.
• These transaction costs generate equity home bias in equilibrium but this is not the purpose of the paper.

6.1.2 Endogenous information acquisition

• The early noisy rational expectations literature when applied to international portfolio choice relies on exogenous information structures.
• Agents are endowed with a small informational advantage on the local asset, which lowers its perceived riskiness.
• Van Nieuwerburg and Veldkamp (2010) apply variations of their rational inattention model to explain investment strategies of investors, varying the specification of the preferences or of the information constraint that they face.
• Such a learning strategy makes sense as the structure of the signal the agents choose in equilibrium depends on their objective function.
• The authors show that persistence of asset pay-offs and increases in information processing capacity tends to magnify home bias.

6.1.3 Empirical evidence on informational frictions

• A number of papers regress portfolio holdings or measures of home bias directly on factors that proxy for information asymmetries.
• Coval and Moskowitz (1999) find that U.S. mutual fund managers prefer to invest in nearby firms even within a country.
• They find that the impact of distance is drastically reduced once the authors control for bilateral goods trade: countries portfolios are strongly biased towards trading partners (see also Lane and Milesi Ferretti (2008) ).
• The empirical evidence on this matter is mixed.
• 35 For a very nice exposition of the applications of rational inattention to invesment choice see Velkamp (2001).

6.2 Behavioural biases

• Some recent papers have put forward a behavioral explanation for the equity home bias.
• In the same vein, Shiller et al. (1991) document large differences in expected returns of Japanese and U.S. investors for the same stockmarkets.
• They show how home bias can emerge when investor feels incompetent in understanding the benefits and the risks of investing in foreign assets.
• It remains difficult to disentangle empirically informational frictions linked to distance and/or institutional differences from behavioural biases such as 'familiarity' and/or 'competence' effects.

7.1 Aggregate Data on Portfolio Holdings

• The Open Economy Financial Macroeconomics models have implications for international bond holdings as well as international equity holdings.
• Hence, the authors provide some measure of the extent of risk-sharing through international bonds holdings (public and private) for a large cross section of countries by using data on cross-border bond holdings.
• For completeness the authors also present some data on international bank lending, as they expect that the Open Economy Financial Macroeconomics will soon incorporate formally bank intermediation in their models.
• For these data, the authors rely on CPIS and IFS data from the IMF and the data from the BIS .

7.1.1 Cross-Border Bond Holdings

• The authors use data form IFS and BIS to compute international bond holdings for selected countries.
• Unfortunately, available data do not allow to disaggregate data of foreign bond holdings across types of bonds (corporate versus public, across maturities, across currency denomination) 36 and the authors had to focus on bond holdings aggregates.
• Portfolios exhibit a home bias in bond holdings of a slightly larger magnitude than the one documented for equity in section (2).
• In figure (4), the authors show the degree of bond home bias for emerging markets: emerging markets have even much less diversified bond portfolios than developed countries and it has barely decreased over the last decade.

7.1.2 Cross-Border Bank Loans

• The authors use data from the BIS and the OECD to compute cross-border banking assets for selected countries.
• Like for bonds, available data do not allow to disaggregate data of foreign bank loans across types of loans and the authors had to focus on aggregate foreign asset holdings by banks.
• To measure the degree of international diversification of banks' portfolios, the authors compute a measure of home bias in bank loans comparable to the one used for equities and bonds.
• Figure (6) shows the same statistic but using BIS statistics .
• The magnitude of the home bias in banking assets is similar to the one observed for equity holdings.

7.2 Institutional Investors Home Bias Data

• An increasing share of capital flows are intermediated through institutional investors.
• A limitation of the data is that they do not include any information on cash holdings, financial leverage, investments in fixed income instruments or investments in derivative contracts.
• Table 2 gives the breakdown of the average number of funds also by country of fund origin.
• As mentioned in Hau and Rey (2008), there is a positive correlation between the number of sectors and the number of countries funds invest in, suggesting that more diversified funds diversify both across sectors of activity and across countries.
• Explaining home bias or indeed investment strategies at the fund level thus probably requires a theory of fund mandates.

8 Conclusions and leads for future research

• The authors view is that the home bias puzzle is now less of a puzzle.
• The Open Economy Financial Macroeconomics literature has an interesting set of predictions on the holdings of a broad menu of financial assets.
• If the authors want to take a more detailed microeconomic view of the home bias, they have to recognize the large heterogeneity of investment strategies both at the household level and at the fund manager level.
• Among families that held equity, either directly or indirectly, in 2007, ownership through a tax-deferred retirement account was most common (84% of families), followed by direct holdings of stocks (35% of families), direct holdings of pooled investment funds (21% of families), and managed investment accounts (8% of families).
• It is likely that the literature on endogenous information acquisition and delegated investment still misses an important component of the incentives of fund managers, which could be marketing and product differentiation.

Figures

Figure 4: Measures of Home Bias in Bonds across emerging countries (the country measure BHBi is Market Capitalization-weighted for each region; source: BIS and IFS. See appendix for the list of countries included)

Figure 3: Measures of Home Bias in Bonds across developed countries (the country measure BHBi is Market Capitalization-weighted for each region; source: BIS and IFS. See appendix for the list of countries included)

Figure 1: Home Bias in Equities measures across developed countries (the country measure EHBi is Market Capitalization-weighted for each region; source: IFS and FIBV. See appendix for the list of countries included)

Figure 2: Home Bias in Equities measures across emerging countries (the country measure EHBi is Market Capitalization-weighted for each region; source: IFS and FIBV. See appendix for the list of countries included)

Figure 5: Measures of Home Bias in Banking assets across OECD countries (in each region, the country measure LHBi is weighted by the share of oustanding loans of the country in the region; source: OECD. See appendix for the list of countries included)

Figure 6: Measures of Home Bias in Banking assets across Regions (in each region, the country measure LHBi is weighted by the share of oustanding loans of the country in the region; source: BIS. See appendix for the list of countries included)

Full text

NBER WORKING PAPER SERIES
HOME BIAS IN OPEN ECONOMY FINANCIAL MACROECONOMICS
Nicolas Coeurdacier
Hélène Rey
Working Paper 17691
http://www.nber.org/papers/w17691
NATIONAL BUREAU OF ECONOMIC RESEARCH
1050 Massachusetts Avenue
Cambridge, MA 02138
December 2011
Ebrahim Rahbari, Nelson Costa-Neto and Francesca Monti provided excellent research assistance.
Coeurdacier thanks the Chaire d’Excellence de l’Agence Nationale pour la Recherche for financial
support. Rey gratefully acknowledges the European Research Council grant number 210584 on ³Countries’
external balance sheets, dynamics of international adjustment and capital flows” for financial support.
The views expressed herein are those of the authors and do not necessarily reflect the views of the
National Bureau of Economic Research.
NBER working papers are circulated for discussion and comment purposes. They have not been peer-
reviewed or been subject to the review by the NBER Board of Directors that accompanies official
NBER publications.
© 2011 by Nicolas Coeurdacier and Hélène Rey. All rights reserved. Short sections of text, not to
exceed two paragraphs, may be quoted without explicit permission provided that full credit, including
© notice, is given to the source.

Home Bias in Open Economy Financial Macroeconomics
Nicolas Coeurdacier and Hélène Rey
NBER Working Paper No. 17691
December 20115HYLVHG-XO\
JEL No. F21,F3,F32,F4,F41,G11
ABSTRACT
Home bias is a perennial feature of international capital markets. We review various explanations
ofthis puzzling phenomenon highlighting recent developments in macroeconomic modelling that
incorporateinternational portfolio choices in standard two-country general equilibrium models. We
refer to thisnew literature as Open Economy Financial Macroeconomics. We focus on three broad
classes of explanations:(i) hedging motives in frictionless financial markets (real exchange rate and
non-tradable income risk), (ii) asset trade costs in international financial markets (such as transaction
costs or differences in taxtreatments between national and foreign assets), (iii) informational frictions
and behavioural biases.Recent theories call for new portfolio facts beyond equity home bias. We
present new evidence oncrossborder asset holdings across different types of assets: equities, bonds
and bank lending and newmicro data on institutional holdings of equity at the fund level. These data
should inform macroeconomicmodelling of the open economy and a growing literature of models
of delegated investment.
Nicolas Coeurdacier
SciencesPo
Department of Economics
28 rue des Saint Pères 75006 Paris, France
ncoeurdacier@gmail.com
Hélène Rey
London Business School
Regents Park
London NW1 4SA
UNITED KINGDOM
and NBER
hrey@london.edu

1 Introduction
Home bias is a perennial feature of international capital markets. Nineteenth century economists called it
“the disinclination of capital to migrate”
1
. Standard finance theory predicts that investors hold a diversified
portfolio of equities across the world if capital is fully mobile across borders. Because foreign equities provide
great diversification opportunities, a point made early on in Grubel (1968), Levy and Sarnat (1970)
and Solnik (1974), falling barriers to international trade in financial assets over the last thirty-years should
have led investors across the world to re-balance their portfolio away from national assets towards foreign
assets. The process of ‘financial globalization’ fostered by capital account liberalizations, electronic trading,
increasing exchanges of information across borders and falling transaction costs has certainly led to a large
increase in cross-border asset trade (Lane and Milesi-Feretti (2003)). However, investors seem still
reluctant to reap the full benefits of international diversification and hold a disproportionate share of local
equities, a phenomenon referred to as the ‘home bias in equities’. Since the seminal paper of French and
Poterba (1991), the home bias in equities has continuously intrigued and fascinated financial economists
and international macroeconomists. Despite better financial integration, the home bias has not decreased
sizably: in 2007, US investors still hold more than 80 percent of domestic equities, a much higher proportion
than the share of US equities in the world market portfolio. Indeed, home bias in equities is still observed
in most countries and tends to be higher in emerging markets.
Many explanations have been put forward in the literature to explain this very robust portfolio fact. We
do not intend to provide a definite answer nor choose among alternative explanations, as they probably all
contribute to part of the gap. Our goal is to review where theory has led us, provide relevant empirical facts
and take a stand at what might be the next challenges ahead. We distinguish between three broad classes
of explanations : (i) hedging motives in frictionless financial markets (real exchange rate and non-tradable
income risk), (ii) asset trade costs in international financial markets (such as transaction costs, differences
in tax treatments between national and foreign assets or differences in legal frameworks) and (iii) infor-
mational frictions and behavioural biases. We will review these explanations highlighting important recent
developments in macroeconomic modelling of the open economy, referred to as Open Economy Financial
Macroeconomics. We will also discuss asymmetric information models, including the recent literature with
endogenous information acquisition.
We put some emphasis on recent developments in the macroeconomics literature, which has embedded
non-trivial portfolio choices in standard two country general equilibrium macro models. Explaining equity
home bias has been one of the main motivations for this literature, which has first focused on models with
equities only. But the importance of considering portfolio choices across a broader class of assets (bonds,
corporate debt, equity...) is now widely recognized. We develop a standard two country/two good DSGE
model with endogenous portfolio choice and allow for equity trade in first instance, as in the early literature,
1
Reported in Flandreau (2006).
1

and then generalize the set up to accommodate trade in bonds and equity. This allows us both to present
recent methodological developments to fully characterize portfolios in this class of models and to show the
limitations of the early literature. These new models also call for new portfolio facts. Accordingly, we present
some new evidence on international holdings across different types of assets: equities, bonds and banking
assets. We focus on portfolio investment and abstract from Foreign Direct Investment, as its determinants
may be of a different nature and are studied extensively in the trade literature. Finally, we present some new
micro data on institutional holdings of equity at the fund level. These data should inform macroeconomic
modelling of the open economy as well as models of delegated investment, which belong to a fast-growing
literature: a large share of international investment is intermediated.
In section 2, we present the standard definition of home equity bias and some recent measures across
countries and across time. In section 3, we focus on the recent methodological developments in the macro-
economics literature. In section 4, we focus on the role of hedging motives as a source of equity home bias.
We use standard dynamic models of the Open Economy Financial Macroeconomics literature. In section 5,
we present the literature on trade costs in financial markets (transaction costs, international taxation, legal
frameworks). In section 6, we review the finance literature on information asymmetries and behavioural bi-
ases. In section 7, in line with recent theoretical work, we present some new evidence on aggregate portfolio
holdings across a wider range of assets. We also present new portfolio facts at the fund level and discuss
leads for future research. Section 8 concludes.
2 The Equity Home Bias : Definition and Measure
2.1 Definition
French and Poterba (1991) w
ere the first to our knowledge to document domestic ownership shares
across countries. Using data for the US, Japan, UK, France and Germany, they show that investors hold a
disproportionate share of domestic assets in their equity portfolios. In 1989, 92% of the US stock market was
held by US residents. Analogous numbers for Japan, UK, France and Germany are respectively 96%, 92%,
89% and 79%. They label this lack of cross border diversification the equity home bias.
2
This is a well-known
puzzle in international finance: in a world with frictionless financial markets, the most basic International
CAPM model with homogenous investors across the world would predict that the representative investor of
a given country should hold the world market portfolio. In other words, the share of his financial wealth
invested in local equities should be equal to the share of local equities in the world market portfolio, a
prediction that contradicts the most casual observation of the data on portfolio holdings. As a result, the
measure of equity home bias (EHB) that is most commonly used is the difference between actual holdings
of domestic equity and the share of domestic equity in the world market portfolio:
3
2
Bohn and Tesar (1996) estimate the share of foreign equities in the U.S. portfolio to be still very low in 1995, equal to
8 percent. Ahearne et al. (2004) estimate it to be slightly above 10% in 2000.
3
See for instance A
hearne et al. (2004). Another commonly used measure in finance is a deviation from a benchmark
mean-variance portfolio. Benchmark portfolio weights are calculated from a mean variance optimisation problem with sample
2

EHB
i
= 1 −
Share of Foreign Equities in Country i Equity Holdings
Share of Foreign Equities in the World Market Portfolio
(1)
When the home bias measure for country i EHB
i
is equal to one, there is full equity home bias; when it
is equal to zero, the portfolio is optimally diversified according to the basic International CAPM.
2.2 Evidence across time and across countries
While one could argue that, at the end of the eighties, international capital markets were far from being
frictionless and this could contribute to rationalize home bias, this line of explanation seems more doubtful
today. Despite increased financial integration, the equity home bias remains a pervasive phenomenon across
countries and across time (classic surveys include Lewis (1999) and Karolyi and Stulz (2003). See
also Sercu and Vanpee (2008) for recent evidence). In figure (1), we show the evolution of home bias
measures in developed countries across regions of the world:
4
it has decreased over the last twenty years
w
ith the process of ‘financial globalization’ but remains high in most countries (see also table (1) for a recent
snapshot of home bias measures for selected countries). On average, the degree of home bias across the world
is 0.63 (lower in Europe where monetary union after 1999 seem to have had an effect).
5
For the developed
world, this means that the share of foreign equities in investors portfolios is roughly a 1/3 of what it should
be if the benchmark is the basic International CAPM. In figure (2), we construct a similar indicator for
emerging markets. Emerging markets have a less diversified equity portfolios than developed countries and
do not exhibit any clear downward trend in home bias. The average degree of home bias in these countries
is 0.9 (smaller in emerging Asia and larger in Latin America) and investors in these countries hold 1/10 of
the amount of foreign equities they should be holding according to the basic International CAPM model.
This robust stylized fact has received considerable attention from both the finance literature and the
macroeconomics literature. The main difference between these two sets of literature relies on some modelling
assumptions. To simplify, the traditional finance literature has tried to rationalize the equity home bias in
multi-country models of portfolio choice where asset prices and their second moments are given (in particular
in these models the risk-free interest rate is exogenously given). The macro literature has tried to integrate
international portfolio decisions in otherwise standard Dynamic Stochastic General Equilibrium (DSGE)
models of the international economy. These models have a fully general equilibrium structure and asset
prices and their second moments are endogenously determined.
6
The motivation is however the same:
estimates of the means and variance-covaraince matrix of returns. The main issue in the existing literature adopting the finance
approach are how to measure returns and covariance matrices. Papers differ in the extent to which they use real or nominal
returns, how they estimate expected returns and how they deal with structural breaks and nonstationarity. As a result, there
is a degree of heterogeneity in the estimates of expected returns and second moments.
4
See appendix for data description and country samples.
5
See Coeurdacier and Martin (2009) and Fidora et al. (2007) for studies on the impact of the Monetary Union on
cross-border equity diversification.
6
The dichotomy, which historically seems relevant, appears increasingly artificial as more papers bridge the two strands of
literature.
3

Frequently asked questions

Q1. What contributions have the authors mentioned in the paper "Nber working paper series home bias in open economy financial macroeconomics" ?

The authors review various explanations of this puzzling phenomenon highlighting recent developments in macroeconomic modelling that incorporate international portfolio choices in standard two-country general equilibrium models. The authors present new evidence on crossborder asset holdings across different types of assets: equities, bonds and bank lending and new micro data on institutional holdings of equity at the fund level. 

Q2. What are the future works in "Nber working paper series home bias in open economy financial macroeconomics" ?

There is little doubt however that understanding better delegated investment strategies and the constraints and incentives of fund managers would be an important step to study the extent and time series variation of international risk sharing, the international transmission of financial shocks and the propagation of crises. More generally, a promising avenue of research would be to study optimal regulatory or taxation policies in environment featuring endogenous asymmetric information or principal agent problems due to delegated investment. 

Q3. Why are the forward positions used to hedge the nominal exchange rate changes?

Due to price rigidity, nominal exchange rate fluctuations are related to real exchange rate fluctuations and the forward positions are used to hedge the nominal exchange rate changes, leaving only a part of the relative price risk to be hedged by equity positions. 

Q4. What is the reason why the authors estimate a vector error correction model?

The authors estimate a vector error correction model that allows the correlation between labour and capital returns to vary over time and be imperfect, while maintaining the assumption that the ratio of labour to capital income is stationary. 

Q5. What is the main reason why households exhibit home bias towards local (government) nominal bonds and local?

In order to hedge fluctuations in taxes, households exhibit home bias towards local (government) nominal bonds and local equities. 

Q6. What are the main advantages of the perturbation methods developed by Devereux and Sutherland?

The main advantage of the perturbation methods developed by Devereux and Sutherland (2008a) and Tille and van Wincoop (2008) are: (1) they are very easy to implement as they are close to standard approximation methods used in DSGE models; (2) they can be applied to broad range of environments (complete and incomplete markets models, potentially large number of shocks and/or securities) (3) they provide (approximate) closed-form expressions for portfolios in many cases. 

Q7. How do they solve for the first order dynamics of the portfolio?

In a companion paper, Devereux and Sutherland (2010), the authors show that in order to solve for the first order dynamics of the portfolio, a second order approximation of the non portfolio equations of the model is needed, while the portfolio equations need to be approximated to the third order. 

Q8. How much equity is held indirectly by households in the US?

In the US, according to the Survey of Consumer Finance (see Polkovnichenko (2004)), the share of equity held indirectly by households through mutual funds, pension funds or other investment vehicles has risen from about 46% in 1989 to close to 62% in 2001. 

Q9. What is the typical portfolio of households who participate in the financial market?

The typical portfolio of households who participate in the financial market is therefore very dichotomic: it contains a very small number of stocks, which are directly held and a more diversified stock portfolio, which is usually managed by a third party. 

Q10. What is the main drawback of the class of models with endogenous portfolio decisions?

An important related shortcoming of this class of models with endogenous portfolio decisions is that the allocation under perfect risk-sharing is replicated, or at least up to the degree of the approximation. 

Q11. What are the main advantages of the perturbation methods developed in Evans and Hnatkovska?

The methods developed in Evans and Hnatkovska (2008) and Judd et al. (2002) can be applied to very general classes of models, but are quite complex and present significant departures from standard DSGE solution methods. 

Q12. What is the key empirical counterpart of portfolio bias in this class of models?

from an empirical perspective, a series of papers have provided estimates of the covariance between relative equity returns and relative returns to human wealth which is the key empirical counterpart of portfolio biases in this class of models. 

Q13. What is the link between information disadvantages and the expectation of a market?

The authors show that there is a link between information disadvantages and the expectations (degree of bullishness) about a market: foreign investors tend to become more bullish about a certain market following a positive return on that market. 

Q14. How does the degree of home bias in bond portfolios differ between emerging markets?

In figure (4), the authors show the degree of bond home bias for emerging markets: emerging markets have even much less diversified bond portfolios than developed countries and it has barely decreased over the last decade.