International investment positions revisited: Investor heterogeneity and individual security characteristics

• Published date: 01 May 2020
• Author: Martijn A. Boermans,Robert Vermeulen
• Date: 01 May 2020
• DOI: http://doi.org/10.1111/roie.12460

466

|

wileyonlinelibrary.com/journal/roie Rev Int Econ. 2020;28:466–496.

© 2019 John Wiley & Sons Ltd

1

|

INTRODUCTION

This paper contributes to the literature on international investment patterns by empirically docu-

menting the importance of investor heterogeneity and security level characteristics for cross-border

bond and stock holdings. To date the empirical literature has devoted much attention to cross-border

frictions in explaining why investors deviate from holding internationally well-diversified portfolios.

Major stylized facts emerged showing the importance of market size and geographical distance as

‘gravity factors’ shaping global assets allocations (Ahearne, Griever, & Warnock, 2004; Buch, 2005;

Lane & Milesi-Ferretti, 2008; Portes & Rey, 2005). These macro studies, most of which based on IMF

Coordinated Portfolio Investment Surveys (CPIS) data, assume that the investment decision stems

from a representative investor that allocates its investment toward a single representative asset in a

Received: 21 March 2019

|

Revised: 30 October 2019

|

Accepted: 4 November 2019

DOI: 10.1111/roie.12460

ORIGINAL ARTICLE

International investment positions revisited:

Investor heterogeneity and individual security

characteristics

Martijn A.Boermans1,2

|

RobertVermeulen1

1De Nederlandsche Bank

2Utrecht University

Correspondence

Robert Vermeulen, De Nederlandsche

Bank, Amsterdam, The Netherlands.

Email: r.j.g.vermeulen@dnb.nl

Abstract

This paper contributes to the literature by empirically

documenting the importance of investor heterogeneity and

security level characteristics for international investment

positions. We estimate a standard gravity model to explain

holdings in individual bonds and stocks for euro area inves-

tors covering 2013Q4–2018Q4. First, our results confirm

the importance of market size and distance, but there are

significant differences between banks, insurance companies

and pension funds, investment funds, households and other

investors. Second, we show the statistical significance and

economic relevance of security characteristics such as cur-

rency, bond maturity and dividend yield. Third, the results

are robust to including domestic holdings.

JEL CLASSIFICATION

F36; G11; G15; G20

|

467

BOERMANS ANd VERMEULEN

given foreign country. In this paper, we revisit these implicit assumptions by first focusing on portfolio

choice decisions that take into account that individual bond and stock characteristics determine inter-

national portfolio positions and second, that the determinants differ across investors by distinguishing

between five investor sectors: banks, insurance companies and pension funds (ICPFs), investment

funds, households and other investors.

We build on a small number of papers that analyze international investment patterns without a

representative investor (Giofré, 2013; Galstyan & Velic, 2018; Galstyan, Lane, Mehigan, & Mercado,

2016; Roque & Céu Cortez, 2014). These studies use disaggregated data from the CPIS to show how

the determinants of bilateral asset holdings differ across investor sectors. For example, Roque and

Céu Cortez (2014) find that households have a stronger tendency to hold assets in nearby countries

compared to more sophisticated financial investors. However, these studies estimate their models

based on aggregated bond and equity holdings at the country-level, thereby assuming investors hold a

representative asset in that country.

In parallel, an expanding literature shows the importance of individual security characteristics

when explaining international portfolio holdings. For example, recent studies show how the currency

denomination of a bond shapes investor holdings of individual bonds (Boermans & Vermeulen, 2016;

Burger, Warnock, & Warnock, 2018; Maggiori, Neiman, & Schreger, 2019). In particular, Burger

etal. (2018) show that a country attracts more investments from other countries when it issues bonds

in the investor’s home country currency. Besides currency, other studies highlight the importance

of bond maturity (Vayanos & Vila, 2009), sovereign bonds (Buch, Koetter, & Ohls, 2016), whether

a bond is eligible collateral for central bank transactions (Hildebrand, Rocholl, & Schulz, 2012),

its credit rating (Becker & Ivashina, 2015), its issuer sector (Koijen, Koulischer, Nguyen, & Yogo,

2017) and for stocks the industrial sector (Schumacher, 2017). Finally, recent work by Boermans and

Vermeulen (2018) and Bergant, Fidora, and Schmitz (2018) analyzes the impact of euro area monetary

policy on bond holdings using a similar dataset as we do.

In this paper, we combine these insights by analyzing investor heterogeneity in international port-

folio allocations at the individual bond and stock level. Doing so offers new insights on the importance

of security characteristics and the role of different investor sectors when explaining international

portfolio choice. To achieve this, we use a unique security-by-security holdings dataset of euro area

investors in domestic and foreign bonds and stocks for the period 2013Q2–2018Q4 at the quarterly

frequency. The data comprise over 290,000 individual bonds and over 33,000 unique stocks with

portfolio investment of on average 17 trillion euro per quarter yielding over 17 million observations

in total. The granularity of the data allows us to identify which security characteristics are relevant for

investment decisions across investor sectors.

Our empirical methodology closely follows the standard gravity model as introduced by Martin

and Rey (2004) in international finance. We include established variables in the model such as the dis-

tance between the country of the investor and the country of the asset’s issuer. In addition, we extend

the model with relevant security level variables. Because our unit of analysis is an individual security,

we use the market value of an individual bond or the firm’s stock market capitalization as the gravity

attractor variable instead of macro proxies such as a country’s stock market capitalization or GDP. In

addition, our model is able to capture cross-border frictions and information asymmetries between

different investor sectors and issuing entities in line with Galstyan etal. (2016).

The results document key differences in the determinants of international portfolio holdings across

investor sectors. Our estimates show that the adoption of investor heterogeneity and security-by-

security data still allow us to use a standard gravity model and obtain general results broadly comparable

to the literature on international investment patterns. The results revisit prior literature by highlighting

the importance of gravity factors like market size, distance and common borders by showing that these