The demand for safe assets in emerging economies and global imbalances: New empirical evidence

• Author: Rudiger Ahrend,Cyrille Schwellnus,Alessandro Saia
• DOI: http://doi.org/10.1111/twec.12526
• Date: 01 February 2018
• Published date: 01 February 2018

ORIGINAL ARTICLE

The demand for safe assets in emerging economies

and global imbalances: New empirical evidence

Rudiger Ahrend

1

|

Alessandro Saia

1,2

|

Cyrille Schwellnus

1

1

OECD, Paris Cedex 16, France

2

Department of Economics, University of Lausanne, Lausanne, Switzerland

1

|

INTRODUCTION

The emergence of large global current account imbalances before the global crisis of 2008–09

sparked concerns, particularly as regards their implications for financial stability. While the

causes of global imbalances remain a subject of debate, one influential view sees them as dri-

ven by a mismatch between the supply and the demand of safe financial assets. According to

this view, emerging and commodity-producing countries have seen fast increases in disposable

income without corresponding increases in financial development and supply of safe financial

assets. The resulting excess demand would have been redirected to financially developed coun-

tries capable of producing vast amounts of financial assets perceived as safe (Caballero, 2008;

Caballero, Farhi, & Gourinchas, 2008; Mendoza, Quadrini, & Rios-Rull, 2009). This would

have raised current account surpluses of emerging countries and put downward pressure on

long-term interest rates in safe-asset producing countries, thereby contributing to increased

financial leverage. In the current economic environment of a declining supply of safe assets

(Chen & Imam, 2014; Gorton, Lewellen & Metrick, 2012; IMF, 2012), the asset mismatch

hypothesis remains highly relevant; the excess demand for safe assets may induce asset price

bubbles for both safe assets and assets further down the safety scale, with adverse conse-

quences for financial stability.

The asset mismatch hypothesis has two empirically testable implications. In the model of

Caballero et al. (2008), a positive shock to growth in financially less-developed countries leads

to excess demand for safe assets.

1

In equilibrium, this results in lower world interest rates and

in financially less-developed countries importing assets from financially developed countries.

While financially developed countries fully satisfy their safe-asset demand domestically, the port-

folio of financially less-developed countries includes foreign assets. A first empirically testable

implication of the asset mismatch hypothesis is thus that financially less-developed countries

should hold a higher share of their total portfolio in foreign securities than financially more-

developed countries; that is, they should display a weaker home bias in portfolio allocation.

1

In Caballero et al. (2008) there is no difference in the risk structure of the assets produced by financially developed and

financially less-developed countries, but the latter are assumed to produce a lower share of assets relative to income.

DOI: 10.1111/twec.12526

World Econ. 2018;41:573–603. wileyonlinelibrary.com/journal/twec ©2017 John Wiley & Sons Ltd

|

573

Second, they should hold a particularly high share of their foreign portfolio in financially devel-

oped countries, as ostensibly safe assets are predominantly produced by the latter. Following

Forbes (2010), this paper uses these theoretical predictions to assess the practical relevance of

the asset mismatch hypothesis.

Forbes (2010) finds that financially less-developed countries display both weak home bias

and a strong preference for US portfolio equity and debt. Insofar as the United States is

among the countries producing a large amount of ostensibly safe assets, this would support

the asset mismatch hypothesis. However, in stark contrast to Forbes (2010), the empirical

literature on the determinants of home bias in portfolio allocation usually finds that finan-

cially less-developed countries display particularly strong home bias (Chan, Covrig, & Ng,

2005). Moreover, in the sample of Forbes (2010), the United States is the only destination

country, which makes it impossible to assess whether financially less-developed countries

generally hold higher portfolio shares in other countries producing assets perceived as safe,

such as those English-speaking countries explicitly mentioned by Caballero et al. (2008)

(Australia, United Kingdom) or Germany and Switzerland. The result could therefore reflect

a US-specific effect, perhaps related to the US dollar’sroleastheworld’s main reserve and

transaction currency, instead of a general preference of financially less-developed countries

for the assets of financially developed countries, as implied by the asset mismatch hypothe-

sis.

Contrary to Forbes (2010), this paper finds strong evidence against the asset mismatch

hypothesis. The main result is that financially less-developed countries generally hold smaller

shares of their total portfolio in financially more-developed countries than the latter do among

themselves, mainly reflecting the stronger home bias of financially less-developed countries.

Even when analysing only the composition of financially less-developed countries’foreign port-

folios, there is no evidence of a general preference for the assets of financially developed

countries. These results are consistent with previous empirical results on the determinants of

home bias in portfolio allocation and with theoretical models predicting that financially devel-

oped countries engage in larger cross-investments because of lower financial transaction costs

(Martin & Rey, 2004).

2

The remainder of this paper is structured as follows. The first section describes the under-

lying data and the methodology adopted for the empirical analysis. Section 3 presents

descriptive statistics on asset allocation across selected origin and destination countries, with

a particular focus on emerging countries. It is shown that asset allocation in emerging coun-

tries exhibits a particularly strong home bias with no generalised preference for the assets of

financially developed countries. Section 4 presents the results from the econometric analysis

ofdebtsecurities,bothinafullybilateralsetupandfocusingonselectedsafe-assetproducing

countries. In line with the descriptive statistics, no evidence for overinvestment of financi ally

less-developed countries in their financially more-developed counterparts is detect ed. The sen-

sitivity checks in Section 5 show that these results are robust to different measures of finan-

cial development and different estimation samples. Sectio n 6 concludes. The Appendices A–C

shows that similar results are obtained when conducting the analysis for equity instead of

debt securities.

2

The results are also consistent with a strand in the finance literature that emphasises that more sophisticated investors tend

to invest more in unfamiliar environments than unsophisticated investors, presumably because of their superior information-

gathering capabilities (Grinblatt & Keloharju, 2001).

574

|

AHREND ET AL.