FINANCIAL CRISES AND INTERNATIONAL PORTFOLIO DYNAMICS
Fernando Broner
Tatiana Didier
Aitor Erce
Sergio L. Schmukler
*
June 28, 2010
Abstract
This paper analyzes the behavior of international capital flows by foreigners and
domestic agents, especially during financial crises. We show that gross capital flows by
foreigners and domestic agents are very large and volatile, especially relative to net
capital flows. This is because when foreigners invest in a country domestic agents tend to
invest abroad and vice versa. Gross capital flows are also pro-cyclical. During
expansions, foreigners tend to bring in more capital and domestic agents tend to invest
more abroad. During crises, there is retrenchment, i.e. a reduction in capital inflows by
foreigners and an increase in capital inflows by domestic agents. This is especially true
during severe crises and during systemic crises. The evidence can shed light on the nature
of shocks driving international capital flows. It seems to favor shocks that affect
foreigners and domestic agents asymmetrically –e.g. sovereign risk and asymmetric
information– over productivity shocks.
JEL Classification Codes: F21, F32
Keywords: capital flows, domestic investors, foreign investors, crises
*
We received very valuable comments from Eduardo Fernandez-Arias, Carmen Reinhart, Frank Warnock,
and participants at presentations held at the AEA Annual Meetings (Atlanta), CREI, the LACEA Annual
Meetings (Buenos Aires), and the SED Annual Meetings (Istanbul). For excellent research assistance, we
especially thank Leandro Brufman, and also Francisco Ceballos, Ana Maria Gazmuri, Laura Fernandez,
and Virginia Poggio. Schmukler acknowledges financial support from the World Bank, Development
Economics. Broner is with the CREI and Universitat Pompeu Fabra. Erce is with the Bank of Spain. Didier
and Schmukler are with the World Bank. The views expressed here do not necessarily represent those of
the Bank of Spain or the World Bank.
1. Introduction
During the last decades, international capital flows have played an increasingly
important role in the business cycles of developed and developing countries,
particularly during episodes of financial crises. Many studies have analyzed the
behavior of net capital flows, or capital inflows by foreigners (CIF) plus capital inflows
by domestic agents (CID). It is known that net capital flows are highly pro-cyclical in
developing countries and, to a lesser extent, in developed countries. It is also known that
crises are associated with a large reduction in, or even a reversal of, net capital inflows.
1
This paper focuses on the behavior of gross capital flows, i.e. capital flows by
foreign and domestic agents (CIF and CID). In part, we are motivated by the behavior
of capital flows prior and during the global financial crisis of 2008. During the years
before the crisis, gross capital flows had been very large. They reflected, among other
things, portfolio diversification among developed countries, foreign direct investment
and portfolio flows to developing countries, and accumulation of safe foreign assets by
developing countries. The crisis, however, was accompanied by a larger reduction in
inflows by foreigners and outflows by domestic agents, prompting a debate on the
possibility of entering into a phase of de-globalization.
In this paper, we analyze systematically the cyclical behavior of gross capital
flows.
2,3
To do so, we construct measures of CIF and CID using Balance of Payments
data from the International Financial Statistics from the International Monetary Fund
(IMF). CIF equals net purchases of domestic assets by non-residents and is thus equal to
1
See for example Kaminsky, Lizondo and Reinhart (1998), Calvo, Izquierdo and Mejia (2002), and
Broner and Rigobon (2006).
2
Perhaps surprisingly, there are few analyses of the behavior of gross capital flows at business-cycle
frequencies. Three exceptions are Faucette, Rothenberg, and Warnock (2005), Rothenberg and Warnock
(2006), and Cowan, De Gregorio, Micco, and Neilson (2008), who use information on gross capital flows
to determine whether in each particular crisis reductions in net capital inflows is driven by foreigners or
domestic agents. A few papers also look at the distinction between domestic and foreign investors during
specific events or specific markets. See, for example, Frankel and Schmukler (1996).
3
For evidence on long-run trends in gross capital flows, see Lane and Milesi-Ferretti (2001, 2007),
Kraay, Loayza, Servén, and Ventura (2005), Devereux (2007), and Gourinchas and Rey (2007a, 2007b).
2
the sum of all “liability” inflows. CID equals net sales of foreign assets by domestic
agents and is thus equal to the sum of all “asset” inflows, including international
reserves.
4
We explore the behavior of these capital flows from 1970 to 2009 in 103
developed and developing countries.
Figures 1 and 2 show the behavior of CIF and CID as a fraction of GDP for a
number of developed and developing countries. Positive values of CIF indicate that
foreigners are increasing their holdings of domestic assets, while positive values of CID
indicate that domestic agents are reducing their holdings of foreign assets. The figures
illustrate the long-run process of financial globalization, as CIF has been consistently
positive and CID consistently negative and they have both been large and growing
fractions of GDP. The figures also suggest that the process of globalization has not been
smooth, as there have been large variations in CIF and CID from year to year.
Particularly during periods of financial turbulence, such as during the global financial
crisis of 2008, there seems to have been retrenchment in capital flows, reflected in
reductions in CIF and increases in CID.
In this paper, we formally document these patterns by analyzing the dynamics of
CIF and CID at business-cycle frequencies, as well as around periods of financial
distress. Our main findings are as follows.
(i) Over the last four decades, there has been a larger increase in the volatility
of CIF and CID than that of net flows, especially in developed countries.
Moreover, the larger volatility of CIF and CID has not been accompanied
by more volatile net flows, perhaps because of the high correlation
between CIF and CID.
4
Note that although we refer to CIF and CID as gross capital flows, they reflect net transactions in
domestic and foreign assets, respectively, between foreign and domestic agents. Also note that CIF and
CID are not equal to changes in foreign liability and foreign asset positions as these flow measures do not
take into account valuation effects.
3
(ii) Gross capital flows are pro-cyclical, as CIF is pro-cyclical and CID is
counter-cyclical. In other words, during expansions foreigners increase
their investment in domestic assets and domestic agents increase their
investment in foreign assets. This is true regardless of whether expansions
are measured in terms of real GDP growth or deviations in GDP from
trend.
(iii) During crises, especially during severe ones, there is a reduction of total
gross capital flows (CIF-CID), with significant reductions in CIF and
increases in CID. Systemic crises are associated with more retrenchment
than crises that occur in a smaller set of countries.
(iv) A decomposition of CIF and CID reveals interesting heterogeneity in the
behavior of their components during crises. The reduction in CIF is due to
reductions in liabilities in equity portfolio investment and other
investments in developed countries and in debt portfolio investment and
other investments in developing countries. The increase in CID is due to
reductions in assets in equity portfolio investment, debt portfolio
investment, and other investments in developed countries and in reserves
and debt portfolio investment in developing countries.
The behavior of capital inflows by foreigners and domestic agents can shed light
on the sources of fluctuations and international capital flows. First, we find no evidence
that, on average, gross capital flows are driven by fire sales of domestic assets to
foreigners and/or domestic capital flight, as we find retrenchment during crises, with
foreigners leaving and domestic agents coming back home. Also, the evidence runs
contrary to the view that capital flows are driven mostly by productivity shocks, since
such shocks would imply a similar behavior by foreigners and domestic agents towards
4
domestic assets. For instance, if crises were only due to negative productivity shocks,
we would expect both foreign and domestic investors to reduce their investments in
domestic assets, resulting in a decrease in both CIF and CIR.
5
Similar implications
would be obtained if crises were associated with a worsening of investor property rights
that affected equally domestic and foreign creditors.
The evidence suggests that crises affect foreigners and domestic agents
asymmetrically. If, for example, crises were associated with a worsening of investor
property rights that affected foreign creditors more than domestic creditors, we would
expect the type of retrenchment observed in the data. We would also expect
retrenchment, if crises increased the informational asymmetry between foreigners and
domestic agents.
6
The rest of the paper is organized as follows. Section 2 describes the data.
Section 3 analyzes the comovement of capital flows by foreigners and domestic agents
and their behavior over the business cycle. Section 4 analyzes the behavior of gross
capital flows during crises. Section 5 concludes.
2. Data
To document worldwide patterns of capital flows by domestic and foreign
agents, we assemble a comprehensive dataset on both aggregate flows and the
subcomponents. The data come from the “analytic” presentation of the IMF’s Balance
5
We are not aware of papers that explicitly model the effects of productivity shocks on gross capital
flows over the business cycle. Some related models include Kraay and Ventura (2000), Kraay, Loayza,
Servén, and Ventura (2005), Devereux (2007), Coeurdacier, Kollmann, and Martin (2010), and Devereux
and Sutherland (forthcoming).
6
The large literature on sovereign risk and “sudden stops” has, for the most part, only concerned itself
with net capital flows. See, for example, Mendoza (forthcoming). Broner, Martin, and Ventura
(forthcoming) show that during periods of financial distress in which the probability of default on
foreigners increases domestic agents have an incentive to repurchase domestic assets in the hands of
foreigners, thereby reducing sovereign risk ex-ante. Brenan and Cao (1997) present a model based on
asymmetric information. Tille and van Wincoop (2010) suggest an alternative mechanism based on time
variation in asset risk.
5
