THE MISSING WEALTH OF NATIONS:
ARE EUROPE AND THE U.S. NET DEBTORS OR
NET CREDITORS?
Gabriel Zucman
Paris School of Economics
∗
February 25, 2013
†
Abstract
This paper shows that official statistics substantially underestimate the net for-
eign asset positions of rich countries because they fail to capture most of the assets
held by households in offshore tax havens. Drawing on a unique Swiss dataset
and exploiting systematic anomalies in countries’ portfolio investment positions,
I find that around 8% of the global financial wealth of households is held in tax
havens, three-quarters of which goes unrecorded. On the basis of plausible assump-
tions, accounting for unrecorded assets turns the eurozone, officially the world’s
second largest net debtor, into a net creditor. It also reduces the U.S. net debt
significantly. The results shed new light on global imbalances and challenge the
widespread view that, after a decade of poor-to-rich capital flows, external assets
are now in poor countries and debts in rich countries. I provide concrete proposals
to improve international statistics.
Keywords: Tax havens, International investment positions, Global imbalances
JEL classifications: F32, H26, H87.
∗
48 boulevard Jourdan, 75014 Paris, France, zucman@pse.ens.fr. A detailed Appendix is available
online.
†
I am very grateful to my advisor Thomas Piketty for his continuous support and guidance. I
thank the editor (Robert Barro), five anonymous referees, Jean-Edouard Colliard, Mihir Desai, Nicolas
Frémeaux, Lucie Gadenne, Pierre-Olivier Gourinchas, Philip Lane, Gian Maria Milesi-Ferretti, Richard
Sylla, Daniel Waldenström, Edward Wolff, and participants at numerous seminars. Paul Farello, Christo-
pher Gohrband, Steve Landefeld, and Robert Yuskavage from the Bureau of Economic Analysis also
provided helpful comments. Part of this paper was drafted while I was visiting NYU; I am grateful to
Jess Benhabib who made this visit possible.
1
I. Introduction
There are two puzzles in international investment statistics. The first is a set of sta-
tistical anomalies. At the global level, liabilities tend to exceed assets: the world as a
whole is a net debtor (Lane and Milesi-Ferretti, 2007). Similarly, the global balance of
payments shows that more investment income is paid than received each year. Since the
problem was identified in the 1970s, the International Monetary Fund has commissioned
a number of reports to investigate its causes, and national statistical agencies have put
considerable resources into improving their data. Yet despite a great deal of progress,
large anomalies remain; many European securities, in particular, have no identifiable
owner (Milesi-Ferretti, Strobbe, and Tamirisa, 2010).
The second puzzle is a theoretical challenge. Since the latter half of the 1990s, capital
has been flowing from poor to rich countries. As a result, the rich world now appears to
be a sizeable net debtor in the official data, dragged down by the U.S. and Europe. While
the literature has put forward possible explanations for the U.S. net debt and the rise in
China’s assets,
1
the negative net positions of Europe and the overall rich world remain
largely unexplained. Despite this, many observers have grown accustomed to the view
that external assets are now in poor countries and debts in rich countries. In the public
debate, the view that “China owns the world” has become particularly popular. Should
it be correct, the implications for policymaking and open-economy modeling would be
far-reaching.
My paper challenges this view. The negative net foreign asset position of the rich
world, I argue, is an illusion caused by tax havens. International statistics fail to capture
most of the assets held by households through tax havens: they overlook the portfolios
of equities, bonds, and mutual fund shares that households own via banks in Switzerland
and other countries with strict bank secrecy rules. This coverage gap explains many
of the long-standing anomalies in global data. My computations find that around 8%
1
See Dooley, Folkerts-Landau, and Garber (2003), Bernanke (2005), Dollar and Kraay (2006), Engel
and Rogers (2006), Caballero, Farhi, and Gourinchas (2008), Mendoza, Rios-Rull, and Quadrini (2009),
Carroll and Jeanne (2009), Ma and Haiwen (2009), Obstfeld, Shambaugh, and Taylor (2010), Aguiar
and Amador (2011), Song, Storesletten, and Zilibotti (2011), and Alfaro, Kalemli-Ozcan, and Volosovych
(2011) among others.
2
of households’ financial wealth is held through tax havens, three-quarters of which goes
unrecorded. This stock of unrecorded assets is double the recorded net debt of the rich
world (Figure I). Since a body of evidence suggests that most of the wealth in tax havens
belongs to residents of rich countries, accounting for it turns the rich world into a net
creditor. Despite a decade of global imbalances, therefore, external wealth is still probably
in rich countries overall: China does not own the world yet. Back in the 1980s-1990s the
rich world had a large positive net position; over the last decade it has eaten some of
its claims away; but today poor countries are still repaying their debts to advanced
economies.
These findings have direct implications for core issues in international macroeco-
nomics. On the basis of plausible assumptions, accounting for the wealth in tax havens
turns the eurozone, officially the world’s second largest net debtor, into a net creditor.
It also improves the U.S. net position. Now, the net foreign asset position is a key state
variable in dynamic macroeconomic models. Accurate net positions are essential to assess
the merits of the different views put forward on the causes of global imbalances and they
are important to monitor financial stability. A large body of literature has questioned the
sustainability of global imbalances.
2
If indeed the net positions of Europe and the U.S.
are higher than in the official statistics, the required international adjustment is smaller
than commonly thought. Domestic imbalances and public finance issues may be more
serious today for rich countries than global imbalances: rich countries taken as a whole
are richer than we think, but some of their wealthiest residents hide part of their assets in
tax havens, which contributes to making governments poor. So far, tax havens have been
ignored by the literature that studies the evolution of top income shares around the world
(Atkinson, Piketty, and Saez, 2011).
3
My findings, therefore, also have implications for
this strand of research: my macro-based estimate of the funds held through tax havens
2
See Obstfeld and Rogoff (2005), Blanchard, Giavazzi, and Sa (2005), Gourinchas and Rey (2007a),
the papers in Clarida (2007), Hausmann and Sturzenegger (2007), Curcuru, Dvorak, and Warnock (2008),
and Blanchard and Milesi-Ferretti (2009) among others.
3
The two exceptions are Roine and Waldenström (2009) who use anomalies in Sweden’s balance of
payments to approximate capital flight, and Dell, Piketty, and Saez (2007) who use Swiss tax data to
put an upper bound on the amount of capital income earned in Switzerland by non-resident taxpayers.
Tax data, however, are not an appropriate source in this case, because the bulk of income earned by
foreigners in Switzerland does not have to be declared to Swiss tax authorities.
3
could be used as a first step to include these funds into micro-based studies of income
and wealth distributions.
The paper proceeds as follows. Section II begins with a brief primer on the activities
that take place in tax havens and the statistical issues involved. Section III analyses a
previously unused official dataset from the Swiss National Bank. A considerable amount
of wealth is held unrecorded in Swiss accounts, and contrary to popular belief, this wealth
mostly belongs to residents of rich countries. Section IV then presents a novel method
to estimate the personal wealth in all the world’s tax havens, using anomalies in the
aggregate portfolio stock data of countries (the key source here is Lane and Milesi-Ferretti,
2007). My method is indirect and relies on data with known imperfections, so it is subject
to some margin of error. Section V presents consistency and robustness checks, based on
bilateral and flow data from the IMF, suggesting that the order of magnitude I find is
reliable. The many datasets used in this paper all paint the same picture: households
own a large amount of mutual fund shares through unrecorded accounts in tax havens. In
Section VI, I propose scenarios as to how including the unrecorded assets in the statistics
would affect published international investment positions. I discuss the implications for
global imbalances and the uncertainties that remain. The conclusion provides concrete
proposals to improve the official data. There are numerous intricacies in the financial
activities of tax havens and the international statistics. The most important ones are
discussed in the paper; others are detailed in a comprehensive Online Appendix.
II. Tax Havens and Their Implications for International Statistics
First, let’s look at the basic concepts that will be used throughout the paper. A country’s
foreign assets and liabilities are recorded in its international investment position (IIP).
The IIP is the stock equivalent of the financial account of the balance of payments: the
IIP shows the stock of existing cross-border investments at the end of each year, while
the balance of payments shows the yearly flow of new investments. There are three broad
categories of cross-border claims: direct investments (holdings of over 10%), portfolio
4
securities (equities and bonds that do not qualify as direct investment), and other assets
(mainly loans and deposits).
4
At the end of 2008, as shown by Table I, securities were
the largest category: they accounted for $40tr out of $90tr.
Tax havens host numerous financial activities. About 40% of the world’s foreign
direct investments are routed through tax havens such as the British Virgin Islands.
5
Many investment funds and financial vehicles are incorporated offshore. Luxembourg is
the second largest mutual fund center in the world after the U.S; a great deal of the
world’s money market funds are incorporated in Ireland; and most hedge funds are in
the Cayman Islands. Multinational corporations routinely use tax havens for treasury
operations and group insurance. Some of these activities have legitimate roles and are
satisfactorily covered in the statistics.
6
My paper focuses on one specific tax haven
activity: personal wealth management or “private banking”. This activity is present in
many but not all tax havens. Leaders include countries with strict bank secrecy rules such
as Switzerland, the Cayman Islands, the Bahamas, Hong Kong, Singapore, and Jersey.
Banks incorporated in these countries – which are often subsidiaries of large global banks
– attract foreign individuals and provide them with investment advice and services. In
the IIPs of countries, the personal wealth management activities of tax havens do not
affect direct investment data, slightly affect “other assets”, but cause large, systematic
errors for portfolio securities.
II.A. How Cross-Border Securities Should be Recorded in Principle
To see what errors occur in portfolio data, denote A
ij
the amount of securities issued
by country j, owned by residents of country i 6= j. To measure A
ij
, the data collection
system of each country i covers some agents directly and others indirectly (IMF, 2002).
Financial corporations such as banks, investment funds, and insurance companies, are
4
On the asset side of official IIPs, statisticians isolate a fourth category, reserve assets, which includes
the portfolio securities and other assets held by central banks. In this paper, “securities” will always
include the fraction of reserve assets invested in securities.
5
See data gathered by the IMF for its Coordinated Direct Investment Survey. In 2011 for instance, 30%
of India’s inward direct investments came from Mauritius; 25% of Brazils’ came from the Netherlands;
60% of China’s came from Hong Kong and the British Virgin Islands.
6
See for instance IMF (2000).
5
