Volume 1, Issue 2 2010 Article 2
Journal of Globalization and
Development
Contagion, Liberalization, and the Optimal
Structure of Globalization
Joseph E. Stiglitz, Columbia University
Recommended Citation:
Stiglitz, Joseph E. (2010) "Contagion, Liberalization, and the Optimal Structure of
Globalization," Journal of Globalization and Development: Vol. 1: Iss. 2, Article 2.
DOI: 10.2202/1948-1837.1149
Brought to you by | Columbia University Law Library New York (Columbia University Law Library New York)
Authenticated | 172.16.1.226
Download Date | 6/18/12 8:19 PM
Contagion, Liberalization, and the Optimal
Structure of Globalization
Joseph E. Stiglitz
Abstract
Advocates of capital market liberalization argue that it leads to greater stability: countries
faced with a negative shock borrow from the rest of the world, allowing cross-country smoothing.
There is considerable evidence against this conclusion. This paper explores one reason: integration
can exacerbate contagion; a failure in one country can more easily spread to others. It derives
conditions under which such adverse effects overwhelm the putative positive effects. It explains
how capital controls can be welfare enhancing, reducing the risk of adverse effects from
contagion. This paper presents an analytic framework within which we can begin to address
broader questions of optimal economic architectures.
KEYWORDS: contagion, liberalization, globalization, capital markets, financial crisis
Author Notes: The author is indebted to Arjun Jayadev, Jose Antonio Ocampo, Jonathan
Dingel, Thuy Lan Nguyen, Sebastian Rondeau and an anonymous referee for helpful comments.
Financial support from the Hewlett and Ford Foundations is gratefully acknowledged.
Brought to you by | Columbia University Law Library New York (Columbia University Law Library New York)
Authenticated | 172.16.1.226
Download Date | 6/18/12 8:19 PM
Advocates of capital market liberalization have long argued that it would
lead to greater stability. Countries that are integrated into the global financial
system could, if they faced a negative shock, borrow from the rest of the world.
This would allow cross-country smoothing (Stiglitz 2000).
There is, by now, considerable evidence against this conclusion. (See
Kose et al. 2003.) Indeed, the IMF seemed surprised when the empirical evidence
contradicted this theoretical proposition. (See Kose et al. 2006). But they should
not have been. It should have been evident that developing countries that
liberalized have been more subject to crises and volatility. (See Ocampo et al.,
2008).
One reason for these adverse outcomes is that capital flows have not
behaved in the way that was hypothesized. Capital flows to and from developing
countries are often pro-cyclical. Given this, it would have been difficult to see
how capital market liberalization could have reduced variability.
One explanation of these procyclical capital flows is that “bankers don’t
like to lend to those who need the money.” Adverse shocks induce them to
reduce their lending.
There are other possible explanations for why liberalization would be
associated with an increase in, say, the volatility of consumption and the lowering
of expected utility. For instance, Stiglitz (2008) has developed a life cycle model
in which, without capital market liberalization, positive productivity shocks lead
to intertemporal smoothing; individuals save more, thus increasing wages of
future generations. In effect, the benefits of a productivity shock at time t are
shared with future generations. Capital market liberalization, then, may result in
greater volatility in consumption, as the generation in which the positive
productivity shock occurs gets to reap the full benefits for itself.
This paper explores a third set of explanations for the adverse
consequences of capital market liberalization, illustrated by global financial
crises, such as those that occurred in 1998 and 2008. A failure in one or more
countries can quickly spread elsewhere. Since the adverse effects of a downturn
can be great, it raises the question: can these periodic but highly negative effects
overwhelm the putative positive effects of income smoothing?
The existence of these adverse effects was itself a puzzle—at least within
the standard models that had assumed that by sharing risks, the effect of any
shock would be mitigated. Indeed, the notion that risk sharing would lead to a
more stable global financial system was one of the reasons that certain regulators
believed we were in a new era of the “Great Moderation.” For the first time, risk
was so widely shared that the world could undertake more risk, growing more
rapidly and more stably. Needless to say, things haven’t turned out the way that
was anticipated.
1
Stiglitz: Contagion, Liberalization and Globalization
Brought to you by | Columbia University Law Library New York (Columbia University Law Library New York)
Authenticated | 172.16.1.226
Download Date | 6/18/12 8:19 PM
There are, in turn, two possible explanations for why risk diversification
didn’t work in the way that many had hoped. The first is that the financial
institutions didn’t understand the risks that they confronted and/or that their
deceptive accounting practices, designed to mislead regulators, investors, and the
tax collector alike, also deceived themselves. The result was that they held on to
many of the toxic mortgages; risk was not in fact diversified and spread out.
There is a second explanation, and that is that the central models
employed by macro- and financial economists were fundamentally flawed.
1
They
assume a structure that says that if risk is widely diversified, the system will be
more stable and expected utility will be higher.
Curiously, the advocates of liberalization have never fully believed this.
For they have always worried about the risk of contagion once a crisis starts.
But, as we have noted, most of the mechanisms by which contagion occurs are
associated with financial market interlinkages.
2
This implies that there is a cost
associated with greater financial or capital market integration—the risk that a
problem in one country will lead to problems elsewhere. Even without, say,
financial market interlinkages, there can be extensive interdependencies through
which a shock in one part of the system can be transmitted to others. Any bank
making a real estate loan would have been affected by the breaking of the real
estate bubble. But financial market interlinkages can exacerbate the contagion of
problems from one economic unit to another.
The word “contagion” itself is associated with the transmission of
diseases; and the traditional way of reacting to worries about contagion is
“quarantining,” that is, breaking the links between the diseased individual and the
rest of society. The more integrated a society, the more rapidly can diseases
spread.
A coherent analysis of the desirability of financial and capital market
liberalization should, accordingly, take into account the benefits of risk sharing
when things work well—and the costs through contagion, when things don’t.
Remarkably, most of the literature has not done so, treating the benefits of
integration and the management of the risks of contagion as if they were
separable.
A moment’s reflection suggests one of the reasons that standard models
have gone astray: they make strong mathematical assumptions under which risk
sharing is always desirable. With convex technologies and concave utility
1
The first explanation focuses on a different flaw in the standard model employed by macro- and
financial economists—the assumption of rationality. Obviously, if the banks had been fully
rational—in the way that term is usually used-- they would not have retained so much of the risk.
2
There are other mechanisms, e.g. trade, but these work more slowly and are more muted; if trade
were the only mechanism for contagion, IMF intervention in the East Asia crisis would have taken
on a markedly different form.
2
Journal of Globalization and Development, Vol. 1 [2010], Iss. 2, Art. 2
DOI: 10.2202/1948-1837.1149
Brought to you by | Columbia University Law Library New York (Columbia University Law Library New York)
Authenticated | 172.16.1.226
Download Date | 6/18/12 8:19 PM
functions, risk sharing is always beneficial. Thus, the more globally integrated
the world economy, the better are risks “dispersed.” But if technologies are not
convex, then risk sharing can lower expected utility. While simplistic models
typically employed in economics assume convexity, the world is rife with non-
convexities. Information structures and externalities themselves give rise to a
natural set of convexities.
3
Learning processes (e.g. associated with learning by
doing) and R&D are naturally associated with non-convex technologies.
Bankruptcy, too, introduces a key non-convexity, as do the constraints associated
with information imperfections (moral hazard and adverse selection)
4,5
. The credit
market imperfections (in part arising from information imperfections) in turn give
rise to the financial accelerator, which in turn implies that the effect of a shock
can be amplified.
6
Concerns about bankruptcy can also give rise to a process of
trend reinforcement (Battiston, et al. 2009). For instance, a firm experiencing a
negative shock— pushing it closer to the bankruptcy brink—will have to pay
higher interest rates, implying that the likelihood of a further decline in net worth
has increased. Similarly, liquidity crises are associated with “forced” sales of
assets, leading to price declines, adversely affecting any bank lending on the basis
of collateral. But the declining value of assets induces a reduction in asset-based
lending, with consequent macroeconomic effects (Miller and Stiglitz,
forthcoming).
The natural models of contagion illustrate the role of non-convexities.
Assume a proportion of the population p has a disease, and that an infected person
communicates the disease with probability z to any non-infected person he comes
into contact with. Assume the degree of integration of the society is measured by
1/g, which measures how long it takes him to “bump into” another person, chosen
3
Information/knowledge can be viewed as a “fixed cost”—the greater the scale of production, the
greater the saving from a cost saving idea. Radner and Stiglitz (1984) describe a natural non-
convexity that arises in the “quantification” of information. Starrett (1972) describes a natural
non-convexity associated with externalities.
4
See, for instance, Arnott and Stiglitz (1988).
5
Credit market imperfections play a key role in the instability associated with financial market
integration on the part of developing countries. As we have noted, credit flows are pro-cyclical, in
contrast with the pattern predicted by “standard” theory, which suggests that they should be
countercyclical.
6
See also Greenwald, Stiglitz, and Weiss (1984), who show that information imperfections can
give rise to equity rationing. Greenwald and Stiglitz (1993) showed how this could led to a
financial accelerator: If firms’ production or demand is limited by their access to capital, the
effect of a positive shock that increases equity is amplified as, say, investment increases by a
multiple of that amount, and then multiplied further through the usual multiplier (Bernanke and
Gertler 1995). There is a growing literature on how these credit and collateral constraints can
give rise to bubbles and economic fluctuations. See Kiyotaki and Moore (1997), Gellegati and
Stiglitz (1992), and Miller and Stiglitz (forthcoming).
3
Stiglitz: Contagion, Liberalization and Globalization
Brought to you by | Columbia University Law Library New York (Columbia University Law Library New York)
Authenticated | 172.16.1.226
Download Date | 6/18/12 8:19 PM
