Globalization and Emerging Markets

Globalization and Emerging Markets: With or Without Crash?
Author(s): Philippe Martin and Hélène Rey

Source: The American Economic Review , Dec., 2006, Vol. 96, No. 5 (Dec., 2006), pp. 1631-
1651
Published by: American Economic Association
Stable URL: https://www.jstor.org/stable/30034988
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
is collaborating with JSTOR to digitize, preserve and extend access to The American Economic
Review
This content downloaded from

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC
All use subject to https://about.jstor.org/terms
Globalization and Emerging Markets: With or Without Crash?
By PHILIPPE MARTIN AND HELENE REY*
We analyze the effects of financial and trade globalization on the likelihood of

financial crashes in emerging markets. While trade globalization always makes
crashes less likely, financial globalization may make them more likely, especially
when trade costs are high. Pessimistic expectations can be self-fulfilling and lead to
a collapse in demand for goods and assets. Such a crash comes with a current
account reversal and drops in income and investment. Lower-income countries are
more prone to such demand-based financial crises. A quantitative evaluation shows
our model is consistent with the main stylized facts of financial crashes in emerging
markets. (JEL F12, F32, F37, F41, 016)
Do emerging markets reap the benefits of of emerging market financial systems to capital
financial globalization, enjoying increased in- mobility. Charles Wyplosz (2001) finds that ex-
vestment and a better ability to diversify risk? ternal financial liberalization is considerably
Or do they face a higher likelihood of financial more destabilizing in developing countries than
crash as more capital flows in? The empirical in developed economies. Graciela Kaminski
literature supports both possibilities. On the one and Sergio Schmukler (2001) show that stock
hand, a number of papers in finance show that markets become more volatile in the three years
financial opening in emerging markets leads to following financial liberalization but stabilize in
a decrease in the cost of equity capital and can the longer run.
have a positive effect on domestic investment.' Interestingly, recent empirical work shows
On the other hand, a voluminous literature sur- that goods trade openness also influences the
veyed by Joshua Aizenman (2004) emphasizes frequency of crashes in emerging markets, but
the risks of liberalization and the vulnerability in the opposite direction to financial openness.
Eduardo A. Cavallo and Jeffrey A. Frankel
(2004) find that trade openness (instrumented
* Martin: University of Paris 1 Pantheon Sorbonne by gravity variables) reduces the vulnerability
Economie, Paris School of Economics, 106-112 bd de of countries to sudden stops. The Argentina of
I'H6pital, 75647 Paris Cedex 13, France, and Centre for the 1990s is often presented as a typical exam-
Economic Policy Research (CEPR) (e-mail: philippe. ple of a financially open economy relatively
martin @univ.paris 1.fr); Rey: Department of Economics and
Woodrow Wilson School, 307 Fisher Hall, Princeton Uni- closed to goods trade. It has suffered heavily
versity, Princeton, NJ 08544, CEPR, and National Bureau from sudden stops (see Guillermo A. Calvo et
of Economic Research (e-mail: hrey@princeton.edu). We al., 2003; Calvo and Ernesto Talvi, 2004).
thank two anonymous referees for very helpful comments These contradictory effects of financial and
on a previous version, Daniel Cohen, Pierre-Olivier Gourin-
trade globalizations are illustrated in Table 1.
chas, Gene Grossman, Galina Hale, Olivier Jeanne, Enrique
Mendoza, Richard Portes, Lars Svensson, Aaron Tornell, as We report the average number of financial
well as participants at many seminars. We also thank Gra- crashes per year for developed and emerging
ciela Kaminsky and Sergio Schmukler for the stock market economies, dividing each group along the di-
data. Rachel Polimeni provided excellent research assis-
mensions of financial and trade openness.2
tance. This paper is part of a Research Training Network on
"The Analysis of International Capital Markets: Under-
standing Europe's Role in the Global Economy," funded by
the European Commission (Contract No. HPRN-CT-1999- 2 More precisely, emerging markets are defined as those
0067). with GDP per capita equal or below that of South Korea.
1'See, for example, Geert Bekaert et al. (2005), Peter The sample coverage for those countries starts at the earliest
Blair Henry (2000), and Anusha Chari and Henry (2002). in 1975 and ends in 2001. A crash is defined as a monthly
The macroeconomic literature finds more tenuous evidence drop in the stock index (in dollars) larger than two standard
that financial opening contributes to long-term growth. See deviations of the average monthly change. We divided the
Sebastian Edwards (2001), for example. sample into periods for which countries were financially
1631

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC
1632 THE AMERICAN ECONOMIC REVIEW DECEMBER 2006
TABLE 1-FREQUENCY OF CRASHES AND OPENNESS
Trade in goods
Emerging Developed
Closed Open Closed Open
Financially closed 0.40a 0.35b 0.09a 0.15b 0.07b 0.09a 0.10b
Financially open 0.78a 0.76b 0.55a 0.57b 0.05b 0.06a 0.14b
a Sachs-Warner measure of openness.

b Trade/GDP measure of openness.
Table 1 suggests that opening to capital impact

differential move-
of financial globalization on
emerging
ments is very positively correlated withmarkets and developed economies.
the fre-
We emphasize
quency of crashes for emerging markets, butthenotkey role of demand and
for industrialized countries. Trade openness
market size in driving both the positive effect of
(whether measured by the trade-to-GDP ratio or
financial integration on an emerging economy
following Jeffrey Sachs and Andrew
and its negative
Warner,
consequences.
In our
1995), however, is associated with a model,
largethe world consists of one
de-
crease in the frequency of crashes emerging market and one developed economy
for emerging
markets.3 Hence, according to Table 1, being
which differ an
only in their productivity level. In
emerging market open to financial flows entrepreneurs
both countries, while operating in mo-
closed to goods flows maximizes the frequency
nopolistic goods markets decide whether to fi-
of crashes. nance risky fixed-sized investments, sell shares
The contribution of our paper is to present a of these investments on the stock exchange, and
general framework in which these contradictory acquire shares in other risky ventures developed
effects of financial and trade liberalizations can at home or abroad. Entrepreneurs may turn pes-
be reconciled. We can also make sense of the simistic and expect low levels of aggregate in-
vestment. Due to home bias in goods trade,
negative prospects regarding investment trans-
open and financially closed, following Kaminsky and late into low expected income and demand for
Schmukler (2001). Hence, among our 62 countries (34 goods, low profits, and hence low demand for
emerging countries), 31 appear twice as they changed status
during the sample years. We also classified countries in
domestic assets. This validates their pessimistic
terms of their openness to trade. We chose two widely used priors and deters them from developing risky
measures of trade openness: (a) The classification by Sachs investments. The home bias in financial markets
and Warner (1995) extended to the 1990s by Romain Wac- in turn implies that the fall in income in the
ziarg and Karen H. Welch (2003), who provide liberaliza-
tion dates for a broad set of countries. This measure of emerging market also leads to a fall in domestic
asset
openness is based on trade policies. Some countries have demand and prices. In this equilibrium,
liberalized financial and trade flows at different dates and asset prices and investment collapse, income
hence appear in different cells of the table. (b) The average decreases, and a capital flight occurs since do-
of exports plus imports over GDP during the period con- mestic agents buy shares in the developed coun-
sidered. This measure of openness captures the degree of
independence of the economy on local demand. We call
try stock exchange. The circular causality is
open (respectively closed) a country whose openness ratio is magnified if trade costs are high, since firms'
above (respectively below) the median of its group. The profits and dividends in more closed economies
trade openness ratio cutoff for financially open emerging are more dependent on the level of local de-
countries is 63 percent of GDP. For the group of financially
mand. They are therefore more at risk when
closed countries, the trade openness ratio cutoff is 54 per-
cent. For more details, see the Data Appendix at http:// expectations turn pessimistic.
team.univ-paris 1 .fr/teamperso/martinp/table.pdf. The likelihood of a crash is higher at an
3 This is not the case for developed countries, for which intermediate degree of financial segmentation.
the frequency of crashes is low overall. For developed When financial markets are perfectly integrated,
economies, we use the ratio of trade to GDP as our measure
of openness. According to the Sachs-Warner measure, only
no financial home bias exists and arbitrage
two industrialized countries are classified as closed at some equates asset prices, so that local income con-
point, so we do not report the result. ditions do not alter the cost of capital in the

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC
VOL. 96 NO. 5 MARTIN AND REY: GLOBALIZATION AND EMERGING MARKETS 1633
emerging market. Symmetrically, crisis

ifisfinancial
driven by a collapse in demand when
asset markets are very segmented goods and financial markets are segmented by
internation-
ally, emerging market agents have trading costs and
no choice asset markets are incomplete.
but
to invest at home. This rules outOur theory flight
capital is therefore complementary to the
existing
and multiple equilibria, at the cost literature on financial crisis.6 Our
of ineffi-
ciency in capital markets. model has multiple rational expectations equi-
libria, like in in-
In our setting, financial globalization Chang and Velasco (2001), for
creases asset prices, investment, example,
and income where internationally
in illiquid banks
the emerging market, but only when may be interna-
subject to a run. But in our setup,
self-fulfilling
tional trade costs are low. When emerging expectations operate through
mar-
kets start opening their financial investment
account but behavior
are and endogenously incom-
closed to trade in goods, they are plete
more asset markets.
prone to Our framework has poten-
tially from
financial crashes. This comes chiefly important policy implications: only once
their
having a lower income than developed coun-are well integrated in world
emerging markets
goods
tries and from their dependence onmarkets should they increase signifi-
local de-
mand due to market segmentation. cantly
Thetheir degree of financial openness. We
demand-
based mechanism also implies that ourpoint
make this modelin a formal model, where any
degree
has the potential of generating quick of frictions
recovery inon the goods and financial
the aftermath of crises. markets and their interactions can be analyzed.
Our work is related to the literature
Weon finan-
present the model in Section I. Section II
cial crises in emerging markets and the
describes sudden
properties of the equilibrium in
stops. Calvo (1998) explores the"normal
role oftimes," while Section III investigates
credit
frictions to explain sudden stops.the conditions
Enrique necessary for a financial crash to
Men-
doza (2004) and Mendoza and Katherine
occur. Section IVA.performs a quantitative eval-
Smith (2002, 2004) show within an uation of our model. Finally, we draw some
equilibrium
conclusions
business cycle framework that small in Section V.
productiv-
ity shocks can trigger sudden stops in the pres-
ence of credit constraints when an economy I.is Model
highly leveraged.4 Philippe Aghion et al. (2004)

also use a model with credit frictions and find Ours is the only model known to us that
that countries with intermediate levels of do- analyzes jointly home market effects in goods
mestic financial development and free capital and asset markets and their interactions. Firms
movements are more prone to macroeconomic sell a monopolistic good in international mar-
volatility. In contrast to these papers and mostkets where trade is costly. They also sell claims
of the existing literature, however, a financialon their expected (risky) profits on international
crisis in our model does not come from the stock markets segmented by financial trading
existence of credit constraints on capital mar- costs. Our modeling strategy is simple enough
kets and/or balance sheet effects." Neither isto it handle both types of frictions in a tractable
caused by moral hazard (as in Ronald I.
way.
McKinnon and Huw Pill, 1999, and Giancarlo
Corsetti et al., 2001). Instead, in our setup, the A. Technology and Trading Costs
There are two countries, E (emerging) and I

(industrialized), and two periods. All decisions
4 See also Mendoza (2002) and the survey of Cristina
Arellano and Mendoza (2003). For a view of Asian crises
are taken in the first period. At the beginning of
based on implicit fiscal liabilities, see Craig Burnside et al.
the first period, L identical agents per country
(2001). Kiminori Matsuyama (2004) presents a model in
are each endowed with one unit of labor and
which borrowing constraints interact with financial global-
one
ization to produce an endogenous degree of inequality firm. There are two sectors: a perfectly
across otherwise identical countries.
5 See Paul Krugman (1999), Ricardo J. Caballero and
Arvind Krishnamurthy (1998), Lawrence J. Christiano et al. 6 We emphasize that the other channels studied in the
(2004), Martin Schneider and Aaron Tornell (2004), Rob- literature may be important, as well, to explain emerging
erto Chang and Andres Velasco (2001), Bernhard Paasche market crises. Our model is certainly compatible with all of
(2001), and Luis Felipe Cespedes et al. (2004). them.

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC
competitive constanttoreturns-to-scale
all other agents in the economy sosector
that ag-
with zero trade cost, which serves
gregate income as the
is not affected nu-
by it.9 The value
meraire, and a monopolistically
of a firm is therefore thecompetitive
expected payoff of the
sector with iceberg trade investment
costs rE =7'r.
PEZE Transport
- 1/2 zZQ - F. The
costs and trade policies investment both affect good is produced
77. Eachwith a firm
Cobb-
corresponds to one variety, so function
Douglas production thatwith the total
a share (1 -
number of varieties in the world is 2L. Both a) for labor and a for the composite good made
sectors use labor as their only input. The only
of all varieties of the monopolistic sector (see
below).
difference between the two countries is labor pro-
ductivity, which we assume is equal in both sec-In the second period, there are N exogenous
and equally likely states of nature, and the re-
tors and higher in the industrialized country than
in the emerging market. Free intersectoral laboralization is revealed at the beginning of that
period after all decisions have been taken. As in
mobility, perfect competition, and free trade in the
constant returns to scale good imply that wage Daron Acemoglu and Fabrizio Zilibotti (1997)
and Martin and Rey (2004), the technology im-
rates wl (in the industrialized country) and wE (in
the emerging market) are equal to the marginal plies that each investment gives dividends (the
productivity of labor. In the monopolistic good operating profits of the first period) in only one
state of nature. In all other states of nature, the
sector, labor productivity is also given by w1 and
wE, so that the marginal cost of production in operating profits of the first period become zero.
numeraire units is equal in both countries. The payoff structure is such that an investment
In the monopolistically competitive sector, in country E yields dE if the corresponding state
firms earn operating profits in the first period. of the world is realized, and zero otherwise.
To create a diversification incentive at both the Hence, investments in the two countries have ex
national and international level, we introduce antea expected dividends, dE/N and d/N. All
simple source of uncertainty. This will inducerisky claims to operating profits are traded on
agents to diversify in equilibrium their owner-the stock market at the end of period one, so that
each claim corresponds to an Arrow-Debreu
ship of firms.7 We assume that first-period prof-
its of monopolistic firms do not always asset. This gives agents in both countries a
materialize in dividends to shareholders in thestrong incentive to diversify and buy shares of
second period. Without firm-specific invest- both foreign and domestic investments. We as-
ments, these profits vanish, due, for example, sume
to that the number of states of nature N is
mismanagement at the firm level. When invest- large enough so that N > Z" where Z"
ment is performed by the firm, profits are dis-L(ZE + ZI) is the total number of investments/
assets issued in the world. N - Z" is therefore
tributed to shareholders with some positive
the endogenous degree of incompleteness of
probability. The price of a share that is a claim
to risky profits is given by pE. The total cost financial
of markets. No duplication occurs in
investment is F + 1/2 ZQ, where zE is the num- equilibrium, so that each investment/asset in the
ber of investments undertaken by a firm in the world is unique.10 This modelling introduces a
emerging market and Q is the price of the in- simple incentive for agents to diversify their
vestment good.8 The marginal cost of undertak- portfolios across firms in an otherwise standard
ing investments rises as the firm decides to do
more investments. In addition, a fixed cost F has
to be paid to start investing. We assume that this 9 If the fixed cost has an impact on aggregate income, the
main results of the model are unaffected. However, the
fixed cost is paid individually by each investor
results are analytically less tractable.
1O This is because as long as some states of nature have
not been covered, the price of an asset associated with these
7 Foreign agents cannot operate production technologies states will always be higher than if the agent were to
in the domestic country; hence, there is no FDI in ourreplicate an existing investment/asset. This could obviously
model. They can, however, invest in claims to domestic lead to some exercise of monopolistic power in the asset
risky profits. market, but we assume that investment developers do not
8 Industrialized country agents face a similar investment exploit it. The issue of "financial" monopolistic competition
cost function. We discuss in the working paper version in this type of framework is dealt with in Martin and Rey
(Martin and Rey, 2002) how our results would be affected (2004), who show that it creates another source of financial
by a more general convex cost function. home bias.

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC
where cE;(n) denotes in

monopolistic competition framework the second-period
thecon-
goods market. sumption min one of the N states of the world and
E is the expectation sign.takes
At the end of the first period, consumption cEY is the consump-
place and shares are sold on each of
tion ofthe
the CRS goodstock
with a share 1 - / in the
markets. These shares can be traded internation-
utility wants
ally, but an agent in either country who function to
while /. is the share of the com-
posite good CE1 made of all varieties produced
in thea world:
buy assets in the other market must pay financial
trading cost. This cost, essential for our results,
may capture government regulations (2) on capital
flows, differences in regulations in accounting,
banking and commission fees, exchange rate
transaction fees, and information costs. The pres-
CE1
ence of these costs translates into a home = clin
bias /l' -
i=1 cj=1
]j-1/"
asset transactions and holdings." We denote the
transaction costs on financial assets '7,oand> assume
1 is the elastici
that they take the form of an iceberg cost.
goods, This while cEij an
implies that the transaction fee is paid of domestic and im
in shares.
Agents have to buy (1 + r7,) > 1 units This
of sharescomposite goo
tion
to receive one share.12 We interpret financial and investmen
glob-
alization as a process through which these The first-period bu
trans-
action costs are reduced. in E is
B. Utility and Budget Constraints (3)
L L
We assume that the utility of an agent in each
country is given by the nonexpected utility func-
tion introduced by Larry G. Epstein and Stanley E.
YE
=1 j=1
= C
Zin (1989) and Philippe Weil (1990). This allows
LZE LzI
the intertemporal elasticity of substitution (which
we assume to be one for simplicity) to be different + PEkSEk + (1 + 7F)PISEI
from the coefficient of relative risk aversion 1/e. k=l l=1
In the emerging market, the utility of a represen-

tative agent is given by: = wE+ 'rE+ T,
where YE is the emerging market per capita

(1) ElCI income of the first period, VEi is the price of the
ith variety produced in the E market, vii is the

price of the jth variety produced in the I market,
+ 3 In and rE is the investment payoff. Asset prices

nil N CEz(2n) 1-le are denoted by pEk and p, and sEk and sE1 are
demands for shares of risky investments devel-
oped in the emerging market and in the indus-
" There is strong empirical evidence for home bias and
trialized country, respectively. T is the transfer
for the role of such costs in generating at least part of the
bias. See Richard Portes and Rey (2005) for the importance (in equilibrium equal to F). The budget con-
of information costs, and Mendoza and Smith (2004) for straint in the industrialized country is analo-
another model featuring trading costs on asset markets. gous. In period 2, income and consumption
12 Iceberg transaction costs are borrowed from the trade
come only from dividends of shares purchased
and geography literature. See Martin and Rey (2004) for a
more precise description. This modelling allows the elastic-
in the first period. Hence, the budget constraint
ity of substitution between assets to be the same for all for an agent in E is:
agents, and does not require the formal introduction of an
intermediation sector. Roger H. Gordon and Lans Boven-
(4) cE2 = dESEk, if k E [1, LZEI;
berg (1996) use a similar type of proportional transaction
cost on capital flows, and focus on the cost of acquiring
information about foreign countries. dsE, if IE[1, Lz]; 0 otherwise.

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC
We can therefore rewrite the are

As investments utility of
ex ante symmetric, thean
de- agent
in the emerging marketmands
as for each asset in a given country are
identical.16 We call SEE (SEI) the demand for
1 shares of a "typical" asset in the E (I) market by
(5) E(UE) = ln[c C,] + /( - - e) an emerging market agent. Similarly, we denote
by cEE and cEl the first-period demand by an
LZE emerging market agent for a good produced in E
and I, respectively. Because of symmetry,
+/3 In 1 (dEsEk)- 1/s
within each country, all assets have the same
k=1
price, denoted
marginal costs by PE and
in units ofpI,
therespectively.
numeraire areSince
Lz' ] /(-l/e)
equal to one in both countries, and the elasticity
-+ E (dlsEI)1- l/e
/= 1
of substitution between varieties o- is the same
for consumers and firms, all firms in the world
choose the same price for the monopolistically
The utility and budget constraint of an competitive goods. That price, equal to the mar-
agent in the industrialized country are sym-
ginal cost multiplied by the markup, is given by
vE = -(O- - 1). For notational simplicity,
metric. In the second period, this utility func-
tion is similar to the one introduced by we drop the expectational sign in what follows.
Avinash K. Dixit and Joseph E. Stiglitz to
represent preferences for differentiated prod- C. Definition of Equilibrium
ucts. In fact, e can be interpreted as the elas-
ticity of substitution between assets. In what An equilibrium is defined by a set of good
follows, we impose e > 1 to have financial and asset prices [vE, V1, PE, PI], consumption and
home bias and realistic asset demands.'3 This
investment allocations [CE1, CI, CEY, cyy, e, zi,
restriction on e mirrors the standard assump-
tion in the differentiated products literature cE2(n), Cl2(n)], and portfolio shares [SEE, SEI, Si,
SlE] such that:
that the elasticity of substitution between dif-
ferent varieties oa is greater than one. This (a) [CEl, CEy, SEE, SEI, CE2(nl)] maximize UE
restriction also implies that assets are substi- subject to E's budget constraints (equations
tutes rather than complements, as in Acemo- (3) and (4)) taking prices as given.
glu and Zilibotti (1997).14 Imposing e > 1 has the (b) [C,1, cy, st,, sIE, c2(n)] maximize U, sub-
additional benefit of ruling out any problem for ject to I's budget constraints (the analogue
the states in which consumption is zero in the of equations (3) and (4)) taking prices as
second period due to market incompleteness."5 given.
Agents in both countries choose consumption
(c) [vE, VI, ZE, ZI]
investment maximize
payoffs profits
of firms andprices
taking the
(cEY, CE, and cty, C/1), and firms choose
ment (the number of investments per firm are zE invest- and investment decisions of other firms as
and zi) at the beginning of the first period. They given. A firm invests if and only if its ex-
form expectations about the number of invest- pected investment payoff ri = Pii -
ments in which other firms will engage, since
/2zZQ - F is nonnegative for i = {E, I}.17
this will have an impact on the price of the (d) Asset markets clear: LSEE + L(1 + F)SIE =
assets they will sell at the end of the first period. I and LS, + L(1 + F7,)sEI = 1.
(e) The world resource constraint is verified,
13 See Section II. The demand for foreign assets de-

which implies: L[CEY + cy + LcEE +
creases with transaction costs 7F for any e. But using
iceberg trading costs (paid for in shares) implies that the
demand inclusive of transaction costs (which determines the 16 In each country, agents are different in the sense that
they hold different assets but they choose identical portfo-
equilibrium on the stock market) would increase with 7, if
e were to be smaller than one. lios and consumption patterns.
14 In Section IV, we review the existing empirical esti- 17 We focus on symmetric equilibria in which all or no
mates for e: they range from 1 to 12. firms in a country invest. Equilibria in which only a portion
15 When we introduce a safe asset (see Section IV), this
of firms invest are studied in the working paper version of
issue of course does not arise any longer. Martin and Rey (2002).

f:ffff:ffff:ffff:ffff:ffff:ffff on Thu, 01 Jan 1976 12:34:56 UTC
Le (1 + 77) + LcI + LCIE(1elusive

+ of TT)
shares +
that are used to pay the transac-
tion costs) can be written for each asset as:
(ZE
WI). 8 z )Q( - a)Ir + dE + di] = L(wE +
(f) Expectations are rational. 1 /3 YE
(7) 1 =
II. When Things Go Well (7) PE 1 + E + ZIlF(q/d)E- 1
We first solve the model in the optimistic

ZI +yz,!(q/d)1
ZE4F(q/d) - -E )
case, when firms of the emerging market expect
others to invest in a positive number of projects.
We define q = PEPI as the relative asset price,
and d = dEd, as the relative dividend. The 1 (3 ( yl
Pi 1 + (3 z! + ZE4F(q/d) -E
budget constraints and the first-order conditions
of an emerging market agent imply the optimal YE F(q/d) -I
consumption demands: ZE + Z14F(q/d)E-
There are L(zE + z1) such equilibrium condi-

(6) EY =(1(1+3) '
- 1-)yE tions. In the parentheses, the first term repre-
sents the demand coming from domestic agents
/L(o" - 1)yE and the second term foreigners' demand (inclu-
sive of transaction costs). These equations im-
CEE = L(1 + 3)[1 + ,1
ply a financial home-market effect: local
income has a more important impact on the
( - 1)yE(l + 1T
CEI = L(+3)+4; local asset market than foreign income, as long
as 4, is less than one, i.e. as long as some
transaction costs exist.
IYE 1 The dividends of the second period are the

operational profits of the first period. Hence,
SEE = 1 + LpE [ZE + (FI1(q/Id) ]'
they are equal to sales (to consumers and firms)
divided by the elasticity of substitution:
P3yE (1 + rF)-E(q/d)
SEtl 1 + p LpI[ZE
(8) + FZt(q/d)E- ]'
where
of trade0openness
, = (1 + d /(YE
7)1-
dE and ,-f=YIkT)
1 1+
(1isa(z +a measu
7F)1- a
a measure of financial openness. The dem
for foreign shares (sE)=o-(1 + P)(1 + +r) +2 o with
decreases (1 + +r) finan
transaction costs.
At the optimum, the marginal cost of invest- +_ P)(1
dl '(Y, + YEPT)
-- 1 + 4Tze)Q
+2 o'(1 + 7,)
ing equals the marginal benefit: zEQ = PE.19
These equations imply a trade hom
The demands for shares sEE and sE1 increase
with income and decrease with the total number effect: local income and investment ha
of investments/assets. Analogous conditions important impact on sales and profit
hold for the industrialized country. For all firms firms than foreign income and inves
in the economy to invest, the expected payoff long as k, is less than one, i.e., as lon
costs exist. Because our theoretical
must be positive:
We normalize PEZE of
the number - /2 Z -so1/2
shares thatpthe
__ Q F. requires only one source of trade hom
stock market equilibria in the two countries (in- effect, we assume from now on that
that the investment good requires on
Q = 1, and profits come only from
18 We have used the cost minimization program of firms
consumers. This allows us to derive all results
to derive their demands for the investment good.
1 Q = a-a(l - a)"-'[r/(o - 1)],[L(1 + T)l-a]c- 1) analytically. We come back to the more general
is the price of the investment good. case with a > 0 in the quantitative section.

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC
A. Equilibrium Relationship
home biasbetween Asset
in goods. In turn, the surge in local
Prices, Dividends, and Income
demand Shares
increases relative operating profits and
dividends. Lower trade costs raise relative prof-
As world income is fixed,20
its and d asit proves
long as s, < a/2.conve-
Together, (10) and (11) provide a nonlinear
nient to define sr = yE(yE +between yI) as the share of
the emerging market in relation
world income. the share ofFrom
income in the
the
budget constraint and the optimal investment
emerging
rule, we get the first equilibrium
q. market
We call this sr andrelation
positively the relative
sloped relationasset price
the be-
qq
tween the relative income and the relative asset schedule. Two effects are at work: first, an
price q, which we call the yy schedule: increase in income raises demand (mostly) for
locally produced goods due to home bias in
s,(2 + p) P3 trade (4, < 1), thereby increasing profits and
(9) s, = 2(1 + 13) 2(1 + P3)(1 + dividends
q-2)' (trade home-market effect). This, in
turn, increases the demand for assets and their
relativeof
where s, = wE(wE + wI) K2 is the share price. Second, an increase in income in
the world
the emerging market wage income in the emerging market leads to an increase in
wage income. The equilibrium yy relation savingim-
which, as long as markets are segmented
(4F <price
plies that an increase in the relative asset 1), falls disproportionately on domestic
q generates an increase in relative income sy.
assets (financial home-market effect). This also
The reason is that emerging market investments
increases the relative price of emerging market
assets.
are sold at a higher price and more investments
are started.
Using the optimal investment rule, equation B. Globalization and Asset Prices
(7) of the stock market equilibrium gives
In this section, we show that trade and finan-
cial liberalizations may have very different ef-
sy(1 - q4) + q4)F(q/d) - + 44
(10) q = F(q/d)E syq(1 increasing
- 4) trade openness is always positive,
fects on asset prices and income. Whereas
If 4F = 1 (zero transaction costs on assetopening

trade), the capital account has an ambiguous
effect.
then q = d -'a". This implies quite intuitively
that without any financial segmentation, theIn Figure
rel- 1, we illustrate the equilibrium as
ative price of assets depends only on thethe intersection of the yy and qq schedules. The
relative
dividend and the elasticity of substitution relative price of assets q is less than one as long
but
not on local demand. as the financial or goods markets are not per-
Using (8), we can derive the relative dividend
as fectly
a/2. Theintegrated
difference (4F + 1 price
in asset or , 4is1)higher,
and swthe<
larger the differential in productivity. The two
sy(1 - T) + 4'T curves cross only once, so that only one "good"
(11) d - sY(1-)
1 - st(1 - abr)" equilibrium exists. Trade integration (an in-
crease in 4) is easily analyzed. As long as s, <
a/2, the fall in trade costs implies a rightward
If 4, = 1 (zero transaction costs on goods
shift of the qq curve (asY/8h, < 0 for a given q
trade), then d = 1. This implies, also quitealong the qq curve). The yy curve, meanwhile,
intuitively, that in the case of perfect goods,is unaffected. The effect, shown in panel A of
market integration operating profits and there-Figure 1, is an increase of the emerging market
fore dividends do not depend on local incomes. relative asset price and income share, for any
level of financial integration.
An increase in the relative income of the emerg-
ing market raises local demand more if there is Intuitively, lower trade costs increase profits
and dividends of firms in the emerging market:
from (11), ad/la, > 0 as long as s, < a/2. This
20 From the stock market equilibria we obtain that
in turn increases the demand for emerging mar-
PEZE + PIZI = 3(YE + yI)/(1 + P3). Using the optimal
ket assets
investment rule and the definition of world income, weand their relative price, which gener-
ates a rise in relative income. Due to the
therefore have L(yE + YI) = 2L(1 + P3)(wE + w1i)/(2 + P3).

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC
Sy of perfect goods market integration, as d = 1

and q < 1 (whenever financial integration is not
1/2 perfect). Intuitively, in that case, financial open-
ing enables agents in the industrialized country
to buy the cheaper emerging market assets. For
high trade costs, however, the profits of emerg-
ing market firms are lower than in the industri-
A
B alized country, making emerging market assets
relatively unattractive. The relation between asset
yy
prices and financial liberalization is U-shaped,
so that financial opening may actually lead to a
decrease in the demand for emerging market
qq, assets (capital outflows), a decrease of their
qq' q price, and more market incompleteness. Finan-
cial liberalization with low trade costs has
1
A the same positive effects on asset prices and
SY income as trade integration. It can also, there-
fore, be illustrated by panel A in Figure 1. In
1/2 contrast, when trade costs are high, financial
liberalization leads to lower asset prices and
income in the emerging market, as illustrated in
qq' panel B.
qq
A
C. Globalization and the Current Account
yy We now study the impact of globalization on

the first-period current account of the emerging
market. The current account is the difference
q between the country's production and the mar-
ket value of investment and consumption:
1
B
(12) CAE =L(YE- E1+j)

FIGURE 1. LIBERALIZATION IN THE EMERGING MARKET
convexity of the investment cost function, the

total number of assets is increasing in q. Hence
:L [3+/3 1+q
(w - wtq21
trade integration also alleviates financial market
incompleteness, as measured by N - Z", and The current account deteriorates as the relative
therefore reduces the volatility of consumption asset price in the emerging market increases.
in the second period. Hence, trade integration (an increase in 4)
In contrast, a fall in financial transaction costs always implies an increase in the current ac-
has an ambiguous effect on asset prices, relative count deficit. Financial integration (an increase
income, and market incompleteness. In Appen- in F) also leads to a current account deficit if
dix A we give the exact condition for which an trade costs are not too high. In that case (see
increase in 4, (increase in financial openness) previous section), liberalizing capital move-
leads to a rise in q. A sufficient condition is that ments generates net capital inflows in the
the relative return of the emerging market asset emerging market as agents in the industrialized
d/q is more than one. Interestingly, this will be economy take advantage of the lower asset
the case for low enough trade costs. The condi- prices in the emerging market. If trade and
tion is verified, for example, in the extreme case financial transaction costs are sufficiently low,

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC
In a crash,account
the emerging market current the emerging market
is relative div-
in deficit
in normal times. idend increases with lower trade costs on goods
markets and with labor productivity in the
III. Self-Fulfilling Expectations and Financial emerging market.
Integration: When Things Go Wrong A crash occurs if the expected payoff of
investing is negative:
Until now, we focused on the equilibrium in
which there is positive investment in both coun-
Ec2 (wE+w)qF<.
tries. The decision to invest depends, however, (15) 2 + pE )q~ - F < 0.
on the expected price of assets at the end of the
period, and therefore on the strategies of all
other firms. We now investigate under what The investment payoff is U-shaped as a function
conditions a crash driven by self-fulfilling ex- of F. Inequality (15) can therefore be satisfied
pectations can occur. We define a crash as an for intermediate levels of financial transaction
equilibrium in which no single firm has an in- costs.
centive to invest, given that no other firm is Multiple equilibria exist if and only if q2 <
investing. The condition for this to happen is q21(1 + q2). This guarantees that, for a given set
of parameter values, a "good" equilibrium ex-
E(TFc)
the = EE(PFcZc-
index 1/2crash
c denotes the ZEc equilibrium.
- F) a 0, where
In ists whenever ZE > 0 and a crash equilibrium
that case, the expected asset price is low enough
exists whenever ZEc = 0.22, 23 For this condition
that no firm deviates from the zero-investment to be verified, the fall in price during a crash
equilibrium.21 must be large enough. Using (13), it can be
Expected aggregate income in the emerging checked that the crash equilibrium cannot occur
market in a crash is T(LyEc) = LwE, since in the absence of capital flows (4, = 0), as qc
expected financial wealth is zero. This affects goes to infinity because agents can save only by
the expected relative demands for assets in the buying domestic assets.2 This puts a floor on
emerging and industrialized economies. Using the demand for domestic assets and on their
the stock market equilibrium (7), we show that expected price since capital flight is impossible.
the expected relative asset price in crash is At the other end, in a situation without frictions
(4F T ), qc = 1, so arbitrage implies that
agents in the industrialized country would rush
(13) qc to buy assets in the emerging market in the
event of a crash. This rules out the possibility of
a crash in the emerging market altogether.
- s,(2 + p3)(GF - F2(1+
F) + 2(1 + p)I3(d P_
dl /
Hence, a crash is possible only for intermediate
levels of the financial frictions and for high
where we drop
enough levels the
of trade costs. expecta
now on. The Circular causation is at work. If firms
relative believe
price
cial that other firms will undertake
globalization atno investment,
low l
increases then theyglobalization
with expect aggregate income in the
,F. The emerging market
relative at the end of the period to be
dividend is
low. Lower expected income entails lower sav-
(14) ings and a lower demand for assets. When fi-
s, (2 + P3)(1 - 'T) + 2(1 + P)3b> 22 As mentioned before, we are limiting our analysis to
d= 2(1 + 3) - sw(1 - ,)(2 + 13) symmetric equilibria in which all investors in each country
behave similarly.
23 In the absence of an equilibrium selection device, our
model has nothing to say about the transition between
21 ZEc in this condition is the investment that would be equilibria. We also cannot perform meaningful welfare
made by a single "pessimistic" firm if it anticipates that nocomparisons. These drawbacks are common to all multiple
other firm will invest. The optimal investment rule ZEc = equilibrium models.
24 This also implies that an equilibrium where both coun-
E(PEc) still applies. This firm is small (L is large) so that its
decision does not affect aggregate income or investment. tries are in crash is not possible.

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC
nancial markets are segmented and assets are

imperfect substitutes, this fall in demand for
assets affects local assets disproportionately.
'i.c
This in turn generates a low relative asset price

in the emerging market (financial home bias
',,low trade costs
effect). Trade costs magnify this effect since a
o QSF
crash that lowers income in the emerging mar- I
ket also lowers demand for goods. This falls

more than proportionately on goods produced in
the emerging market, so that expected n't,,high trade costs
operating
profits in the emerging market alsoFIGURE fall. This
2. CRASH AND TRANSACTION COSTS
home bias in trade in goods also contributes to
the fall in dividends and asset prices.
Is the emerging market more vulnerable to a
Figure economy?
financial crash than the industrialized 2 depicts payoff functions in crash as
We can compare the payoff level ofof afinancial
a function single openness 4F. Crashes can
occur in the
"pessimistic" investor in the emerging area below the zero line, whose
market
exact
(ZEc = 0) given in equation (15) to itsposition
analoguedepends on the level of F. For a
given level of trade openness, countries with
in the industrialized country (zc = 0). We find
higher levels of productivity (higher wages) are
that 'rl > rEc as long as ', or
"pessimist" payoff function of the
+F < 1. toThe
lessindustrialized
vulnerable crashes. For a given level of
productivity,
country is always above the emerging countries that are more open to
market
one. Due to the dual home biastrade (inintrade
goods areandless vulnerable to crashes.
finance), the demand for assets in theA financial
rich crash
mar- in the emerging market is
characterized
ket, even when depressed by pessimistic by low asset prices, investment,
expec-
tations, is always higher than inincome, and consumption (both in first and in
the emerging
secondfor
market. This implies a higher price periods). The total number of assets at
assets
even when bad times are expected: the world
the level decreases since it is an increas-
indus-
trialized country can never be ing as function
pessimisticof q. Hence, both market incom-
about its own demand-and therefore its asset pleteness and the volatility of second-period
prices-as the emerging market. Hence, if the consumption are higher. It can be shown that
productivity differential is sufficiently high, theper capita income in the emerging market is
industrialized country can never experience a lower in a financial crash (wE) than in autarky.
crash. The negative relation between income Contrary to what occurs in the "good" equi-
per capita and the vulnerability to crashes ap- librium, the emerging market experiences a cur-
pears only when countries are sufficiently openrent account surplus given by LwE/(1 + /3). In a
to capital movements, a fact that accords well crash, agents can only buy foreign assets from
with Table 1. the industrialized country to save and diversify
The analysis of asset prices in a crash alsorisk, so that capital flight occurs.
shows that countries more open to trade in
goods (larger T) are less vulnerable to financial IV. Quantitative Analysis
crashes. Indeed, these countries' operating prof-
its and dividends (equation (14)) are less depen- This section assesses the potential of our de-
dent on local income and therefore less affected mand-based theory of financial crisis to match
by the crash. Hence, the crash itself is less key stylized facts of emerging market crashes,
likely. This implies that the set of parameters such as a drop in asset prices, income collapse,
for which a crash occurs is smaller for countries and current account reversal. Table 2 (taken
more open to trade. We therefore find a funda- from Mendoza and Smith, 2004) reproduces
mental asymmetry in the effect of trade and data for four emerging markets, Argentina, Ko-
financial openness on the vulnerability of coun- rea, Mexico, and Russia. Table 3 presents the
tries to financial crashes. Whereas trade open- parameter values used in the calibrations. Panel
ness unambiguously decreases this vulnerability, A of Table 4 provides the quantitative implica-
financial openness may increase it. tions of the exact model we described in the

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC
TABLE 2-CRISES IN FOUR EMERGING ECONOMIES
Real equity prices Current account/GDP Ind. production

(percent change) (percent points change) (percent change)
Argentina (94.4-95.1) -27.82 4.05 -9.26

Korea (97.4-98.1) -9.79 10.97 -7.20
Mexico (94.4-95.1) -28.72 5.24 -9.52
Russia (98.3-98.4) -59.37 9.46 -5.20
Notes: Real equity prices are deflated by the CPI, ex

U.S. dollar terms. The change in the current account
to the second quarter of 1995. Industrial production
(Mendoza and Smith, 2004).
TABLE 3A-PARAMETERS
Subst. Subst. Manuf.

Wages goods assets Discount Risk Safe share
WE/I o e 13 premium a ,., a
Base case 1/5 5 5 0.99 1.05 1 0.4
High 1/3 10 8 - 1.10 - 0.6
Low 1/8 4 3 0.95 1.02 0.1 0.3
TABLE 3B-FRICTIONS
Financial cost Trade cost Participation

7, percent 7T percent y percent
No crisis Base case 5 40 36

High 10 80 50
Low 1 20 10
Crisis Base case 6 50 36
High 12 10 50
Low 1.2 25 10
previousquantitative
sections.properties of our model. Appendix W
model." And, indeed,
B provides the key equations of this augmented
stylized, model.25
we augment
features to get our "b
Table 4): (a) A. Calibration
Agents h
return technology tha
tion a of the states of nature covered in normal The most important parameters of our model
times, i.e., without a crash. We experiment with
are the trade costs 7T, financial costs 'F, ratio of
different degrees of international tradability of wages wEwI, elasticities of substitution for goods
this technology. We interpret our safe technol- oT and assets e, and the share of households
ogy as any alternative way used by agents to participating in the stock market which we de-
save their income during financial crises, such note as 7. The interaction of trade costs and
as purchases of durable goods or cash hoard-
ings. (b) We allow for limited participation in
the stock market. Neither of these two new 25 They are not analytically solvable, unlike their coun-
terparts
of Sections II and III, but carry the same effects and
features alters significantly the qualitative prop-
intuitions. This is why we chose to discuss the more stylized
erties of our model, nor do they change the model in the core of the paper and present this more general
fundamental mechanisms presented in the pre-version in the quantitative section. The programme used to
vious sections. But they notably improve thesolve the model is available from the authors.

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC
TABLE 4--QUANTITATIVE RESULTS
Aq ACA~yE AYE
Asset price Current account Income
percent percent point percent
Panel A change change change
1 Stylized model -22.8 +46.2 -30.7

2 With safe technology -22.4 +36.5 -30.7
3 With limited participation -12.1 + 19.7 -13.1
4 With trade disruption -29.5 +46.2 - 30.7
5 With financial disruption -23.2 +46.2 -30.7
Panel B
6 Baseline model -20.5 + 15.4 -13.1

7 Tradable safe asset -20.1 +8.8 -13.1
8 High 7, -20.8 +13.9 -12.7
9 Low TF -20.2 +16.8 -13.5
10 High 7, -18.9 +8.1 -6.9
11 Low 7T no crash
12 High 7,t (on imports) -22.2 +17.1 -14.4
13 Low 7,t (on exports) no crash
14 Low wEJwI -21.1 +20.3 -15.8
15 High wEwI -20.0 +11.7 -11.5
16 High e -20.9 +13.0 -11.8
17 Low e -18.3 +19.1 -15.6
18 High r -17.1 +5.9 -5.0
19 Low o- -18.5 + 19.0 - 16.3
20 High /, a -20.5 +15.4 -13.1
21 Low /L, a -20.6 + 15.5 -13.1
22 High participation -23.1 +20.8 -17.5
23 Low participation - 14.1 +4.6 -3.9
percent as our plausible

financial costs is key to getting upper estimate and 40 percent as
quan-
titative results. A period our
is base
a case.
quarter. The values
of the parameters we useCrises
are are discussed
accompanied by the collapse of
below
and summarized in Table 3.
trade credits, increased exchange rate uncer-
tainty, information asymmetries and higher in-
Trade Costs.-We base our estimates of surance costs. All these elements are exogenous
to our
trade costs 7T mainly on James E. Anderson andmodel. Unfortunately we do not have any
Eric van Wincoop (2004). According toreliable
theseestimates of the increase in trade costs
authors, "the pure international component in crisis
of time to calibrate our model precisely.26
We
trade barriers, including transport costs and assume in our stylized model of Table 4
bor-
(panelisA, lines 1-3, 5) that trade costs are in-
der barriers but not local distribution margins,
variant between normal and crisis times. Then,
estimated to be in the range of 40-80 percent
for industrialized countries." The estimate is
in the baseline model of Table 4, panel B, and in
based on both direct evidence and indirect evi- panel A, line 4, we assume that trade
dence stemming from the gravity literature.
This estimate roughly breaks down as a 21-
percent transportation cost, an 8-percent policy 26 Zihui Ma and Leonard Cheng (2005) document the
barrier, a 7-percent language barrier, a 14-per- disruption of trade during a financial crisis. Using a gravity
cent currency barrier, a 6-percent information equation framework, they find a significant decline in trade
cost barrier, and a 3-percent security barrier. We flows, even after controlling for economic fundamentals.
Their analysis does not allow a precise quantification of the
pick 20 percent as the low estimate of our
effect, however. In the narrower case of sovereign defaults,
trading costs on the goods market; this roughly Andrew K. Rose (2005) and Jos6 V. Martinez and Guido
corresponds to the pure transport cost estimate Sandleris (2004) also document a decrease in trade flows,
of Anderson and van Wincoop. We choose 80 even after controlling for fundamentals.

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC
Elasticities,
costs, both on imports and Relative Wages,
exports, and Manufac-
increase in
crisis time by 25 percent
turingfrom
Shares.-Wetheir baseofvalue
pick an elasticity substi-
tution for goods of 5 in the base case, in the
7aT in normal times. We call this the trade dis-
ruption case. middle of the range of the estimates of the trade
literature. We experiment with values of 4 and 10,
Financial Costs and Limited Participation.-
thereby covering the estimates surveyed in Ander-
The choice of an estimate
sonfor financial
and van transaction
Wincoop. We calibrate the elasticity
costs is more difficult, as there between
of substitution is noassets
consensus in
using Jeffrey A.
the literature. Financial costs should include the Wurgler and Ekaterina V. Zhuravskaya (2002).
cost of government regulations on capital flows, They report the results of several studies, as well
the cost of differences in regulations in account- as their own estimates, for U.S. stocks. The elas-
ing, banking and commission fees, foreign ex- ticity ranges from 1 (Andrei Shleifer, 1986) to
change transaction fees, and, most importantly, their own: 6, 8, and 12 depending whether stocks
information costs between emerging markets and have close substitutes or not. Given that the im-
industrialized countries. Reviewing the literature, portant elasticity in our context is the one between
John Heaton and Deborah J. Lucas (1996, p. 467) equities of the emerging market and equities of the
argue that for the U.S. equity market, "transaction industrialized country, which are less substitutes
costs as high as 5 percent are reasonable." Given than domestic ones, we choose a rather low elas-
the lack of precise data for emerging markets, we ticity for the base case, i.e., 5. We also experiment
choose again a wide interval of transaction costs with 8 and 3.
ranging from 1 percent to 10 percent, with a base We calibrate the wage ratio wEwI between
case set at 5 percent. the emerging market and the industrialized
During crises, however, volatility on the for- country at 1/5. The Bureau of Statistics of the
eign exchange market increases, and there is U.S. Department of Labor (2002) reports hourly
more information asymmetry and adverse selec- compensation costs for production workers in
tion. International financial transaction costs aremanufacturing for a selected group of countries.
therefore also likely to increase.27 We take this For Mexico and Brazil, these were 12 percent of
possibility into account and call it the financial those of the United States. For Korea, these
disruption case. In that scenario, financial costs amounted to 42 percent and for Asian newly
go from our baseline case of 5 percent in normal industrialized economies, 34 percent. We ex-
times to 6 percent. We also allow for the case of periment with 1/8 in the low case and 1/3 in the
joint financial and trade disruption, where both high case.
financial and trade costs increase during a crash. We choose p and a, the share of the manu-
Data on limited stock market participation factured good in the utility function and in the
are not available for emerging markets. For the production function of the investment good, to
United States, Annette Vissing-Jorgensen (2002) be equal to 0.4. This number is usually the one
documents household participation rates in the picked in the trade literature for the share of the
stock market of 36 percent in 1994. We pick manufacturing sector. We also experiment with
this number as our baseline case. higher (0.6) and lower (0.3) values.28
Safe Technology.-We set a to 1 in the base

case, implying that agents are able to use the
safe technology to save during the crash for all
27 For example, for the forex market alone, the 1998 states of nature covered in a noncrash equilib-
International Monetary Fund (IMF) International Capital rium. We also experiment with low levels of a
Markets report mentions that: "Prior to the crisis, bid-ask
(0.1), implying that the "safe asset" gives a
spreads on these (Asian) currencies had been similar, per-
haps modestly higher, than those for the major currencies. dividend in 10 percent of the states of the world
Following the crisis, these spreads widened by factors of covered in normal times. We also vary the
between 6 (ringgit) and 13 (rupiah), implying, for example,
a hefty 1.7-percent average cost of carrying out a rupiah-
dollar transaction on the spot market since the crisis, rising
on occasion to as much as 10 percent. Higher volatility and 28 The other constraint is that the nonmanufacturing sec-
transaction costs were also associated with a drying up of tor should always exist in both countries. This requires that
liquidity." I and a not be too large.

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC
degree of tradability of the safe asset,

aggregate from
income. The stylized model generates
nontraded to a transaction cost as for the other capital inflows into the emerging market in tran-
assets. We set the discount rate /3 to 0.99 in the quil times and outflows in crisis times. But it
base case. We calibrate the safe technology produces too large current account reversals
parameters in order to match the risk premium (from -11.7 percent of GDP in normal times to
at 5 percent (annualized). The latter is defined +34.5 percent in crisis times). This comes, in
as the expected difference in return between a particular, from the absence of a safe technol-
risky asset in the emerging market and the safe ogy: during the crisis, emerging market agents
technology. The return of the safe technology is can save only by purchasing foreign risky as-
low enough that agents who have access to sets, implying large capital outflows. In lines 2
financial markets have no incentive to use it in to 5 of Table 4, we alter the stylized model by
absence of a crash. adding each time only one of the following
features: safe technology; limited participation;
B. Results increase in trade costs during the crisis (trade
disruption); increase in financial costs during
We are interested in the change in three vari- the crisis (financial disruption).
ables summarizing the state of the economy:If we add the safe technology (line 2), we do
equity prices in dollars, the current account rel- not change much the drop in asset price nor the
ative to income, and income.29 collapse in output; but the swing in the current
We start with a calibration of the exact styl- account is lower, which helps bring the model
ized model described in Sections I to III. There somewhat closer to the data.30 Adding limited
is no safe technology and no limited participa- participation to the stylized model (only 36 per-
tion. All parameters are set to their base value ofcent of households participate in the stock mar-
Table 3A. Furthermore, trading costs and finan- ket) decreases the effect of financial wealth on
cial costs are equal in normal and crisis times. the economy (line 3). The drop in asset price is
The only difference is that the investment sector smaller because, in this case, 64 percent of the
uses manufactured goods so that a is not equal economy is effectively insulated from the crash.
to zero. Results are displayed in Table 4 (line 1). In contrast, if we introduce trade disruption
The stylized model has qualitatively correct (line 4), we are able to match the data as far as
predictions. High enough trade costs insure thatthe drop in asset price is concerned (-29.5
the emerging market asset dividends are depen-percent) but the drop in output and the current
dent on local conditions, which in turn makes account reversal are still too extreme. This dra-
possible self-fulfilling demand collapses. Con- matic effect on the asset price comes from the
versely, when trade costs are reduced to 20 decrease in profits of the emerging market
percent, for example, the possibility of a crash is firms, which have to rely even more on domes-
eliminated. Also, multiple equilibria do not ex- tic demand in crisis times. The ensuing decrease
ist whenever financial costs are higher than 60in dividends is magnified by the income effect
percent. These results confirm the interactions and creates a sharp drop in the emerging market
between trade and financial costs we put for- asset price. The same type of mechanism, i.e.,
ward in the theoretical section. an increased reliance on domestic demand to
Quantitatively, the model is able to generate sell assets in crisis times, explains the effect of
large drops in asset prices (-22.8 percent) but financial disruption on asset prices (line 5), but
produces far too large a drop in income (-30.7 quantitatively the effect is much smaller.
percent). The reason is that, in the crash, the
entire financial wealth of the emerging market is
wiped out. Since all our agents participate in the 30 Adding a safe technology makes possible the exis-
stock market, this generates a dramatic drop in tence of crashes in autarky, since it may not be worthwhile
for agents to invest in risky assets if everyone else coordi-
nates on the safe technology. This does not, however, alter
29 Quarterly data on GDP are not available for these
the logic behind the existence of multiple equilibria for
intermediate levels of financial costs. Emerging market
countries. Industrial production is therefore used as a proxy
for income. In our model, income and consumption are now invest both in foreign risky assets and in the safe
agents
perfectly correlated due to the log utility, so we doasset not during crashes. In this experiment, the safe technology
report changes in consumption. is nontraded.

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC
Our baseline model incorporates all the

percent, line 12) because these fea-
asset price in E is
tures. Panel B of Tablehigher
4 (line 6) equilibrium.
in the no-crash presents the
Protectionism
model when the safe technology, increases profits of limited
firms in the goodpartic-
equilibrium
ipation, trade, and financial disruption
when domestic income is large. It are all
does not, how-
present at the same time. All
ever, sever the
the link betweenparameters
asset prices and pes-
have been set to their base value of Table simistic expectations affecting domestic demand.
3A and 3B. As before, the only difference Hence,
be- protectionism does not decrease the
tween the emerging market and the developedlikelihood of a financial crash. On the other
economy is their productivity level. Thishand,
base- lower export costs make sales of E
line model is closer to the data. Asset firms prices more dependent on world income,
drop by 20.5 percent, income drops by which
13.1is more stable than domestic income
percent, and the current account goes from and-4.1
therefore weakens the circular causality
percent in normal times to +11.3 percent in
mechanism at the origin of a crash. When
crisis times, i.e., a reversal of 15.4 pointsexportofcosts are down to 20 percent, the pos-
income. We have checked that the industrial- sibility of a crash is eliminated (line 13).
ized country cannot be subject to a crash withA higher productivity differential between the
these parameters. rich country and the emerging market exacerbates
We now subject the baseline model to sensi- all the characteristics of the crash since our mech-
tivity experiments. If the safe technologyanismsis are based on demand: a relatively poorer
internationally tradable, the current account emerging
re- market will experience, ceteris paribus,
versal becomes smaller. For example, if 30 per-
a sharper drop in asset prices and income and a
cent of the safe projects are internationally
larger current account reversal (lines 14 and 15). If
tradable (with the same transaction coststhe
as difference in wage between the emerging mar-
other assets), then (see line 7) the current ac-
ket and the industrialized country is small enough,
count reversal is only 8.8 points of GDP be- the possibility of a crash disappears. This is the
cause the emerging market sells these assets to
case if wE is only 50 percent smaller than wi. This
the industrialized country. The drops in assetconfirms that our mechanism is able to explain
price and income remain similar because why
theemerging markets are more prone to crashes
international tradability of the safe asset leads
than high-income countries.
the industrialized country to buy less of bothA high elasticity of substitution across as-
sets tends to increase the extent of the crash
risky assets in a crash so that their relative price
does not change much. (lines 16 and 17). Since the transformations
In lines 8 to 13, we perform some sensitivity of the financial costs 4F = (1 + F)1 -e and
analysis of the magnitude of the frictions. Varying
of the trade
effective costsof4,financial
measures = (1 + 7a)1- are
and trade the
financial costs (high and low cases in Table 3B)
openness in the model, an increase in e is like
affect the magnitude of the current account rever-
sal. If a < 1 (the safe asset gives in less than 100
an increase in 71". An increase in o is analo-
percent of states of nature covered in a noncrash
gous to a increase in 7-', but it also decreases
profits in the monopolistic sector and there-
equilibrium), then increasing financial costs suffi-
fore the role of demand on dividends. Hence,
ciently eliminate the possibility of a crash. Chang-
ing trade costs alter both the domain of existence
the effect of oa on the magnitude of the crash
of multiple equilibria and the magnitude of the is ambiguous (lines 18 and 19).
crash. Because (symmetrically) higher trade costs In lines 20 and 21, we check that the manu-
in the goods market generate lower asset prices facturing sector share does not change the re-
and income in both the no-crash and the crash sults. We have also checked that changing the
equilibria, they may lead to a smaller crash (line risk premium, the discount factor, or the number
10). Lower trade costs, however, always make the of states covered by the safe technology, a, does
domain of multiple equilibria smaller: with trade not alter our results. In lines 22 and 23, we find
costs at 20 percent, a crash is not possible (line that higher participation in stock markets, which
11). But we can also investigate the impact of can be interpreted as a higher dependence of the
asymmetric trade costs. For high import costs (due economy on financial wealth, leads to larger
to higher tariffs or less efficient port facilities, for crashes, income drops, and current account
example), the crash is more pronounced (-22.2 reversals.

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC
Overall, our baseline model matches the segmentation of the goods and asset markets
stylized facts of Table 2 reasonably well. In plays a key role. Our framework is the first one,
order to get a smaller current reversal, we would to our knowledge, that analyzes jointly home
need some degree of international tradability ofmarket effects in the financial and goods mar-
the safe technology (see line 7). There are differ-kets and their interactions. Relatively high trade
ent plausible mechanisms to get a larger drop in costs on the goods market make profits and
asset prices with similar drops in income and dividends very dependent on domestic demand.
current account reversals. First, a larger trade dis- Financial globalization makes coordination on
ruption (trade costs increasing from 40 percent tocapital flight possible. Emerging market income
60 percent in crash) would generate a 26.9-percent itself depends on investment, which is affected
crash in asset prices. Similarly, a high degree ofby asset prices, in turn dependent on domestic
financial disruption (transaction costs increase income and demand. This circularity makes our
from 5 percent to 15 percent in crash) also gener- demand channel quantitatively powerful. Our
ates a larger crash (-23.7 percent). The model ismechanism of financial crisis is very general,
flexible enough to allow for domestic trade costssince it is at work whenever there is a sizable
on goods markets. If we assume that those trade difference in income between countries and
costs go from 0 to 20 percent in crisis time, this alonethere are trading costs in goods and financial
would generate a crash of -25.8 percent. Domestic markets.
trade and international trade disruptions reinforce We see our approach as complementary to
each other so that we can generate a sharp drop inexisting views on the links between financial
asset prices with relatively small levels of trade dis- globalization and crises. So far, the literature
ruption in domestic and international markets. has emphasized that financial globalization, by
The assumption that all assets give dividends making borrowing on world financial markets
in only one state of nature (as opposed to sev- easier, strengthens market failures prevalent in
eral) is immaterial for the results on relative emerging markets. In particular, moral hazard
asset prices. But relaxing the assumption that and credit constraints have been shown to facil-
the risk of assets is identical in the two countries itate the advent of financial crises. Our paper
and/or across the no-crash and crash equilibria suggests that such market failures are not a
is interesting. If E assets are riskier (they give necessary condition for emerging markets to
dividends in fewer states of nature), then the become vulnerable to a crash when capital flows
crash is less pronounced: the price of the asset are liberalized. Trade costs on international
in normal times is lower so that the difference trade in goods and assets will themselves gen-
between no-crash and crash is also lower.31 If, erate that vulnerability.
however, during a crash assets become more Both the potential benefit of globalization
risky in the sense that the number of states they (in terms of cost of capital, investment, and
cover is 10 percent lower than in the no-crash income) and the higher vulnerability of
equilibrium, then the drop in asset price is more emerging markets to a crash come from the
pronounced (-28.6 percent). The introduction same factor that differentiates emerging mar-
of a fixed cost in the production of goods also kets and industrialized countries in our
makes the crash more pronounced. If the fixed model: their productivity and income lev
cost is proportional to wage costs (at around 10 The higher vulnerability is not necessari
percent of the value of sales), this increases due to bad institutions, bad incentives (ba
profitability in the E market, and the magnitude outs), or bad exchange rate regimes. This
of the crash becomes larger at -26 percent. not to say that these problems do not consti
tute important channels through which fina
V. Conclusion cial globalization can make emerging market
more vulnerable to a financial crisis.32 The
Our model puts forward a demand-based existing literature has logically recommended
mechanism of crisis in emerging markets where
32 The inclusion of credit constraints on investment in our

31 For example, if the number of states covered by E assets
model would certainly reinforce the possibility of a crash, as
the fall in asset prices would reduce the value of collateral.
is 10 percent lower than in I, then the crash is 18.8 percent.

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC
policies addressing the informational and insti-

emerging market on dome
tutional frictions at the origin
creases of
the the credit
domain of ex
market imperfections it describes.
equilibria. More
This alsotrans- sugg
markets should
parency, better information, liberalize
and their trade account
better banking
regulation have been advocated.
before their capital Similarly,
account. Although cur- such a
rency mismatches in prescription
fixed is exchange
sometimes heard inrate re-
policy cir-
gimes are listed as prime suspects
cles, we believe our paper is the to explain
first analytical
crises of these countries. Our
work giving paper
an economic shows
rationale that
to support it.
these policies and institutional changes
Ultimately, to analyze may
precisely policy not
implica-
be sufficient to preventtions on the timing
crises in of reforms, it would be
intermediate-
income countries and that
necessaryfinancial
to quantify a dynamic crises may
infinite horizon
be a more general phenomenon
version of the model.for those
This would require,econ-
how-
omies. A possible policy
ever, animplication of our
equilibrium selection mechanism to
model is that trade openness
pick the crash orhas a beneficial
no-crash equilibrium. We
role, since it mitigatesleave
thethis for future work.
dependence of the
APPENDIX A: THE EFFECT OF FINANCIAL LIBERALIZATION ON EMERGING MARKET ASSET PRICES
An increase in 4F affects only the qq curve. It will lead to an increase in q if the intersection point of the
qq curve and the YY curve shifts right when 4F increases. This will be the case if
asr q[(q/d)"-' - (q/d)'-6] - 2a4,(1 - s,) S<0.
S (1 + q2)(1 - 42) + (e - 1)v,[q2(qld)-" + (q/d)"]

[1 - (1-
sy(12 - 4T)]
A sufficient condition for this is that q/d < 1.
APPENDIX B: KEY EQUATIONS OF THE MODEL USED IN THE QUANTITATIVE SECTION
The model includes a safe asset which gives a dividend in a share a of the states of the world covered in
normal times, and has a return r. We also introduce a parameter y describing the extent of participation in the
stock market (only 1 - y households participate). The stock market equilibrium with limited participation in
normal times becomes
S( YE- YWE (Y - ywI)4F(q/d)1-6

PE = 1 + p3 (1 - Y)ZE + (1 - Y)zF(q/d)E-1 (1 - y)z, + (1 - )ZE F(q/d)1
/ y w, (yE - YWE))F(q/d)-
PI1 +/3 (1 - )z,+ - y) - Y)ZEF(q/d) 1-+ (1
Income in the emerging market in normal times is now given by yE
In crash, the stock market equilibrium becomes
0 / (1 - y)wE (Yic -wI)F

+ /3 (1 - y)OZE(rpEc/dEc)6-
S(yic - Tw - y +)wE+F(1c/d)
(1 - F)ZlcdF(qc/dc)-l (1 - y)zlc
Ptc = 1 + p (1 - y)zI (1 y)azE(rpE/dE 6- 1 - F(qcdc)e-1
The dividends are given by equation (8) adjusted for limited participation and its symmetric in a crash. The
value of the emerging market demand of the safe asset in a crash is
w3E (rPEc)a-1
1 + /3 (1 - y)azE(rpEc)e1 -+ (1 - Y)ZlcF(PEcdic/Plc)e-l

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC
REFERENCES of Economic Research Working Paper

11153.
Cespedes, Luis Felipe, Roberto Chang, and An-
Acemoglu, Daron, and Fabrizio Zilibotti. 1997.
"Was Prometheus Unbound by Chance?dris Velasco. 2004. "Balance Sheets and Ex-
change Rate Policy." American Economic
Risk, Diversification, and Growth." Journal
of Political Economy, 105(4): pp. 709-51.Review, 94(4): 1183-93.
Chang, Roberto, and Andris Velasco. 2001. "A
Aghion, Philippe, Philippe Bacchetta, and Abhijit
Banerjee. 2004. "Financial Development and Model of Financial Crises in Emerging Mar-
the Instability of Open Economies." Journal kets." Quarterly Journal of Economics,
of Monetary Economics, 51(6): 1077-1106. 116(2): 489-517.
Chari, Anusha, and Peter Blair Henry. 2002.
Aizenman, Joshua. 2004. "Financial Opening:
"Capital Account Liberalization: Allocative
Evidence and Policy Options." In Challenges
to Globalization: Analyzing the Economics,Efficiency or Animal Spirits?" National Bu-
ed. Robert E. Baldwin and L. Alan Winters,reau of Economic Research Working Paper
473-94. Chicago: University of Chicago8908.
Press. Christiano, Lawrence J., Christopher Gust, and
Anderson, James E., and Eric van Wincoop. Jorge Roldos. 2004. "Monetary Policy in a
2004. "Trade Costs." Journal of Economic Financial Crisis." Journal of Economic The-
Literature, 42(3): 691-751. ory, 119(1): 64-103.
Arellano, Cristina, and Enrique G. Mendoza. Corsetti, Giancarlo, Paolo Pesenti, and Nouriel
2003. "Credit Frictions and 'Sudden Stops' in Roubini. 2001. "Fundamental Determinants
Small Open Economies: An Equilibrium of the Asian Crisis: The Role of Financial
Business Cycle Framework for Emerging Fragility and External Imbalances." In Re-
Markets Crises." In Dynamic Macroeco- gional and Global Capital Flows: Macroeco-
nomic Analysis: Theory and Policy in Gen- nomic Causes and Consequences, ed.
eral Equilibrium, ed. Sumru Altug, Jagjit S. Takatoshi Ito and Anne O. Krueger, 11-41.
Chadha, and Charles Nolan, 335-405. Cam- Chicago: University of Chicago Press.
bridge: Cambridge University Press. Edwards, Sebastian. 2001. "Capital Mobility
Bekaert, Geert, Campbell R. Harvey, and Chris- and Economic Performance: Are Emerging
tian Lundblad. 2005. "Does Financial Liber- Economies Different?" In The World's New
alization Spur Growth?" Journal of Financial Financial Landscape: Challenges for Eco-
Economics, 77(1): 3-55. nomic Policy, ed. Horst Siebert, 219-44.
Burnside, Craig, Martin S. Eichenbaum, and Ser-New York: Springer-Verlag.
Epstein, Larry G., and Stanley E. Zin. 1989.
gio Rebelo. 2001. "Prospective Deficits and
the Asian Currency Crisis." Journal ofPolit- "Substitution, Risk Aversion, and the Tem-
ical Economy, 109(6): 1155-97. poral Behavior of Consumption and Asset
Caballero, Ricardo J., and Arvind Krishnamur- Returns: A Theoretical Framework." Econo-
thy. 1998. "Emerging Market Crises: An As- metrica, 57(4): 937-69.
Frankel, Jeffrey A., and Eduardo A. Cavallo.
set Markets Perspective." National Bureau of
Economic Research Working Paper 6843. 2004. "Does Openness to Trade Make Coun-
Calvo, Guillermo A. 1998. "Capital Flows andtries More Vulnerable to Sudden Stops, or
Capital-Market Crises: The Simple Econom- Less? Using Gravity to Establish Causality."
ics of Sudden Stops." Journal of Applied National Bureau of Economic Research
Economics, 1(1): 35-54. Working Paper 10957.
Calvo, Guillermo A., Alejandro Izquierdo, and
Gordon, Roger H., and A. Lans Bovenberg. 1996.
Luis-Fernando Mejia. 2004. "On the Empirics "Why Is Capital So Immobile Internation-
of Sudden Stops: The Relevance of Balance- ally? Possible Explanations and Implications
Sheet Effects." National Bureau of Economic for Capital Income Taxation." American Eco-
Research Working Paper 10520. nomic Review, 86(5): 1057-75.
Calvo, Guillermo A., and Ernesto Talvi. 2005. Heaton, John, and Deborah J. Lucas. 1996.
"Sudden Stop, Financial Factors and Eco- "Evaluating the Effects of Incomplete Markets
nomic Collapse in Latin America: Learning on Risk Sharing and Asset Pricing." Journal of
from Argentina and Chile." National Bureau Political Economy, 104(3): 443-87.

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC
Henry, Peter Blair. 2000. Stop." In"Do Stock

Preventing CurrencyMarket
Crises in
Liberalizations Cause Investment Booms?" Emerging Markets, ed. Sebastian Edwards
Journal of Financial Economics, 58(1-2): and Jeffrey A. Frankel, 335-83. Chicago:
301-34. University of Chicago Press.
International Monetary Fund. Various Years.
Mendoza, Enrique G. 2004. "Sudden Stops in an
Annual Report on Exchange ArrangementsEquilibrium Business Cycle Model with
and Exchange Restrictions. Washington, DC: Credit Constraints: A Fisherian Deflation of
International Monetary Fund Publications. Tobin's Q." Unpublished.
International Monetary Fund. 1998. Interna- Mendoza, Enrique G., and Katherine A. Smith.
tional Capital Markets: Developments, Pros- 2002. "Margin Calls, Trading Costs, and As-
pects, and Key Policy Issues: II, the Asianset Prices in Emerging Markets: The Finan-
Crisis: Capital Market Dynamics and Spill-cial Mechanics of the 'Sudden Stop'
over. Washington, DC: International Mone-Phenomenon." National Bureau of Economic
tary Fund Publications. Research Working Paper 9286.
Kaminsky, Graciela L., and Sergio Schmukler.
Mendoza, Enrique G., and Katherine A. Smith.
2001. "On Booms and Crashes: Financial Forthcoming. "Quantitative Implication of
Liberalization and Stock Market Cycles." a Debt-Deflation Theory of Sudden Stops
Unpublished. and Asset Prices." Journal of International
Krugman, Paul. 1999. "Balance Sheets, the Economics.
Transfer Problem, and Financial Crises."Paasche,
In Bernhard. 2001. "Credit Constraints
International Finance and Financial Crises: and International Financial Crises." Journal
Essays in Honor of Robert P. Flood, Jr. of Monetary Economics, 48(3): 623-50.
Portes, Richard, and H6lne Rey. 2005. "The
(Partly reprinted from International Tax and
Public Finance, vol. 6, no. 4 (1999)), ed. Determinants of Cross-Border Equity Flows."
Peter Isard, Assaf Razin, and Andrew K. Journal of International Economics, 65(2):
Rose, 31-44, Dordrecht: Kluwer Academic. 269-96.
Ma, Zihui, and Leonard Cheng. 2005. "The Ef- Rose, Andrew K. 2005. "One Reason Countries
fects of Financial Crises on International Pay Their Debts: Renegotiation and Intemrna-
Trade." In International Trade in East Asia. tional Trade." Journal of Development Eco-
Vol. XIV, ed. Takatoshi Ito and Andrew K. nomics, 77(1): 189-206.
Rose. Chicago: University of Chicago Press.Sachs, Jeffrey D., and Andrew M. Warner. 1995.
Martin, Philippe, and H6lne Rey. 2002. "Finan-"Economic Reform and the Process of Global
cial Globalization and Emerging Markets:Integration." Brookings Papers on Economic
With or Without Crash?" National Bureau of Activity, 1: 1-95.
Economic Research Working Paper 9288. Schneider, Martin, and Aaron Tornell. 2004.
Martin, Philippe, and Helene Rey. 2004. "Finan- "Balance Sheet Effects, Bailout Guarantees
cial Super-Markets: Size Matters for Asset and Financial Crises." Review of Economic
Trade." Journal of International Economics, Studies, 71(3): 883-913.
64(2): 335-61. Shleifer, Andrei. 1986. "Do Demand Curves for
Martinez, Jose V., and Guido M. Sandleris. 2004. Stocks Slope Down?" Journal of Finance,
"Is It Punishment? Sovereign Defaults and 41(3): 579-90.
the Declines in Trade." Unpublished. U.S. Department of Labor, Bureau of Labor Sta-
Matsuyama, Kiminori. 2004. "Financial Market tistics. 2004. International Comparisons of
Globalization, Symmetry-Breaking and En- Hourly Compensation Costs for Production
dogenous Inequality of Nations." Economet- Workers in Manufacturing, SIC Basis, 1975-
rica, 72(3): 853-84. 2002. Washington, DC: U.S. Department of
McKinnon, Ronald I., and Huw Pill. 1999. "Ex- Labor, Bureau of Labor Statistics.
change-Rate Regimes for Emerging Markets: Vissing-Jorgensen, Annette. 2002. "Towards an
Moral Hazard and International Overborrow- Explanation of Household Portfolio Choice
ing." Oxford Review of Economic Policy, Heterogeneity: Nonfinancial Income and Par-
15(3): 19-38. ticipation Cost Structures." National Bureau
Mendoza, Enrique G. 2002. "Credit, Prices, and of Economic Research Working Paper 8884.
Crashes: Business Cycles with a Sudden Wacziarg, Romain, and Karen Horn Welch.

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC
2003. "Trade Liberalization and Growth: 2002. "Does Arbitrage Flatten Demand
New Evidence." National Bureau of Eco- Curves for Stocks?" Journal of Business,
nomic Research Working Paper 10152. 75(4): 583-608.
Weil, Philippe. 1990. "Nonexpected Utility Wyplosz,
in Charles. 2001. "How Risky Is Finan-
Macroeconomics." Quarterly Journal ofEco-cial Liberalization in the Developing Coun-
nomics, 105(1): 29-42. tries?" Centre for Economic Policy Research
Discussion Paper 2724.
Wurgler, Jeffrey A., and Ekaterina V. Zhuravskaya.

193.243.196.158 on Thu, 22 Dec 2022 09:22:34 UTC

Globalization and Emerging Markets

Uploaded by

Copyright:

Available Formats

Globalization and Emerging Markets

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Globalization and Emerging Markets

Uploaded by

Copyright:

Available Formats

Globalization and Emerging Markets: With or Without Crash?

Author(s): Philippe Martin and Hélène Rey

Stable URL: https://www.jstor.org/stable/30034988

This content downloaded from

By PHILIPPE MARTIN AND HELENE REY*

We analyze the effects of financial and trade globalization on the likelihood of

This content downloaded from

TABLE 1-FREQUENCY OF CRASHES AND OPENNESS

a Sachs-Warner measure of openness.

Table 1 suggests that opening to capital impact

This content downloaded from

emerging market. Symmetrically, crisis

highly leveraged.4 Philippe Aghion et al. (2004)

There are two countries, E (emerging) and I

This content downloaded from

This content downloaded from

where cE;(n) denotes in

B. Utility and Budget Constraints (3)

In the emerging market, the utility of a represen-

where YE is the emerging market per capita

ith variety produced in the E market, vii is the

+ 3 In and rE is the investment payoff. Asset prices

This content downloaded from

We can therefore rewrite the are

13 See Section II. The demand for foreign assets de-

This content downloaded from

Le (1 + 77) + LcI + LCIE(1elusive

We first solve the model in the optimistic

There are L(zE + z1) such equilibrium condi-

IYE 1 The dividends of the second period are the

This content downloaded from

If 4F = 1 (zero transaction costs on assetopening

This content downloaded from

Sy of perfect goods market integration, as d = 1

yy We now study the impact of globalization on

(12) CAE =L(YE- E1+j)

convexity of the investment cost function, the

This content downloaded from

This content downloaded from

nancial markets are segmented and assets are

This in turn generates a low relative asset price

ket also lowers demand for goods. This falls

This content downloaded from

TABLE 2-CRISES IN FOUR EMERGING ECONOMIES

Real equity prices Current account/GDP Ind. production

Argentina (94.4-95.1) -27.82 4.05 -9.26

Notes: Real equity prices are deflated by the CPI, ex

Subst. Subst. Manuf.

Financial cost Trade cost Participation

No crisis Base case 5 40 36

This content downloaded from

TABLE 4--QUANTITATIVE RESULTS

1 Stylized model -22.8 +46.2 -30.7

6 Baseline model -20.5 + 15.4 -13.1

percent as our plausible

This content downloaded from

Safe Technology.-We set a to 1 in the base

This content downloaded from

degree of tradability of the safe asset,

This content downloaded from