Trade Linkages and Crisis Spillovers - 麻省理工学院人工智能研究专业

Trade Linkages and Crisis Spillovers
Trade Linkages and Crisis Spillovers

Kornkarun
Cheewatrakoolpong
Faculty of Economics,
Chulalongkorn University
Phayathai Road, Bangkok
10330, Thailand
Kornkarun.k@chula.ac.th

Somprawin Manprasert
Faculty of Economics,
Chulalongkorn University
Phayathai Road, Bangkok
10330, Thailand
somprawin.m@chula.ac.th

Trade Linkages and Crisis Spillovers*

抽象的
Many empirical studies find little evidence to support trade link-
ages as a channel for crisis spillovers during the 2008–09 global
financial crisis, although trade linkages were one of the most
important crisis transmission channels during 1971–97. A reason
that may explain why trade linkages play a less important role in
recent years is the changing composition of trade. 尤其, 这
increasing formation of international production networks implies
that trade increasingly involves indirect trade linkages. 因此,
direct trade statistics may fail to accurately to capture the total
trade exposure. In our study, we estimate total trade linkages by
including indirect trade linkages obtained from the construction of
a trade matrix. When we account for indirect trade linkages, 我们
find that export dependency on the U.S. market still helps to ex-
plain crisis severity for developed countries.

1. 介绍

The sub-prime crisis that originated in the United States in
2008 initiated a ªnancial crisis that spread globally, 和
has been deemed the worst ªnancial catastrophe since the
Great Depression. 的确, due to strong ªnancial and
economic linkages between the United States and many
countries all over the world, as well as the relatively
large size of the U.S. 经济, 美国. sub-prime
crisis was transmitted worldwide. One feature of the
downturn was a sharp contraction in global trade. 许多
presume that trade linkages provide the most important
transmission channel of the sub-prime crisis to many
developing countries.

* The authors are grateful to the Chulalongkorn Economics

Research Center for providing funds for this project and anony-
mous reviewers for their valuable comments.

Asian Economic Papers 13:1

Institute of Technology

土井:10.1162/ASEP_a_00253

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

1
3
1
8
4
1
6
8
3
3
8
1
A
s
e
p
_
A
_
0
0
2
5
3
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Trade Linkages and Crisis Spillovers

出奇, recent empirical studies of the 2008–09 global ªnancial crisis (GFC)
ªnd little evidence to support trade linkages as a channel for crisis spillovers. Rose
and Spiegel (2011) show that the countries with high export dependence on the U.S.
market experienced signiªcantly less severe crises than did other countries when
they use the level of export dependency on the U.S. destination as an explanatory
variable for crisis transmission. This result implies that trade linkages with the U.S.
market played no role or a mitigating role in crisis transmission. 相似地, Berkmen
等人. (2012) do not ªnd evidence that trade channels were an important crisis trans-
mission channel for emerging markets during the sub-prime crisis. 实际上, 什么时候
their the study includes interaction terms between the share of exporting manufac-
turing goods to those shipped to advanced economies and a dummy for emerging
markets to explain crisis severity, the negative coefªcient on the interaction term
yields an opposite-sign (negative) coefªcient in explaining crisis severity. 相似地,
the economic analysis of Berkmen et al. is unable to ªnd statistically signiªcant evi-
dence linking the degree of trade openness with the transmission of the sub-prime
crisis to emerging markets. The lack of connection is also noted by Blanchard, 这,
and Faruqee (2010), as their study does not ªnd a connection between country
export to GDP ratios and country GDP growth disruption during the GFC.

Apart from trade dependency, the international trade literature points out that both
geographical and commodity trade concentrations are considered to be important
factors that may amplify the effect of crisis transmission via an international trade
channel. Da Costa Neto and Romeu (2011) consider the impact of export, destina-
的, and intra-industry concentration on export performance of Latin American
countries during the sub-prime crisis using a gravity model and ªnds that export
concentration signiªcantly ampliªes trade collapse during the crisis whereas desti-
nation concentration insigniªcantly helps relieve the impact of crisis spillovers
on trade.

The general message of recent work on the transmission of the sub-prime crisis is
that it cannot be explained by the degree of openness, trade dependency on the U.S.
市场, or export destination concentration. 简而言之, trade linkages have not been
implicated as important transmission channels of the GFC. 相比之下, the empirical
studies explaining the causes of crises during 1971–97 ªnd that trade linkages are
one of the most important crisis transmission channels. Eichengreen, Rose, 和
Wyplosz (1997) and Glick and Rose (1999) ªnd that international trade is one of the
most important explanations of crisis transmission in the years from 1971 到 1997. 在
事实, many empirical studies during the 1990s concluded that the ªnancial crises
during 1971–97 tended to be regional phenomena because trade linkages were
among the major transmission channels. Kali and Reyes (2005) also ªnd that the

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

1
3
1
8
4
1
6
8
3
3
8
1
A
s
e
p
_
A
_
0
0
2
5
3
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Trade Linkages and Crisis Spillovers

country that is a center of trade network experiences more severe effects from the
crisis. Perry and Lederman (1998) studied the crisis transmission via international
trade and found that demand contraction and substitution effect played an impor-
tant role in the spillover of the crisis from one country to another when the origina-
tor’s counterparts were export-dependent. Demand contraction in the country expe-
riencing the crisis leads to a decrease in imports and affects its trading partners’
current account, especially when those partners are export-oriented. 此外,
the crisis may lead to currency depreciation in a country where the crisis originates,
because that country will experience a drop in domestic demand. 反过来, 作为
crisis country’s depreciation makes its exports cheaper, competitors in other coun-
tries encounter export demand as substitution effects shift purchases towards the
crisis country’s products. 因此, the crisis country’s competitors may face
transmission of crisis via the mentioned demand contraction.

The major puzzle in this paper is to explain why trade linkages, whether measured
by trade dependency with a crisis originator or export destination concentration,
seem to have been much less important in helping to transmit the recent sub-prime
crisis. Due to recent changes in composition of trade, it is possible that the change in
inºuence may stem from the changing nature of trade linkages. Thanks to the emer-
gence and growth of international production networks in many regions, 包括
东亚, many crisis effects transmitted by trade will be transmitted indirectly. 为了
例子, a substantial share of industrial production in East Asia, in products such
as automobiles and parts, computers and parts, and electronic and electrical appli-
ances, are in form of the international production networks (IPN), where production
is fragmented into several stages, each of which is located in different countries, 交流电-
cording to the countries’ comparative advantages. 因此, a fall in demand for
ªnal goods in the crisis originator also results in a drop in the export of parts and
成分 (or indirect trade) via the IPN channel.

As indirect trade linkages increase in importance, analyses that are based solely on
direct trade statistics will provide an increasingly incomplete view of international
trade linkages. Countries that have a low level of direct trade dependency with the
United States may actually have high trade exposure if indirect trade linkages are
包括. 因此, our study tries to bridge this gap by recalculating trade de-
pendency and export destination concentration using trade measures that include
indirect trade linkages.

A few other studies including Sato and Shrestha (2012) have sought to measure the
transmission of the ªnancial crisis to other countries and the dispersion when indi-
rect trade linkages are added to the analysis. In this work the authors build a com-

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

1
3
1
8
4
1
6
8
3
3
8
1
A
s
e
p
_
A
_
0
0
2
5
3
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Trade Linkages and Crisis Spillovers

prehensive Global Input–Output (GIO) system that tracks the transmission of
shocks via international supply chain linkages. Due to the input–output setup, 关于-
sults from the GIO represent “pure” effects stemming from production linkages.
相似地, Cheewatrakoolpong and Manprasert (2012) also use the GTAP framework
to capture both direct and indirect trade linkages and their roles in crisis transmis-
sion in case of Thailand. Their study shows that Thailand still has high trade expo-
sure with the G3 economies when indirect trade is included. Their results show that
indirect trade channels were more important than direct trade channels in the trans-
mission of the sub-prime crisis from the G3 economies to Thailand.

To our knowledge, 然而, there are no studies that consider whether trade de-
pendency and export destination concentration are explanatory factors of crisis
spillovers when indirect trade linkages are included. This is due to the fact that pre-
vious studies used trade statistics that capture only direct trade linkages to measure
the level of trade dependency and export destination concentration.

因此, our study will consider whether trade dependency with a crisis origina-
托尔 (即, 美国) and export destination concentration play a role as
important crisis transmission channels when indirect trade linkages are taken into
帐户. 这样做, we will ªrst calculate total trade exposure that accounts for both
direct and indirect trade linkages. The indirect trade linkages will be measured by
the construction of an “international trade matrix” using data from the GTAP and
UN COMTRADE to measure indirect trade linkages. Then we will utilize cross-
country regressions to see whether trade dependency with a crisis originator and
export destination concentration help explain crisis transmission.

The paper will proceed as follows. 部分 2 explains the conceptual framework and
方法论. 部分 3 provides the pattern of trade linkages when indirect trade is
包括. 部分 4 portrays the results of the study, and Section 5 concludes.

2. 方法

The analysis in this study is divided into three parts: 我) the computation of the trade
matrix to measure indirect trade linkages; 二) the measurement of trade concentra-
的; and iii) the use of cross-country regression to examine how the sub-prime crisis
was transmitted.

2.1 Calculation of the total trade linkages
International trade linkages may go beyond those that are reported in bilateral
统计数据, because international trade transactions among trading countries may be

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

1
3
1
8
4
1
6
8
3
3
8
1
A
s
e
p
_
A
_
0
0
2
5
3
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Trade Linkages and Crisis Spillovers

both directly and indirectly linked. 例如, indirect trade linkages arise when
ASEAN exports computer parts to China, which are used as intermediate inputs for
computer notebooks and are assembled in China and re-exported to the United States.
在这种情况下, ASEAN computer parts are “indirectly linked” to the United States
(through China). To overcome the distortion arising from reported statistics, the Bank
for International Settlements (BIS) (2007) tried to measure total trade linkages through
both direct and indirect channels. 这样做, indirect exposure was calculated by multi-
plying export shares for each segment of the bilateral trades. 例如, an estimate
of the indirect exposure of ASEAN exports to the United States through China can be
calculated by multiplying the export share of ASEAN exports to China by the export
share of China to the United States. The BIS results reveal that “a signiªcant propor-
tion of China’s imports were accounted for by production inputs, particularly for the
export sector. 关于 70 percent of Chinese imports consist of intermediate goods, 和
57 percent of these goods come from emerging Asia and Japan. 同时,
consumption and capital goods make up 72 percent of Chinese exports to the United
States and 68 percent of those to the European Union” (BIS 2007, 53). Although the BIS
calculation highlights the importance of international trade linkage measurements,
the study examined only two legs of the linkages.

Another attempt to measure real connection between trading partners was done by
Koopman, 王, and Wei (2008), who proposed a general framework for the assess-
ment of “value-added exports.” Because imports of intermediate goods are usually
exempt from taxation when used in export-oriented production processing trade,
processing trade data can be combined with international trade data to impute
value-added export ªgures for a given country. It is worth noting that this method
can only be applied in cases where the operation of tariff-favored production net-
works provides additional insight into the production process. When the method
was applied to Chinese data the results showed that the share of domestic content
for exported manufacturing products increased continuously during the ªrst ªve
years of China’s WTO membership. The analysis also showed that high-tech indus-
tries were more likely to have high shares of imported content than was the case for
labor intensive production activities such as textiles and clothing.

Work by Johnson and Noguera (2009) derives a “VAX ratio”—which refers to the ra-
tio of value-added exports to gross exports—that attempts to measure the size of
trade connections through inªnite legs of relationships. The generation of this meas-
ure uses the GTAP 6.0 Data Base along with input–output techniques, which ac-
count for the inªnite segment of indirect linkages. The ratio can be interpreted as a
level of vertical specialization. 更重要, this work shows that the real value-
added of an export may differ signiªcantly from reported gross export due to the

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

1
3
1
8
4
1
6
8
3
3
8
1
A
s
e
p
_
A
_
0
0
2
5
3
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Trade Linkages and Crisis Spillovers

activities of international production networks. 平均而言, value-added exports
是 27 percent lower than gross exports. The larger values of value-added to ex-
port ratio were found in the agricultural, natural resources, and services sectors. 在
对比, manufacturing has lower VAX ratios. 最后, high-income countries often
have lower ratios of value-added to gross export than do lower-income countries.

Recently Sato and Shrestha (2012) constructed a comprehensive GIO system and
used it to measure the effects caused by external shocks to domestic country and the
dispersion through global supply chain network. The GIO has details on 35 produc-
ing sectors, 在哪里 27 countries are endogenous to the system and the remaining
65 countries are exogenous. The table was applied to examine effects of negative im-
port demand shock from the United States to Japan. They found that a high level
of domestic content for export-oriented activities could lead Japan to be severely
affected by a decline in external shocks.

在本文中, we proposed a simple algorithm that allows us to compute the “total”
trade linkages for 129 countries and regions that are based on the complete set of
global trade transactions for a given year. In contrast with the input–output tables
produced by the Institute of Developing Economies-Japan External Trade Organiza-
tion and the GIO from Sato and Shrestha (2012), the trade matrix used in this study
is based on aggregate trade data that encompass all global transactions rather than a
subset limited to certain countries or regions. Because we use data from the GTAP
8.0 Data Base, 全部 129 countries in our analysis are treated as endogenous actors in
global trade transactions. By adopting the input–output technique, our results
deliver measures that are based on the inªnite sum of indirect trade relationship
between countries.

Let matrix T consists of international trade transactions among countries in the
世界. Each element of Tij refers to bilateral export from country i to country j. 因此
the row sum of Tij captures the total export of country II, which we deªne as Ei. 在
另一方面, the column sum of Tij’s yields the total import by country j, 哪个
we denote by the matrix S, consisting of elements sij. In world equilibrium, 全部的
world exports equals total world imports.

(西德:2)Mj

(西德:3) (西德:2)Ei

(1)

For each country, produced goods are used to meet domestic ªnal consumption and
by export demands. Let A be a diagonal matrix containing ratios of domestic ªnal
good demand relative to total production available in a country.

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

1
3
1
8
4
1
6
8
3
3
8
1
A
s
e
p
_
A
_
0
0
2
5
3
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Trade Linkages and Crisis Spillovers

⎡
⎢
⎢
⎢
⎢
⎣

11
0

0
A

二
0

0
(西德:2)

毫米

⎤
⎥
⎥
⎥
⎥
⎦

In this matrix, the elements aii represent the proportion of domestic ªnal use to
total demands of the country i. 所以, each country imports goods for con-
sumption and re-exports according to proportions of aii and (1 (西德:4) aii), 分别.
We deªne the total trade linkages (lj) to be equal to direct linkages (dj) plus indirect
linkages (ij),

(西德:3) dj

(西德:5) ij

(2)

(西德:3) Tij/Ei, which refers to the export share of country i to j. By design, 全部的

现在, let sij
indirect imports of country j must be equal to the sum of all re-exports from other
countries to j. 所以,

(西德:3) 我

1 −∑
(

我

a s
)
ij
二

(3)

我

In this formula the term (1 (西德:4) aii)sij on the right-hand side of equation (3) refers to the
re-export share of i to j. 因此, lj(1 (西德:4) aii)sij is the total re-export value from i to j, 和
the summation adds up to all other countries’ re-exports to j. 所以, 方程
(3) can be rewritten as:

(西德:4) dj

(西德:3) 我

1 −∑
(

我

a s
)
ij
二

(4)

The left-hand side of equation (4) refers to indirect imports of country j and the
right-hand side represents the sum of all re-exports from all other countries to j.
While the creation of this trade measure implies that re-export value is proportional
to indirect trade linkages, the absence of data on inputs used for export, outside of
cases such documentation of processing trade activities as in Koopman, 王, 和
Wei (2008), requires us to make this proportionality assumption. For simplicity,
consider a world economy consisting of three countries trading to each other.
方程 (4) could be written in matrix terms as:

⎡
⎢
⎢
⎢
⎣

我

时间
⎤
⎥
⎥
⎥
⎦

⎡
⎢
⎢
⎢
⎣

时间
⎤
⎥
⎥
⎥
⎦

⎡
⎢
⎢
⎢
⎣

我

时间
⎤
⎥
⎥
⎥
⎦

⎡
1
⎢
⎢
⎢
⎣

0
-
A

0
0
-
A

⎤
⎥
⎥
⎥
⎦

⎡
s
⎢
s
⎢
21
⎢
⎣
s
31

s
22
s
32

⎤
⎥
⎥
⎥
⎦

s
23
s
33

(5)

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

1
3
1
8
4
1
6
8
3
3
8
1
A
s
e
p
_
A
_
0
0
2
5
3
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Trade Linkages and Crisis Spillovers

所以, we can write:

L (西德:4) D (西德:3) L(我 (西德:4) A)S

(6)

where L is a row vector of lj, D is a row vector of dj, A and S are the domestic ªnal
consumption share and export share matrices deªned earlier. For simplicity, deªne:

问 (西德:3) (我 (西德:4) A)S

Total linkages, in terms of direct linkages and the re-export share, can be given as
the following equation:

L (西德:3) D(我 (西德:4) 问)(西德:4)1

or L (西德:3) D (西德:5) DQ (西德:5) DQ2 (西德:5) DQ3 (西德:5) . . .

where D is direct linkages

DQ are the linkages from ªrst round re-export

(7)

(8)

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

1
3
1
8
4
1
6
8
3
3
8
1
A
s
e
p
_
A
_
0
0
2
5
3
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

DQ2 are the linkages from second round re-exports, 等等.

From equation (8), the calculation of trade linkages follows the input–output model
convention that total linkages between two countries are equal to direct transactions
plus inªnite indirect activities.1

To calculate total linkages according to above method, we used data from two
different sources. Bilateral trade transactions between each pair of 129 国家
were obtained from the United Nations Comtrade. These trade transactions are
used to form the vector D and matrix S. Elements in matrix A were obtained from
GTAP Version 8.0 coefªcients. 桌子 1 summarizes the list of GTAP variables used in
the calculation.

2.2 Measurement of trade concentration
This study follows the previous literature such as Hesse (2008) and Da Costa
Neto and Romeu (2011) in utilizing the standard Herªndahl index to identify the
level of trade concentration. Meilak (2011) and Samen (2010) review several concen-
tration ratios used to measure export concentration and ªnd that the Hirschman

1 Miller and Blair (1985) provide further background on input–output analysis.

Asian Economic Papers

Trade Linkages and Crisis Spillovers

桌子 1. GTAP coefªcients used in the calculation of matrix A

VDPA(n, 我)
VIPA(n, 我)
VDGA(n, 我)
VIGA(n, 我)
VXWD(n, 我, j)

二

∑

private consumption expenditure on domestic sector n in country i
private consumption expenditure on imported sector n in country i
government consumption expenditure on domestic sector n in country i
government consumption expenditure on imported sector i
exports of sector n from country i to country j valued FOB
∑

∑

VIPA n i
( , )
∑

VDGA n i
( , )
∑

VIGA n i
( , )
∑

∑

VDPA n i
( , )
∑

VDPA n i
( , )

VIPA n i
( , )

VDGA n i
( , )

VIGA n i
( , )

XWD n i j
, )

( ,

n j
,

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

1
3
1
8
4
1
6
8
3
3
8
1
A
s
e
p
_
A
_
0
0
2
5
3
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

index2 and the Herªndahl index are the most widely used measurements of destina-
tion concentration and export (or commodity) concentration. Meilak (2008) explains
that the Herªndahl index is the simplest and is friendly with data availability. 还,
it also captures relative size variation among sectors. Meilak (2011) portrays three
desirable features one would like to ªnd in concentration measures, and shows that
only the Herªndahl index and the Hannah and Key index possess these properties.
因此, we use the Herªndahl index for the measurement of export destination
concentration and export sector concentration.

A total of 129 destinations are included in the calculations. The destination
Herªndahl index for country j at time t is given by:

j
t

129

=∑ 2
s
我

我

(9)

where si

2 is the share of the exports from the country j to the country i.

2.3 Causes of crisis
To identify the channels for crisis spillovers, we follow the approach of Rose and
Spiegel (2011), Berkmen et al. (2012), and Blanchard et al. (2010), which means using
cross-country regressions to investigate the causes of differences in country crisis
severity (or output impact) from the GFC.

Some studies such as Rose and Spiegel (2009) use real GDP growth over 2008–09 to
measure crisis severity. Rose and Spiegel (2011), 然而, argue that such a measure
does not capture underlying growth trends. 所以, our study follows Blanchard,
这, and Faruqee (2010) and Lane and Milesi-Ferretti (2011) in using the real GDP
growth difference between 2008–09 and 2005–07 instead.

2 The Hirschman index is similar to the Herªndahl index. It is the square root of the

Herªndahl index.

Asian Economic Papers

Trade Linkages and Crisis Spillovers

Berkmen et al. (2012) suggest that crisis transmission may follow four groups,
即, 我) trade linkages, 二) ªnancial linkages, 三、) vulnerabilities, and iv) 政策
frameworks. The important measures for trade linkages according to Rose and
Spiegel (2011), Da Costa Neto and Romeu (2011), and Blanchard et al. (2010) are the
degree of openness, the share of export to GDP, the share of export to the U.S. 三月-
ket, commodity composition, export sector concentration, and export destination
concentration. All of the mentioned studies ªnd that trade linkages were not a
signiªcant spillover channel for the sub-prime crisis. According to Da Costa Neto
and Romeu (2011) and Berkmen et al. (2012), only export composition matters;
export sector concentration could signiªcantly lessen crisis severity.

For ªnancial linkages or ªnancial measures, the previously mentioned studies use
short-term external debt, total foreign debt, short-term foreign debt, and lending
from advanced economies as important indicators. Blanchard et al. (2010) notes that
short-term external debt signiªcantly explains the incidence of crisis.

The measures of economic vulnerabilities are measured by growth in bank credit,
credit-to-deposit ratio, house price appreciation, the ratio of current account to GDP,
and public debt as a share of GDP. Claessens et al. (2010) ªnd that house price ap-
preciation, bank credit growth, and the magnitude of current account signiªcantly
explain the crisis spillovers. 相似地, Rose and Spiegel (2011) and Lane and Milesi-
Ferrertti (2011) uncover a connection between of the current account and the sever-
ity of crisis transmission.

On the policy dimension, Berkmen et al. (2012) suggest that exchange-rate policy re-
gime can affect the severity of crisis. More ºexible exchange rates may help to buffer
the external shocks. 此外, credit market regulation is another important pol-
icy factor that explains the incidence of crisis. Giannone Lenza, and Reichlin (2010)
and Rose and Spiegel (2011) ªnd that the countries with more liberal credit market
regulation suffered more from the ªnancial crisis.

In our study, we model the channels of transmission as follows. To measure crisis se-
verity, we focused on the difference between 2008–09 and 2005–07 real GDP growth.
In choosing explanatory variables, we follow Berkmen et al. (2012) and divide trans-
mission channels into their four previously mentioned categories, 即, 贸易
linkages, ªnancial linkages, vulnerabilities, and policy framework. Since Rose and
Spiegel (2011) note that the 2007 values of variables might already be contaminated
by the subprime crisis, we follow their decision to use 2006 values for all explana-
tory variables in the regressions. Our explanatory variables are as follows.

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

1
3
1
8
4
1
6
8
3
3
8
1
A
s
e
p
_
A
_
0
0
2
5
3
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Trade Linkages and Crisis Spillovers

我) Trade linkages

To measure trade linkages, we follow Spiegel (2011) and Da Costa Neto and Romeu
(2011) using trade dependency on the U.S. 市场, measured by a fraction of exports
to the United States to the total exports, and export destination concentration as ex-
planatory variables. Export concentration is included to enable us to test whether
trade diversiªcation lessens crisis severity. In contrast with the previous literature,
when we generate our measures of trade dependency and export destination con-
centration our study includes indirect trade linkages using in addition to the com-
monly included direct trade statistics. Our calculation of total trade linkages follows
the method described in Section 2.1. Due to the importance of this measure, we por-
tray patterns of trade linkages in the next section.

二) Financial linkages

Following Blanchard et al. (2010), we use a fraction of short-term external debt to
GDP as our explanatory variable for ªnancial linkages, since short-term debt is more
sensitive to crisis than long-term debt.

三、) Vulnerabilities

To measure economic vulnerabilities we follow Claessens et al. (2010) and Rose and
Spiegel (2011) using house price appreciation, the ratio of current account, and bank
credit growth.

四号) Policy framework

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

1
3
1
8
4
1
6
8
3
3
8
1
A
s
e
p
_
A
_
0
0
2
5
3
p
d

Because many countries adopted a ºexible exchange rate regime after the 1997–98
Asian ªnancial crisis, we only credit market regulation as our measure of policy.

3. Pattern of trade linkages

To examine the pattern of trade linkages obtained from our calculation, we used the
L1 norm, 即, the Manhattan norm, to calculate distance between vectors of re-
ported bilateral trade ªgures and the calculated total linkages. For each exporting
country i, the L1 distance between a vector of bilateral and a vector of total linkages,
||希||1, could be written as:

||希||1

(西德:3)

∑

全部的
|

direct

|
ij

(10)

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Asian Economic Papers

Trade Linkages and Crisis Spillovers

桌子 2. Proportion (%) of indirect trade transactions in international trade

2000

2004

2008

2009

2010

2011

2000–11

世界
阿尔巴尼亚
阿根廷
Armenia
澳大利亚
奥地利
Azerbaijan
Bahrain
Bangladesh
Belarus
比利时
玻利维亚, Plurinational Republic of
博茨瓦纳
巴西
Bulgaria
Cambodia
Cameroon
加拿大
Caribbean
Central Africa
智利
中国
哥伦比亚
Costa Rica
Cote d’Ivoire
Croatia
塞浦路斯
捷克共和国
丹麦
厄瓜多尔
埃及
El Salvador
爱沙尼亚
埃塞俄比亚
芬兰
法国
乔治亚州
德国
加纳
0希腊
危地马拉
Honduras
香港
匈牙利
印度
印度尼西亚
伊朗, Islamic Republic of
爱尔兰
以色列
意大利
日本
Kazakhstan
肯尼亚
Korea Republic of
Kuwait
Kyrgyzstan
Lao People’s Democratic Republic
拉脱维亚
立陶宛
卢森堡
马达加斯加
马拉维
Malaysia
马耳他
Mauritius
墨西哥
蒙古

11.8
19.7
11.7
23.3
12.6
13.6
18.3
22.4
7.2
19.4
9.3
13.7
22.6
7.8
11.1
11.7
18.1
8.8
7.4
10.7
9.1
12.4
7.4
8.1
13.2
15.2
16.4
15.8
12.0
10.1
14.3
20.5
18.1
17.2
8.8
9.4
18.5
9.7
17.1
10.0
10.1
9.5
14.6
14.1
11.0
12.3
21.0
7.8
8.3
7.7
12.4
18.2
14.1
10.9
25.2
24.0
24.9
14.8
12.5
11.3
14.2
12.3
15.7
15.1
10.4
9.0
23.4

12.2
20.9
13.0
18.8
12.1
13.1
21.3
22.7
7.3
16.9
10.0
13.2
20.9
9.3
10.6
12.8
15.1
8.4
8.8
13.3
10.1
12.4
8.2
8.1
14.6
15.2
13.9
14.5
11.3
10.0
14.3
20.7
17.6
15.7
10.3
9.8
19.2
9.6
18.4
9.5
12.4
9.3
16.9
12.7
12.6
12.5
19.9
9.6
8.5
7.8
12.9
19.7
15.2
12.3
24.7
27.0
23.2
14.0
12.5
11.7
11.8
12.0
14.5
15.9
11.8
9.4
20.3

12.6
22.5
12.4
18.0
13.1
12.4
15.3
22.8
6.5
18.2
10.1
15.0
20.5
10.3
10.3
13.5
16.9
7.7
9.6
13.7
11.7
11.7
9.7
9.4
13.7
14.2
13.6
13.2
11.1
11.0
14.1
9.6
16.3
15.9
9.7
9.6
19.9
9.8
19.5
9.5
9.9
8.1
18.6
12.1
13.5
12.8
16.5
9.7
20.8
8.1
13.3
18.3
15.5
13.9
24.1
26.7
29.5
14.9
12.7
12.2
12.6
15.8
13.8
16.1
13.9
8.5
28.5

12.6
20.2
12.3
17.2
14.0
12.9
17.2
23.0
6.6
18.6
9.9
15.2
20.7
11.2
10.6
21.5
18.1
7.3
9.3
14.4
12.9
11.8
10.3
10.5
14.1
14.0
14.2
13.4
11.1
11.5
14.5
9.6
16.8
17.0
9.2
9.9
20.2
10.2
21.4
9.9
9.4
9.1
18.8
11.4
12.5
13.1
16.6
10.1
9.2
8.0
14.1
17.1
15.6
14.2
24.0
30.1
29.2
15.4
13.0
12.4
11.6
18.3
14.1
16.6
12.8
8.3
30.6

12.8
18.4
12.2
18.6
14.6
13.0
18.1
23.9
6.7
17.7
9.7
14.5
21.2
11.7
10.0
18.1
18.3
7.6
9.3
20.2
13.3
11.2
8.9
10.0
14.6
15.0
14.1
13.3
11.2
10.8
14.0
9.3
16.8
18.1
10.0
10.2
18.3
10.3
21.2
9.8
9.4
7.7
19.3
11.7
13.0
12.9
27.9
9.9
8.9
8.1
13.9
17.1
15.6
13.6
15.3
27.3
30.9
15.2
13.5
12.8
13.7
16.5
13.4
17.5
12.2
8.3
30.8

Asian Economic Papers

13.0
18.4
12.1
19.2
14.8
13.0
17.6
27.0
7.2
18.6
9.7
13.9
20.8
11.7
10.4
14.7
18.8
7.6
9.7
23.1
12.5
11.4
9.0
10.1
15.1
14.7
14.1
13.4
11.1
11.2
13.7
9.8
16.2
16.7
10.2
10.3
20.7
10.4
16.9
11.1
8.8
7.9
20.0
12.6
12.7
13.0
29.0
9.5
9.2
8.2
13.4
18.5
14.1
13.3
15.5
30.8
31.0
15.1
13.5
12.8
15.3
15.0
12.7
15.6
13.0
8.3
31.1

1.1
(西德:4)1.3
0.4
(西德:4)4.1
2.2
(西德:4)0.6
(西德:4)0.8
4.6
0.1
(西德:4)0.8
0.4
0.2
(西德:4)1.8
3.9
(西德:4)0.7
3.0
0.8
(西德:4)1.2
2.3
12.5
3.4
(西德:4)1.0
1.6
2.0
1.9
(西德:4)0.5
(西德:4)2.3
(西德:4)2.4
(西德:4)0.8
1.2
(西德:4)0.6
(西德:4)10.7
(西德:4)1.9
(西德:4)0.5
1.4
0.9
2.1
0.7
(西德:4)0.2
1.1
(西德:4)1.3
(西德:4)1.6
5.4
(西德:4)1.4
1.7
0.7
8.0
1.7
0.9
0.5
1.1
0.2
0.0
2.3
(西德:4)9.7
6.8
6.1
0.3
1.0
1.5
1.1
2.6
(西德:4)3.0
0.4
2.6
(西德:4)0.7
7.7

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

1
3
1
8
4
1
6
8
3
3
8
1
A
s
e
p
_
A
_
0
0
2
5
3
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Trade Linkages and Crisis Spillovers

2000

2004

2008

2009

2010

2011

2000–11

10.8
20.1
18.1
13.5
11.0
9.4
9.7
10.6
11.1
21.8
8.9
13.1
14.6
9.8
13.9
13.7
10.0
19.9
12.5
8.6
22.0
16.7
11.2
12.0
21.2
11.3
25.7
19.2
17.4
17.8
13.9
15.6
21.5
14.6
20.0
11.1
13.2
19.6
14.7
18.6
15.6
13.6
13.7
16.5
8.1
7.8
9.0
8.0
11.8
12.3
11.5
13.2
10.3
20.0
15.0
23.0
9.8
14.2
11.2
9.3
13.5
21.3
15.7

11.1
26.6
17.5
12.7
10.6
9.1
9.2
10.5
11.1
21.8
9.6
11.7
15.6
9.3
15.0
12.8
12.3
18.7
12.8
10.2
26.0
16.9
12.0
11.7
18.6
11.1
25.9
14.6
18.5
19.4
20.7
16.7
22.3
15.4
17.7
10.2
14.3
21.3
16.6
16.9
15.3
12.5
12.8
18.4
8.3
8.3
9.3
8.0
15.2
14.5
11.8
13.3
10.2
20.6
13.9
22.6
9.8
14.1
9.7
10.4
10.8
22.0
16.7

11.7
24.9
18.5
15.2
10.9
10.9
10.1
10.1
11.3
20.6
11.0
15.6
15.6
13.4
14.0
11.9
12.8
21.8
25.7
13.1
28.0
15.5
14.4
13.4
17.8
11.1
25.4
14.8
21.4
20.6
19.2
18.9
25.2
17.2
16.8
10.9
13.4
19.7
18.3
17.3
12.8
12.0
10.9
15.7
8.1
8.5
10.0
7.9
16.6
18.3
12.3
12.9
12.1
19.8
15.7
23.3
10.3
13.0
12.2
15.0
9.8
28.0
20.2

11.6
22.3
14.4
15.3
10.9
10.3
11.3
11.1
11.2
17.1
10.5
13.9
16.3
15.1
14.6
11.6
12.0
19.3
25.9
19.5
27.3
15.1
14.6
13.3
17.9
10.9
21.9
12.3
16.2
19.8
22.4
18.5
27.4
18.2
17.2
10.8
13.1
21.4
17.2
17.3
13.3
12.1
11.6
17.8
8.6
9.0
9.9
7.7
16.5
18.3
12.5
13.9
12.3
18.1
15.6
23.4
9.9
12.8
12.0
25.2
10.1
28.2
20.9

11.3
25.6
14.0
15.9
11.1
10.4
11.2
10.7
11.6
21.8
10.3
15.1
16.8
14.2
15.1
11.8
11.8
18.4
12.2
21.1
24.3
15.2
14.8
14.1
19.8
11.3
21.9
16.2
14.3
20.1
19.6
28.5
28.1
20.3
14.5
10.5
13.3
21.0
18.7
17.1
13.5
11.9
11.5
19.3
8.7
9.2
10.0
7.9
16.6
19.2
12.3
13.0
12.4
18.2
15.7
27.0
9.9
12.8
13.1
25.3
9.2
30.2
20.7

11.4
22.5
17.6
15.9
11.4
10.6
11.9
10.7
11.4
18.4
10.2
13.9
14.5
15.3
13.1
12.0
12.2
17.2
17.6
26.1
29.2
16.3
15.1
14.6
22.4
10.8
20.1
14.4
16.5
21.0
19.4
17.0
24.2
21.1
18.9
11.1
14.7
20.1
18.2
17.0
14.0
12.3
12.9
18.5
8.6
10.3
10.4
8.8
16.9
19.4
12.5
14.0
12.8
18.8
16.0
27.6
9.8
12.7
13.4
26.7
8.5
28.9
22.2

0.6
2.3
(西德:4)0.5
2.4
0.4
1.1
2.2
0.1
0.3
(西德:4)3.4
1.4
0.8
0.0
5.5
(西德:4)0.8
(西德:4)1.7
2.2
(西德:4)2.7
5.1
17.6
7.2
(西德:4)0.4
3.9
2.6
1.2
(西德:4)0.5
(西德:4)5.6
(西德:4)4.8
(西德:4)0.9
3.2
5.4
1.4
2.8
6.4
(西德:4)1.1
0.0
1.5
0.5
3.5
(西德:4)1.6
(西德:4)1.6
(西德:4)1.2
(西德:4)0.8
2.0
0.5
2.5
1.3
0.7
5.1
7.1
1.0
0.8
2.6
(西德:4)1.2
1.0
4.7
0.1
(西德:4)1.5
2.2
17.4
(西德:4)5.0
7.6
6.5

桌子 2. (Continued)

摩洛哥
Mozambique
纳米比亚
Nepal
荷兰
新西兰
Nicaragua
尼日利亚
Norway
Oman
巴基斯坦
Panama
Paraguay
秘鲁
菲律宾
波兰
Portugal
Qatar
Rest of Central America
Rest of East Asia
Rest of Eastern Africa
Rest of Eastern Europe
Rest of EFTA
Rest of Europe
Rest of Former Soviet Union
Rest of North Africa
Rest of North America
Rest of Oceania
Rest of S.African Customs Union
Rest of South America
Rest of South Asia
Rest of Southeast Asia
Rest of the World
Rest of Western Africa
Rest of Western Asia
罗马尼亚
Russian Federation
Saudi Arabia
塞内加尔
新加坡
斯洛伐克
Slovenia
南非
South Central Africa
西班牙
斯里兰卡
瑞典
瑞士
台湾
Tanzania United Republic of
Thailand
Tunisia
Turkey
乌干达
Ukraine
United Arab Emirates
英国
美国
Uruguay
委内瑞拉
Viet Nam
赞比亚
Zimbabwe

来源: 作者的计算.

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

1
3
1
8
4
1
6
8
3
3
8
1
A
s
e
p
_
A
_
0
0
2
5
3
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Trade Linkages and Crisis Spillovers

Where totalij is our calculated value of total linkages from i to j, and directij is the bi-
lateral export from i to j. The right-hand side of equation (10) implies that we sum
all indirect transactions between country i and other countries. The summation dou-
ble-counts transactions, 然而, because indirect trades appear on both sides—for
both exporter and importer. 换句话说, indirect exports from country i to j will
be the indirect imports of country j from i. To arrive at a more intuitive indicator, 我们
computed the value||希||1/2 to represent the proportion of indirect transactions that
involve in country i exports. 桌子 2 displays the calculated values of||希||1/2 对全部
129 countries over the period 2000–11.

平均而言, the world’s indirect trade transactions have been increasing over
time—from 11.8 percent of the total trade transactions in 2000 到 13.0 百分比在
2011. This provides evidence that production-sharing activities have grown in im-
portance over the past decade. 的 129 国家, 有 76 countries where the
proportion of indirect trade exceeds the world’s average, and the other 50 国家
have a lower proportion. 例如, large shares of indirect trades are found in
蒙古, Lao PDR, Kyrgyzstan, 伊朗, and Zambia. 对于这些国家, the indirect
transactions involve as much as 30 percent of their international trade activities. 在
另一方面, countries like Bangladesh, 加拿大, Honduras, 意大利, 和墨西哥
have only small indirect transactions with the proportions well below 10 百分.
The largest growth in indirect activities during 2000–11 were recorded in Venezuela,
伊朗, 蒙古, and Zambia.

4. 结果

In this section we use regression analysis to search for insights into the international
transmission crisis.

4.1 Data description and descriptive statistics
Our study uses country-level data on 108 国家:3 33 developed countries and
75 developing countries.4 The limitation comes from the data of house price appreci-
ation that we could collect only for 35 国家. Our dependent variable, data on
country GDP growth, is obtained from the World Bank. To give information on our

3 We use data from the World Bank’s database, composed of 214 国家. Some countries,
然而, have missing data in some categories such as the short-term external debt. 还,
仅有的 127 countries have public debt data. 因此, 108 countries is the largest panel data
set we can use in our regression.

4 This list of countries follows the World Bank’s World Development Indicators database with
deduction of the countries with missing data. Our developed countries here are listed fol-
lowing the deªnition of IMF’s advanced economies.

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

1
3
1
8
4
1
6
8
3
3
8
1
A
s
e
p
_
A
_
0
0
2
5
3
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Trade Linkages and Crisis Spillovers

桌子 3. Descriptive statistics of variables in the model

Variable

不. of observations

Growth difference (%)
Trade dependency, direct (%)
Trade dependency, 全部的 (%)
Export concentration, direct (%)
Export concentration, 全部的 (%)
Public debt/GDP (%)
Credit market regulation (%)
Current account/GDP (%)
ST external debt / GDP (%)
Bank credit growth (%)
House price appreciation (%)

来源: 作者的计算.

108
108
108
108
108
108
104
109
109
109
35

意思是
(西德:4)4.33022
0.131501
0.153346
0.157809
0.118635
0.4206
8.567202
0.438073
5.347162

12.97766
10.3904

Std. Dev.

4.260868
0.165912
0.149439
0.149672
0.10003
0.015576
1.005234
10.02169
4.634809
14.33622
14.20482

最小
(西德:4)22.0365
0
0.012439
0.045986
0.041025

(西德:4)3.924
5.0982
(西德:4)22.6817
0
(西德:4)36.5489
(西德:4)19.0998

Max

3.318626
0.848658
0.766414
0.901686
0.589822
7.5335
10
44.61526
11.23899
64.02757
65.18899

桌子 4. Correlations of variables in the model

Explanatory variables

Trade dependency, direct
Trade dependency, 全部的
Export concentration, direct
Export concentration, 全部的
Public debt/GDP
Credit market regulation
Current account/GDP
ST external debt / GDP
Bank credit growth
House price appreciation

来源: 作者的计算.

Correlations with growth difference
(西德:4)0.0323
(西德:4)0.0451
(西德:4)0.0668
(西德:4)0.0612
(西德:4)0.0492
(西德:4)0.3148
(西德:4)0.0921
(西德:4)0.0314
(西德:4)0.4438
(西德:4)0.6621

数据集, 桌子 3 shows the average, 标准差, minimum and maximum
of each variable. Variable correlations with our dependent variable—growth
difference—are shown in Table 4. The highest correlation comes from house price
appreciation at –0.66, implying that bubble prices in housing sector may aggravate
crisis severity. Other notably explanatory variables with high correlations are the
growth of bank credit to private sector and credit market regulation.

4.2 Regression results
This section reports the results from our estimation. To provide a comparison
with the earlier literature on this topic, 桌子 5 displays the regression results for our
project when we limit our measure of trade linkages to the common variable that is
calculated using direct trade statistics alone. The results in Table 5 are in line with
the previous literature, because we ªnd that trade linkages, both as a fraction of
export to the U.S. market and export destination concentration, play no role in ex-
plaining the severity of spillovers from the subprime crisis. In comparison, export
destination concentration has more impact on crisis incidence than dependency on
美国. 市场.

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

1
3
1
8
4
1
6
8
3
3
8
1
A
s
e
p
_
A
_
0
0
2
5
3
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Trade Linkages and Crisis Spillovers

桌子 5. Regression results when using direct trade linkages

Variables

Export dependency on U.S., direct linkages

Export concentration, direct linkages

Export dependency (西德:6) dummy DC

Export concentration (西德:6) dummy DC

Credit market regulation

Dummy of being developed countries

Current account/GDP

Short-term external debt/GDP

Bank credit growth

Public debt/GDP

House price appreciation

持续的

观察结果
R-squared

0.908
(0.499)

2.589
(0.327)

(西德:4)1.187*
(0.067)
(西德:4)2.457*
(0.112)
(西德:4)0.00181
(0.00944)
(西德:4)0.211
(0.0226)
(西德:4)0.132
(0.0124)
(西德:4)0.160*
(0.0078)

8.942*
(0.566)

102

0.321

2
(西德:4)0.68
(3.116)
(西德:4)0.558
(0.319)

(西德:4)0.766
(1.082)

0.94
(0.342)

0.0383
(0.00628)

0.264*
(0.0166)
(西德:4)0.0835
(0.0284)
(西德:4)0.648
(0.0853)
(西德:4)0.142
(0.035)
2.978
(9.308)

0.632

1.075
(0.219)

2.388*
(0.111)
(西德:4)2.837
(0.252)

3.986
(0.656)
(西德:4)1.196*
(0.0534)
(西德:4)2.619**
(0.0177)
(西德:4)0.000723
(0.0112)
(西德:4)0.216*
(0.0131)
(西德:4)0.133
(0.0133)

(西德:4)15.51

(1.518)

9.070*
(0.331)

102

0.321

来源: 作者的计算.

笔记: Robust standard errors in parentheses.

**Statistically signiªcant at the 1% 等级; *statistically signiªcant at the 5% 等级.

After running our initial speciªcation, we augment our regression with a developed
country dummy variable, which is interacted with the trade exposure measures.
Such a modiªcation is relevant if the transmission of crisis differed for developed
and developing countries. The new regression result is shown in model 3 of Table 5.
We ªnd that trade concentration may have a role in crisis severity when we separate
trade linkages between developed countries and developing countries. 相比之下
with previous literature such as Grossman and Helpman (1991) and Matsuyama
(1992), which illustrate that trade diversiªcation stimulates economic growth, 我们的
ªndings are similar to Da Costa Neto and Romeu (2011) in ªnding that export desti-
nation concentration lessens crisis severity.

Similar to Giannone et al. (2010) and Rose and Spiegel (2011), we ªnd that the coun-
tries with more liberal credit market regulation suffered more severely from crisis
spillovers. 还, we ªnd that the countries with a high fraction of public debt and
short-term external debt, which represent reliance on foreign credit and ªnancial
linkages, experienced worse crisis outcomes than other countries that had better
ªnancial conditions. This result is similar to Claessens et al. (2010), Lane and Milesi-
Ferrertti (2011), and Rose and Spiegel (2011).

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

1
3
1
8
4
1
6
8
3
3
8
1
A
s
e
p
_
A
_
0
0
2
5
3
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Trade Linkages and Crisis Spillovers

此外, the regression results indicate that developed countries were affected by
the sub-prime crisis more heavily than developing countries.

最后, we observe that even though house price appreciation has high negative
correlation with crisis severity, the regression estimates do not reveal evidence that
house price appreciation explained the incidence of crisis. It is important to point
出去, 然而, that we have limited observations in non-house price appreciation
(35 国家). 因此, the regression result on this issue may be weak due to the
small sample size.

总共, the results in Table 5 point out that ªnancial vulnerability plays the most
important role as a crisis transmission channel.

表中 6 we replace the trade measures that are based solely on direct trade with
our new measures of total trade linkages, and repeat the analysis of Table 5. As be-
fore we ªnd that a share of public debt to GDP and credit market regulation are still
the most important explanatory variables for crisis severity.

Although the full sample results in Table 6 suggest that trade linkages are
insufªcient to explain the transmission of crisis for our full sample countries, 贸易
linkages are found to be relevant when we allow for differential transmission to de-
veloping versus developed countries. 尤其, the results in Table 6 显示
developed countries with a higher share of exports to the U.S. market suffer from
ªnancial crisis during 2008–09 more severely than the developed countries that were
less closely tied by trade. 因此, in contrast with Rose and Spiegel (2011), we ªnd
that trade dependency to the U.S. market could explain crisis severity of developed
countries when total trade linkages are used instead of direct trade statistics. 那里-
gression results conªrm our hypothesis that direct trade statistics may fail to accu-
rately capture trade pattern for countries that are deeply connected via international
production networks.

As for trade diversiªcation, 桌子 6 shows that developed countries with higher ex-
port concentration experienced less severe crises that did countries whose exports
were more diversiªed.

It is also important to point out that trade linkages could not explain the crisis sever-
ity of developing countries. One possible explanation for the lack of regression evi-
dence for a connection is that the GFC was a global phenomenon, unlike the more
regional crises during 1971–97. 因此, trade linkages may play a less important
transmission role than they did in the regional crises. The developing countries hit

100

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

1
3
1
8
4
1
6
8
3
3
8
1
A
s
e
p
_
A
_
0
0
2
5
3
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Trade Linkages and Crisis Spillovers

桌子 6. Regression results when using total trade linkages

Variables

Export dependency on U.S., total linkages

Export concentration, total linkages

Export dependency (西德:6) dummy DC

Export concentration (西德:6) dummy DC

Credit market regulation

Dummy of being developed countries

Current account/GDP

Short-term external debt/GDP

Bank credit growth

Public debt/GDP

House price appreciation

持续的

观察结果
R-squared

0.355
(1.204)

4.757
(1.581)

(西德:4)1.198*
(0.0691)
(西德:4)2.453
(0.223)
(西德:4)0.00141
(0.00731)
(西德:4)0.212
(0.0343)
(西德:4)0.133
(0.014)
(西德:4)0.153**
(0.00234)

8.958*
(0.675)

102

0.323

来源: 作者的计算.

笔记: Robust standard errors in parentheses.

**Statistically signiªcant at the 1% 等级; *statistically signiªcant at the 5% 等级.

2
(西德:4)5.29
(1.358)

6.008
(4.558)

(西德:4)0.843
(1.024)

1.185
(0.384)

0.0388
(0.0145)

0.293
(0.0322)
(西德:4)0.0985
(0.0387)
(西德:4)0.623
(0.102)
(西德:4)0.14
(0.0257)

3.624
(8.9)

35
0.634

0.696
(0.567)

4.522
(1.153)
(西德:4)3.074*
(0.157)

3.656*
(0.125)
(西德:4)1.212*
(0.0472)
(西德:4)2.551*
(0.180)
(西德:4)0.000693
(0.00830)
(西德:4)0.219
(0.0225)
(西德:4)0.133
(0.0145)

(西德:4)14.82*
(1.148)

9.094*
(0.445)

102

0.324

more severely by the GFC are the countries with more vulnerability such as higher
public debt ratio, weak credit market regulation, or a larger fraction of short-term
external debt.

5. 结论

Previous studies such as Berkmen et al. (2012), Rose and Spiegel (2011), Da Costa
Neto and Romeu (2011), and Blanchard et al. (2010) point out that trade linkages do
not explain cross-country crisis severity stemming from the GFC. We suggest that
direct trade linkages may play a less important role in a crisis transmission follow-
ing the GFC than they did in the earlier crises during 1971–97, since the composition
of trade has changed. 尤其, due to the growth of international production net-
作品, direct trade statistics fail to capture trade exposure. 所以, to improve the
measures of total trade linkages, we use a trade matrix to construct trade depend-
ence measures that include both direct and indirect trade connections. When we in-
clude indirect trade linkages, our ªndings show that export dependency on the

101

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

1
3
1
8
4
1
6
8
3
3
8
1
A
s
e
p
_
A
_
0
0
2
5
3
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Trade Linkages and Crisis Spillovers

我们. market signiªcantly crisis severity in case of developed countries. 因此,
trade linkages are still one of the important causes of crisis when indirect trade
linkages are taken into account. 因此, while our ªndings conªrm that trade statis-
tics alone cannot fully explain trade exposure and trade pattern nowadays, our anal-
ysis shows that it is important to include indirect trade linkages to analyze trade-
related issues. 最后, similar to previous work we ªnd that vulnerabilities such as a
high public debt ratio, high fraction of short-term foreign debt, and weak credit
market regulation remain important in explaining the international transmission of
the crisis.

参考

Berkmen, Pelin S., Gaston Gelos, Robert Rennhack, and James P. Walsh. 2012. The Global
Financial Crisis: Explaining Cross-country Differences in the Output Impact. 杂志
International Money and Finance 31(1):42–59.

BIS. 2007. 77th Annual Report 2006/07. 巴塞尔: Bank for International Settlements.

Blanchard, 奥利维尔, Mitali Das, and Hamid Faruqee. 2010. The Initial Impact of the Crisis on
Emerging Market Countries. Brookings Papers on Economic Activity 41(1):263–323.

Cheewatrakoolpong, K., 和S. Manprasert. 2012. Trade Concentration and Crisis Spillover:
Case Study of Transmission of the Subprime Crisis to Thailand. ARTNeT Working Paper Series
不. 112. Bangkok: UNESCAP.

Claessens, S。, G. Dell’Ariccia, D. Igan, 和L. Laeven. 2010. Cross-Country Experience and
Policy Implications from the Global Financial Crisis. Economic Policy 62:269–293.

Da Costa Neto, 纳尔逊, and Rafael Romeu. 2011. Did Export Diversiªcation Soften the Impact
of the Global Financial Crisis? IMF Working Paper No. WP/11/99. 华盛顿, 直流: 内特纳-
tional Monetary Fund.

Eichengreen, Barry, Andrew Rose, and Charles Wyplosz. 1997. Contagious Currency Crises.
NBER Working Paper Series No. 5681. 剑桥, 嘛: National Bureau of Economic
研究.

Giannone, D ., 中号. Lenza, 和L. Reichlin. 2010. Market Freedom and the Global Recession.
CEPR Discussion Paper No. 7884. 华盛顿, 直流: Center for Economic and Policy Research.

Glick, Reuven, and Andrew Rose. 1999. Why Are Currency Crises Contagious? Review of World
经济学 134(4):664–691.

Grossman, 基因, and Elhanan Helpman. 1991. Trade, Knowledge Spillovers, and Growth.
European Economic Review 35(2–3):517–526.

Hesse, Heiko. 2008. Export Diversiªcation and Economic Growth. Commission on Growth and
发展. Working Paper No. 21. 华盛顿, 直流: 世界银行.

约翰逊, 右. C。, and Guillermo Noguera. 2009. Accounting for Intermediates: Production
Sharing and Trade in Value-Added. Working Paper. Dartmouth, NH: Dartmouth College.

102

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

1
3
1
8
4
1
6
8
3
3
8
1
A
s
e
p
_
A
_
0
0
2
5
3
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3

Trade Linkages and Crisis Spillovers

Kali, Raja, and Reyes, Javier A. 2005. Financial Contagion on the International Trade Network.
Working Paper, 经济系. Fayetteville: University of Arkansas, Sam M.
Walton College of Business.

Koopman, R。, Z. 王, 和S. Wei. 2008. How Much of Chinese Exports is Really Made In
中国? Assessing Domestic Value-Added When Processing Trade is Pervasive. NBER Working
Paper No. 14109. 剑桥, 嘛: 国家经济研究局.

Lane, Philip R., and Gian M. Milesi-Ferretti. 2011. The Cross-country Incidence of the Global
Crisis. IMF Economic Review 59(1):77–110.

Matsuyama, Kiminori. 1992. Innovation and Growth in the Global Economy: Gene M.
Grossman and Elhanan Helpman. Journal of International Economics 33(3–4):383–386.

Meilak, Chris. 2011. Measuring Export Concentration: The Implications for Small States. Bank
of Valletta Review 37(春天 2008):35–48.

磨坊主, 右. E., 和P. D. 布莱尔. 1985. Input-Output Analysis: Foundations and Extensions. Englewood
Cliffs, 新泽西州: Prentice Hall.

Perry, Guillermo, and Daniel Lederman. 1998. Financial Vulnerability, Spillover Effects, and Conta-
只园: Lessons from the Asian Crises for Latin America. World Bank Latin American and Caribbean
学习.

Rose, 安德鲁, and Mark Spiegel. 2009. Cross-country Causes and Consequences of the 2008
Crisis: Early Warning. NBER Working Paper No. 15357. 剑桥, 嘛: 国家统计局
经济研究.

Rose, 安德鲁, and Mark Spiegel. 2011. Cross-country Causes and Consequences of the Crisis:
An Update. European Economic Review 55(3):309–324.

Samen, Salomon. 2010. A Primer on Export Diversiªcation: Key Concepts, Theoretical Under-
pinnings and Empirical Evidence. Growth and Crisis Unit, World Bank Institute, 可能.

Sato, Kiyotaka, and Nagendra Shrestha. 2012. Global and Regional Shock Transmission—New
Evidence from Globally Integrated Input–Output Table. Paper presented at the 13th Interna-
tional Convention of the East Asian Economic Association, 十月 2012, 新加坡.

103

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

1
3
1
8
4
1
6
8
3
3
8
1
A
s
e
p
_
A
_
0
0
2
5
3
p
d

乙
y
G
你
e
s
t

哦
n
0
8
S
e
p
e
米
乙
e
r
2
0
2
3
下载pdf