How Does Terms-Of-trade Behavior Shape International Finan

Int. Fin. Markets, Inst.
and Money 33 (2014) 335353
Contents lists available at ScienceDirect
Journal of International Financial

Markets, Institutions & Money
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / i n t fi n
How does terms-of-trade behavior shape international

nancial integration in primary-commodity exporting
economies?
Almukhtar Saif Al-Abri *
Department of Economics and Finance Sultan Qaboos University, PO Box 20, Alkod, Muscat 123, Oman
A R T I C L E I N F O
A B S T R A C T
Article history:
Received 4 December 2013
Received in revised form 29 June 2014
Accepted 4 September 2014
Available online 16 September 2014
This paper presents empirical evidence suggesting that the volatility, trend-growth, and
shock-duration of terms-of-trade (TOT) are important drivers of the degree and
composition of international nancial integration (IFI). Our results are based on a panel
of 55 primary-commodity exporting countries during 19802007. The ndings reveal that
TOT trend-growth has larger impact on IFI compared to TOT volatility. Also, higher TOT
volatility is robustly associated with greater cross-holdings of foreign assets and lower
cross-holdings of foreign liabilities. Another notable nding is that longer duration of TOT
shocks seems to shift IFI towards equity assets.
2014 Elsevier B.V. All rights reserved.
JEL classication:
F210
F320
Keywords:
Terms-of-trade
Financial globalization
International investment positions
Foreign direct investment
Portfolio investment
Developing countries
1. Introduction
Over the last three decades the developing countries have become more nancially globalized1. However, the degree and
speed of this development is not uniform across the developing countries (Chinn and Ito, 2008; Kose et al., 2009a; Lane and
Milesi-Ferretti, 2007). To date, little research has been done to understand the drivers of nancial globalization across the
developing countries, as research investigating this issue has been largely conned to the developed countries (for example,
Furceri et al., 2011; Lane, 2000; Lane and Milesi-Ferretti, 2008)2 . The diversity of economic structures, institutional
environments, and level of country risk between developed and developing countries warrant a new consideration of the
determinants of nancial globalization.
* Tel.: +968 24142942; fax: +968 24414043.

E-mail address: asalabri@squ.edu.om (A. Saif Al-Abri).
1
The literature uses nancial openness and international nancial integration interchangeably with nancial globalization. Financial openness
usually refers to de jure measures of legal restrictions of cross-border capital ows, while international nancial integration refers to volume-based de facto
measures of stocks and ows of cross-border assets and liabilities (Chinn and Ito 2008; Lane and Milesi-Ferretti, 2007; Miniane, 2004; Quinn, 2003).
2
A related literature considers the determinants of international capital ows. Broto et al. (2011) has also examined the determinants of volatility of
capital ows in emerging markets. Terms-of-trade volatility has also been found to affect changes in capital ows. Blattman et al. (2007) nd a negative
effect of terms-of-trade volatility on capital inow during 18701940 for a large sample of developed and developing countries. Calvo et al. (2004) nd that
negative terms-of-trade shocks increase the likelihood of a sudden stop in capital inows and nancial crises.
http://dx.doi.org/10.1016/j.intn.2014.09.002
1042-4431/ 2014 Elsevier B.V. All rights reserved.
336
A. Saif Al-Abri / Int. Fin. Markets, Inst. and Money 33 (2014) 335353
A common feature of most developing countries is the dominance of a few primary commodities in their exports. The
empirical literature suggests that developing countries are fundamentally more exposed to TOT uctuations (Cashin et al.,
2004a; Hausmann et al., 2006). Adverse shocks to a countrys TOT not only disrupt the economys growth, but may also
introduce some level of instability. For example, Mendoza (1995) and Kose (2002) nd that TOT shocks account for at least
half of the output volatility in developing countries. Although TOT volatility is a key determinant of macroeconomic
performance in a majority of the developing countries, the empirical evidence suggests that there are wide differences across
these countries in terms of TOT volatility, trend, and shock persistence. For example, Cashin et al. (2004b) nd the TOT shock
persistence to vary widely across 42 Sub-Saharan African countries.
In light of this inherent heterogeneity of TOT uctuations and IFI in developing countries, it is important for trade and
nancial policy development to investigate the role of TOT uctuations in shaping IFI. The anticipation is that the ndings of
such a study would provide more meaningful and robust understanding of the response of cross-holdings of foreign assets
and liabilities to TOT uctuations. This would help the developing countries determine the sustainability and likelihood of
future trends in IFI. In particular, the ndings of this paper can be used as guidelines in designing more effective capital
controls and other domestic policies in order to enhance the stability of their international investment positions and attract
more productive nancing.
Although the dynamics of TOT are often the centerpiece in open economy models of international adjustment, theoretical
models do not provide a clear guide to the effects of TOT volatility on IFI. We draw two hypotheses from the different strands
of the literature. One hypothesis suggests that, to smooth domestic consumption and investment, countries with more
volatile environments seek more nancial integration with the rest of the world. In this case, greater uctuations in TOT
increase the level of international risk-sharing and IFI (Svensson, 1988; Obstfeld and Rogoff, 1996; Lane, 2001). On the
contrary, international portfolio models suggest that higher volatility in TOT could result in uctuations in real returns to
foreign investment, which might deter foreign investors and, thus, lead to lower IFI (Devereux and Sutherland, 2009). Also,
the classic risk-sharing model of Cole and Obstfeld (1991) suggests that TOT changes can act as an automatic stabilizer when
they are negatively correlated to output uctuations, reducing the motive for international risk-sharing. The empirical
literature that attempted to verify the implications of these two hypotheses mostly concentrate on the effects of TOT
volatility on capital ows (for example, Blattman et al., 2007; Broto et al., 2011; Eichengreen, 1996; Mendoza and Terrones,
2008). We deviate from this literature by examining the effects of TOT volatility, trend changes, and shock persistence on the
size of different forms of cross-holdings of foreign assets and liabilities. As discussed in Kose et al. (2009a), the use of gross
stocks is preferable to annual capital ows in measuring international integration, as the latter tend to be more volatile and
prone to measurement errors. In addition, the use of gross stocks, compared to net stocks, provides a better measure of
integration and efcient risk-sharing as it captures two-way interactions between economies with different risk portfolios.
The paper contributes to the literature by addressing the TOTIFI nexus in a number of distinct ways. First, the paper
focuses on a more homogenous panel and emphasizes the long-run nature of the relationship by taking 5-year
non-overlapping windows of the data. Previous empirical studies were based on cross section samples (for example, Lane,
2000). Second, the relationship is analyzed using sub-measures of nancial integration, including foreign direct investment
(FDI - hereafter), foreign portfolio equity, and foreign debt. The association is also analyzed separately with respect to
cross-holdings of foreign assets and foreign liabilities for these sub-measures. These detailed investigations provide deeper
insights into the nature and channels of association of TOT movements and IFI. This is important because differences in
the composition of foreign assets and liabilities have important effects on real activity (Bosworth and Collins, 1999), on the
probability of sudden stops (Calvo, 2007), on banking and currency crises (Furceri et al., 2012), and on the real exchange rate
volatility (Al-Abri, 2013).
Third, while previous studies mainly used TOT volatility as their only measure of TOT uctuations, this paper employs two
other measures of TOT uctuations: trend-growth and shock-duration. The theoretical literature and the international
evidence suggest that cross-country average growth rates depend on the mean and the variance of the rate of change of TOT
(Blattman et al., 2007; Mendoza, 1997; Turnovsky and Chattopadhyay, 2003)3 . Since the impact of TOT on cross holdings of
foreign assets and liabilities should depend on the effects of the former on national income and savings (Devereux and
Sutherland, 2009; Obstfeld and Rogoff, 1996), we include both the trend-growth and volatility of TOT in our empirical model.
For the TOT shock-duration, the inter-temporal approach to the current account and the consumption smoothing behavior
predict that if TOT shocks are short-lived then it is a sensible strategy to use external nancing to smooth consumption.
However, if TOT shocks are long-lived then relying on external nancing is not sustainable and should eventually lead to
long-term changes in consumption and saving (Cashin and McDermott, 2003; Obstfeld, 1982; Ostry and Reinhart, 1992).
Thus, these studies suggest that the impact of TOT on IFI depends crucially on the expected duration of the TOT shock. The
trend-growth is calculated using the smoothing procedure of HodrickPrescott (HP) lters. The shock-duration is measured
using the half-life of shocks, as proposed by Andrews (1993). To our knowledge, no previous study has analyzed the impact
of TOT trend changes and shock persistence on IFI or its sub-components. Fourth, the paper analyzes the possibility of
non-linearities by allowing the impact of TOT volatility on IFI to vary with a number of factors including: trade openness,
export diversication, nancial development, economic development, and nominal exchange rate exibility.
Mendoza (1997) shows that TOT average growth and volatility are both needed for the model to be properly specied.
337
2.0
0.4
1.8
0.2
1.6
0.0
1.4
-0.2
1.2
-0.4
1.0
-0.6
0.8
-0.8
0.6
INDEX
% of GDP
-1.0
80
82
84
86
88
90
92
94
96
98
00
02
04
06
Year
IFI
KAOPEN
Fig. 1. Overall trend of nancial globalization in selected developing countries, 19802007.

IFI is the international nancial integration measured as the sum of total foreign assets and total foreign liabilities as a % of GDP (source: Lane and
Milesi-Ferretti (2007)); KAOPEN is Chinn and Ito (2008)s de jure index of capital account restrictions, where larger values indicate less restrictions
(source: Chinn and Ito (2008)); countries included are listed in Table A1
The paper also provides a notable contribution to the literature on international risk-sharing in the developing countries.
The literature that examines the level of risk sharing focuses on the correlation of consumption and output. For example,
Kose et al. (2009b) nd that despite greater nancial globalization, developing countries have little ability to share risk due to
the dominance of debt assets in their portfolio. Theoretically, greater nancial integration would help a country stabilize its
national consumption and investment against adverse shocks through international risk-sharing and inter-temporal
substitution (Obstfeld and Rogoff, 1996). Our analysis provide an alternative examination of the degree of risk-sharing
in the developing countries. Since TOT are dominant and exogenous to primary-commodity countries, the response of
cross-holdings of foreign assets to these exogenous idiosyncratic shocks reects the degree of risk-sharing.
Using a sample of 55 primary commodity-exporting countries for the period 19802007, the results of this paper suggest
that TOT volatility, trend, and shock-duration are important long-term drivers of IFI in primary commodity-exporting
countries4,5 . This is robust across the different components of foreign assets and foreign liabilities (FDI, debt, and portfolio
equity investments) and different sub-samples of countries and time periods. The results are also robust to alternative
estimation methods, and to possible variations in the denitions of volatility and growth of TOT. The ndings uncover a
number of interesting patterns. Although TOT has a negative impact on the aggregate IFI measure (consistent with Lane
(2000)s ndings for developed counties), higher TOT volatility is robustly associated with greater holdings of foreign assets
and lower holdings of foreign liabilities. This evidence helps bridging the two hypotheses, outlined above, which predict the
effects of TOT changes on foreign cross-holdings. The size of parameter estimates suggests that TOT trend-growth has larger
impact on IFI compared to TOT volatility. In addition, cross-holdings of portfolio assets seem more responsive to TOT
volatility and trend changes compared to FDI. Furthermore, the ndings suggest that the effects of TOT volatility on IFI are
increasing in the level of trade openness and nancial development, and decreasing in the level of nancial openness and
economic development. Another notable nding is that longer duration of TOT shocks seems to shift nancial globalization
towards equity assets and away from debt assets.
The structure of the paper is as follows. Section 2 provides some salient features of IFI and TOT movements. Section 3
discusses the conceptual frameworks linking IFI to TOT volatility. Section 4 describes the methodology and data. Section 5
discusses the results, and Section 6 concludes.
The list of primary-commodity exporting countries is taken from Cashin et al. (2004a).
We limit the sample to 2007 to avoid the period of dramatic collapse in international capital ows and the sharp depreciations in nancial assets caused
by the international nancial crises 20072009 (Milesi-Ferretti and Tille, 2011).
5
338
2.0
% of GDP
1.6
1.2
0.8
0.4
0.0
80
82
84
86
88
90
92
94
96
98
00
02
04
06
Year
TD
TFDI
IFI
Fig. 2. Overall trend of nancial globalization for selected developing countries using different forms of foreign investment, 19802007.
IFI is the international nancial integration measured as the sum of total foreign assets and total foreign liabilities as a % of GDP; TFDI is the sum of foreign
direct investment assets and liabilities as a % of GDP; TD is the sum of foreign debt assets and liabilities as a % of GDP; source: Lane and Milesi-Ferretti
(2007); countries included are listed in Table A1.
2. Financial globalization and TOT volatility

The last three decades have witnessed signicant changes in the nancial globalization map of the developing countries.
During the 1980s and 1990s, many developing countries had liberalized their capital accounts, experiencing dramatic
growth in their gross cross-holdings of foreign assets and liabilities6 . For a group of 55 developing countries, Fig. 1 depicts the
overall trend in two measures of nancial integration: a de facto volume-based measure of total foreign assets and total
foreign liabilities as a share of GDP (International nancial integration IFI) and a de jure index representing the degree of
capital account liberalization (KAOPEN) of Chinn and Ito (2008). In both measures, larger values indicate higher degree of
nancial globalization.
According to the patterns noted in Fig. 1, it is clear that the two measures do not follow the same pattern. The de jure
measure shows a progressive and steady relaxation of capital controls in these countries after 1987 (except for a short period
following the Asian nancial crisis of 1997). However, the volume-based measure of cross-holdings of foreign assets and
liabilities (IFI) shows a substantial acceleration during 1980s, a mixed path during 1990s, and a negative trend during 2000s.
To gain further insights into the extent and composition of nancial globalization in the developing countries, Fig. 2
depicts the overall trend in gross cross-holdings of FDI (TFDI) and foreign debt (TD) for a group of 55 developing countries.
Two notable patterns stand out. First, although debt assets (portfolio debt, bank loans and deposits, and other debt
instruments) are still the dominant form of cross-holdings of foreign assets and liabilities, it shows a downward trend after
19907. Second, the share of FDI shows a steady increase throughout the period 19802007, with higher speed after 1990.
These patterns indicate that nancial globalization in the developing countries has increasingly been favoring long-term
equity integration (i.e., FDI). Further decomposition of nancial globalization in the developing countries is provided in the
three panels of Fig. 3. These depict the trend in aggregate foreign assets and foreign liabilities, FDI assets and liabilities, and
debt assets and liabilities, respectively. At the aggregate level, the share of foreign assets is increasing and the share of foreign
liabilities is decreasing, suggesting that net foreign position in the developing countries is slowly improving. The same
pattern is also observed in the changes of cross-holdings of debt assets and liabilities. However, this observation is not
carried forward to FDI as foreign assets have shown slower growth compared to foreign liabilities. In other words, the net FDI
(FDI assets less FDI liabilities) of developing countries has decreased. These observations support the notion that developing
countries participate in cross country risk-sharing by accumulating debt assets claim on advanced countries, while also
issuing claims in the form of FDI liabilities that are held by residents of advanced countries (Devereux and Sutherland, 2009;
Lane and Milesi-Ferretti, 2007).
Despite these overall trends of nancial globalization in the developing countries, it is important to note that countries
exhibit considerable heterogeneity in terms of the size and composition of their IFI. Table 1 reports the total foreign assets
and foreign liabilities for the aggregate measure of nancial integration and its subcomponents, averaged for the period
6
Capital ows to developing countries have also witnessed important changes in terms of size and composition. Capital ows to developing countries
have become mostly private (compared to ofcial ows-grants and loans from governments and multilateral institutions), have increased dramatically, and
have been dominated by foreign direct investment (Kose et al., 2009a).
7
As dened by Lane and Milesi-Ferretti, 2007 (pp. 228), the debt category is the sum of portfolio debt securities, bank loans and deposits, and other debt
instruments.
339
2.0
1.6
% of GDP
1.2
0.8
0.4
0.0
-0.4
-0.8
80
82
84
86
88
90
92
94
96
98
00
02
04
06
Year
IFI
NFA
TFA
TFL
.5
% of GDP
.4
.3
.2
.1
.0
80
82
84
86
88
90
92
94
96
98
00
02
04
06
Year
TFDI
FDIL
FDIA
2.0
% of GDP
1.6
1.2
0.8
0.4
0.0
80
82
84
86
88
90
92
94
96
98
00
02
04
06
Year
TD
DA
DL
Fig. 3. Decomposition of different forms of international nancial integration for selected developing countries, 19802007.
In the upper panel, IFI is the international nancial integration measured as the sum of total foreign assets (TFA) and total foreign liabilities (TFL); NFA is the
net foreign assets measured as total foreign assets minus total foreign liabilities; in the middle panel, TFDI is the sum of foreign direct investment assets
(FDIA) and foreign direct investment liabilities (FDIL); in the lower panel, TD is the sum of foreign debt assets (DA) and debt liabilities (DL); all measures are
expressed as a % of GDP; source: Lane and Milesi-Ferretti (2007); countries included are listed in Table A1.
19802007 and grouped by geographic locations. As can be seen, for the aggregate measure, IFI, and its subcomponents, the
degree of nancial globalization varies widely across these countries. In summary, there is clear evidence not just of an
increase in nancial globalization for developing countries, but also of a movement in components of foreign cross-holdings.
The trend and volatility of TOT in the developing countries have also showed dramatic changes over the past three
decades. In general, the TOT average volatility had peaked during 19851987, showed a downward trend subsequently,
before it turned upward again after 2004 (Fig. 4, lower panel). On the other hand, the TOT average trend-growth showed a
downward trend during the period 19801989 and a strong upward trend after 1990 (Fig. 4, upper panel). Despite these
340
Table 1
Different forms of cross-holdings of foreign assets and liabilities in selected developing countries, 19802007.
African countries
IFI
TFA
TFL
TD
DA
DL
PEA
TFDI
FDIA
FDIL
Burundi
Cameroon
Central African Republic
Cote dIvoire
Ethiopia
Gabon
Ghana
Kenya
Madagascar
Malawi
Mali
Mauritania
Mauritius
Mozambique
Niger
Nigeria
Senegal
South Africa
Sudan
Tanzania
Togo
Uganda
Zambia
1.294
0.728
0.976
1.641
0.786
0.976
1.064
0.946
1.160
1.296
1.220
2.230
0.686
1.685
1.030
1.444
1.050
0.867
1.323
1.377
1.744
0.835
2.375
0.300
0.113
0.193
0.239
0.127
0.227
0.216
0.233
0.166
0.155
0.187
0.245
0.272
0.237
0.174
0.324
0.202
0.345
0.096
0.289
0.375
0.182
0.205
0.994
0.615
0.783
1.402
0.660
0.749
0.848
0.712
0.995
1.141
1.033
1.985
0.414
1.448
0.856
1.120
0.848
0.522
1.227
1.089
1.369
0.653
2.170
1.152
0.553
0.713
1.439
0.652
0.671
0.803
0.763
1.046
1.080
1.010
1.882
0.407
1.424
0.750
0.912
0.846
0.298
1.221
1.161
1.192
0.663
1.834
0.208
0.073
0.061
0.186
0.077
0.130
0.129
0.159
0.110
0.090
0.074
0.169
0.073
0.130
0.064
0.147
0.127
0.067
0.083
0.197
0.194
0.104
0.150
0.943
0.480
0.652
1.253
0.575
0.541
0.674
0.604
0.937
0.990
0.936
1.712
0.334
1.294
0.686
0.764
0.719
0.231
1.138
0.964
0.998
0.558
1.684
0.000
0.000
0.000
0.010
0.000
0.001
0.000
0.003
0.000
0.000
0.000
0.000
0.008
0.002
0.000
0.000
0.004
0.103
0.000
0.000
0.004
0.000
0.000
0.052
0.151
0.155
0.148
0.084
0.252
0.161
0.117
0.058
0.151
0.103
0.274
0.087
0.152
0.208
0.403
0.140
0.328
0.087
0.124
0.339
0.094
0.483
0.001
0.017
0.024
0.002
0.000
0.055
0.008
0.011
0.001
0.002
0.011
0.001
0.016
0.000
0.040
0.052
0.018
0.148
0.000
0.000
0.012
0.000
0.000
0.051
0.134
0.131
0.146
0.084
0.197
0.153
0.106
0.056
0.149
0.092
0.273
0.071
0.152
0.169
0.351
0.122
0.180
0.087
0.123
0.326
0.094
0.483
Asian countries
Bangladesh
India
Indonesia
Malaysia
Pakistan
Papua New Guinea
Philippines
Sri Lanka
Thailand
0.505
0.364
0.901
1.571
0.612
1.397
1.109
0.901
0.955
0.074
0.090
0.185
0.593
0.096
0.278
0.270
0.209
0.274
0.431
0.273
0.716
0.977
0.515
1.119
0.839
0.692
0.681
0.404
0.224
0.679
0.632
0.493
0.751
0.829
0.708
0.504
0.034
0.018
0.080
0.175
0.048
0.107
0.149
0.115
0.075
0.370
0.206
0.600
0.457
0.445
0.644
0.680
0.594
0.429
0.000
0.001
0.001
0.015
0.001
0.005
0.008
0.000
0.002
0.061
0.035
0.093
0.486
0.062
0.440
0.130
0.095
0.175
0.001
0.005
0.009
0.113
0.005
0.039
0.018
0.002
0.016
0.061
0.029
0.084
0.373
0.056
0.401
0.112
0.093
0.159
Latin American countries

Argentina
Bolivia
Brazil
Chile
Colombia
Costa Rica
Dominica
Dominican Republic
Ecuador
Guatemala
Honduras
Mexico
Nicaragua
Paraguay
Peru
St. Vincent and the Grenadines
Uruguay
Venezuela, RB
1.156
1.285
0.657
1.501
0.742
1.123
1.689
0.660
1.105
0.575
1.406
0.779
3.422
0.885
1.072
1.301
1.355
1.284
0.454
0.249
0.160
0.505
0.256
0.330
0.643
0.097
0.107
0.228
0.347
0.177
0.319
0.376
0.246
0.327
0.563
0.607
0.702
1.036
0.497
0.996
0.486
0.793
1.045
0.564
0.998
0.348
1.059
0.602
3.103
0.509
0.826
0.974
0.792
0.677
0.862
0.923
0.384
0.758
0.476
0.770
1.024
0.423
0.830
0.408
1.081
0.500
3.119
0.605
0.729
0.525
1.201
0.905
0.329
0.164
0.060
0.191
0.130
0.225
0.547
0.060
0.042
0.156
0.246
0.104
0.223
0.218
0.095
0.204
0.502
0.409
0.533
0.759
0.323
0.567
0.346
0.545
0.477
0.364
0.788
0.253
0.835
0.396
2.896
0.387
0.634
0.321
0.699
0.496
0.029
0.000
0.004
0.082
0.006
0.003
0.000
0.002
0.000
0.000
0.000
0.007
0.000
0.001
0.024
0.000
0.003
0.015
0.179
0.279
0.166
0.452
0.152
0.249
0.569
0.201
0.211
0.097
0.223
0.155
0.212
0.144
0.165
0.653
0.129
0.196
0.030
0.003
0.044
0.067
0.021
0.007
0.000
0.002
0.003
0.003
0.001
0.012
0.005
0.023
0.007
0.000
0.005
0.042
0.149
0.276
0.122
0.385
0.132
0.243
0.569
0.200
0.207
0.095
0.222
0.143
0.206
0.122
0.159
0.653
0.124
0.154
Middle Eastern countries

Algeria
Morocco
Saudi Arabia
Tunisia
Turkey
0.758
1.083
1.390
1.292
0.647
0.261
0.264
1.051
0.148
0.158
0.497
0.819
0.340
1.144
0.489
0.511
0.784
1.025
0.621
0.493
0.074
0.132
0.890
0.048
0.089
0.437
0.652
0.135
0.573
0.404
0.000
0.004
0.005
0.011
0.003
0.064
0.165
0.229
0.538
0.072
0.005
0.009
0.027
0.001
0.007
0.059
0.156
0.202
0.537
0.065
Notes: IFI is the international nancial integration measured as the sum of total foreign assets (TFA) and total foreign liabilities (TFL); TFDI is the sum of
foreign direct investment assets (FDIA) and foreign direct investment liabilities (FDIL); TD is the sum of foreign debt assets (DA) and debt liabilities (DL); all
measures are expressed as a % of GDP; source: Lane and Milesi-Ferretti (2007).
341
.015
.010
dlog(hptrend)
.005
.000
-.005
-.010
-.015
-.020
-.025
80
82
84
86
88
90
92
94
96
98
00
02
04
06
00
02
04
06
Year
Average TOT trend growth
Standard deviation of trend growth
.28
.24
.20
.16
.12
.08
.04
80
82
84
86
88
90
92
94
96
98
Year
Average TOT volatility
Fig. 4. Average terms-of-trade trend-growth and volatility for selected developing countries, 19802007.
In the upper panel, the trend-growth is calculated as the average percentage change of the HodrickPrescott (HP) smoothed trend; in the lower panel, the
TOT volatility is the standard deviation of TOT uctuations around the HP smoothed trend; source: Authors calculation based on the World Banks World
Development Indicators (WDI) data; countries included are listed in Table A1.
general patterns, the developing countries have also exhibited different experiences of TOT trend-growth and volatility.
Table 2 summarizes three measures of TOT changes: TOT volatility, TOT trend-growth, and TOT shock-duration during the
period 19802007 for a group of 55 developing countries. The three measures show the large heterogeneity in TOT behavior
across the developing countries.
In light of this inherent heterogeneity of TOT uctuations and nancial globalization in the developing countries, this
paper examines whether TOT movements, in particular TOT volatility, trend changes, and shock persistence, are a major
determinant in shaping IFI. However, before we discuss the empirical methodology, we turn next to review the theoretical
framework linking TOT uctuations to IFI.
3. The conceptual framework and hypotheses
Although the dynamics of TOT are often the centerpiece in open economy models of international adjustment, these
models do not provide a clear guide to the effects of TOT volatility on IFI. Nonetheless, we summarize the relationship
between TOT uctuations and IFI, as predicted by the different strands of the literature, in two main hypotheses. On one
hand, to cope with uctuations in TOT and to smooth domestic consumption and investment, countries with more volatile
environment seek more integration with the rest of the world. In this case, greater uctuations in TOT lead to greater
cross-border investment, which increases the level of international risk-sharing (Svensson, 1988; Obstfeld and Rogoff, 1996;
Lane, 2001). The empirical literature documents a pro-cyclicality of capital ows to developing countries(for example,
342
Table 2
Terms-of-trade volatility, average trend-growth, and duration of shocks in selected developing countries, 19802007.
African countries
TOT volatility
Average TOT trend-growth
TOT shock-duration
Burundi
Cameroon
Cote d'Ivoire
Ethiopia
Gabon
Ghana
Kenya
Madagascar
Malawi
Mali
Mauritania
Mauritius
Mozambique
Niger
Nigeria
Senegal
South Africa
Sudan
Tanzania
Togo
Uganda
Zambia
0.290
0.104
0.190
0.213
0.139
0.282
NA
0.067
0.077
0.123
0.079
0.080
0.046
0.104
0.145
0.186
0.096
0.050
0.132
0.070
0.145
0.342
0.363
0.008
0.018
0.037
0.005
0.000
0.012
0.001
0.002
0.009
-0.024
0.011
0.024
0.006
0.024
0.007
0.005
0.015
0.009
0.027
0.001
0.066
0.027
0.008
3.1
0.7
3.7
2.4
1.1
4.7
2.1
2.6
1.6
1.0
3.6
4.9
4.9
2.4
11.6
4.4
2.2
8.1
5.4
1.7
3.0
1.6
2.2
Asian countries
Bangladesh
India
Indonesia
Malaysia
Pakistan
Papua New Guinea
Philippines
Sri Lanka
Thailand
0.063
0.078
0.114
0.042
0.091
0.201
0.085
0.039
0.050
0.035
0.022
0.012
0.014
0.040
0.055
0.009
0.034
0.011
2.8
1.4
4.1
5.1
7.3
0.7
2.1
2.1
1.2
Latin American countries

Argentina
Bolivia
Brazil
Chile
Colombia
Costa Rica
Dominica
Dominican Republic
Ecuador
Guatemala
Honduras
Mexico
Nicaragua
Paraguay
Peru
St. Vincent and the Grenadines
Uruguay
Venezuela, RB
0.073
0.109
0.071
0.142
0.081
0.072
0.057
0.101
0.118
0.135
0.103
0.134
0.132
0.097
0.127
0.097
0.084
0.180
0.012
0.003
0.020
0.000
0.009
0.002
0.006
0.018
0.014
0.012
0.009
0.031
0.022
0.014
0.010
0.016
0.007
0.015
1.0
5.3
1.5
14.7
3.1
2.5
0.3
2.1
3.8
0.3
1.3
4.9
0.3
1.4
4.5
4.0
1.0
7.3
Middle Eastern and North African countries

Algeria
Morocco
Saudi Arabia
Tunisia
Turkey
0.208
0.066
0.297
0.039
0.047
0.005
0.016
0.091
0.007
0.004
5.6
1.2
0.6
3.2
1.7
Notes: TOT trend-growth and TOT volatility are dened in Table A2; TOT shock-duration is the half-life of TOT shocks. For an AR(1) process the
half-life is calculated as TOT_hf = abs(ln (0.5)/ln (b)), where b is the autoregressive parameter obtained from the regression TOTt a bTOTt1
p1
P
di DTOTti et ; source: authors calculation based on the World Banks World Development Indicators (WDI) data.
ct
t1
343
Kaminsky et al., 2005; Gavin et al., 1996; Mendoza and Terrones, 2008). Also, the empirical evidence suggests that the
correlation between TOT and business cycles in developing countries is lower with greater nancial globalization
(Bhattacharya et al., 2013). For example, foreign direct investment is seen to ease nancing constraints, especially for the
low-income countries (Harrison et al., 2004). In addition, Blalock and Gertler (2005) nd that FDI mitigates the effects of
adverse shocks by helping rms maintain continuous access to credit through their parent companies.
On the other hand, higher TOT volatility can work in the opposite direction and reduce the level of IFI. That is, higher
volatility in terms-of-trade could result in uctuations in real return to foreign investment, which might deter foreign
investors and thus lead to lower IFI. For example, international portfolio models, such as Devereux and Sutherland (2009),
postulate that relative price volatility would affect IFI negatively. In addition, the seminal two good-two country trade model
of risk-sharing of Cole and Obstfeld (1991) shows that TOT changes can act as an automatic stabilizer when they are
negatively correlated with output uctuations, reducing the motive for international risk-sharing. In this case, TOT
uctuations provide insurance rather than being a source of risk and, in effect, reduce IFI. However, the empirical ndings
suggest that TOT are only weakly correlated with output (Backus et al., 1994). Also, Coeurdacier et al. (2010) show that the
level of international diversication does not strongly depend on countercyclical TOT. Further, Berka et al. (2012) nd that
TOT uctuations offset commodity-specic shocks, but not country-specic shocks, thus, leaving room for international
asset accumulation to stabilize consumption. As for the empirics, Eichengreen (1996) nds that both negative trend and
volatility in the TOT depressed capital inows for many developing countries.
Since the theoretical models do not provide a clear guide to the effects of TOT volatility on IFI, this is essentially an
empirical question. The empirical literature that attempted to verify the implications of these two hypotheses mostly
concentrate on the effects of TOT volatility on capital ows. We deviate from this literature by examining the effects of TOT
volatility, trend changes, and shock persistence on the size of different forms of cross-holdings of foreign assets and
liabilities.
4. Data and methodology
4.1. Empirical model
The theoretical and empirical literature on IFI and international investment positions suggest a number of factors that
determine this relationship. These include the level of economic and nancial development, capital account restrictions,
trade openness, economic stability and policy environment, institutional quality, and the size of the market (Alfaro et al.,
2008; Blattman et al., 2007; Lane, 2000; Lane and Milesi-Ferretti, 2008; Martin and Rey, 2004). The primary focus of this
study is to quantify the long-run association of patterns of TOT changes and IFI. Our baseline model takes the following
specication:
IFIit c g TOTit dTOTGit fX it mi u t eit
(1)
where i refers to each of the 55 countries in the sample and t refers to the time period from 1980 to 2007. In the estimations,
IFIit represents either aggregate or a component of international nancial integration: FDI, portfolio equity holdings, or
foreign debt; each expressed as a percentage of the GDP. For a comprehensive examination, we analyze separately the stocks
of foreign assets and the stocks of foreign liabilities for each of these components as well as for the aggregate level. The
TOTGit is the TOT trend-growth measured by the percentage change in the TOT trend over the 5-year interval, while TOTVit is
the TOT volatility measured by the standard deviation of departure from the trend over the 5-year interval. Xit is a vector of
other control variables including: initial real output per capita, ination rate, level of trade openness, capital account
restrictions, level of nancial development, share of primary-commodities in total exports (export diversication), level of
institutional quality (policy and regulatory environment), education, and the degree of exibility in the exchange
rate regime. mt is unobserved country-specic effect, ut is unobserved period-specic effect, while eit is random error. All
variable measures and sources are presented in the Appendix A. The model is estimated using OLS with country xed effects
and time period dummies. All standard errors are heteroskedasticity-robust.
TOT is exogenous to the countries included in the sample as they are price-takers in the world market, they cannot
inuence the prices of the goods they import and the prices of commodities they export (Broda, 2004; Frankel, 2010)8 . Since
TOT can be assumed exogenous to these countries, we do not need to include the other control variables beside the TOT
variables. In other words, the empirical model is not vulnerable to omitted variable bias. However, we nd that the standard
errors are lower when we add independent variables.
The empirical estimations are carried out using 5-year non-overlapping windows to lter out business cycle uctuations
and to mitigate the problem of reverse causality. We also nd 5-year windows better account for the commodity cycles that
shape the behavior of TOT, as suggested by Cashin et al. (2004b). The results remain robust to using shorter or longer
time-windows.
8
Saudi Arabia might be an exception given its large size in world crude oil markets. We address this issue in the Diagnosis and robustness part of
Section 5.
344
Table 3
Descriptive statistics of the full sample, 5-year non-overlapping windows, 19802007, 55 countries.
Mean
Median
Maximum
Dependent variable
International nancial integration IFI (logs)
Total foreign assets TFA (logs)
Total foreign liabilities TFL (logs)
Total debt TD (logs)
Debt assets DA (logs)
Debt liabilities DL (logs)
Portfolio equity assets PEA (logs)
Total foreign direct investment TFDI (logs)
Foreign direct investment assets FDIA (logs)
Foreign direct investment liabilities FDIL (logs)
Minimum
Std. Dev.
Observations
0.062
1.437
0.280
0.339
2.103
0.637
3.643
1.847
3.387
1.858
0.053
1.475
0.287
0.351
2.095
0.646
3.904
1.817
3.672
1.858
3.238
2.584
2.504
3.195
2.525
2.479
0.385
0.920
0.050
0.405
1.758
3.455
3.052
2.193
5.027
3.232
3.912
4.284
3.913
3.737
0.631
0.892
0.654
0.689
1.001
0.746
0.647
0.949
0.691
0.852
335
335
335
335
335
335
335
335
335
335
TOT variables
TOT volatility
TOT average trend-growth
0.094
0.004
0.063
0.003
0.714
0.137
0.003
0.139
0.093
0.039
329
335
Other determinants
Initial GDP per capita
Trade openness (logs)
Private credit (logs)
Exchange rate exibility
Financial openness
Export diversication
Population
Ination volatility
Education
Real output volatility
Real exchange rate volatility
10.245
4.080
3.195
2.321
0.347
2.320
16.439
8.786
3.606
0.031
0.127
10.309
4.095
3.168
2.000
0.949
2.460
16.487
5.229
3.805
0.018
0.074
10.646
6.065
5.012
6.000
2.456
4.596
21.014
775.468
4.697
1.501
5.624
6.902
2.481
0.643
1.000
1.856
3.493
11.124
0.000
1.037
0.003
0.006
0.392
0.567
0.784
1.142
1.274
1.516
1.780
41.834
0.781
0.094
0.326
335
335
335
335
335
320
336
320
325
335
320
Notes: for detailed description and sources of these variables, see Table A2.
4.2. Data
The measures of IFI that we use are drawn from the dataset of Lane and Milesi-Ferretti (2007), which comprises stocks of
gross foreign liabilities and foreign assets categorized into foreign direct investment, portfolio equity investment, debt and
other investment, and ofcial reserves9 . As discussed in Kose et al. (2009a), the use of gross stocks is preferable to annual
capital ows in measuring international integration, as the latter tend to be more volatile and prone to measurement errors.
In addition, the use of gross stocks, compared to net stocks, provides a better measure of integration and efcient
risk-sharing as it captures two-way interactions between economies with different risk portfolios. All other variables are
taken from the World Banks World Development Indicators (WDI) database.
For economic development we use initial level of real GDP per capita (expressed relative to the US real GDP per capita).
The GDP per capita controls also for market size. For capital account restrictions, we use the de jure index, KAOPEN, of Chinn
and Ito (2008). This index embodies four binary dummy variables on restrictions taken from the IMFs Annual Report on
Exchange Arrangements and Exchange Restrictions (AREAER). Due to the lack of a consistent measure of institutional quality
that provides a good coverage for our sample period, we use the volatility of ination to proxy the uncertainty about the
policy environment and the institutional quality10 . For education, we use high school enrollments as a percentage of
population. Following the literature, education is a proxy for human capital (Alfaro et al., 2008). The exibility of the nominal
exchange rate is measured by the de facto categorical measure of Reinhart and Rogoff (2004). Devarajan and Rodrick (1991)
show that countries faced with higher volatility in their TOT would benet by having more exible exchange rate regimes.
Table 3 shows the descriptive statistics.
4.3. Decomposition of terms-of-trade uctuations
We decompose TOT movements into two components, trend and volatility, using a HodrickPrescott (HP) lter to
produce a smoothed trend and stationary deviations. An alternative measure is to use the TOT growth rate and its standard
9
Changes in international investment positions reect both transactional and revaluation components. This is to say that changes in the position could
occur without an accompanying change in the ow. The dataset of Lane and Milesi-Ferretti (2007) is adjusted for exchange rate uctuations and changes in
stock market values.
10
For example, the World Banks Country Policy Institutional Assessments (CPIA) and other World Banks governance indicators are only available from
2005 for most of the countries in our sample.
345
Table 4
OLS panel regression of full sample, 19802007 dependent variable: the alternative measures of international nancial integration.
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
Dependent variable
Independent variables
TOT volatility
IFI
TFA
TFL
TD
DA
DL
PEAa
TFDI
FDIA
FDIL
0.373
(0.112)***
0.344
(0.078)***
0.707
(0.165)***
0.817
(0.205)***
0.442
(0.226)*
1.083
(0.224)***
0.418
(0.057)***
0.196
(0.059)***
0.251
(0.101)**
0.024
(0.075)
TOT trend-growth
1.099
(0.339)***
0.645
(0.317)**
1.57
(0.391)***
2.066
(0.449)***
0.982
(0.474)**
2.398
(0.576)***
0.442
(0.146)***
0.167
(0.197)
0.307
(0.292)
0.153
(0.311)
Economic development
0.581
(0.075)***
1.725
(0.062)***
0.163
(0.090)*
0.110
(0.108)
1.353
(0.098)***
0.276
(0.132)**
0.390
(0.031)***
1.376
(0.092)***
0.602
(0.058)***
1.178
(0.071)***
Trade openness
0.265
(0.056)***
0.402
(0.057)***
0.28
(0.067)***
0.005
(0.088)
0.164
(0.066)**
0.035
(0.102)
0.095
(0.028)***
0.951
(0.065)***
0.149
(0.055)***
0.784
(0.060)***
Financial development
0.098
(0.034)***
0.028
(0.029)
0.136
(0.040)***
0.168
(0.044)***
0.074
(0.036)**
0.197
(0.055)***
0.022
(0.011)**
0.026
(0.040)
0.022
(0.023)
0.021
(0.033)
0.058
(0.016)***
0.031
(0.018)*
0.074
(0.020)***
0.096
(0.023)***
0.061
(0.016)***
0.09
(0.030)***
0.017
(0.008)**
0.019
(0.022)
0.007
(0.014)
0.021
(0.018)
Financial openness
0.017
(0.012)*
0.042
(0.015)***
0.005
(0.016)
0.042
(0.019)**
0.026
(0.017)
0.055
(0.026)**
0.036
(0.008)***
0.121
(0.019)***
0.003
(0.012)
0.126
(0.016)***
Countries/Observations:
R-squared:
55/299
0.821
55/287
0.868
55/299
0.787
55/299
0.769
55/287
0.810
55/299
0.795
55/295
0.802
55/299
0.846
55/299
0.852
55/299
0.811
Notes: the model is estimated using OLS with country xed effects and time period dummies; robust standard errors are in brackets; regressions (1)(10)
correspond to using alternative measures of international nancial integration for the dependent variable but the same set of independent variables; IFI is
the international nancial integration measured as total foreign assets plus total foreign liabilities as share of GDP; TFA (TFL) is the ratio of total foreign
assets (liabilities) to GDP; TD is the ratio of total foreign debt to GDP; DA (DL) is the ratio of foreign debt assets (liabilities) to GDP; PEA is the ratio of foreign
portfolio equity assets to GDP; TFDI is the ratio of total foreign direct investment assets and liabilities to GDP; FDIA (FDIL) is the ratio of foreign direct
investment assets (liabilities) to GDP; the independent variables are dened in Table A2; the parameter estimates for the other control variables
secondary schooling, export diversication, and ination volatility are not signicant and not reported.
*
Signicance at 10%.
**
Signicance at 5%.
***
Signicance at 1%.
a
Portfolio equity liability is not estimated due to the lack of sufcient observations.
deviation. As discussed by Blattman et al. (2007), if the estimation is carried for separate intervals (e.g., 5-year
non-overlapping windows), then such measures are sensitive to pronounced changes at the end points of the intervals,
leading to measurement errors and biased estimates, especially for more volatile countries. Also, it is difcult to analyze the
effects of structural changes in the TOT or of discrete changes in its growth rate as they are reected in the trend and the
volatility. Our results are robust to using alternative measures of TOT growth and volatility.
5. Results and discussion
5.1. Total foreign assets and total foreign liabilities
The estimations based on Eq. (1) with the aggregate measures of foreign cross-holdings: international nancial
integration (IFI), total foreign assets (TFA), and total foreign liabilities (TFL) are reported for the full sample in Table 4
(columns 13). The signs and magnitude of the parameter estimates for the other determinants are generally in line with
similar studies (for example, Lane, 2000; Lane and Milesi-Ferretti, 2008; Martin and Rey, 2004)11.
More importantly, the parameter estimates of the TOT volatility is negatively related to the aggregate measure of IFI. The
point estimate is both statistically and economically signicant, which indicates that raising one standard deviation in TOT
volatility (an increase from its average of 0.0940.187) decreases the IFI position from its mean of 106% of GDP to 100%.
However, the impact of TOT volatility is not the same across total foreign assets (TFA) and total foreign liabilities (TFL). The
parameter estimates associated with the TOT volatility for TFA (a parameter estimate of 0.344) and for TFL (a parameter
11
The effects of the control variables on the IFI and its subcomponents have theoretically the expected sign. The gross cross-foreign-asset-holdings are
positively associated with trade openness, economic development, and exchange rate exibility. It is also negatively associated with capital account
restrictions (nancial openness) and nancial development. Although the latter result contradicts those for advanced countries, Lane (2000) points out that
larger nancial sector might not necessarily lead to greater IFI, especially for countries with capital restrictions and lower institutional quality.
346
estimate of 0.707), columns 2 and 3 of Table 4, indicate that greater TOT volatility leads to greater cross-holdings of foreign
assets and lower cross-holdings of foreign liabilities (i.e., higher TOT volatility improves the aggregate net foreign assets
position). On the other hand, positive TOT trend changes have larger impacts on IFI, TFA, and TFL compared to TOT volatility.
This indicates that external assets and liabilities are responding more to trend changes compared to uctuations around the
trend. Interestingly, the patterns we document here explain some of the stylized facts discussed in Section 2. That is, the
parameter estimates (Table 4) and the observed behavior of the TOT trend and volatility (Fig. 4) can be reconciled to explain
the patterns observed in the IFI and the net foreign assets position (Fig. 3, upper panel).
5.2. Composition of gross foreign assets and liabilities
The estimates of Eq. (1) using various subcomponents of the IFI, namely: foreign debt, portfolio equity, and FDI, using the
same set of independent variables, are reported in columns 410 of Table 4. A number of interesting observations emerge.
The rst observation is that, in all cases, TOT volatility is associated with higher levels of foreign assets and lower levels of
foreign liabilities. This nding reconciles the two hypotheses outlined in section 3. To cope with uctuations in TOT and to
smooth domestic consumption and investment, domestic agents seek more integration with the rest of the world and
acquire more foreign assets (consistent with the theoretical propositions of Lane, 2001; Obstfeld and Rogoff, 1996; Svensson,
1988). On the other hand, higher volatility in TOT results in uctuations in real return to foreign investment, which deters
foreign investors and lowers foreign liabilities (as suggested by the international portfolio models, such as, Devereux and
Sutherland, 2009).
A second notable observation is that, across the different forms of external assets and liabilities, the response of foreign
liabilities is consistently larger than the response of foreign assets. A third observation is that the parameter estimates
associated with foreign debt and portfolio equity are larger than those associated with FDI. The varying response of different
components of international nancial integration to TOT volatility can be explained in view of the nature of these
investments. That is, the higher responses of portfolio investments and bank debt is due to their greater volatility and their
being subject to sudden stops (Calvo, 2007), while the lower response of FDI to TOT volatility is due to its costly and
pro-longed reversals. Along this line, Lane (2000) nds that the TOT volatility affects portfolio equity and not FDI. The fourth
observation is that the results for debt are very similar to those of aggregate cross-assets holdings discussed above. This
observation is due to the dominance of debt in cross asset holdings of developing countries (Figs. 2 and 3).
Table 5
OLS panel regression of full sample with interaction effects, 19802007 dependent variable: the international nancial integration (IFI).
(1)
(2)
(3)
(4)
TOT volatility
5.604
(0.820)***
1.664
(0.212)***
1.403
(0.184)***
1.513
(0.389)***
1.363
(0.316)***
0.217
(0.124)*
1.209
(0.344)***
8.574
(3.764)**
1.068
(0.337)***
TOT trend-growth
TOT volatility trade openness
TOT volatility nancial development
0.519
(0.106)***
TOT volatility nancial openness
0.285
(0.120)**
TOT volatility economic development

Trade openness
Financial openness
Countries/Observations
R-squared
0.574
(0.064)***
0.195
(0.058)***
0.114
(0.032)***
0.066
(0.018)***
0.006
(0.014)
55/299
0.827
0.601
(0.072)***
0.306
(0.057)***
0.171
(0.035)***
0.064
(0.017)***
0.002
(0.014)
55/299
0.824
0.572
(0.074)***
0.288
(0.055)***
0.108
(0.034)***
0.055
(0.017)***
0.015
(0.016)
55/299
0.818
0.855
(0.359)**
0.671
(0.085)***
0.27
(0.055)***
0.105
(0.033)***
0.058
(0.016)***
0.001
(0.012)
55/299
0.820
Notes: the dependent variable, IFI, is the same in all four regressions, (1)(4), and measured as total foreign assets plus total foreign liabilities as share of
GDP; each regression corresponds to using alternative interaction effects; the model is estimated using OLS with country xed effects and time period
dummies; robust standard errors are in brackets; the independent variables are dened in Table A2; the parameter estimates for the other control variables
*
Signicance at 10%.
**
Signicance at 5%.
***
Signicance at 1%.
Years
8
10
347
12
14
16
Algeria
Argentina
Bahrain
Bangladesh
Bolivia
Brazil
Burundi
Cameroon
Chile
Colombia
Costa Rica
Cote d'Ivoire
Dominica
Dominican Republic
Ecuador
Ethiopia
Gabon
Ghana
Guatemala
Honduras
India
Indonesia
Kenya
Madagascar
Malawi
Malaysia
Mali
Mauritania
Mauritius
Mexico
Morocco
Mozambique
Nicaragua
Niger
Nigeria
Pakistan
Papua New Guinea
Paraguay
Peru
Philippines
Saudi Arabia
Senegal
South Africa
Sri Lanka
St. Vincent and the
Sudan
Suriname
Tanzania
Thailand
Togo
Tunisia
Turkey
Uganda
Uruguay
Venezuela, RB
Zambia
Zimbabwe
Fig. 5. The half-life of TOT shocks (years), selected developing countries, 19802007.
Notes: For an AR(1) process, the half-life
is calculated as TOT_hf = abs[ln (0.5)/ln (b)], where b is the autoregressive parameter estimated in the following
p1
P
regression: TOTt a bTOTt1 ct
di TOTt1 et ; For higher order processes, the half-life is calculated from the impulse response function
as the length of time it takes for a i1
unit impulse to dissipate by half (Andrews, 1993); source: authors calculation based on the World Banks
World Development Indicators (WDI) data.
5.3. Examining the interaction effects

To capture possible non-linearities in the relationship between TOT uctuations and IFI, we augment Eq. (1) with
interaction factors. The model takes the following specication:
348
Table 6
Cross-section regression, full sample, 19802007 dependent variable: the alternative measures of international nancial integration.
Dependent variable
Constant
TOT shock-duration
Trade openness
Real output volatility
Ination volatility
Observations:
R-squared:
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
IFI
TFA
TFL
TD
DA
DL
PEAa
TFDI
FDIA
FDIL
2.584
(1.216)**
0.029
(0.054)
0.606
(0.107)***
0.210
(0.062)***
0.067
(0.115)
1.170
(0.497)**
0.448
(0.176)**
55
0.577
0.833
(1.831)
0.086
(0.081)
0.469
(0.161)***
0.110
(0.093)
0.467
(0.174)***
1.575
(0.748)**
0.335
(0.265)
55
0.588
6.876
(1.399)***
0.071
(0.062)
0.686
(0.123)***
0.315
(0.071)***
0.463
(0.133)***
0.883
(0.571)
0.596
(0.203)***
55
0.552
2.126
(1.534)
0.010
(0.068)
0.501
(0.135)***
0.371
(0.078)***
0.082
(0.145)
1.081
(0.627)*
0.368
(0.222)
55
0.442
2.197
(2.560)
0.247
(0.113)***
0.441
(0.225)*
0.080
(0.130)
0.583
(0.243)**
1.384
(1.046)
0.344
(0.371)
55
0.428
8.439
(1.612)***
0.089
(0.071)
0.551
(0.142)***
0.442
(0.082)***
0.680
(0.153)***
1.078
(0.658)
0.525
(0.233)**
55
0.548
5.158
(0.953)***
0.091
(0.042)***
0.071
(0.084)
0.174
(0.048)***
0.094
(0.090)
3.684
(0.389)***
0.167
(0.138)
55
0.776
4.898
(1.995)**
0.006
(0.088)
1.010
(0.175)***
0.069
(0.101)
0.146
(0.189)
0.211
(0.815)
0.818
(0.289)***
55
0.583
2.454
(1.712)
0.146
(0.075)*
0.002
(0.150)
0.245
(0.087)***
0.203
(0.162)
2.750
(0.699)***
0.013
(0.248)
55
0.544
5.661
(1.861)***
0.052
(0.082)
0.957
(0.164)***
0.000
(0.095)
0.021
(0.176)
0.132
(0.760)
0.792
(0.269)***
55
0.552
Notes: the model is estimated using OLS; robust standard errors are in brackets; IFI is the international nancial integration measured as total foreign assets
plus total foreign liabilities as share of GDP; TFA (TFL) is the ratio of total foreign assets (liabilities) to GDP; TD is the ratio of total foreign debt assets and
liabilities to GDP; DA (DL) is the ratio of foreign debt assets (liabilities) to GDP; PEA is the ratio of foreign portfolio equity assets to GDP; TFDI is the ratio of
total foreign direct investment assets and liabilities to GDP; FDIA (FDIL) is the ratio of foreign direct investment assets (liabilities) to GDP; the independent
variables are dened in Table A2; the parameter estimates for the other control variables secondary schooling and export diversication are not
signicant and not reported.
*
Signicance at 10%.
**
Signicance at 5%.
***
Signicance at 1%.
a
IFIit c g TOTVit dTOTGit bTOTV Fit fX it mi ut eit
(2)
where variables and parameters are as dened before for Eq. (1) and F is the interaction factor. Eq. (2) is estimated with
alternative interaction factors including: trade openness, nancial sector development, nancial openness (or capital
account restrictions), exchange rate exibility, economic development, and export diversication. The results of these
estimations are reported in Table 5, where columns 14 denote alternative interaction effects. The following factors seem to
matter in the relationship between TOT volatility and IFI12 .
Trade openness. On theoretical grounds, and for a given negative TOT shock that requires current account adjustments,
countries with higher trade openness are less likely to default on their debt (Calvo et al., 2004; Lane and Milesi-Ferretti,
2008). That is, with more trade linkages, default risk is lower, which increases the propensity to invest in foreign assets. The
interaction effect of trade openness (column 1 of Table 5) is positive and statistically signicant, supporting this
hypothesis. This nding also supports the notion that trade linkages may improve information ows and thereby increase
the willingness to invest in foreign assets.
Financial development. Deeper domestic nancial markets and banking sector facilitate asset trade among local residents
and could reduce the need of external nancing. At the same time, domestic nancial development facilitates foreign
demand for domestic liabilities, thus, increasing IFI. Kose et al. (2011) nd empirically that there are certain threshold
levels of nancial development that an economy should attain before it can realize the benets and reduce the risk of
nancial openness. The interaction estimates of nancial development (column 2 of Table 5) shows that TOT volatility
increases the complementarity between domestic nancial development and external asset holdings.
Financial openness. Historical experiences of developing countries suggest that capital account restrictions has a negative
impact on the accumulated stocks of foreign assets and liabilities (Chinn and Ito, 2008; Lane and Milesi-Ferretti, 2007;
Quinn, 2003; Miniane, 2004). In view of the interaction estimates (column 3 Table 5), we nd that greater nancial
openness mitigates the effects of TOT volatility on IFI.
Economic development. Theoretical propositions suggest that larger economies allow for more cross-border asset trade
(Lane and Milesi-Ferretti, 2008). For example, the country risk is perceived to be lower with higher level of economic
development. In addition, given the xed costs and learning costs associated with foreign asset trade, the degree of IFI is
positively related to the level of income. Also, a countrys level of income can be used to proxy the overall institutional
12
The interaction terms of export diversication and exchange rate exibility were insignicant and not reported in Table 5.
349
development (Kose et al., 2011). The interaction effect of economic development is negative and statistically signicant,
suggesting that economic development decreases the effect of TOT volatility on IFI (column 4 of Table 5).
5.4. Does the persistence of TOT shocks matter in international investment positions?
The literature on the macroeconomic effects of TOT shocks strongly distinguishes between transitory and permanent
shocks (Kose, 2002; Obstfeld and Rogoff, 1996; Ostry and Reinhart, 1992). Given the wide range of commodities in our
sample, we thus expect the duration of TOT shocks to affect domestic investors behavior as well as foreign investors. For
example, if an adverse TOT shock is perceived as transitory, then the country would use external nancing to smooth the
effects of the shock. However, if the shock is perceived as permanent, then external nancing would not be sustainable. Thus,
shocks that are mean-reverting are transitory and would have more impact on short-term IFI. However, shocks that are not
mean-reverting are permanent and would have more impact on long-term IFI and might also lead to fundamental
adjustments in the economy (i.e., permanent changes in consumption and investment).
To analyze the role of persistence of TOT shocks on the cross-holding of assets, we calculate the scalar estimate of
persistence, the half-life of TOT shocks proposed by Andrews (1993) 13 . The half-life is the time it takes for a unit shock to
dissipate by 50%. To calculate the half-life, we rst estimate an AR(p) process with an intercept, a time trend, and lagged
differences for the TOT series of each individual country as:
TOTt a bTOTt1 ct
p1
X
di DTOTt1 et fort 1; . . . ; T
(3)
i1
The half-life of a shock for an AR(1) process is calculated as TOT_hf = abs(ln (0.5)/ln (b)), while for higher order processes,
the half-life is calculated from the impulse response functions, and dened as the time it takes for a unit impulse to dissipate
permanently by one-half from the occurrence of the initial shock14 . The half-life estimates of the TOT shocks are shown in
Fig. 5. The results indicate that the duration of TOT shocks varies widely (the average half-life is 3.4 years and the range
varies between 0.3 and 14.7 years)15 . We then augment the TOT shock-duration to the cross-section reduced-form equation
as:
IFIt c g TOTh f i fX i mi eit
(4)
The control variables are the level of economic and nancial development, capital account restrictions, trade openness,
population, the volatility of ination (see cross-section studies, for example, Alfaro et al. (2008),Lane (2000), and Lane and
Milesi-Ferretti (2008)). All variable measures and sources are presented in the Appendix A. The estimation of Eq. (4) is shown
in Table 6. As reported in Table 6, the duration of TOT shocks is signicant for the case of debt assets, portfolio equity assets,
and FDI assets, columns 5, 7, and 9, respectively. For cross-holding of foreign liabilities, the shock-duration is insignicant.
Cross-holdings of debt assets are inversely related to the duration of TOT shocks, while cross-holdings of portfolio equity
assets are positively related to the duration of TOT shocks. FDI assets are also positively related to the duration of TOT shocks,
with larger parameter estimates compared to portfolio equity assets. This suggests that if shocks are perceived to be longlived then domestic agents decrease their holding of foreign debt assets and increase their holdings of portfolio equity assets
and FDI assets. This nding is consistent with the predictions of international portfolio models such as Devereux and
Sutherland (2009) and with the prediction of open economy real business cycle models such as Baxter and Crucini (1995).
5.5. Diagnosis and robustness
We performed a number of exercises to check the robustness of our ndings. We rst probe the sensitivity of the results to
different country groups. Since our sample encompasses countries with varying levels of institutional environments and
history of nancial integration, we allow the TOT volatility coefcient in Eq. (1) to differ for two country groups more
nancially integrated (MFI) economies and less nancially integrated (LFI) economies, as dened in Kose et al. (2006). The
MFI economies essentially constitute the group of emerging markets economies16 . Table 7 reports the estimation. The results
of the standard hypothesis-testing suggest that the relationship between TOT volatility and IFI is not signicantly different
between the group of emerging markets economies and other developing countries. To further unveil any differences across
13
See also Andrews and Chen (1994) and Cashin et al. (2004b) for a discussion on the calculation of the half-life.
Standard time series tests suggest that for all countries included in the sample the appropriate number of lagged rst difference of the TOT (DTOT) to be
included to account for serial correlation is zero. Thus, all TOT series are estimated with AR(1) model. The time trend is included for the majority of countries.
15
Cashin et al. (2004b) nd an average half-life of 6 year for the TOT shocks of 42 Sub-Saharan African countries for the period 19601996. For the
24 African countries include in our sample, we nd the average half-life of TOT shocks during 19802007 to be approximately 4 years.
16
We thank an anonymous referee for this suggestion.
14
350
Table 7
OLS panel regression, country-group dummies, 19802007 dependent variable: the alternative measures of international nancial integration
Dependent variable
TOT volatility MFI
TOT volatility LFI
TOT trend-growth
Trade openness
Financial openness
Ho: is the TOT volatility coefcient
equal between MFI and LFI? (pvalues)
Countries/Observations:
R-squared:
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
IFI
TFA
TFL
TD
DA
DL
PEAa
TFDI
FDIA
FDIL
0.256
(0.334)
0.365
(0.127)***
1.176
(0.338)***
0.578
(0.075)***
0.275
(0.057)***
0.1
(0.033)***
0.055
(0.017)***
0.008
(0.013)
0.757
0.338
(0.222)
0.249
(0.154)*
0.662
(0.373)*
1.684
(0.070)***
0.42
(0.060)***
0.037
(0.030)
0.027
(0.018)
0.027
(0.015)*
0.756
0.876
(0.342)**
0.687
(0.182)***
1.617
(0.395)***
0.158
(0.090)*
0.273
(0.067)***
0.138
(0.040)***
0.074
(0.020)***
0.001
(0.016)
0.609
0.532
(0.467)
-0.877
(0.216)***
2.03
(0.455)***
0.107
(0.111)
0.007
(0.090)
0.171
(0.044)***
0.094
(0.024)***
0.043
(0.019)**
0.486
0.896
(0.080)***
0.076
(0.253)
1.539
(0.377)***
1.295
(0.075)***
0.208
(0.052)***
0.055
(0.025)**
0.048
(0.012)***
0.035
(0.012)***
0.001
1.197
(0.566)**
1.079
(0.240)***
2.419
(0.573)***
0.282
(0.133)**
0.032
(0.103)
0.198
(0.055)***
0.09
(0.029)***
0.053
(0.026)**
0.846
0.113
(0.150)
0.371
(0.062)***
0.472
(0.140)***
0.371
(0.032)***
0.09
(0.028)***
0.018
(0.011)
0.011
(0.009)
0.042
(0.008)***
0.101
0.701
(0.411)*
0.359
(0.101)***
0.255
(0.317)
0.615
(0.065)***
0.148
(0.055)***
0.016
(0.023)
0.004
(0.013)
0.000
(0.012)
0.154
0.025
(0.193)
0.034
(0.096)
0.12
(0.326)
1.171
(0.082)***
0.786
(0.063)***
0.022
(0.033)
0.021
(0.018)
0.126
(0.016)***
0.010
0.113
(0.150)
0.371
(0.062)***
0.472
(0.140)***
0.371
(0.032)***
0.09
(0.028)***
0.018
(0.011)
0.011
(0.009)
0.042
(0.008)***
0.803
55/299
0.821
55/287
0.868
55/299
0.787
55/299
0.769
55/287
0.870
55/299
0.726
55/295
0.802
55/299
0.846
55/299
0.826
55/299
0.812
Notes: the model is estimated using OLS with country xed effects and time period dummies; robust standard errors are in brackets; regressions (1)(10)
correspond to using alternative measures of international nancial integration for the dependent variable but the same set of independent variables; MFI is
a dummy for more nancially integrated countries including Argentina, Brazil, Chile, Colombia, India, Indonesia, Malaysia, Mexico, Pakistan, Peru,
Philippines, South Africa, Thailand, Turkey, and Venezuela; LFI is a dummy for less nancial integrated countries which include all other countries in our
sample; IFI is the international nancial integration measured as total foreign assets plus total foreign liabilities as share of GDP; TFA (TFL) is the ratio of total
foreign assets (liabilities) to GDP; TD is the ratio of total foreign debt to GDP; DA (DL) is the ratio of foreign debt assets (liabilities) to GDP; PEA is the ratio of
foreign portfolio equity assets to GDP; TFDI is the ratio of total foreign direct investment assets and liabilities to GDP; FDIA (FDIL) is the ratio of foreign direct
investment assets (liabilities) to GDP; the independent variables are dened in Table A2; the parameter estimates for the other control variables
*
Signicance at 10%.
**
Signicance at 5%.
***
Signicance at 1%.
a
the geographic groups for our main analysis, we estimate Eq. (1) for three groups: Africa, Asia, and Latin America17.
The results are generally within the same line of the full sample18 .
For the sample construction, the results are robust to using 3-year and 7-year averages of variables instead of 5-year
averages. We use the results of shock persistence as a guide to possible division of the sample into two groups: high
persistence and transitory. However, there is not much difference in the results between the two groups. In addition, the
results are robust to possible alterations in the control variables (additional control variables) and sample coverage. First,
the results are robust to removing Chile and Nicaragua as outliers on the IFI measures. We also removed Saudi Arabia as the
price taking condition may not safely apply due to its large size in world crude oil markets. Second, beside the estimation for
the periods 19802007, estimation is also carried out for the period 19902007, which witness the dominance of private
ows to developing countries. The results are mostly preserved.
The results are also robust to using alternative estimation methods. For example, to address the issue of joint endogeneity
of explanatory variables with the error term and the potential biases caused by country-specic effects and omitted
variables, we also estimated Eq. (1) using a generalized method of moments (GMM) dynamic panel data model. Our results
are also robust to alternative methods of decomposing the TOT series into trend and uctuations and alternative denitions
of TOT volatility.
6. Conclusions and extensions
This paper signicantly contributes to the literature on the determinants of nancial globalization. We examine the
effects of TOT movements on the size of different forms of cross-holdings of foreign assets and liabilities using a sample of
55 primary-commodity exporting countries for the period 19802007. The ndings of this paper suggest a robust and
signicant role of TOT volatility, trend-growth, and shock persistence in shaping the size and composition of nancial
17
18
The model is not estimated for the Middle East and North Africa due to the lack of enough observations.
All results are available from the author upon request.
351
globalization of primary commodity-exporting countries. To provide deeper insights into the nature and channels of
association of TOT movements and IFI, we investigate different aspects of this relationship. First, beside the aggregate
measures of gross external assets and liabilities, the relationship is analyzed using sub-measures of IFI, including foreign
direct investment, foreign portfolio equity, and foreign debt. The association is also analyzed separately with respect to gross
cross-holdings of foreign assets and foreign liabilities for these sub-measures. Second, while previous studies mainly used
TOT volatility as their only measure of TOT uctuations, this paper employs two other measures of TOT uctuations: trendgrowth and shock-duration. Our ndings suggest that trend-growth is more important than volatility in explaining changes
in IFI. Higher TOT volatility is robustly associated with greater cross-holdings of foreign assets and smaller cross-holdings of
foreign liabilities. On the other hand, cross-holdings of portfolio assets seem more responsive to TOT volatility compared to
FDI assets. A notable nding is that longer duration of terms-of-trade shocks seems to shift nancial globalization towards
equity assets and away from debt assets.
The ndings of this paper also contribute to the literature on international risk-sharing in the developing countries. For
example, Kose et al. (2009b) nd that, despite greater nancial globalization, the degree of international risk-sharing in
developing countries is limited. They attribute their ndings to the fact that the developing countries rely more on less stable
foreign cross-holdings, such as bank loans, that may not allow for efcient risk-sharing. Our analysis provides an alternative
examination of the degree of risk-sharing in developing countries. Since TOT shocks are dominant and exogenous to
primary-commodity countries, the response of cross-holdings of foreign assets and liabilities to these exogenous
idiosyncratic shocks reects the degree of risk-sharing. Our results show that the TOT volatility and shock-duration have
larger impact on debt assets compared to other forms of foreign cross-holdings, supporting Kose et al. (2009b). Further
investigation using dis-aggregated measures of IFI can contribute to the literature on international risk-sharing in the
developing countries. Another direction is to explore the effects of TOT on international nancial integration and intranational nancial integration due to countries being a part of an economic grouping (e.g., The Association of Southeast Asian
Nations (ASEAN); The South Asian Association for Regional Cooperation (SAARC); The Southern African Customs Union
(SACU)).
Acknowledgments
The comments and suggestions of an anonymous referee are gratefully acknowledged. The author also received useful
comments on this paper from Narjess Boubakri, Arthur Denzau, Azmat Gani, and the participants at the Economics Seminar
Series of the Economics Department of the American University of Sharjah. The author is responsible for any remaining
errors.
Appendix A.
See Tables A1 and A2
Table A1
List of countries included in the sample.
Algeria, Argentina, Bangladesh, Bolivia, Brazil, Burundi, Cameron, Central African Republic, Chile, Colombia, Costa Rica, Cte d'Ivoire, Dominica,
Dominican Republic, Ecuador, Ethiopia, Gabon, Ghana, Guatemala, Honduras, India, Indonesia, Kenya, Madagascar, Malawi, Malaysia, Mali,
Mauritania, Mauritius, Mexico, Morocco, Mozambique, Nicaragua, Niger, Nigeria, Pakistan, Papua New Guinea, Paraguay, Peru, Philippines, Saudi
Table A2
List and denitions of variables.
Variable name
Denition and construction
Source
Terms-of-trade
trend-growth
Terms-of-trade
volatility
GDP growth
volatility
Ination volatility
GDP per capita gap
Average growth rate of the HodrickPrescott (HP) smoothed trend of the net-barter-termsof-trade during the 5-year interval
Standard deviation of the log change in net-barter-terms-of-trade away from the Hodrick
Prescott (HP) smoothed trend
Standard deviation of the log change in real GDP
World Bank, WDI and authors

calculation
World Bank, WDI and authors
calculation
World Bank, WDI
Standard deviation of the log change in the CPI

Country GDP per capita minus US GDP per capita, both in PPP terms (constant 2005
international $) and in log
Log total domestic credit to private sector (% of GDP)
World Bank, WDI

Authors calculation based on
World Bank, WDI
World Bank, WDI
KAOPEN index of Chinn and Ito (2008)
Chinn and Ito (2008)
Log of exports plus imports (% of GDP)
World Bank, WDI

Reinhart and Rogoff (2004)
Financial
development
Capital account
restrictions
Trade openness
352
Table A2 (Continued)
Variable name
Denition and construction
Exchange rate
exibility
International
nancial
integration (IFI)
Total foreign
direct investment
Portfolio
equity assets
Total
debt investment
Education
Export
diversication
A categorical measure of exchange rate regime exibility; 1 = x, 2 = peg, 3 = managed oat,

4 = oat
Log total foreign assets plus total foreign liabilities (% of GDP)
Lane and Milesi-Ferretti (2007)
Source
Log total foreign direct investment, assets plus liabilities (% of GDP)
Log foreign portfolio equity assets (% of GDP)
Log total foreign assets and liabilities of portfolio debt, bank loans and deposits, and other Lane and Milesi-Ferretti (2007)
debt instruments (% of GDP)
Total secondary enrollments (% of population)
World Bank, WDI
Share of primary commodities (raw materials and crude oil) in total exports
Authors calculation based on
World Bank, WDI
References
Al-Abri, A., 2013. Real exchange volatility, terms-of-trade shocks, and nancial integration in primary-commodity exporting countries. Econ. Lett. 120,
126129.
Alfaro, L., Kalemli-Ozcan, S., Volosovych, V., 2008. Why doesnt capital ow from rich to poor countries? An empirical investigation. Rev. Econ. Stat. 90,
347368.
Andrews, D., 1993. Exactly median-unbiased estimation of rst-order autoregressive/unit root models. Econometrica 61, 139165.
Andrews, D., Chen, H., 1994. Approximately median-unbiased estimation of autoregressive models. J. Bus. Econ. Stat. 12, 187204.
Backus, D., Kehoe, P., Kydland, F., 1994. Dynamics of the trade balance and the terms of trade: the J-curve. Am. Econ. Rev. 84, 84103.
Baxter, M., Crucini, M., 1995. Business cycles and the asset structure of foreign trade. Int. Econ. Rev. 36 (4), 821854.
Berka, M., Crucini, M., Wang, C.-W., 2012. International risk sharing and commodity prices. Can. J. Econ. 45 (2), 417447.
Bhattacharya, R., Patnaik, I., Pundit, M., 2013. Emerging economy business cycles: nancial integration and terms of trade shocks. Working Papers 13/120,
National Institute of Public Finance and Policy.
Blalock, G., Gertler, P., 2005. Foreign direct investment and externalities: the case of the public intervention. In: Moran, T.H., Graham, E.M., Blomstrom, M.
(Eds.), Does Foreign Direct Investment Prompt Development?. Institute of International Economics, Washington, DC, pp. 73106.
Blattman, C., Hwang, J., Williamson, J., 2007. Winners and losers in the commodity lottery: the impact of terms-of-trade growth and volatility in the
Periphery 18701939. J. Dev. Econ. 82, 156179.
Bosworth, B., Collins, S., 1999. Capital ows to developing economies: implications for saving and investment. Brookings Papers on Economic Activity, Vol.
30 (1. The Brookings Institution, pp. 143180.
Broda, C., 2004. Terms of trade and exchange rate regimes in developing countries. J. Int. Econ. 63 (1), 3158.
Broto, C., Daz-Cassou, J., Erce, A., 2011. Measuring and explaining the volatility of capital ows to emerging countries. J. Bank. Financ. 35 (8), 19411953.
Calvo, G., 2007. Crises in Emerging Market Economies: A Global Perspective. Working Papers, Central Bank of Chile, No. 441.
Calvo, G., Izquierdo, A., Mejia, L.,2004. On the empirics of sudden stops: The relevance of balance sheet effects. NBER Working Papers 10520, National Bureau
of Economic Research, Inc.
Cashin, P., Cespedes, L., Sahay, R., 2004a. Commodity currencies and the real exchange rate. J. Dev. Econ. 75, 239268.
Cashin, P., McDermott, C., 2003. Intertemporal substitution and terms-of-trade shocks. Rev. Int. Econ. 11 (4), 604618.
Cashin, P., McDermott, C., Pattillo, C., 2004b. Terms-of-trade shocks in Africa: are they short-lived or long-lived. J. Dev. Econ. 73, 727744.
Chinn, M., Ito, H., 2008. A new measure of nancial openness. J. Comp. Policy Anal. 10 (3), 309322.
Coeurdacier, N., Kollmann, R., Martin, P., 2010. International portfolios, capital accumulation and foreign assets dynamics. J. Int. Econ. 80 (1), 100112.
Cole, H., Obstfeld, M., 1991. Commodity trade and international risk sharing: how much do nancial markets matter. J. Monetary Econ. 28, 324.
Devarajan, S., Rodrick, D., 1991 Do the benets of xed exchange rates outweigh their costs? The Franc Zone in Africa. Policy Research Working Paper Series
777, The World Bank.
Devereux, M., Sutherland, A., 2009. A portfolio model of capital ows to emerging markets. J. Dev. Econ. 89 (2), 181193.
Eichengreen, B., 1996. Globalizing Capital: A History of the International Monetary System. Princeton University Press, Princeton.
Frankel, J., 2010. The natural resource curse: A survey. NBER Working Papers 15836, National Bureau of Economic Research, Inc.
Furceri, D., Guichard, S., Rusticelli, E., 2011. Medium-term determinants of international investment positions: The role of structural policies. OECD
Economics Department Working Papers, No. 863, OECD Publishing.
Furceri, D., Guichard, S., Rusticelli, E., 2012. Episodes of large capital inows, banking and currency crises, and sudden stops. Int. Financ. 15, 135.
Gavin, M., Hausmann, R., Talvi, E.,1996. Saving behavior in Latin America: Overview and policy issues. Research Department Publications 4070,
Inter-American Development Bank, Research Department.
Harrison, A., Love, I., McMillan, M., 2004. Global capital ows and nancing constraints. J. Dev. Econ. 75 (1), 269301.
Hausmann, R., Panizza, U., Rigobon, R., 2006. The long-run volatility puzzle of the real exchange rate. J. Int. Money Financ. 25, 93124.
Kaminsky, G., Reinhart, C., Vgh, C.-A., 2005. When it rains, it pours: pro-cyclical capital ows and policies. NBER Macroeconomics Annual 2004, Vol. 19. MIT
Press, pp. 1182.
Kose, A., 2002. Explaining business cycles in small open economies: how much so world price matter. J. Int. Econ. 56 (2), 299328.
Kose, M., Prasad, E., Rogoff, K., Wei, S.-J., 2009a. Financial globalization: a reappraisal. IMF Staff Papers, Vol. 56 (1. Palgrave Macmillan, pp. 862.
Kose, M., Prasad, E., Terrones, M., 2006. How do trade and nancial integration affect the relationship between growth and volatility. J. Int. Econ. 69 (1),
176202.
Kose, M., Prasad, E., Terrones, M., 2009b. Does nancial globalization promote risk sharing. J. Dev. Econ. 89, 258270.
Kose, M., Prasad, E., Taylor, A., 2011. Thresholds in the process of international nancial integration. J. Int. Money Financ. 30, 147179.
Lane, P., 2000. International investment positions: a cross-sectional analysis. J. Int. Money Financ. 19, 513534.
Lane, P., 2001. The new open economy macroeconomics: a survey. J. Int. Econ. 54 (2), 235266.
Lane, P., Milesi-Ferretti, G., 2007. The external wealth of nations mark II: revised and extended estimates of foreign assets and liabilities, 19702004. J. Int.
Econ. 73 (2), 223250.
Lane, P., Milesi-Ferretti, G., 2008. The drivers of nancial globalization. Am. Econ. Rev. 98 (2), 327332.
Martin, P., Rey, H., 2004. Financial super-markets: size matters for asset trade. J. Int. Econ. 64 (2), 335361.
353
Mendoza, E., 1995. The terms-of-trade, the real exchange rate, and economic uctuations. Int. Econ. Rev. 36 (1), 101137.
Mendoza, E., 1997. Terms-of-trade uncertainty and economic growth. J. Dev. Econ. 54, 323356.
Mendoza, E., Terrones, M., 2008. An anatomy of credit booms: Evidence from macro aggregates and micro data. NBER Working Papers 14049, National
Bureau of Economic Research, Inc.
Milesi-Ferretti, G., Tille, C., 2011. The great retrenchment: international capital ows during the global nancial crisis. Econ. Policy 26, 289346.
Miniane, J., 2004. A new set of measures on capital account restrictions. IMF Staff Pap. 51 (2) .
Obstfeld, M., 1982. Aggregate spending and the terms of trade: is there a LaursenMetzler effect. Q. J. Econ. 97 (1), 251270.
Obstfeld, M., Rogoff, K., 1996. Foundations of International Macroeconomics. MIT Press, Cambridge, MA.
Ostry, J., Reinhart, C., 1992. Private saving and terms-of-trade shocks. IMF Staff Pap. 39 (September), 495517.
Quinn, D., 2003. Capital account liberalization and nancial globalization, 18901999. Int. J. Financ. Econ. 8, 189204.
Reinhart, C., Rogoff, K., 2004. The modern history of exchange rate arrangements: a reinterpretation. Q. J. Econ. 119 (1), 148.
Svensson, L., 1988. Trade in risky assets. Am. Econ. Rev. 78, 375394.
Turnovsky, S., Chattopadhyay, P., 2003. Volatility and growth in developing economies: some numerical results and empirical evidence. J. Int. Econ. 59,
267295.

How Does Terms-Of-trade Behavior Shape International Finan

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

How Does Terms-Of-trade Behavior Shape International Finan

Hochgeladen von

Copyright:

Verfügbare Formate

Int. Fin. Markets, Inst.

and Money 33 (2014) 335353

Contents lists available at ScienceDirect

Journal of International Financial

How does terms-of-trade behavior shape international

* Tel.: +968 24142942; fax: +968 24414043.

Fig. 1. Overall trend of nancial globalization in selected developing countries, 19802007.

2. Financial globalization and TOT volatility

Latin American countries

Middle Eastern countries

Standard deviation of trend growth

Average TOT trend-growth

Latin American countries

Middle Eastern and North African countries

Exchange rate exibility

TOT volatility nancial openness

TOT volatility economic development

5.3. Examining the interaction effects

IFIit c g TOTVit dTOTGit bTOTV Fit fX it mi ut eit

Denition and construction

World Bank, WDI and authors

Standard deviation of the log change in the CPI

World Bank, WDI

KAOPEN index of Chinn and Ito (2008)

Chinn and Ito (2008)

Log of exports plus imports (% of GDP)

World Bank, WDI

Denition and construction

A categorical measure of exchange rate regime exibility; 1 = x, 2 = peg, 3 = managed oat,

Log total foreign direct investment, assets plus liabilities (% of GDP)

Lane and Milesi-Ferretti (2007)

Log foreign portfolio equity assets (% of GDP)

Lane and Milesi-Ferretti (2007)

Das könnte Ihnen auch gefallen