WPS3733
Comments welcome
DOES OPENNESS IMPLY GREATER EXPOSURE?*
César Calderón Norman Loayza Klaus Schmidt-Hebbel**
The World Bank The World Bank Central Bank of Chile
Abstract
External exposure can be measured by the sensitivity of first and second moments of economic
growth to openness and foreign shocks. This paper provides an empirical evaluation of external
exposure using panel data methods for a worldwide sample of countries. Controlling for domestic
conditions, the paper examines the growth and volatility effects of outcome measures of trade and
financial integration as well as four types of foreign shocks: terms of trade changes, trading
partners' growth rates, international real interest rate changes, and net regional capital inflows.
The paper analyzes the possibility of non-linearities by allowing the growth and volatility effects
of openness to vary with the general level of economic development and by letting the effects of
foreign shocks depend on the degree of trade and financial integration. The findings point toward
strong non-monotonic effects of openness and external shocks on growth and volatility.
Moreover, all in all, the results contradict the view that international integration increases external
vulnerability by hurting growth and increasing volatility or by amplifying the adverse effect of
external shocks.
Key Words: External Shocks, Openness, Volatility, Growth, Non-linearities
JEL Classification: F36, F41, F43
World Bank Policy Research Working Paper 3733, October 2005
The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the
exchange of ideas about development issues. An objective of the series is to get the findings out quickly,
even if the presentations are less than fully polished. The papers carry the names of the authors and should
be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely
those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors,
or the countries they represent. Policy Research Working Papers are available online at
http://econ.worldbank.org.
*For useful comments and discussion we thank Pierre-Richard Agénor, Ricardo Caballero, Luis Felipe
Céspedes, William Easterly, Linda Kaltani, Mauricio Larraín, Claudio Raddatz, Luis Servén, Raimundo
Soto, and participants at the IFPRI/Cornell Conference on Threshold Effects and Non-Linearities in
Growth and Development, the Annual Conference of the Central Bank of Chile, the Econometric Society
World Congress, the Latin American Meetings of the Econometric Society, the Annual Meetings of the
Chilean Society of Economics, and seminars at the Bank of International Settlements, Catholic University
of Chile, University of Chile, University of La Plata, and Central Bank of Portugal.
**Calderón: The World Bank. Address: 1818 H St. N.W., Washington, DC 20433. Email:
ccalderon@worldbank.org. Loayza: The World Bank. Address: 1818 H St. N.W., Washington, DC 20433.
Email: nloayza@worldbank.org. Schmidt-Hebbel: Central Bank of Chile, Economic Research Division.
Address: Agustinas 1180, Santiago de Chile. E-mail: kschmidt@bcentral.cl.
I. Introduction
A central dimension of globalization is the world trend toward larger financial and
trade openness, observed in most industrial and developing economies. Financial openness
has increased from a world median of 5% of GDP in 1970 to 45% of GDP in 2000 and
trade openness has grown from a world median of 44% of GDP in 1970 to 70% of GDP in
2000 (see Figure 1 for annual data and definitions).
More openness implies higher integration of world goods and capital markets,
contributing to potential gains in growth and welfare. However, more international
integration could also lead to heightened external exposure, measured by the sensitivity of
first and second moments of economic growth to openness and foreign shocks. This
vulnerability may be particularly important in poor countries, due to their production
specialization, non-diversified sources of income, unstable policies, incomplete financial
markets, and/or weak institutions.
A growing empirical literature is addressing the links between openness and
macroeconomic performance, uncovering complex relationships but offering only partial
perspectives on them. In fact, the existing literature does not provide a systematic and
symmetric empirical analysis of the relationship between financial and trade openness,
financial- and trade-related foreign shocks, and macroeconomic performance as represented
by the average and volatility of economic growth. This paper attempts to fill this void by
analyzing the experience of a large sample of developed and developing countries in the
decades spanning 1970-2000.
The plan of the paper is the following. Section II provides a comprehensive review
of the relevant literature. Section III presents the empirical contribution of the paper,
including its methodology, worldwide data sample, and panel-data regression results for
average growth and growth volatility. In this section we report first the simple linear effects
of measures of trade and financial openness as well as four types of foreign shocks: terms
of trade changes, trading partners' growth rates, foreign real interest rate changes, and net
regional capital inflows. Then we analyze empirically the possibility of non-linearities by
assessing the dependence of the effects of trade and financial openness on the level of per
1
capita income. Third, we measure the amplification or dampening of the effects of external
shocks depending on the degree of trade and financial openness. Section IV concludes.
II. Review of the Empirical Literature
A growing empirical literature has analyzed the effects of financial openness, trade
openness, and foreign shocks on growth levels and growth volatility. In this section we
proceed to review briefly the analytical underpinnings and existing empirical results on the
core relations that are the focus of this paper: those between financial openness, trade
openness, foreign shocks, GDP growth, and GDP growth volatility.
A. Financial Openness and Growth
The empirical literature on financial openness (FO) as well as the research on
trade openness (TO) reviewed below is based on two classes of openness measures.
Policy or legal measures reflect policy and regulatory restrictions or barriers imposed
domestically on international trade volumes and financial flows and/or holdings. In
contrast, outcome or de facto measures reflect actual trade volumes and financial flows or
stocks between the domestic economy and the rest of the world. On one hand, the strength
of the first class of FO / TO measures is that it reflects policy restrictions while the second
class is influenced by country-specific features including size, distance, production
specialization, and risk in addition to domestic restrictions. Hence the first class may
represent more exogenous policy conditions while the second class of FO/TO measures is
likely to be endogenous to variables that are often explained by integration, including
growth. On the other hand, outcome measures may reflect more truly country integration
into world markets, while policy measures may not reflect binding policy restrictions and
may be subject to measurement bias. Moreover, outcome indicators represent continuous
variables that are more easily measured and are more widely available than the discrete
measures of policy restrictions.
The policy measure of FO is largely based on information on capital controls in the
IMF's Annual Report on Exchange Arrangements and Exchange Restrictions, which has
spawned construction of several discrete indicators of capital-flow restrictions on cross-
2
border or cross-residence flows.1 The outcome measure reflects an actual measure of FO,
based on cross-country capital flows or capital stocks, defined as ratios to GDP.2 3
There is a growing literature on the effects of capital-account liberalization and FO
on growth Edison, Klein et al. (2002) provide an excellent survey. Here we discuss the
previous work only briefly, in those dimensions that are relevant to this paper. (See Table 1
for a selective summary).
Preceding studies diverge significantly in FO measures, empirical methods, data
samples, and results. The studies conducted during the past decade most of them on
measures based on the de jure IMF restrictions show mixed results. While Quinn (1997)
and Edison, Klein et al. (2002) report positive significant growth effects of IFI for the
world at large, many others do not find any evidence or reject robust evidence of FO
growth effects, including Grilli and Milesi-Ferretti (1995), Kraay (1998), Rodrik (1998),
O'Donnell (2001), Edison, Levine et al. (2002), and Razin and Rubinstein (2004).
Other studies have tested for FO measures and their interaction with third variables,
in order to reflect non-linearities and non-monotonicities in the relationship between
growth and FO. Here the general finding is that FO and external financial liberalization
tend to reduce growth in countries that are either not industrialized (Klein and Olivei 2000),
feature ethnic heterogeneity (Chanda 2001), have low income (Edwards 2001), or exhibit
high black-market premiums (Arteta et al. 2001), while FO raises growth in countries with
the opposite features. Klein (2003) reports quadratic interaction terms of FO with
government quality and with per capita GDP, implying that FO only raises growth in
middle-income countries and reduces growth in low- and high-income countries.
1 Policy or legal measures of capital-flow restrictions include a dummy variable based directly on the IMF
measure, the Share measure reflecting the number of years of IMF-measured restrictions, and the Quinn index
that allows varying degrees of restrictions based on the IMF information (Quinn 1997), all available for
varying country and annual samples. In addition, the OECD computes a measure of capital-flow restrictions
for 21 OECD countries. Finally, Montiel and Reinhart (1999) compute an index of restrictions for 15
emerging economies. For detailed discussion of legal and actual measures of capital restrictions see Edison,
Klein et al. (2002).
2 Actual measures of FO include large country samples for capital flows (Kraay 1998) and stocks on gross
capital flows accumulated from flows (Lane and Milesi-Ferretti 2001, 2003, Kose, Prasad and Terrones
2004).
3 More partial dimensions of financial restrictions and liberalization, even if they have some bearing for FO,
are not considered in this paper. Among the latter are measures of stock market liberalization and the
literature on the effects of the latter on growth, reviewed by Edison, Klein et al. (2002).
3
Therefore the latter evidence suggests strongly that FO and growth display a non-
monotonic relationship. At low levels of development for several measures of
development financial liberalization and/or FO tend to lower growth, while the opposite
is observed in developed nations, where FO contributes to growth. This result should not
come as surprise. It reflects that international financial liberalization in non-OECD
countries frequently hampered by low-quality governments, poor institutions, or lack of
effective banking regulation and supervision takes place jointly with domestic financial
liberalization. After the subsequent boom phase of massive credit growth, capital inflows,
and domestic spending, frequently a hard landing ensues, reflected in a banking crisis and
domestic recession, and hence depressing medium-term growth.
Finally Kose et al. (2005) do not find robust growth effects of FO separately but,
interestingly, show that FO interaction with growth volatility turns the negative effect of
volatility on growth into a positive one. This suggests that FO may alleviate adverse growth
effects stemming from high GDP volatility by strengthening access to external counter-
cyclical lending and allowing for international portfolio diversification by domestic agents.
B. Trade Openness and Growth
Like FO measures, policy or legal measures of TO are based on measures of trade
policy or trade restrictions and the standard measure of outcome or de facto TO is the GDP
share of total trade (exports plus imports).
As compared to the empirical literature on FO and growth, there is a longer and
broader cross-country growth literature on the effects of trade liberalization and TO.4
Again, here we discuss the previous work only briefly, in those dimensions that are relevant
to this paper. (See Table 1 for a selective summary.)
Previous work differs strongly in many dimensions dependent variable (GDP
growth, per capita income level, per capita income convergence), TO measures, controls,
data samples, and econometric techniques. Earlier work finds significant, positive, and
often very large effects of TO on growth, income levels, or income convergence (Dollar
1992, Ben-David 1993, Sachs and Warner 1995, Edwards 1998, Frankel and Romer 1999).
Much of the latter research has been subsequently criticized for possible bias stemming
4Edwards (1993) provides a survey of the older literature.
4
from endogeneity of trade to income levels or GDP growth, lack of robustness due to
exclusion of relevant controls and/or use of inadequate data samples and estimation
techniques. Rodrik and Rodriguez (2000) report that TO effects on growth are not robust to
inclusion of geographic latitude and Rodrik et al. (2004) find that TO is not robust to
inclusion of institutional quality. Rigobon and Rodrik (2005), based on a technique of
identification through heteroskedasticity, is the only recent study reporting negative
significant effects of TO on per capita income levels, controlling for institutions and
geography. A subsequent study based on the same technique, by Lee, Ricci, and Rigobon
(2004), shows that TO has positive but small effects on growth.
In contrast to the latter studies, there is recent research that finds significant and
robust effects of TO on growth or income levels (Wacziarg 2001, Irwin and Terviö 2002,
Alcala and Ciccone 2004, Kose et al. 2005). Some other recent work reports more qualified
results: TO effects are not robust in cross-section estimations but they are significant in
panel studies and robust to inclusion of institutional variables (Dollar and Kraay 2003,
Wacziarg and Welch 2003). The latter study, focusing on trade liberalization country
episodes, shows that trade shares and growth increase significantly and substantially after
trade is liberalized. Finally, two recent studies look at interaction effects between TO
measures and other variables. Kose et al. (2005), complementing their research on FO and
growth, report robust positive effects of TO on growth and find that TO turns the negative
effect of volatility on growth into a positive one. Alesina et al. (2005), controlling for
country size and interaction effects between TO and size, find that TO has large effects in
small countries but these effects become zero as country size tends to maximum size.
C. Foreign Shocks and Growth
Foreign variables relevant to open economies comprise financial and real variables
associated to capital flows and trade flows, respectively. They include price variables
(international interest rates, terms of trade) and quantity variables (capital flows to
emerging economies). For truly small countries (those facing infinite demand/supply
elasticities for their exports/imports of capital and goods), only price variables matter for
determining domestic performance, including growth. For countries that have some
monopolic or monopsonic power in international markets reflecting their size and/or their
5
specialization in trading differentiated goods under conditions of monopolistic competition
quantities matter for domestic performance too. However, here it is important to isolate
the exogenous determinant (or the instrument) of quantities as the relevant predetermined
quantity variable. In this vein, the world or regional supply of capital could be an adequate
instrument for the supply of capital to the domestic economy, while average growth of all
trading partners could be an adequate instrument for the foreign demand of exports by the
domestic economy.
Foreign shocks are measured in two ways: the rate of growth or deviation of a
foreign variable from its preceding level and the standard deviation of the variable in a
given time period. Most empirical growth studies include one or two foreign shock
variables based on the first measure, typically as control variables. Here we refer only
selectively to recent cross-country empirical studies that quantify the effects of terms of
trade, the foreign interest rate, capital inflows, and trade partner growth on domestic growth
and growth volatility (Table 1).
The rate of growth of the terms of trade is the most widely used measure of foreign
shocks (among representative studies are Easterly, Loayza, and Montiel 1997, Fernández-
Arias and Montiel 2002, Loayza, Fajnzylber, and Calderón 2004, Barro and Sala-i-Martin
2004). In most studies terms-of-trade shocks turn out to be significantly positive
determinants of growth.
In a highly integrated world, open economies are also influenced by growth in the
rest of the world. Arora and Vamvakidis (2005) find that a 1% increase in economic growth
of the countries' trading partners leads to higher domestic growth by 0.8 percentage points.5
Other studies have included the ratio of private capital inflows to GDP as a growth
determinant and have also evaluated the impact of different types of capital flows on
growth (Calderón and Schmidt-Hebbel 2003). Most of these studies have found a positive
impact of private capital inflows on growth, with a stronger effect in the case of foreign
direct investment.6
5They argue that this result is consistent with the literature on the impact of cross-country spillovers (Arora
and Vamvakidis 2004, Ahmed and Loungani 1999).
6Most of these studies instrumentalize capital inflows using lagged values, legal origin variables or investor
protection measures in order to avoid endogeneity bias from the response of capital flows to growth.
6
D. Growth and Growth Volatility
Since the influential work by Ramey and Ramey (1995) that pointed out the
negative cross-country relation between average GDP growth and GDP volatility, research
has focused on providing explanations and more evidence on this negative relation found
for the world sample and among low-income countries but not among industrial economies
(Table 2). Elbadawi and Schmidt-Hebbel (1998) provide evidence that measures of
macroeconomic volatility and low macroeconomic performance have a negative effect on
growth, hypothesizing that the latter is behind the changing sign in the volatility-growth
relation between low-income and high-income countries. Hnatkovska and Loayza (2004)
provide robust evidence for the negative effect of GDP volatility on growth. By including
interaction effects between volatility and relevant variables, they report that the negative
effect of growth volatility on output is larger in economies with low-income levels and
weak institutions and policies, and that the negative effect has grown in recent decades due
to deep recessions.
E. Financial Openness, Trade Openness, and Growth Volatility
Easterly, Islam, and Stiglitz (2000) explore the sources of GDP growth volatility in
industrial and developing countries (Table 2). They find that higher TO leads to larger
growth volatility, especially in developing countries. However they do not find a significant
impact of FO on output volatility. O'Donnell (2001) finds that larger FO is associated with
lower (higher) output volatility in OECD (non-OECD) countries. His results also suggest
that countries with more developed financial sectors are able to reduce output volatility
through financial integration.
Kose et al. (2003) reports that none of four FO and TO measures has any robust
effect on GDP volatility. However terms of trade volatility, financial depth, and M2
volatility raise output volatility in the world. Bekaert et al. (2004) provide cross-section and
time-series evidence of the relationship between FO and GDP growth volatility. Using two
indicators of capital account openness, they find considerable evidence of lower GDP
volatility after capital account liberalization in the world sample, with somewhat weaker
results for developing countries.
7
Cavallo (2005) presents evidence that suggests that TO lowers output volatility in
net terms. According to the author, this is due to two countervailing effects. Larger TO
raises growth volatility through the terms-of-trade channel but this is more than offset by
the finding that TO lowers growth volatility due to lower financial vulnerability to external
shocks, sudden stops, and currency crushes.
Finally, regarding the influence of openness on the effect of growth volatility on
growth levels, we mentioned above the result by Kose et al. (2005) that both FO and TO
turn the latter negative effect into a positive one. Hnatkovska and Loayza (2004), however,
reject an ameliorating influence of TO on the negative volatility-growth effect.
F. Foreign Shocks and Growth Volatility
Foreign shocks measured as deviation of a foreign variable from its trend or
average level, or as the standard deviation of the variable have been shown to be
significant in determining business-cycle fluctuations of GDP or GDP volatility.
Kose (2002) evaluates the importance of fluctuations in world prices --fluctuations
in the prices of primary, capital, and intermediate goods, and in the world interest rate-- in
driving business cycle fluctuations in small open developing countries (Table 2). He finds
that roughly 88% of aggregate output fluctuations can be explained by world price shocks.
Rodrik (2001) shows that GNP volatility in Latin America and the Caribbean is driven by
both external shocks and domestic policy failures with the terms of trade and capital
flows as key contributors on the external front.
Mendoza (1995) was among the first attempts to evaluate the quantitative
importance of terms of trade shocks in explaining business cycles, using a stochastic
dynamic small open economy model. He found that terms of trade disturbances explain
56% of output variation.
Early research found that world interest rates do not have a significant role in
explaining the dynamics of small open economies, including output fluctuations (Mendoza,
1991, Correia et al. 1992, 1995, Schmitt-Grohe 1998). However, using a dynamic
stochastic small open economy model, Blankenau et al. (2001) find that world real interest
rate shocks explain 33% of Canada's output variation. The above mentioned research by
Kose (2002) finds that world interest rate shocks account for roughly 1% of output
8
volatility in developing countries. Neumeyer and Perri (2004) report large effects of
country risk fluctuations and small effects of world interest rate fluctuations on the high
volatility of output in emerging economies.
Finally, Rodrik (2001) suggests that the instability of private capital flows has been
the most important determinant of macroeconomic volatility in Latin America and the
Caribbean during the 1990s. Together with per capita income, capital flow volatility
accounts for close to half of the cross-national variation in GNP volatility in the region
during the 1990s, compared to 20% in the 1980s.
G. Summing Up and Open Questions
A rich empirical literature has developed on the relations between TO, FO, foreign
shocks, growth, and growth volatility. At this stage of the cross-country and panel data
research we conclude that neither FO nor TO does show a linear or even monotonic relation
with economic growth and growth volatility. Financial liberalization and FO tend to lower
growth at low levels of development while the opposite is observed in developed
economies. In the case of TO and trade liberalization, the non-monotonicity with growth is
not quite so strong as for FO and financial liberalization, as all studies, less one, show that
the lower bound of growth effects of TO is zero. Non-monotonicities between openness and
growth show up as strong interaction effects between measures of openness and various
measures of development (per capita income, institutional quality, and growth volatility,
among others) and country size. Yet the discussion about inclusion of adequate controls and
their possible interaction with measures of openness, and their implications for the
robustness of the role of the latter in determining growth, is still raging.
Regarding the behavior of growth volatility and cyclical output fluctuations, the
evidence on the effects of FO and TO is contradictory and inconclusive. There is not even
agreement on the intensifying or ameliorating effects that openness has on the traditionally
negative relation between growth volatility and growth levels.
There is much more agreement regarding growth effects of foreign relevant
variables, particularly the terms of trade. The latter variable is a frequently-used control
variable in many empirical studies, and is typically found to be positive and significant in
explaining growth. Some studies also report positive domestic growth gains from higher
world growth and larger capital inflows.
9
The existing evidence on growth volatility and cyclical output fluctuations shows
generally significant and large positive effects of volatility in relevant external variables.
Therefore, in the light of this review, we conclude that the existing literature does
not provide a systematic and symmetrical empirical analysis of:
· relationships between average economic growth and both financial and trade openness,
· relationships between growth volatility and both financial and trade openness,
· foreign shocks (financial and real; exogenous price- and quantity-related) and their
influence on growth and growth volatility, and
· interaction between openness and foreign shocks in their effect on average growth and
growth volatility.
The rest of the paper addresses these issues.
III. Empirical Analysis
We conduct two analogous empirical analyses. The first focuses on economic
growth and the second on growth volatility. In both cases, the dependent variable is
constructed using the annual per capita real GDP growth rate as the main input. For
economic growth, the dependent variable is the average rate of growth over a medium-run
time window; and for growth volatility, the dependent variable is the standard deviation of
the growth rate over the same time window.
For both empirical analyses, our objective is, first, to study the simple linear effects
of trade and financial openness as well as of various external shocks; second, to examine
how the effects of trade and financial opening vary with the level of per capita income; and
third, to consider whether the effects of external shocks are amplified or reduced by the
degree of trade and financial openness. By conducting these exercises, we aim to provide a
comprehensive empirical assessment of openness and external conditions for
macroeconomic performance.
A. Sample and Methodology
We work with a pooled data set of cross-country and time-series observations. It
consists of 76 countries and, for each of them, at most 6 non-overlapping five-year periods
10
spanning the 1970-2000 period. See Appendix 1 for the list of countries in the sample.
Appendix 2 provides full definitions and sources of all variables used in the paper, and
Appendixes 3 and 4 present basic descriptive statistics for the data used in the economic
growth and growth volatility regressions, respectively.
We use an estimation method that is suited to panel data, deals with static or
dynamic regression specifications, controls for unobserved time- and country-specific
effects, and accounts for some endogeneity in the explanatory variables. This is the
generalized method of moments (GMM) for dynamic models of panel data developed by
Arellano and Bond (1991) and Arellano and Bover (1995).
The general regression equation to be estimated is the following
yi,t = ' Xi,t + µt +i +i,t (1)
where the subscripts i,t represent country and time period, respectively. y is the dependent
variable of interest, that is, economic growth or growth volatility. X is a set of time- and
country-varying explanatory variables that may include a lagged dependent variable,
proxies of trade and financial openness, measures of various external shocks, interaction
terms, and control variables. Finally, µt is an unobserved time-specific effect, i is an
unobserved country-specific effect, and it is the error term.
The method deals with unobserved time effects through the inclusion of period-
specific intercepts. Dealing with unobserved country effects is not as simple given the
possibility that the model is dynamic and contains endogenous explanatory variables.
Unobserved country effects are controlled for by differencing and instrumentation.
Likewise, the method relies on instrumentation to control for joint endogeneity.
Specifically, it allows relaxing the assumption of strong exogeneity of the explanatory
variables by allowing them to be correlated with current and previous realizations of the
error term .
Parameter identification is achieved by assuming that future realizations of the error
term do not affect current values of the explanatory variables, that the error term is
serially uncorrelated, and that changes in the explanatory variables are uncorrelated with
the unobserved country-specific effect. As Arellano and Bond (1991) and Arellano and
Bover (1995) show, this set of assumptions generates moment conditions that allow
estimation of the parameters of interest. The instruments corresponding to these moment
11
conditions are appropriately lagged values of both levels and differences of the explanatory
and dependent variables (the latter if the model is dynamic). Since typically the moment
conditions over-identify the regression model, they also allow for specification testing
through a Sargan-type test.
B. Economic Growth Regressions
As is standard in the literature, the dependent variable is the average rate of real per
capita GDP growth. The regression equation is dynamic in the sense that it includes the
initial level of per capita GDP as an explanatory variable. As additional control variables,
the regression includes the average rate of secondary school enrollment to account for
human capital investment, the average ratio of private credit to GDP as a measure of
financial depth, the average inflation rate to account for monetary discipline, and the
average ratio of government consumption to GDP as a measure of government burden. The
regression equation also allows for both unobserved time-specific and country-specific
effects.
The explanatory variables of interest are measures of trade and financial openness,
measures of external shocks, and various interaction terms. Given that we want to evaluate
the effects of the economy's actual contact with international markets, we work with
outcome measures of trade and financial openness. These measures are related to policies
but are also the result of structural characteristics of the economy, such as size, natural and
social endowments, and public infrastructure. The outcome measures we use are the ratio
of exports and imports to GDP in the case of trade, and the ratio of portfolio and FDI
liabilities to GDP in the case of financial openness.
We consider four types of external shocks; the first two primarily related to trade in
goods and the latter two mainly related to financial transactions. All of them are defined so
that they can be considered as exogenous to the country in question. They are the average
growth of the terms of trade, the average weighted output growth rate of trade partners, the
average amount of capital flows to the region where the country is located, and the average
change of the international interest rate. Whereas the first two variables vary by country
and time period, the third varies only by region and period, and the fourth varies only by
time period. Because of its limited sample variation, the effect of the international interest
12
rate shock cannot be distinguished from the unobserved time-specific effect; however, it is
still possible to evaluate the effect of its interaction with the measures of trade and financial
openness.
Linear Effects of Openness and External Shocks
In the basic case, the effects of openness and shocks on growth are independent
from each other and independent from other characteristics of the economy. This
corresponds to the most common treatment of growth determinants in the literature. The
regression equation we estimate in this case is the following,
yi,t = 0'CVi,t + 1'OPEi,t + 2'EXTi,t + µt +i +i,t (2)
where CV is the set of control variables, OPE is the set of openness variables, and EXT is
the set of foreign-shock variables.
The estimation results are presented in the first column of Table 3. We find that
both financial and trade openness are positively related to economic growth. As mentioned
in the introduction, the effect of financial and trade opening on growth is found to be
ambiguous in the literature. However, ambiguity gives way to positive effects when the
time-series dimension is taken into account: the beneficial impact of opening is most
clearly seen in the experience of countries before and after liberalization (see, for instance,
Wacziarg and Welch 2003). Our panel-data results confirm this finding as they are based
not only on cross-country comparisons but also on over-time changes for individual
countries. Furthermore, since our methodology controls for country-specific effects and the
joint endogeneity of openness, our results are not subject to the criticism that the positive
growth effect of openness is not robust to the inclusion of variables such as geographical
location (see Rodrik and Rodríguez 2000 for the case of trade) or is due to reverse
causation.
Regarding the growth effects of external shocks, they are all significant and carry
the expected signs. That is, increases in favorable terms of trade, in the growth rate of trade
partners, and in capital flows to the region produce a rise in average economic growth.
These results together with the estimated size and significance of the period shifts confirm
the substantial importance of external conditions as growth determinants.
13
All control variables carry significant coefficients of expected signs. The Sargan
and serial-correlation specification tests do not reject the null hypothesis of correct
specification, lending support to our estimation results. This is the case in all exercises
presented below, and to avoid redundancy we only mention it here.
The Effect of Openness Depending on the Level of Income
Increasingly there is the notion that the growth effect of openness may not be
homogeneous across countries. Indeed, in part motivated by the work of Klein and Olivei
(2000) in the case of financial openness, researchers have lately considered the possibility
that the growth effect of opening the economy may depend on country characteristics such
as income and institutional quality (see Edwards 2001 and Klein 2003). In the framework
of our panel-data methodology, we now reassess this possibility by allowing the effect of
each measure of openness to vary with the level of real per capita GDP, which serves as a
proxy for overall development. We do this by interacting each openness measure with
linear and quadratic per capita GDP (Inc) in each country at the start of the corresponding
period. The regression equation we estimate in this case is the following
yi,t = 0'CVi,t + 1'OPEi,t + 2'EXTi,t + 3'OPEi,t * Inci,t + 4'OPEi,t * Inci,t + µt +i + i,t (3)
2
The last two columns of Table 3 present the estimation results. We consider the
interaction between per capita GDP and the financial and trade openness variables one at a
time; we do this in order to both simplify the interpretation of the results and do not
overextend the parameter requirements on the data. Thus, column 2 shows the results when
financial openness is interacted with income, and column 3, when trade openness is
interacted with income. Interestingly, the regression results are qualitatively similar
whether we deal with financial or trade openness and can be summarized as follows. The
coefficient on the corresponding openness indicator by itself is negative and significant,
and the coefficients on the linear and quadratic interaction terms are significantly positive
and negative, respectively. Therefore, the growth effect of opening the economy appears to
depend on per capita income or, taking a broader interpretation, on the overall level of
development. A corollary is that the net growth effect could in theory be positive or
negative, and we should examine which is the case for the actual levels of per capita GDP
found in our sample.
14
Figure 2 (a and b) illustrates what the estimated pattern of coefficients implies for
the change in growth produced by an increase in each openness measure. Specifically,
Figure 2 uses the regression point estimates to plot the growth effect of a one-standard-
deviation increase in openness as a function of per capita GDP for the full range of the
sample. For both financial and trade opening, their growth effect is nearly zero for low
levels of per capita GDP, it increases at a decreasing rate as income rises, and it reaches a
maximum but only at high levels of income (higher in the case of trade openness). The
growth effect of rising openness appears to be economically significant for middle- and
high-income countries. In relation to the previous literature, we also find non-linear growth
effects of opening, but we have some differences in the precise nature of the non-linearity.
For instance, we differ from Edwards (2001) in that we do not find negative effects of
increasing financial openness for low-income countries; and we differ from Klein (2003) in
that we find that the growth effects of financial opening remain positive even for high-
income countries. However, we agree with these papers that middle-income countries can
expect to improve their growth performance as they become more integrated with the rest
of the world.
Our coefficient estimates suggest significant economic effects. For instance, a one-
standard-deviation increase in the degree of financial openness would lead to higher per
capita growth rates by: (a) 0.67 percentage points for country observations at the 25th
percentile of the sample distribution of output per capita during the 1970-2000 period
(corresponding approximately to Pakistan in the mid 1990s), (b) 0.9 percentage points for
observations at the 75th percentile (Malaysia around 1995), and (c) 0.8 percentage points
for countries at the 95th percentile (Italy around 1995). The highest growth effect of larger
international financial integration, 0.91 percentage points, is achieved by middle-income
country observations at the 62th percentile (corresponding approximately to the levels of
income per capita of Chile, Mexico, and Uruguay in 1995).
Similarly, an analogous increase in the degree of trade openness will generate an
increase in the growth rate of GDP per capita of 0.33 percentage points for countries at the
25th percentile of the sample distribution of output per capita, 0.53 for the median country
in the sample (Indonesia around 1995), 0.67 for countries at the 75th percentile, and 0.74
for countries at the 95th percentile of the world distribution of output per capita.
15
The Interaction between Openness and External Shocks
The previous exercises analyze the potential effect that openness can have on
economic growth, controlling for various external shocks. Here we focus on whether
openness makes the economy more or less responsive to external shocks. Ideally for the
beneficial role of openness, it would amplify the growth effect of positive shocks and
dampen the effect of negative ones. However, we find no evidence of asymmetric
interaction effects (in results not presented here) and limit ourselves to the issue of whether
the economic growth effect of shocks depends on the openness of the economy.
We address this question by considering interaction terms between each of the
shocks and the openness variables. The regression equation we estimate in this case is the
following,
yi,t = 0'CVi,t + 1'OPEi,t + 2'EXTi,t + 3'OPEi,t * EXTi,t + µt +i +i,t (4)
There are a large number of possibilities for these interactions, but in order to avoid
overextending the parameter requirements on the data, we consider the interactions between
financial and trade openness indicators with the external shocks one shock at a time. This
will also allow us to simulate the effect of each shock independently. The results are
presented in Table 4, with each column devoted to the interactions with each of the four
external shocks. An interesting pattern emerges. Larger trade openness decreases the
economic growth effect of trade-related shocks --the growth of terms of trade and the GDP
growth of trade partners-- while increasing the economic growth effect of financial market
shocks --international interest rate changes and regional capital inflows. Conversely, larger
financial openness increases the economic growth effect of trade-related shocks while
reducing the impact of regional capital inflows, one of the financial market shocks. The
exception to this pattern is that financial openness seems to increase the economic growth
effect of international interest rate changes.
Since the growth effect of a shock now depends on three coefficients plus the levels
of trade and financial openness, it is not immediately clear what the net effect is. Figure 3
helps to make this assessment by graphing the economic growth effect of one-standard-
deviation increase in each shock as a function of, first, trade openness and, then, financial
16
openness.7 In the former case, we use the sample average of financial openness in the
calculation of the partial effects; and in the latter, the sample average of trade openness.8
The economic growth effect of an increase in terms of trade growth is positive for low and
medium levels of trade openness; but as this increases further, the growth effect declines
and may even become negative. The economic growth effect of the positive terms of trade
shock increases with financial openness and is positive throughout the relevant range.
The economic growth effect of an increase in trade partners GDP growth is always
positive, and, as noted above, it decreases with trade openness and increases with financial
openness. In the case of the international interest rate shock, its direct impact on economic
growth cannot be separated from the time effects; however, based on an exercise where we
compare the period shifts with and without the interest rate shock interactions, we estimate
that the direct impact of an increase in international interest rates is significantly negative.
Once we take into account the interactions, the total economic growth effect of the interest
rate jump continues to be negative, but it approaches zero as either type of openness rises.
Finally, regarding the positive capital flow shock, its effect on economic growth is positive
throughout most of the relevant ranges; but, as indicated above, larger trade openness
increases its economic growth effect, while financial openness reduces it. All in all, these
results challenge the conventional wisdom that trade openness increases the exposure to
trade-related shocks and that financial openness magnifies financial-market shocks. The
channels of transmission seem to be more complex.
We can use the numbers to draw several quantitative implications. Consider the
following, for instance. First let's consider the growth effects of external shocks at various
levels of trade openness and given the sample average level of financial openness
(corresponding approximately to that of India in the late 1990s). A one-standard-deviation
decline in the terms of trade would lead to a fall of growth in real GDP per capita of 0.98
percentage points for country observations at the 25th percentile of the sample distribution
of trade openness (approximately Pakistan in the late 1990s), 0.50 percentage points for the
median country (Italy, 1996-2000), and nearly zero for country observations at the 75th
7Considering their effect on growth, an increase in terms of trade, partners' GDP, and capital inflows are
regarded as beneficial shocks in the following discussion, while an increase in international interest rates is
regarded as a detrimental shock.
8For these simulations, we restrict the range of the financial openness indicator to values where the stock of
foreign liabilities is positive.
17
percentile (Paraguay, 1996-2000). Analogously, a one-standard-deviation decline in the
growth rate of a country's trading partners would lead to a reduction in the growth rate of
0.96 percentage points for country observations at the 25th percentile of the sample
distribution of trade openness and 0.78 percentage points for countries at the 75th
percentile. Now let's consider the growth effects of external shocks at various levels of
financial openness and given the sample average level of trade openness (approximately
corresponding to Indonesia and Ecuador in the late 1990s). A one-standard-deviation
decline in regional capital inflows would generate a reduction in per capita GDP growth of
0.35 percentage points for country observations at the 25th percentile of the sample
distribution of financial openness (approximately Algeria in the late 1990s), 0.28
percentage points for the median country (El Salvador, 1996-00), and 0.2 percentage points
for countries at the 75th percentile (South Korea, 1990-00).
Finally, openness and external shocks may have a significant effect on
macroeconomic volatility, and this in turn has been found to have a harmful influence on
economic growth (see Fatás 2002, and Hnatkovska and Loayza 2003). Therefore, there is
the possibility that the growth effects of openness and external shocks that we just
described occur through their impact on macroeconomic volatility. To consider this
possibility, we add the standard deviation of economic growth as an additional explanatory
variable. The results are presented in Table 5, and although growth volatility carries the
expected negative and significant coefficient, the coefficients on all other relevant variables
retain their sign, significance, and, to a large extent, size. Therefore, the growth effects of
openness, external shocks, and their interactions can be considered independently of their
volatility effects, to which we turn next.
C. Volatility Regressions
In this case, the dependent variable is the standard deviation of annual real per
capita GDP growth, calculated over each 5-year period. The control variables represent
some of the main sources of domestically induced volatility and are calculated over the
same periods. They are the standard deviation of annual inflation, an average index of real
exchange rate overvaluation, and the average number of years under systemic banking
18
crisis. The volatility regression equation also allows for both unobserved time-specific and
country-specific effects.
The explanatory variables of interest are measures of trade and financial openness,
the volatility of external shocks, and various interaction terms. The outcome measures of
trade and financial openness are the same as those described above for the case of growth.
Note, however, that for growth volatility regressions, the relevant indicator of the external
shock is given by its volatility. For this reason, we use the standard deviation of each
external shock as the measure of interest (and not its average value, as we did in the case of
growth regressions).
Linear Effects of Openness and External Shocks
The regression we estimate in this case is analogous to equation (2). The estimation
results are presented in the first column of Table 6. We find that whereas an increase in
financial openness tends to reduce volatility, larger trade openness increases it. This would
indicate that financial opening has an unambiguous beneficial impact on the economy by
both increasing average growth and reducing its volatility. In contrast, the effect of larger
trade openness would be ambiguous as it induces higher but also more volatile growth.
Regarding external shocks, we find that an increase in the volatility of terms of trade
changes, the growth rate of trade partners, and capital flows to the region produce an
increase in the volatility of economic growth. As expected, then, countries facing a more
volatile external environment would undergo higher volatility.
All control variables carry positive and significant coefficients, as expected. The
Sargan and serial-correlation specification tests do not reject the null hypothesis of correct
specification, lending support to our estimation results. This is also the case in all
remaining volatility regressions presented below.
The Effect of Openness Depending on the Level of Income
We consider now the possibility that the volatility effect of openness vary with the
level of real per capita GDP in the country at the start of the corresponding period. As in
the case of growth regressions, we do this by interacting each openness measure with linear
19
and quadratic per capita GDP. Then, the regression we estimate in this group of exercises
is analogous to equation (3).
The estimation results are presented in the last two columns of Table 6. As with
growth regressions, we consider the interaction between per capita GDP and the openness
variables one at a time. Figure 4 (a and b) takes the regression point estimates to graph the
volatility effect of each openness indicator as a function of per capita GDP. There is a
remarkable degree of similarity in the pattern of coefficients related to financial and trade
openness indicators. The coefficient on the corresponding measure of openness by itself is
always negative, and the coefficients on the linear and quadratic interaction terms are
positive and negative, respectively. All of them are statistically significant. The interaction
coefficients indicate that the volatility effect of a change in both types of openness varies
convexly with income. The total volatility effect of a change in openness can be positive or
negative depending on the size of the coefficients, and in this regard the effect of financial
openness is different from that of trade openness, as discussed next.
Figure 4 (a) illustrates that a rise in financial openness leads to a decrease in
volatility at all income levels. The volatility reducing effect of financial opening is smallest
in magnitude in medium-income countries and largest at the both ends of the cross-country
income distribution. The case of trade openness is somewhat different (see Figure 4 (b)).
A rise in trade openness leads to an increase in volatility at all levels of national income.
Poorest countries are the most vulnerable to the volatility inducing impact of trade
openness. This effect decreases gradually as national income rises. Regarding the size of
the volatility effect, the beneficial impact of financial opening is considerably larger than
the detrimental effect of trade opening. This is important to keep in mind when financial
and trade opening are undertaken together and their joint effects on vulnerability are
considered.
The Interaction between Openness and External Shocks
The last exercise considers to what extent the volatility effect of external shocks
depends on the economy's trade and financial openness. To do so, we include interaction
terms between each of the shocks and the openness variables in the volatility regression.
Then, in this case the regression we estimate is analogous to equation (4).
20
The estimation results are presented in Table 7, with each column devoted to the
interactions with each of the four external shocks. Figure 5 illustrates how the volatility
effects of each shock vary with the level of, first, trade openness and, then, financial
openness. We find that the interaction between any shock and financial openness always
carries a significantly negative coefficient. This implies that financial openness helps
reduce the volatility effect of all shocks considered, possibly by helping diversify away
risk. Trade openness also reduces the volatility effect of terms of trade shocks. However,
it appears to increase increases the volatility effect of the growth rate of trade partners,
while it has no significant interaction with interest rate and capital flow shocks.
IV. Conclusions
This paper tries to shed light on the question as to whether international integration
increases a country's external vulnerability. There are two sides to this question. The first
is whether openness by itself hurts macroeconomic performance by decreasing average
economic growth and increasing its volatility. The second one is whether openness
magnifies the impact of adverse foreign shocks and exacerbates the contagion of external
volatility. The cross-country and over-time empirical evaluation conducted in the paper
provides some answers to these questions, and around them we organize these concluding
remarks.
First, we find no evidence that financial or trade openness by itself induces a decline
in economic growth. On the contrary, there is strong indication that financial and trade
opening lead to growth improvement. Granted, this positive impact appears to be small in
poor countries, but it becomes substantial as the country develops.
Second, we find no evidence that higher financial openness increases growth
volatility. In fact, it appears that financial opening helps a country reduce its growth
volatility, particularly if its domestic financial system is not in the transitional turmoil that
characterizes middle-income countries. Trade opening does seem to increase volatility, but
the effect is small and only relevant for low-income countries. For poor and middle-
income countries, financial and trade openness present some trade-offs, but they are mostly
21
in favor of international integration. Moreover, as countries develop, the beneficial impact
of trade and financial openness becomes strong and unambiguous.
Third, on whether financial and trade openness magnify the growth effect of adverse
external shocks, the evidence is mixed. On the positive side, financial openness dampens
the harmful growth impact of financial market shocks (such as higher international interest
rates and lower capital inflows) while trade openness limits the adverse effect of trade-
related shocks (such as unfavorable terms of trade changes and lower foreign growth). On
the negative side, financial openness tends to exacerbate the growth effect of trade-related
shocks while trade openness magnifies the effect of falling capital flows.
Fourth, the evidence regarding the ability of financial and trade openness to limit
the contagion of external volatility is quite in favor of openness. With only one exception,
we find no evidence that openness leads to higher volatility contagion. It is quite the
contrary. Trade openness attenuates the effect of terms-of-trade volatility on domestic
volatility. More impressively, financial openness dampens the volatility inducing effect
coming from all external shocks, both financial- and trade-related. It appears that financial
openness (and to some extent trade openness) provides the means for the economy to
diversify away the external sources of risk.
All in all, the results presented in the paper cast a favorable light on financial and
trade openness as engines of growth and stability. They question the view that international
integration increases external vulnerability by hurting growth and increasing volatility or by
amplifying the adverse effect of external shocks. For future research, there remains to
understand the mechanisms through which development affects the link between openness
and growth and the channels by which integration prepares the economy to deal with
external instability.
22
References
Ahmed, S., and P. Loungani (1999). "Business Cycles in Emerging Market Economies",
Monetaria, XXII (4), October-December, CEMLA.
Alcala, F. and A. Ciccone (2004). "Trade and Productivity", Quarterly Journal of Economics: 119
(2), 613-646.
Alesina, A., E. Spolaore, and R. Wacziarg (2005). "Trade, Growth, and Size of Countries."
Forthcoming in: Aghion, P. and S. Durlauf Eds., Handbook of Economic Growth.
Amsterdam: North-Holland.
Arellano M. and S. Bond (1991). "Some Tests of Specification for Panel Data: Monte Carlo
Evidence and an Application to Employment Equations", Review of Economic Studies 58,
277-297.
Arellano M. and O. Bover (1995). "Another Look at the Instrumental-Variable Estimation of Error-
Components Models", Journal of Econometrics 68, 29-51.
Arora, V. and A. Vamvakidis (2004). "The Impact of U.S. Economic Growth on the Rest of the
World: How much does it matter?," Journal of Economic Integration 19, 1-18.
Arora, V. and A. Vamvakidis (2005). "How much do Trading Partners matter for Economic
Growth?", IMF Staff Papers 52, 24-40.
Arteta, C., B. Eichengreen, and C. Wyplosz (2001). "On the Growth Effects of Capital Account
Liberalization", manuscript, University of California, Berkeley, CA.
Barro, R.J., and X. Sala-i-Martin (2004). Economic Growth. McGraw-Hill, Second Edition.
Bekaert, G., H. Campbell, and C. Lundblad (2004). "Growth Volatility and Financial
Liberalization", NBER Working Paper No. 10560.
Blankenau, W., M.A. Kose, and K.-M. Yi (2001). "Can World Real Interest Rates explain Business
Cycles in a Small Open Economy?", Journal of Economic Dynamics and Control 25, 867-
889.
Beck T., A. Demirguc-Kunt, and R. Levine (2000). "A New Database on Financial Development
and Structure", World Bank Economic Review, 14 (3), pp.597-605.
Ben-David, D. (1993). "Equalizing Exchange: Trade Liberalization and Income Convergence,"
Quarterly Journal of Economics: 108 (3), 653-679.
Calderón, C., A. Chong and N. Loayza (2002). "Determinants of Current Account Deficits in
Developing Countries", Contributions to Macroeconomics, 2 (1), Article 2
(http://www.bepress.com/bejm/contributions/vol2/iss1/art2)
Calderón, C., and K. Schmidt-Hebbel (2003). "Macroeconomic Policies and Performance in Latin
America", Journal of International Money and Finance 22, 895-923.
Caprio, G. and D. Klingebiel (1999). "Episodes of Systemic and Borderline Financial Crises."
Washington: World Bank. Processed.
Cavallo, E. (2005). "Output Volatility and Openness to Trade: A Reassessment", manuscript,
Kennedy School of Government, Harvard University, Cambridge, MA.
Chanda, A. (2001). "The Influence of Capital Controls on Ling-Run Growth: Where and How
Much?", manuscript, Brown University, Providence, RI.
Correia, I., J.C. Neves, and S.T. Rebelo (1992). "Business Cycles 1850-1950: New Facts about Old
Data." European Economic Review 36, 459-467.
Correia, I., J.C. Neves, and S.T. Rebelo (1995). "Business Cycles in a Small Open Economy",
European Economic Review 39, 1089-1113.
Dollar, D. (1992). "Outward-Oriented Developing Economies Really Do Grow More Rapidly:
Evidence from 95 LDCs, 1976-85", Economic Development and Cultural Change, 523-44.
Dollar, D. and A. Kraay (2003). "Institutions, Trade, and Growth", Journal of Monetary Economics
50: 133-62.
23
Easterly, W. (2001). "Growth Implosions, Debt Explosions, and My Aunt Marilyn: Do Growth
Slowdowns Cause Public Debt Crises?" Policy research working paper 2531. Washington:
World Bank.
Easterly, W. and M. Sewadeh (2002). "Global Development Network Growth Database."
Washington: World Bank.
Easterly, W., R. Islam, and J. Stiglitz (2000). "Shaken and Stirred: Explaining Growth Volatility,",
in B. Pleskovic and N. Stern (eds.): Annual Bank Conference on Development Economics.
Washington, DC: The World Bank.
Easterly, W. N. Loayza, and P. Montiel (1997). "Has Latin America's Post-Reform Growth Been
Disappointing?," Journal of International Economics 43, 287-311.
Edison, H. J., M. Klein, L. Ricci, and T. Slok (2002): "Capital Account Liberalization and
Economic Performance: Survey and Synthesis", IMF Working Paper WP702/120, July.
Edison, H. J., R. Levine, L. Ricci, and T. Slok (2002): "International Financial Integration and
Economic Growth", Journal of International Money and Finance 21: 749-76.
Edwards, S. (1993). "Openness, Trade Liberalization, and Growth in Developing Countries",
Journal of Economic Literature 31:1358-93.
Edwards, S. (1998). "Openness, Productivity and Growth: What Do We Really Know?", Economic
Journal 108 (447): 383-98.
Edwards, S. (2001). "Capital Mobility and Economic Performance: Are Emerging Economies
Different?", NBER Working Paper 8076, Cambridge, MA.
Elbadawi, I. and K. Schmidt-Hebbel (1998). "Macroeconomic Policies, Instability, and Growth in
the World", Journal of African Economies 7 (2): 116-68.
Fatás, A. (2002). "The Effects of Business Cycles on Growth", in: Loayza, N. and R. Soto, eds.,
Economic Growth: Sources, Trends, and Cycles, Santiago, Chile: Central Bank of Chile.
Fernández-Arias, E. and P. Montiel (2001). "Reform and Growth: All Pain, No Gain?," IMF Staff
Papers 48, 522-546.
Frankel, J. and D. Romer (1999). "Does Trade Cause Growth?" American Economic Review 89 (3):
379-99.
Grilli, V. and G.M. Milesi-Ferretti (1995). "Economic Effects and Structural Determinants of
Capital Controls", IMF Staff Papers 42: 517-51.
Hnatkovska, V. and N. Loayza (2004). "Volatility and Growth", World Bank Working Paper 3184,
January.
Irwin, D.A. and M. Terviö (2002). "Does Trade raise Income? Evidence from the Twentieth
Century", Journal of International Economics 58: 1-18.
Kaminsky, G. and C. Reinhart (1998). "Financial Crisis in Asia and Latin America: Then and
Now." American Economic Review 88(2): 444-48.
Klein, M.W. (2003). "Capital Account Openness and the Varieties of Growth Experience", NBER
Working Paper 9500, February.
Klein, M. and G. Olivei (2000). "Capital Account Liberalization, Financial Depth, and Economic
Growth", manuscript, Tufts University, Medford, MA.
Kose, M.A. (2002). "Explaining Business Cycles in Small Open Economies: How much do World
Prices matter?", Journal of International Economics 56, 299-327.
Kose, M.A., E.S. Prasad, and M. Terrones (2003). "Financial Integration and Macroeconomic
Volatility", IMF Staff Papers 50 (Special Issue): 119-42.
Kose, M.A., E.S. Prasad, and M. Terrones (2005). "How do Trade and Financial Integration affect
the Relationship between Growth and Volatility?", IMF Working Paper WP/05/19, January.
Kraay, A. (1998). "In Search of the Macroeconomics Effects of Capital Account Liberalization",
manuscript, The World Bank, Washington, DC.
Lane, P. and G.M. Milesi-Ferreti (2001). "The external Wealth of Nations: Measures of Foreign
Assets and Liabilities for Industrial and Developing Countries," Journal of International
Economics 55, 263-294.
24
Lane, P. and G.M. Milesi-Ferreti (2003). "International Financial Integration," IMF Staff Papers 50
(Special Issue), 82-113.
Lee, H.Y., L.A. Ricci, and R. Rigobon (2004). "Once Again, is Openness good for Growth?",
Journal of Development Economics 75, 451-472.
Loayza, N., Fajnzylber, P., and C. Calderón (2004). "Economic Growth in Latin America: Stylized
Facts, Explanations and Forecasts", The World Bank Latin American and Caribbean
Studies Viewpoints, forthcoming.
Mendoza, E.G. (1991). "Real Business Cycles in a Small Open Economy", American Economic
Review 81, 797-818.
Mendoza, E.G. (1995). "The Terms of Trade, the Real Exchange Rate, and Economic Fluctuations".
International Economic Review 36, 101-137.
Montiel, P. and C. Reinhart (1999). "Do Capital Controls and Macroeconomic policies influence
the Volume and Composition of Capital Flows?", Journal of International Money and
Finance 18: 619-35.
Neumeyer, P.A. and F. Perri (2004). "Business Cycles in Emerging Economies: The Role of
Interest Rates", Journal of Monetary Economics 52, 345-380.
O'Donnell, B. (2001). "Financial Openness and Economic Performance", manuscript, Trinity
College, Dublin, Ireland.
Prasad, E., K. Rogoff, S.-J. Wei, and M.A. Kose (2003). "Effects of Financial Globalization on
Developing Countries: Some Empirical Evidence," IMF Occasional Paper 220, August.
Quinn, D. (1997). "The Correlates of Change in International Financial Regulation", American
Political Science Review 91 (3): 531-51.
Ramey, G. and V. Ramey, 1995, "Cross-Country Evidence on the Link Between Volatility and
Growth", American Economic Review 85(5), 1138-1151.
Razin, A. and Y. Rubinstein (2004): "Growth Effects of Exchange Rate Regimes and Capital
Account Liberalization in the Presence of Crises: A Nuanced View", NBER Working Paper
10555, June.
Rigobon, R. and D. Rodrik (2004). "Rule of Law, Democracy, Openness, and Income: Estimating
the Interrelationships", NBER Working Paper 10750, September.
Rodrik, D. (1998). "Who needs Capital Account Convertibility?", manuscript, Harvard University,
Cambridge, Mass., February.
Rodrik, D. (2001). "Why is there so much Economic Insecurity in Latin America?" CEPAL Review
73, 7-29.
Rodrik, D. and F. Rodríguez (2000). "Trade Policy and Economic Growth: A Skeptics Guide to the
Cross-National Evidence", en B. Bernanke y K. Rogoff (eds.): NBER Macroeconomics
Annual 15. Cambridge, MA: MIT Press.
Rodrik, D., A. Subramanian, and F. Trebbi (2004). "Institutions rule: The Primacy of Institutions
over Geography and Integration in Economic Development", Journal of Economic Growth
9, 131-165.
Sachs, J.D. and A.M. Warner (1995). "Economic Reform and the Process of Global Integration",
Brookings Papers on Economic Activity, 1-118.
Schmitt-Grohé, S. (1998). "The International Transmission of Economic Fluctuations", Journal of
International Economics 44, 257-287.
Summers, R. and A. Heston (1991). "The Penn World Table (Mark 5): an expanded set of
international comparisons, 1950-1988", Quarterly Journal of Economics, 106, 327-68.
Tytell, I. and S.-J. Wei (2004). "Does Financial Globalization induce better Macroeconomic
Policies?", IMF Working Paper WP/04/84, May
Wacziarg, R. (2001). "Measuring the Dynamic Gains from Trade", World Bank Economic Review
15 (3).
Wacziarg, R. and K.H. Welch (2003). "Trade Liberalization and Growth: New Evidence", NBER
Working Paper 10152, Cambridge, MA., December.
25
World Bank (2002). World Development Indicators, CD-ROM. Washington, DC: The World Bank.
World Bank (2003). World Development Indicators. Washington, D.C.: The World Bank.
26
Table 1
Literature Review on Trade Openness, Financial Openness, Foreign Shocks, and Growth 1/
Authors Main findings Measures Largest sample
I. Financial Openness (FO) and Growth
Grilli and Milesi-Ferretti (1995) [0]: No robust evidence Share measure, other measures Cross-section: 61 countries,
1971-1994
Quinn (1997) [+] effects Quinn restrictions Cross-section: 58 countries,
1958-1988
Kraay (1998) No effects. [+] effects of FO interactions with Share measures, Cross-section: 100 countries,
capital flows. Quinn restrictions 1985-1997
Rodrik (1998) No evidence Share measure Cross-section: 100 countries,
1975-1995
Klein and Olivei (2000) [+] effect of FO on financial depth (FD) in OECD Share measure Cross-section: 67 countries,
countries, but not in non-OECD countries. FD has 1976-1995
[+] effect on growth
Chanda (2001) Significant interaction of FO and ethnic Share measure Cross-section: 57 non-OECD countries
heterogeneity; FO lowers growth in ethnically
heterogeneous countries
Edwards (2001) [-] effect of FO and [+] effect of FO interaction Share measure (not significant) Cross- section: 62 countries,
with per capita GDP; FO lowers (raises) growth in Quinn restrictions (significant) 1980-1989
low- (high-) income countries.
Arteta et al. (2001) [-] effect of FO and [-] effect of FO interaction with Quinn restrictions Cross-section and pooled data:
black market premium. 59 countries, 1973-1992
FO lowers growth in countries with high black
market premiums
O'Donnell (2001) No robust results for FO and for FO interaction Share measure (not significant) Cross-section: 94 countries,
with financial depth Capital flows (significant) 1971-1994
Edison, Levine et al. (2002) Considering interaction with several variables, no IMF restrictions, Quinn restrictions, Panel: 57 countries, 25 years
robust findings LMF capital flows and stocks
Edison, Klein et al. (2002) [+] in world, OECD, East Asia; IMF restrictions (share); Cross-section: 89 countries,
[-] in non-OECD Quinn restrictions, others. 1976-1995
Klein (2003) Quadratic interaction with per capita GDP and IMF restrictions (share); Cross-section: 85 countries,
government quality, only [+] effect on middle- Quinn restrictions 1976-1995
income countries, [0] for others
Razin and Rubinstein (2004) Level of FO has no robust effects on growth but Weighted avge of several restrictions on Panel: 100 countries, 1970-1998
switching to capital controls reduces growth capital-account transactions, IMF
Kose et al. (2005) FO has no robust effect but FO turns the [-] effect LM and KPT capital flows Cross-section and panel:
of volatility growth into a [+] one 85 countries, 1960-2000
27
Table 1 (continued)
Literature Review on Trade Openness, Financial Openness, Foreign Shocks, and Growth 1/
Authors Main findings Measures Largest sample
II. Trade Openness (TO) and Growth
Dollar (1992) [+] effect of TO Measure based on international price Panel: 95 developing countries,
deviations. 1976-1985
Ben-David (1993) [+] effects on absolute convergence in income Binary measure of trade restrictions Cross section: 1970-1989
levels
Sachs and Warner (1995) TO leads to 2% higher growth than in closed Binary measure of trade restrictions Cross section: 1970-1989
economies
Edwards (1998) [+] effects of eight out of nine TO measures on 9 measures of trade restrictions and trade Panel: 93 countries, 1960-1990
TFP. shares
Frankel and Romer (1999) [+]: a 1% increase in trade volume raises level of Trade shares Cross section: 1985, 100 countries
per capita income by about 2%
Rodrik and Rodríguez (2000) TO effects not robust to inclusion of other Broad set of measures used in the Several samples according to different
variables, including geographic latitude. However, literature authors
lower bound of TO effects is [0]
Wacziarg (2001) [+]: 1 s.d. rise in trade volume to GDP increases Trade shares Panel: 57 countries, 1970-1989
growth by 1%
Irwin and Terviö (2002) Confirms results of Frankel and Romer (1999). Bilateral trade flows 8 annual cross sections: 146 countries,
But TO effect on income per capita income is not between 1913 and 1990
robust to inclusion of latitude
Dollar and Kraay (2003) Cross-section: TO is not robust to inclusion of Trade shares Cross-section and panel: 154
institutions Panel-data: TO effects are [+], more countries, 1970-2000
important than institutions
Wacziarg and Welch (2003) [0] in cross-section. Updates Sachs and Warner's Trade policy measure and trade shares Cross-section, panel, and country case
(1995) liberalization events; trade shares rise by studies: 116 countries, 1950-1998
5 pp, growth by 1.5-2% after liberalization,
compared to pre-liberalization period
Álcala and Ciccone (2004) [+] TO effects, robust to controlling for Trade shares (in PPP $) Cross section: 98 countries, 1985
institutional quality
Rodrik et al. (2004) TO effects are not robust to controlling for Trade shares Cross section: 80 countries, 1995
institutional quality
Rigobon and Rodrik (2004) [-] effect on income pc, after controlling for Trade shares Cross section: 86 countries, 1990s
geography and institutions
Kose et al. (2004) [+] effect. Changes sign of volatility on growth Binary trade policy measure. Cross-section and panel: 85 countries,
into [+] Trade shares: not robust 1960-2000
Lee, Ricci, and Rigobon (2004) TO has positive but small effects on growth 4 measure of openness: trade size Panel: 100 countries, 5-year periods
tariff indicator, import duties, and BMP from 1961-65 to 1996-2000
Alesina et al. (2005) [+] and large in small countries, [0] in large Trade shares (current prices, PPP $) Panel: 113 countries, 4-periods of 10-
countries yr avge 1960-2000
III. Foreign Shocks and Growth
Easterly et al. (1997) [+] and large significant effect on GDP growth Terms of trade changes Panel: LAC and World, 1986-1993
Fernández-Arias et al. (2002) [+] and large significant effect on GDP growth Terms of trade changes Panel: 69 countries, 1961-1995
Calderón and Schmidt-Hebbel (2003) [+] effect of private capital inflows on growth. Capital inflows and components: FDI, Panel: 56 developing countries,
Composition does matter: FDI is the only portfolio equity, debt and official flows 1970-2000
significant component
Barro and Sala-i-Martin (2004) [+] and large significant effect on GDP growth Terms of trade changes Cross-section: 97 countries,
1975-1985
Loayza et al. (2005) [+] and large significant effect on GDP growth Terms of trade changes Panel: LAC and World, 1960-2000
Arora and Vamvakidis (2005) [+] and significant effect on GDP growth Trade partners' growth per capita Panel: 101 countries, 19601999
[+] imples a positive and robust relationship found in the literature, [-] indicates a negative one, and [0] reflects the inability of finding a robust relationship
between the variables. 1/ Quinn restrictions based on Quinn (1967), Share measure is based on number of years of IMF restrictions in place,
LM is Lane and Milesi-Ferretti's (2002) measure of capital flows or stocks, KPT is Kose, Prasad and Terrones' (2004) measure of capital flows or stocks.
28
Table 2
Literature Review on Trade Openness, Financial Openness, Foreign Shocks, and Growth Volatility 1/
Authors Main findings Key Variables Largest sample
I. Growth and Growth Volatility
Ramey and Ramey (1995) Growth and growth volatility are [+]-ly Mean GDP growth and s.d. of GDP Cross-section: 92 countries,
correlated among industrial countries, and growth 1960-1985
[-]-ly correlated among developing
conntries
Elbadawi and Schmidt-Hebbel (1998) Growth and growth volatility are [+]-ly Mean GDP growth and s.d. of GDP Panel: 56 countries,
correlated among industrial countries, and growth s.d. of macro policy 1961-1994
[-]-ly correlated among developing fundamentals and crisis measures
conntries performance and macro volatility
lower growth.
Hnatkovska and Loayza (2004) Growth volatility lowers growth robustly. [-] s.d. of output gap and of growth per Panel: 79 countries,
effect larger in low-income economies, capita 1960-2000
and countries with institutional and policy
weaknesses
II. Financial Openness (FO), Trade Openness (TO), and Growth Volatility
Easterly et al. (2000) TO raises growth volatility, particularly in Growth volatility: s.d. of growth per Panel: 74 countries,
developing countries [0]: FO has no capita 19601997
impact on output volatility
O'Donnell (2001) [-] effect on output volatility of OECD Growth volatility: s.d. of growth per Panel: 93 countries,
countries [+] effect on volatility of non- capita 19711994
OECD countries
Kose et al. (2003) [0]: no effect of FO, TO on volatility [+]: Volatility: s.d. growth. TO, FO: two Panel: 55 countries,
Terms-of-trade volatility raises GDP (policy and outcome) measures 1960-1999
volatility
Bekaert et al. (2004) [-]: Capital accout liberalization lowers Growth volatility: s.d. of growth per Cross-section and time-series:
growth volatility in world sample. Weaker capita. FO: two measures, IMF and 95 countries,
results for emerging market economies Quinn restrictions 1980-2000
Cavallo (2005) [-]: TO reduces growth volatility Growth volatility: s.d. of growth per Cross section: 77 countries,
capita 1960-2000
III. Foreign Shocks and Growth Volatility
Mendoza (1995) Terms-of-trade disturbances explain 56% Terms of trade is the ratio of export to Model calibrated for G7
of output variation import unit values countries and developing
countries
Rodrik (2001) Capital flow volatility accounts for 1/2 of Private capital inflows Panel: LAC countries, 2-
cross-national variation in GNP volatility in periods of 10-yr avge 1980-
LAC 2000
Blankenau et al. (2002) World real interest rate shocks explain Model calibrated to Canadian data Model calibrated for
33% of output variation in Canada Canadian data, 1961-1996
Kose (2002) World interest rate shocks account for 1% World interest rate: LIBOR, deflated Model calibrated for a
of output volatility in developing countries by export unit value index developing, small open
economy
Neumeyer and Perri (2005) Large effects of country risk fluctuations 3-month ex-ante real interest rate at Model calibrated for
and small effects of world rate fluctuations which firms borrow in international Argentinean data,
on output volatility in emerging-market markets 1983-2001
economies
1/ See footnote in Table 1.
29
Table 3
Economic Growth and the Interaction between Openness and Real GDP Per Capita
Sample of 76 Countries, 1970-2000 (5-year period observations)
Dependent Variable: Growth in real GDP per capita
Estimation Method: GMM-IV System Estimator
Measures of Trade and Baseline Non-Linear Effects
Financial Openness: Regression Financial Trade
Constant 7.142 ** 2.105 23.419 **
(2.25) (2.57) (2.37)
Control Variables
Initial GDP per capita -0.177 ** -0.704 ** -2.883 **
(in logs) (0.09) (0.24) (0.27)
Education 1.058 ** 2.443 ** 2.062 **
(Secondary Enrollment, in logs) (0.16) (0.24) (0.15)
Financial Depth 0.631 ** 0.354 ** 0.402 **
(Private Domestic Credit to GDP, in logs) (0.10) (0.15) (0.13)
Lack of Price Stability -2.275 ** -1.434 ** -1.605 **
(inflation rate, in log[100+inf.rate]) (0.37) (0.43) (0.34)
Government Burden -1.488 ** -1.184 ** -1.460 **
(Government Consumption to GDP, in logs) (0.22) (0.24) (0.32)
Openness:
Trade Openness (TO) 0.403 ** 0.449 ** -8.214 **
(Real Exports and Imports to GDP, in logs) (0.13) (0.10) (0.77)
Financial Openness (FO) 0.051 ** -2.274 ** -0.050 **
(Stock Equity-related Foreign liabilities to GDP, logs) (0.01) (0.35) (0.01)
Openness * Inc ... 0.562 ** 1.832 **
(0.10) (0.19)
Openness * Inc squared ... -0.031 ** -0.089 **
(0.01) (0.01)
Foreign Shocks:
Terms of Trade Shocks 0.038 ** 0.041 ** 0.055 **
(Growth rate of Terms of Trade) (0.01) (0.01) (0.01)
Foreign Growth 1.536 ** 1.749 ** 1.666 **
(Growth rate of the country's trading partners) (0.17) (0.12) (0.11)
Regional Capital Inflows 0.098 ** 0.115 ** 0.115 **
(Private capital inflows to the country's region) (0.02) (0.03) (0.03)
Period Shifts
- 76-80 Period: -1.119 ** -1.359 ** -1.110 **
- 81-85 Period: -1.284 ** -1.627 ** -1.099 **
- 86-90 Period: -1.865 ** -2.322 ** -1.873 **
- 91-95 Period: -0.517 * -0.832 ** -0.260
- 96-00 Period: -1.843 ** -2.610 ** -1.609 **
Countries / Observations 76 / 438 76 / 438 76 / 438
Specification Tests (p-values)
- Sargan Test (0.41) (0.18) (0.45)
- 2nd. Order Correlation (0.90) (0.94) (0.79)
Numbers in parenthesis are robust standard errors. * (**) denotes statistical significance at the 10 (5) percent level.
30
Table 4
Economic Growth and the Interaction between Openness and Foreign Shocks
Sample of 76 Countries, 1970-2000 (5-year period observations)
Dependent Variable: Growth in real GDP per capita
Estimation Method: GMM-IV System Estimator (Arellano and Bover, 1995; Blundell and Bond, 1998)
[1] [2] [3] [4]
Terms of Trade World Int. Rate Regional Capital
Foreign Growth
Foreign Shock: Changes Changes Inflows
Constant 5.242 ** 12.005 ** 9.444 ** 10.804 **
(2.41) (2.94) (2.44) (3.06)
Control Variables
Initial GDP per capita -0.138 * -0.280 ** -0.176 * -0.152 *
(in logs) (0.09) (0.13) (0.11) (0.09)
Education 1.284 ** 1.419 ** 1.110 ** 0.977 **
(Secondary Enrollment, in logs) (0.19) (0.19) (0.17) (0.16)
Financial Depth 0.592 ** 0.669 ** 0.578 ** 0.628 **
(Private Domestic Credit to GDP, in logs) (0.07) (0.14) (0.11) (0.10)
Lack of Price Stability -1.786 ** -3.936 ** -2.400 ** -2.733 **
(inflation rate, in log[100+inf.rate]) (0.39) (0.33) (0.42) (0.49)
Government Burden -1.597 ** -1.523 ** -1.547 ** -1.384 **
(Government Consumption to GDP, in logs) (0.24) (0.28) (0.26) (0.23)
Openness:
Trade Openness (TO) 0.133 * 1.227 ** 0.404 ** -0.190
(Real Exports and Imports to GDP, in logs) (0.08) (0.46) (0.12) (0.15)
Financial Openness (FO) 0.080 ** -0.159 ** 0.071 ** 0.146 **
(Stock Equity-related Foreign liabilities to GDP, logs) (0.01) (0.04) (0.01) (0.02)
Foreign Shocks:
Terms of Trade Shocks 1.175 ** 0.033 ** 0.050 ** 0.039 **
(Growth rate of Terms of Trade) (0.12) (0.01) (0.01) (0.01)
Foreign Growth 1.703 ** 2.756 ** 1.499 ** 1.618 **
(Growth rate of the country's trading partners) (0.17) (0.75) (0.16) (0.19)
Regional Capital Inflows 0.025 0.057 ** 0.086 ** -0.374 **
(Private capital inflows to the country's region) (0.02) (0.01) (0.02) (0.12)
Interaction: Openness and Foreign Shock
TO * Foreign Shock -0.276 ** -0.361 * 0.397 ** 0.151 **
(0.03) (0.19) (0.11) (0.03)
FO * Foreign Shock 0.010 ** 0.067 ** 0.118 ** -0.043 **
(0.00) (0.02) (0.02) (0.01)
Period Shifts
- 76-80 Period: -1.239 ** -1.087 ** -5.122 ** -0.993 **
- 81-85 Period: -1.413 ** -1.290 ** -2.605 ** -1.099 **
- 86-90 Period: -2.495 ** -1.807 ** -3.443 ** -1.638 **
- 91-95 Period: -0.564 ** -0.545 * -1.359 ** -0.169
- 96-00 Period: -1.900 ** -1.911 ** -3.075 ** -1.604 **
Countries / Observations 76 / 438 76 / 438 76 / 438 76 / 438
Specification Tests (p-values)
- Sargan Test (0.22) (0.38) (0.37) (0.38)
- 2nd. Order Correlation (0.81) (0.59) (0.96) (0.67)
Numbers in parenthesis are robust standard errors. * (**) denotes statistical significance at the 10 (5) percent level.
31
Table 5
Economic Growth and the Interaction between Openness and Foreign Shocks: Controlling for
Macroeconomic Volatility
Sample of 76 Countries, 1970-2000 (5-year period observations)
Dependent Variable: Growth in real GDP per capita
Estimation Method: GMM-IV System Estimator (Arellano and Bover, 1995; Blundell and Bond, 1998)
[1] [2] [3] [4]
Terms of Trade World Int. Rate Regional Capital
Foreign Growth
Foreign Shock: Changes Changes Inflows
Constant 0.139 1.107 2.011 4.385 **
(1.85) (1.81) (2.08) (1.91)
Control Variables
Initial GDP per capita -0.284 * -0.226 * -0.342 * -0.351 **
(in logs) (0.16) (0.13) (0.18) (0.13)
Education 1.233 ** 0.861 ** 0.821 ** 0.836 **
(Secondary Enrollment, in logs) (0.20) (0.12) (0.19) (0.12)
Financial Depth 0.670 ** 0.826 ** 1.080 ** 0.972 **
(Private Domestic Credit to GDP, in logs) (0.18) (0.15) (0.18) (0.16)
Lack of Price Stability -0.390 -1.138 ** -0.522 * -0.999 **
(inflation rate, in log[100+inf.rate]) (0.32) (0.17) (0.35) (0.23)
Government Burden -1.622 ** -1.583 ** -1.660 ** -1.345 **
(Government Consumption to GDP, in logs) (0.26) (0.22) (0.30) (0.22)
Openness:
Trade Openness (TO) 0.573 ** 1.380 ** 0.616 ** 0.118
(Real Exports and Imports to GDP, in logs) (0.13) (0.33) (0.13) (0.15)
Financial Openness (FO) 0.029 ** -0.203 ** 0.034 ** 0.111 **
(Stock Equity-related Foreign liabilities to GDP, logs) (0.01) (0.03) (0.01) (0.01)
Foreign Shocks:
Terms of Trade Shocks 0.917 ** 0.038 ** 0.042 ** 0.036 **
(Growth rate of Terms of Trade) (0.15) (0.01) (0.01) (0.01)
Foreign Growth 1.457 ** 2.573 ** 1.457 ** 1.477 **
(Growth rate of the country's trading partners) (0.17) (0.50) (0.16) (0.16)
Regional Capital Inflows 0.029 0.049 ** 0.063 ** -0.233 *
(Private capital inflows to the country's region) (0.03) (0.02) (0.03) (0.14)
Interaction: Openness and Foreign Shock
TO * Foreign Shock -0.215 ** -0.346 ** 0.311 ** 0.105 **
(0.04) (0.12) (0.13) (0.04)
FO * Foreign Shock 0.004 ** 0.089 ** 0.101 ** -0.039 **
(0.00) (0.01) (0.02) (0.01)
MacroeconomicVolatility
Std. Dev. Growth -0.380 ** -0.401 ** -0.354 ** -0.395 **
(0.02) (0.02) (0.02) (0.02)
Period Shifts
- 76-80 Period: -1.324 ** -1.187 ** -4.458 ** -1.252 **
- 81-85 Period: -1.704 ** -1.491 ** -2.587 ** -1.486 **
- 86-90 Period: -2.624 ** -2.099 ** -3.512 ** -2.043 **
- 91-95 Period: -0.974 ** -0.629 * -1.306 ** -0.497
- 96-00 Period: -2.457 ** -2.173 ** -3.200 ** -2.093 **
Countries / Observations 76 / 438 76 / 438 76 / 438 76 / 438
Specification Tests (p-values)
- Sargan Test (0.22) (0.27) (0.24) (0.46)
- 2nd. Order Correlation (0.94) (0.62) (0.78) (0.74)
Numbers in parenthesis are robust standard errors. * (**) denotes statistical significance at the 10 (5) percent level.
32
Table 6
Growth Volatility, Trade Openness, Financial Openness and Foreign Shocks
Sample of 76 Countries, 1970-2000 (5-year period observations)
Dependent Variable: Standard Deviation of Growth in Real GDP per capita
Estimation Method: GMM-IV System Estimator (Arellano and Bover, 1995; Blundell and Bond, 1998)
Measures of Trade and Baseline Non-Linear Effects
Financial Openness: Regression Financial Trade
Constant -0.682 ** -1.254 ** 0.773 *
(0.30) (0.37) (0.44)
Control Variables
Inflation Volatility 0.006 ** 0.007 ** 0.005 **
(S.D. annual log differences of CPI) (0.00) (0.00) (0.00)
RER Overvaluation 0.004 ** 0.003 ** 0.003 **
(Proportional index, overvaluation if >100) (0.00) (0.00) (0.00)
Systemic Banking Crises 1.303 ** 1.341 ** 0.973 **
(Frequency of years under crises: 0-1) (0.13) (0.14) (0.14)
Openness:
Trade Openness (TO) 0.543 ** 0.664 ** -0.658 *
(Real Exports and Imports to GDP, in logs) (0.07) (0.08) (0.39)
Financial Openness (FO) -0.088 ** -4.206 ** -0.097 **
(Stock Equity-related Foreign liabilities to GDP, logs) (0.01) (0.32) (0.01)
Openness * Inc ... 0.977 ** 0.277 **
(0.09) (0.11)
Openness * Inc squared ... -0.057 ** -0.020 **
(0.01) (0.01)
Volatility of Foreign Shocks
Volatility of Terms of Trade Changes 0.075 ** 0.075 ** 0.072 **
(S.D. annual log differences of ToT) (0.00) (0.00) (0.00)
Volatility of Foreign Growth Volatility 0.246 ** 0.312 ** 0.109
(S.D. annual log differences of Foreign Growth) (0.07) (0.07) (0.12)
Volatility of Regional Capital Inflows 0.350 ** 0.332 ** 0.322 **
(S.D. ratio of Regional Capital Flows to GDP) (0.04) (0.04) (0.04)
Period Shifts
- 81-85 Period: -0.286 -0.093 -0.081
- 86-90 Period: -0.533 -0.339 ** -0.436 **
- 91-95 Period: -0.223 -0.053 -0.096
- 96-00 Period: -1.101 ** -0.869 ** -0.977 **
Countries / Observations 76 / 371 76 / 371 76 / 371
Specification Tests (p-values)
- Sargan Test (0.21) (0.728) (0.518)
- 2nd. Order Correlation (0.92) (0.855) (0.990)
Numbers in parenthesis are robust standard errors. * (**) denotes statistical significance at the 10 (5) percent level.
33
Table 7
Growth Volatility and the Interaction between Openness and the Volatility of Foreign Shocks
Sample of 76 Countries, 1970-2000 (5-year period observations)
Dependent Variable: Standard Deviation of Growth in Real GDP per capita
Estimation Method: GMM-IV System Estimator (Arellano and Bover, 1995; Blundell and Bond, 1998)
[1] [2] [3] [4]
Terms of Trade World Int. Rate Regional Capital
Foreign Growth
Foreign Shock: Changes Changes Inflows
Constant -2.005 ** 1.995 ** -0.142 -1.917 **
(0.32) (0.67) (0.41) (0.44)
Control Variables
Inflation Volatility 0.007 ** 0.006 ** 0.004 ** 0.008 **
(S.D. annual log differences of CPI) (0.00) (0.00) (0.00) (0.00)
RER Overvaluation 0.002 ** 0.003 ** 0.004 ** 0.002 **
(Proportional index, overvaluation if >100) (0.00) (0.00) (0.00) (0.00)
Systemic Banking Crises 1.346 ** 1.360 ** 1.228 ** 1.391 **
(Frequency of years under crises: 0-1) (0.10) (0.12) (0.14) (0.10)
Openness:
Trade Openness (TO) 0.941 ** -0.266 * 0.438 ** 0.919 **
(Real Exports and Imports to GDP, in logs) (0.08) (0.16) (0.13) (0.11)
Financial Openness (FO) -0.092 ** -0.018 -0.018 -0.138 **
(Stock Equity-related Foreign liabilities to GDP, logs) (0.01) (0.02) (0.01) (0.01)
Volatility of Foreign Shocks
Volatility of Terms of Trade Changes 0.079 ** 0.077 ** 0.074 ** 0.072 **
(S.D. annual log differences of ToT) (0.00) (0.00) (0.00) (0.00)
Volatility of Foreign Growth Volatility 0.237 ** -2.147 ** 0.432 ** 0.315 **
(S.D. annual log differences of Foreign Growth) (0.07) (0.50) (0.10) (0.07)
Volatility of Regional Capital Inflows 0.325 ** 0.413 ** 0.331 ** 0.366 *
(S.D. ratio of Regional Capital Flows to GDP) (0.03) (0.05) (0.04) (0.23)
Interaction: Openness and Vol. Foreign Shock
TO * Vol(Foreign Shock) -0.010 ** 0.727 ** -0.128 0.005
(0.00) (0.12) (0.15) (0.06)
FO * Vol(Foreign Shock) -0.006 ** -0.146 ** -0.152 ** -0.019 **
(0.00) (0.02) (0.01) (0.00)
Period Shifts
- 81-85 Period: -0.361 ** -0.181 ** 0.217 -0.315 **
- 86-90 Period: -0.587 ** -0.461 ** -0.372 ** -0.523 **
- 91-95 Period: -0.366 ** -0.187 ** 0.162 -0.221 **
- 96-00 Period: -1.289 ** -0.904 ** -1.250 ** -1.111 **
Countries / Observations 76 / 371 76 / 371 76 / 371 76 / 371
Specification Tests (p-values)
- Sargan Test (0.22) (0.37) (0.48) (0.19)
- 2nd. Order Correlation (0.85) (0.86) (0.84) (0.88)
Numbers in parenthesis are robust standard errors. * (**) denotes statistical significance at the 10 (5) percent level.
34
Figure 1
Trends in Openness, 1970-2000
75% 50%
45%
70%
40%
65%
35% )P
P) D
D G
G
%( 30% (%
60%
essn essn
25%
enp enp
55% O
O 20% al
e ci
adrT 15%
50% anniF
10%
45%
5%
40% 0%
1970 1975 1980 1985 1990 1995 2000
Trade Openness Financial Openness
Note: Openness measures are defined as the ratio of real exports and imports to GDP (trade) and equity-based foreign liabilities fo GDP (financial).
World medians are calculated from data for 76 countries.
35
Figure 2
Growth Effect of Openness as a function to GDP per capita 1/
(a) Growth Effect of Financial Openness as a function of GDP per capita
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
6.2 6.5 6.8 7.1 7.3 7.6 7.9 8.1 8.4 8.7 9.0 9.2 9.5 9.8 10.0
Real GDP per capita (in logs)
(b) Growth Effect of Trade Openness as a function of GDP per capita
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
6.2 6.5 6.8 7.1 7.3 7.6 7.9 8.1 8.4 8.7 9.0 9.2 9.5 9.8 10.0
-0.1
-0.2
Real GDP per capita (in logs)
1/ Growth effect of a one standard deviation increase in financial and trade openness. See Table A.3 for the computed standard deviations.
36
Figure 3
Growth Effect of External Shocks as a Function of Openness 1/
(a) Growth Effect of Terms of Trade Changes
As a function of Trade Openness As a function of Financial Openness
4.0 0.7
3.0 0.6
0.5
2.0
0.4
1.0
0.3
0.0
2.0 2.3 2.5 2.8 3.0 3.3 3.6 3.8 4.1 4.3 4.6 4.8 5.1 5.3 5.6
0.2
-1.0
0.1
-2.0
0.0
-3.0 -7.1 -6.2 -5.4 -4.5 -3.7 -2.8 -2.0 -1.1 -0.2 0.6 1.5 2.3 3.2 4.0 4.9
Volume of Trade to Real GDP (in logs) Stock of Foreign Liabilities to Real GDP (in logs)
(b) Growth Effect of Foreign Growth
As a function of Trade Openness As a function of Financial Openness
1.4 1.2
1.2 1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.0 0.0
2.0 2.3 2.5 2.8 3.0 3.3 3.6 3.8 4.1 4.3 4.6 4.8 5.1 5.3 5.6 -7.1 -6.2 -5.4 -4.5 -3.7 -2.8 -2.0 -1.1 -0.2 0.6 1.5 2.3 3.2 4.0 4.9
Volume of Trade to Real GDP (in logs) Stock of Foreign Liabilities to Real GDP (in logs)
(c) Growth Effect of World Interest Rate Changes
As a function of Trade Openness As a function of Financial Openness
-4.4 -4.6
2.0 2.3 2.5 2.8 3.0 3.3 3.6 3.8 4.1 4.3 4.6 4.8 5.1 5.3 5.6 -7.1 -6.2 -5.4 -4.5 -3.7 -2.8 -2.0 -1.1 -0.2 0.6 1.5 2.3 3.2 4.0 4.9
-4.6 -4.8
-4.8 -5.0
-5.0 -5.2
-5.2 -5.4
-5.4 -5.6
-5.6 -5.8
Volume of Trade to Real GDP (in logs) Stock of Foreign Liabilities to Real GDP (in logs)
(d) Growth Effect of Regional Capital Flows
As a function of Trade Openness As a function of Financial Openness
1.2 1.4
1.0 1.2
0.8 1.0
0.6 0.8
0.4 0.6
0.2 0.4
0.0 0.2
2.0 2.3 2.5 2.8 3.0 3.3 3.6 3.8 4.1 4.3 4.6 4.8 5.1 5.3 5.6
-0.2 0.0
-7.1 -6.2 -5.4 -4.5 -3.7 -2.8 -2.0 -1.1 -0.2 0.6 1.5 2.3 3.2 4.0 4.9
-0.4 -0.2
Volume of Trade to Real GDP (in logs) Stock of Foreign Liabilities to Real GDP (in logs)
1/ Growth effect of a one standard deviation increase in the corresponding external shock. See Table A.3 for the computed standard deviations of these shocks.
37
Figure 4
Volatility Effect of Openness as a function to GDP per capita 1/
(a) Volatility Effect of Financial Openness as a function of GDP per capita
0.0
6.2 6.5 6.8 7.1 7.3 7.6 7.9 8.1 8.4 8.7 9.0 9.2 9.5 9.8 10.0
-0.2
-0.4
-0.6
-0.8
-1.0
-1.2
-1.4
Real GDP per capita (in logs)
(b) Volatility Effect of Trade Openness as a function of GDP per capita
0.18
0.16
0.14
0.12
0.10
0.08
0.06
0.04
0.02
0.00
6.2 6.5 6.8 7.1 7.3 7.6 7.9 8.1 8.4 8.7 9.0 9.2 9.5 9.8 10.0
Real GDP per capita (in logs)
1/ Volatility effect of a one standard deviation increase in financial and trade openness. See Table A.4 for the computed standard deviations.
38
Figure 5
Volatility Effect of External Shocks as a Function of Openness 1/
(a) Volatility Effect of Volatility in Terms of Trade Changes
As a function of Trade Openness As a function of Financial Openness
0.45 0.8
0.40 0.7
0.35 0.6
0.30
0.5
0.25
0.4
0.20
0.3
0.15
0.2
0.10
0.05 0.1
0.00 0.0
2.0 2.3 2.5 2.8 3.0 3.3 3.6 3.8 4.1 4.3 4.6 4.8 5.1 5.3 5.6 -7.1 -6.2 -5.4 -4.5 -3.7 -2.8 -2.0 -1.1 -0.2 0.6 1.5 2.3 3.2 4.0 4.9
Volume of Trade to Real GDP (in logs) Stock of Foreign Liabilities to Real GDP (in logs)
(b) Volatility Effect of Volatility of Foreign Growth
As a function of Trade Openness As a function of Financial Openness
1.0 0.9
0.8
0.8
0.7
0.6
0.6
0.4 0.5
0.2 0.4
0.3
0.0
2.0 2.3 2.5 2.8 3.0 3.3 3.6 3.8 4.1 4.3 4.6 4.8 5.1 5.3 5.6 0.2
-0.2
0.1
-0.4 0.0
-7.1 -6.2 -5.4 -4.5 -3.7 -2.8 -2.0 -1.1 -0.2 0.6 1.5 2.3 3.2 4.0 4.9
-0.6 -0.1
Stock of Foreign Liabilities to Real GDP (in logs)
(c) Volatility Effect of Volatility of World Interest Rate Changes
As a function of Trade Openness As a function of Financial Openness
0.3 1.2
0.2 1.0
0.2 0.8
0.1
0.6
0.1
0.4
0.0
0.2
2.0 2.3 2.5 2.8 3.0 3.3 3.6 3.8 4.1 4.3 4.6 4.8 5.1 5.3 5.6
-0.1
0.0
-0.1 -7.1 -6.2 -5.4 -4.5 -3.7 -2.8 -2.0 -1.1 -0.2 0.6 1.5 2.3 3.2 4.0 4.9
-0.2
-0.2
-0.4
-0.2
-0.6
Stock of Foreign Liabilities to Real GDP (in logs)
(d) Volatility Effect of Volatility of Regional Capital Flows
As a function of Trade Openness As a function of Financial Openness
0.355 0.6
0.350 0.5
0.345
0.4
0.340
0.3
0.335
0.2
0.330
0.325 0.1
0.320 0.0
2.0 2.3 2.5 2.8 3.0 3.3 3.6 3.8 4.1 4.3 4.6 4.8 5.1 5.3 5.6 -7.1 -6.2 -5.4 -4.5 -3.7 -2.8 -2.0 -1.1 -0.2 0.6 1.5 2.3 3.2 4.0 4.9
Volume of Trade to Real GDP (in logs) Stock of Foreign Liabilities to Real GDP (in logs)
1/ Volatility effect of a one standard deviation increase in the volatility of the external shock. See Table A.4 for the computed standard deviations of these shocks.
39
Appendix 1: Sample of countries
I. Industrial Economies (22 countries)
Australia Germany Norway
Austria Greece Portugal
Belgium Iceland Spain
Canada Ireland Sweden
Denmark Italy Switzerland
Finland Japan United Kingdom
France Netherlands United States
New Zealand
II. Latin America and the Caribbean (21 countries)
Argentina Ecuador Nicaragua
Bolivia El Salvador Panama
Brazil Guatemala Paraguay
Chile Haiti Peru
Colombia Honduras Trinidad and Tobago
Costa Rica Jamaica Uruguay
Dominican Republic Mexico Venezuela, RB
III. East Asia and the Pacific (8 countries)
China Malaysia Singapore
Indonesia Papua New Guinea Thailand
Korea, Rep. Philippines
IV. Middle East and North Africa (7 countries)
Algeria Israel Tunisia
Egypt, Arab Rep. Jordan Turkey
Morocco
V. South Asia (3 countries)
India Pakistan Sri Lanka
VI. Sub-Saharan Africa (15 countries)
Botswana Madagascar Sierra Leone
Cote d'Ivoire Malawi South Africa
Gambia, The Niger Togo
Ghana Nigeria Zambia
Kenya Senegal Zimbabwe
40
Appendix 2: Definitions and Sources of Variables Used in Regression Analysis
Variable Definition and Construction Source
GDP per capita Ratio of total GDP to total population. GDP is in 1985 PPP- Authors' construction using Summers and Heston (1991)
adjusted US$. and The World Bank (2003).
GDP per capita growth Log difference of real GDP per capita. Authors' construction using Summers and Heston (1991)
and The World Bank (2003).
Initial GDP per capita Initial value of ratio of total GDP to total population. GDP is Authors' construction using Summers and Heston (1991)
in 1985 PPP-adjusted US$. and The World Bank (2003).
Education Ratio of total secondary enrollment, regardless of age, to the Easterly and Sewadeh (2002) and The World Bank
population of the age group that officially corresponds to that (2003).
level of education.
Financial Depth Ratio of domestic credit claims on private sector to GDP Author's calculations using data from IFS, the
publications of the Central Bank and PWD. The method
of calculations is based on Beck, Demiguc-Kunt and
Levine (1999).
Trade Openness Log of the ratio of exports and imports (in 1995 US$) to Easterly and Sewadeh (2002) and The World Bank
GDP (in 1995 US$). (2003).
Financial Openness Log of the Stock of Equity-based Foreign Liabilities to GDP Lane and Milesi-Ferreti (2001, 2003), IMF's Balance of
(both expressed in 1995 US$). Following Eichengreen and Payments Statistics
Irwin (1998), we add the value of 1 to the stock in order to
include the cases where the stock of foreign liabilities is 0.
Government Burden Log of the Ratio of government consumption to GDP. The World Bank (2003).
CPI Consumer price index (1995 = 100) at the end of the year Author's calculations with data from IFS.
Inflation rate Log differences of CPI Author's calculations with data from IFS.
Real Exchange Rate Real Effective Exchange Rate, with the level adjusted such Easterly (2001)
Overvaluation that the average for 1976-85 equals Dollar's (1992) index of
overvaluation (based on the ratio of actual to income-
adjusted Summers-Heston purchasing power parity
comparisons).
Terms of Trade Net barter terms of trade index (1995=100) The World Bank (2003).
Terms of Trade Changes Log differences of the terms of trade index Authors' construction using The World Bank (2003).
Foreign Growth Growth of main trading partners calculated as the trade- Authors' construction using Summers and Heston
weighted growth for the main trading partners of the (1991), The World Bank (2003), and the IMF's
corresponding country. Direction of Trade Statistics.
World Nominal Interest G-3 (U.S., Germany and Japan) Money Market Rate (period Author's calculations with data from IFS.
Rate average)
World Inflation G-3 (U.S., Germany and Japan) Consumer Price Index (CPI) Author's calculations with data from IFS.
Inflation rate
World Real Interest Rate World Nominal Interest Rate adjusted by World Inflation. Author's calculations with data from IFS.
Regional Capital Inflows (Gross) Capital Inflows (FDI, portofolio-equity, loans) to the Author's calculations with data from the IMF's Balance
region of the corresponding country, as a percentage to the of Payments Statistics.
corresponding GDP.
Inflation Volatility Standard deviation of the annual log differences of CPI Authors' construction using The World Bank (2003).
Systemic Banking Crises Number of years in which a country underwent systemic Author's calculations using data from Caprio and
banking crisis, as a fraction of years in the corresponding Klingebiel (1999), and Kaminsky and Reinhart (1998).
period.
Volatility of Terms of Standard deviation of the annual log differences of the terms Authors' construction using The World Bank (2003).
Trade Changes of trade.
Volatility of Foreign Standard deviation of the trade-weighted annual growth of Authors' construction using Summers and Heston
Growth the main trading partners of the corresponding country. (1991), The World Bank (2003), and the IMF's
Direction of Trade Statistics.
Volatility of World Real Standard deviation of the world real interest rate annual Author's calculations with data from the IMF's Balance
Interest Rates changes. of Payments Statistics.
Volatility of Regional Standard deviation of the capital inflows to region of the Author's calculations with data from the IMF's Balance
Capital Inflows corresponding country relative to its regional GDP. of Payments Statistics.
Period-specific Shifts Time dummy variables. Authors' construction.
41
Appendix 3: Descriptive Statistics for Growth Regressions
Data in 5-year period averages, 76 countries, 438 observations
(a) Univariate
Variable Mean Std. Dev. Minimum Maximum
Growth rate of GDP per capita 1.422 2.642 -7.944 10.128
Initial GDP per capita (in logs) 8.522 0.989 6.243 10.240
Secondary enrollment (in logs) 3.739 0.788 0.113 4.923
Private domestic credit/GDP (in logs) 3.485 0.844 0.568 5.435
Inflation (in log [100+inf. rate]) 4.743 0.175 4.585 6.135
Government consumption /GDP (in logs) 2.680 0.371 1.475 3.637
Trade Openness 3.948 0.594 2.024 5.787
Financial Openness 1.689 3.779 -21.044 5.536
Terms of Trade Changes -0.424 4.644 -18.859 21.415
Foreign Growth 2.244 0.606 0.834 3.833
World Int. Rate Changes -0.017 0.658 -0.975 1.505
Regional Capital Flows / GDP 3.419 2.359 -1.635 10.336
(b) Bivariate Correlations between Growth and Determinants
Private domestic
Secondary Inflation (in Government Terms of World Int. Regional
Growth rate of Initial GDP per credit/GDP (in Trade Financial Foreign
enrollment (in log [100+inf. consumption /GDP Trade Rate Capital
GDP per capita capita (in logs) logs) Openness Openness Growth
logs) rate]) (in logs) Changes Changes Inflows/GDP
Variable
Growth rate of GDP per capita 1.00
Initial GDP per capita (in logs) 0.19 1.00
Secondary enrollment (in logs) 0.22 0.80 1.00
Private domestic credit/GDP (in logs) 0.25 0.71 0.61 1.00
Inflation (in log [100+inf. rate]) -0.29 -0.10 0.00 -0.35 1.00
Government consumption /GDP (in logs) -0.03 0.36 0.29 0.35 -0.10 1.00
Trade Openness -0.04 -0.14 -0.13 0.01 -0.30 0.27 1.00
Financial Openness 0.15 0.34 0.44 0.28 0.04 0.03 -0.11 1.00
Terms of Trade Changes 0.10 0.07 0.03 0.04 -0.12 -0.04 0.08 0.06 1.00
Foreign Growth 0.28 -0.14 -0.24 -0.12 -0.17 -0.15 -0.12 -0.23 0.08 1.00
World Int. Rate Changes 0.04 -0.01 -0.02 0.00 0.03 0.05 -0.01 0.03 0.20 0.29 1.00
Regional Capital Inflows / GDP 0.19 0.44 0.34 0.41 -0.29 0.18 0.12 0.17 0.11 -0.14 0.06 1.00
42
Appendix 4: Descriptive Statistics for Growth Volatility Regressions
Data in 5-year period averages, 76 countries, 371 observations
(a) Univariate
Variable Mean Std. Dev. Minimum Maximum
Growth Volatility 2.887 2.190 0.314 16.053
Inflation Volatility 8.681 19.193 0.198 168.127
RER Overvaluation 107.417 44.123 47.192 555.027
Systemic Banking Crises 0.138 0.286 0.000 1.000
Trade Openness 3.968 0.569 2.249 5.781
Financial Openness 2.085 3.027 -21.044 5.536
Volatility of Terms of Trade Changes 8.476 8.628 0.000 56.323
Volatility of Foreign Growth 1.000 0.434 0.214 2.438
Volatility of World Int. Rate Changes 1.056 0.737 0.303 2.849
Volatility of Regional Capital Inflows/GDP 1.424 0.969 0.139 4.444
(b) Bivariate Correlations between Growth Volatility and Determinants
Volatility of Volatility of
RER Systemic Financial Volatility of Volatility of Regional
Growth Volatility Inflation Volatility Trade Openness Terms of Trade World Int. Rate
Overvaluation Banking Crises Openness Foreign Growth Capital Inflows/GDP
Changes Changes
Variable
Growth Volatility 1.00
Inflation Volatility 0.24 1.00
RER Overvaluation 0.08 0.13 1.00
Systemic Banking Crises 0.11 0.24 0.04 1.00
Trade Openness 0.05 -0.24 0.09 -0.09 1.00
Financial Openness -0.30 -0.03 -0.09 0.07 -0.05 1.00
Volatility of Terms of Trade Changes 0.36 0.27 0.16 0.03 0.01 -0.30 1.00
Volatility of Foreign Growth 0.18 0.02 -0.05 -0.14 -0.15 -0.18 0.12 1.00
Volatility of World Int. Rate Changes 0.15 0.02 -0.05 -0.16 -0.13 -0.24 0.15 0.87 1.00
Volatility of Regional Capital Inflows/GDP 0.12 0.16 0.10 0.19 0.04 0.10 -0.08 -0.07 -0.23 1.00
43