IS TRADE POLICY OPENNESS GOOD FOR

GROWTH?

CHARLES ACKAH
CREDIT, School of Economics, University of Nottingham, UK

February 2006

ABSTRACT

Very few issues are more contentious today than the effects of trade policy on the rate
of economic growth and poverty in developing countries. The paper investigates the
relationship between trade barriers and growth using a dynamic panel regression
model for data on 48 developing countries over 1980-1999. Trade barriers are
captured by measures of tariffs, import and export taxes. Particular attention is paid to
simultaneity, country-specific effects and the potential contingency of this relation on
income. Our preferred specification for growth includes as an explanatory variable an
interaction term between trade barriers and initial income levels. The interaction term
is meant to capture the non-linearity in the relationship between trade barriers and
economic growth. This specification reveals a significant interaction effect under
which the marginal impact of tariffs on growth is rising in initial income. In
particular, the relationship between tariffs and growth is negative and significant
across all alternative policy measures, but is not uniform across income groups. The
richer the country, the smaller are the growth-reducing effects of trade protection and
the poorer the country the more likelihood that trade protection will affect growth
negatively. This finding is particularly interesting for Sub-Saharan Africa (the world’s
poorest continent) where trade restrictions are still pervasive and where poverty is
widespread.

Author
Charles Ackah is research student in the School of Economics. I am grateful to my
supervisors, Professor Oliver Morrissey and Dr Simon Appleton for guidance and
comments. Contact email: lexcga@nottingham.ac.uk.
1 INTRODUCTION

For many years, economists and policy makers have discussed the impact of trade
barriers on economic performance. Economic theory generally supports the
conclusion that trade liberalization has a positive effect on economic growth, although
there are scenarios in which liberalization might slow economic growth. Empirically,
some studies have identified a positive linkage between a country’s rate of economic
growth and its openness to international trade, while others have failed to demonstrate
this linkage. A general criticism of many of the empirical studies is that they use
measures of trade or indices of openness, rather than using measures of trade policy.
We address this latter concern by using three alternative policy measures: average
unweighted scheduled tariffs, import taxes (as a percentage of import, a measure of
average implicit tariff) and export taxes (as a percentage of exports). These indicators
have limitations as measures of trade policy (see Milner and Morrissey, 1999; Rodrik,
1999), but should capture the broad pattern of trade policy across countries and over
time.

This study investigates the impact of trade policy on economic growth in developing
countries during the period 1980-1999, based on a dynamic panel regression model.
The focus of this study is an empirical question: Does trade policy openness cause
economies which liberalize to grow more rapidly than those which do not? Is the
relationship between trade policy and growth dissimilar for poor and rich countries?
What can we say specifically about Africa?

Even though the focus of this study is on Africa, the strategy is to begin by placing
Africa’s trade policy in a global context where we introduce the ‘traditional’ Africa
dummy in a developing country sample. Then we take a detailed look at trade policy
and growth in an ‘Africa only’ sample. In this study, we address the extremely
important issue of endogeneity by employing the GMM systems estimator. We find
that trade policy has a large and significant effect on economic performance in
developing countries, but this effect is non-universal across countries in different
income groups.

2
The remainder of this paper is organized as follows. The next section provides a brief
review of the theoretical and empirical literature on trade and growth. In Section 3 we
describe a standard growth equation and present some preliminary statistics from the
data. Section 3 further describes the econometric methodology we employ for our
estimations. In Section 4, we present evidence from a dynamic panel data model
based on the Generalised Methods of Moments (GMM) estimator of trade barriers and
growth, and discuss the estimation results for various measures of trade restrictions. In
Section 5, we carry out some sensitivity analysis to test for the robustness of our
results. Section 6 presents concluding remarks.

2 LITERATURE REVIEW
There is a large literature on the empirical and theoretical linkages between the degree
of an economy’s openness to international trade and its growth rate. While theory is
ambiguous regarding the relationship between greater openness to trade and economic
growth (Riveria-Batiz and Romer (1991), Grossman and Helpman (1991), Parente
and Prescott (2002)), most empirical work (Levine and Renelt (1992), Sachs and
Warner (1995), Harrison (1996), Sala-i-Martin (1997), Harrison and Hanson (1999),
and Wacziarg and Welch (2003)) has generally found a positive relationship, although
a fragile one, as pointed out by Rodriguez and Rodrik (2001).

2.1 Theoretical Arguments

Economic theory offers many reasons to expect trade liberalisation to stimulate
economic growth and poverty reduction. There is a very strong claim that trade
liberalisation or openness to trade leads to faster economic growth and that the poor
share (to some extent) in the benefits of growth (Jenkins, 2004). The theoretical
models of endogenous growth suggest positive association between openness and
growth through several channels, e.g. embodied technology, availability of inputs,
technical assistance and learning, and reduced networking costs, e.g. see Bhagwati
and Srinivasan (2001), Grossman and Helpman (1991) and Lucas (1988). However,
there are also endogenous growth models that suggest that trade may be growth-
stunting (Grossman and Helpman, 1991; Srinivasan, 2001). It is possible to develop
endogenous growth models in which protection of the domestic market promotes

3
growth.1 Harrison (1996), for example, has pointed out that the endogenous growth
theorists do not predict that free trade will unambiguously raise economic growth;
increased competition could, for example, discourage innovation by lowering
expected profits. Thus, ‘there should be no theoretical presumption in favour of
finding an unambiguous negative relationship between trade barriers and growth rates
in the types of cross-national data typically analyzed’ (Rodrik and Rodriguez, 2001:
18). Ultimately, the issue of the impact of trade policy on economic growth remains a
matter of empirical testing.

2.2 Empirical Literature

In spite of the numerous critiques of cross-country studies they still have become a
major research tool over the last decade for understanding the links between trade
policy and economic growth. Most of the empirical literature using cross-country data
has found international trade in goods to be growth inducing (see Baldwin, 2003), e.g.
Dollar (1992), Edwards (1993, 1998), Sachs and Warner (1995), Coe et al. (1997),
Dollar and Kraay (2001b) and Mbabazi et al (2002). These studies, specifically or the
general empirical approach, have come under severe criticism from free-trade skeptics
(Rodriguez and Rodrik, 2001) and pro-free-trade economists (Srinivasan and
Bhagwati, 2001). A number of methodological and econometric problems have been
identified that could account for the lack of robustness of these studies. In what
follows, we consider the main limitations of these cross-country studies and note
some of the main arguments advanced by their critics.

2.2.1 Causation and Simultaneity Bias

First, while the studies may have unearthed a positive association between trade and
growth, most are unable to conclude anything about causality per se. It has also been
pointed out that in the case of the relationship between openness and growth, the
direction of causation is by no means straightforward. As Rodrik (1999) argues, it
may well be the case that faster growing economies become more open rather than
economies that become more open grow faster. Rodrik (2000b) holds the view that
both are caused by the quality of institutions. Harrison (1996) concludes that previous

1
See (Romer, 1990; Grossman and Helpman, 1990; Rivera-Batiz and Romer, 1991a,b; Matsuyama,
1992; Young, 1991 and Lucas, 1988).

4
studies on the direction of causality between openness and growth have generated
mixed results, with causality being bi-directional. Rodríguez and Rodrik (2001)
believe that there has been a tendency to overstate the pure tariff effect on growth.

2.2.2 Definition and Measurement Issues

Research must confront the fact that it is very difficult to obtain reliable direct
measures of trade policy openness across countries over time. Measuring the extent of
trade openness is a major challenge for any study involving the analysis of trade
policy (Winters 2003, 2004; Rodrik and Rodriguez 2001; Pritchett 1996; Edwards
1993, 1998; Greenaway et al, 1998; Milner and Morrissey, 1999; Rodrik, 1992, 1998,
1999). Several approaches have been employed to circumvent the problem, especially
the use of indices of trade orientation that are constructed using quantitative and
qualitative judgements, e.g. Dollar (1992), Sachs and Warner (1995), Harrison (1996),
Edwards (1998), and Frankel and Romer (1999). Some studies confuse trade outcome
measures (trade volume or its components) with policy indicators, e.g. those that
interpret the trade volume measure of openness, (X+M)/GDP, as a policy indicator.
Rodriguez and Rodrik (2001) argue that the indicators of openness used by
researchers have crucial shortcomings in measuring the trade orientation of countries
and are therefore problematic as measures of trade policy. Given the problems of
measuring openness we use more direct measures of trade policies - average tariffs,
tax revenue from imports and tax revenue from exports.

Only a handful of studies have looked at the relationship between actual trade barrier
indicators (tariff rates, non-tariff barriers and tariff revenues) and growth in the last
several decades and reported mixed empirical results. Edwards (1992) tested for a
relationship between average tariffs and growth using a sample of 20 countries over
1972-80, and finds a statistically insignificant (negative) relationship. However, Lee
(1993), Harrison (1996), and Edwards (1998) found a significant and negative
relationship between tariff rates and growth. Clemens and Williamson (2002) find that
prior to 1950, the correlation between average tariffs and growth was positive, but
since 1950 the correlation has been negative. Vamvakidis (2002) finds that prior to
1970 there is no evidence of a negative relationship between average tariffs and
growth, but finds evidence of a non-robust negative relationship between average
tariffs and growth between 1970 and 1990. Yanikkaya (2003), looking at the growth

5
of more than 100 countries between 1970 and 1997, finds that trade barriers are
positively correlated with growth. Further, the positive correlation is found to be
stronger for developing countries than developed countries. However, as the results
reported in his Table 6 (p. 82) indicate, this pattern is not robust to the removal of bias
resulting from unobserved fixed effects. DeJong and Ripoll (2004) find that the
relationship between tariffs and growth depends on the level of development. They
find a negative relationship only among the world’s richest countries, and a positive
but not significant relationship among the world’s poorest countries.

2.2.3 Cross-sectional or Panel Data Analysis

Developing countries differ in terms of their colonial history, their political regimes,
their ideologies and religious affiliations, their geographical locations and climatic
conditions, not to mention a wide range of other country-specific variables. Failing to
take this heterogeneity (country-specific effects) into account will bias the results.
Greenaway et al. (2002) argue that dynamic mis-specification is a further limitation of
many empirical studies. The use of panel data techniques, exploiting both the cross-
section and time series properties of the data, offers potential benefits. Panel data
could also offer a potential solution to the endogeneity problem through the use of
lagged levels as instruments for the endogenous variables. Another important
advantage of a panel model is that fixed country and time effects can be used to
control for unobserved heterogeneity that is correlated with both growth and the
regressors.

2.2.4 Specification Issues (Linear or Non-linear Relationship)

DeJong and Ripoll (2004) argue that a potential source of mis-specification in the
literature involves the presence of an un-modelled contingency in the relationship
between trade barriers and growth. Using a non-linear specification, for a global panel
data set comprising 60 developing and OECD countries, they find that the relationship
between tariffs and growth depends on the level of development. They find a
significant interaction between tariff and initial income under which the marginal
impact of tariff is decreasing in income. Most of the empirical literature has not yet
systematically addressed the question of whether trade policy affects growth
differently across countries in different income groups. Rodrik (1999) argues that the
benefits from openness are contingent on the availability of complementary policies

6
and institutions, implying a contingent or nonlinear relationship between openness
and growth.

The picture that emerges is that the effect of trade policy openness may not be
uniform across countries and over time. A general proposition that is consistent with
the cross-country regression studies is that the countries that have grown rapidly on a
sustained basis have almost always done so in the presence of either low or declining
barriers to trade. In a systematic defence of free trade along these lines Panagariya
(2004a) concludes that openness is necessary but not sufficient for sustained rapid
growth. He identifies all countries that have grown at three percent or more in per-
capita terms during the last four decades and shows that these growth “miracles”
uniformly took place in the presence of low or declining barriers to trade. He also
identifies the growth “debacles” - the countries that did not experience any growth in
per-capita terms on a sustained basis or actually declined - and show that they are
rarely the outcome of openness. Thus, in his view, while openness is an important part
of the miracles, it does not lead to debacles.

3 EMPIRICAL METHODOLOGY AND DATA

3.1 Data
We construct an unbalanced panel that consists of data for 48 developing countries
over 20 years (1980-1999),2 averaging the data over five non-overlapping four-year
periods. Not all countries have data for all five time periods, but the use of unbalanced
panels may lessen the impact of self selection in the sample. The final sample consists
of 20 Sub- Saharan African countries, 13 Latin American countries, 7 from East Asia,
5 from South Asia and 3 from the Middle East and North Africa. The data comprise a
heterogeneous group of countries in terms of size, level of income, degree of
openness, population, resource endowments and so on. Our dependent variable
(GROWTH) is the (period) growth of real per capita GDP (detailed data definitions
and sources are provided in Appendix A).

2
We began with a sample of 64 countries for which data on per capita GDP growth and average tariffs
were available. However, 16 of these countries were eliminated from the sample due to lack of data for
the other two alternative trade policy measures.

7
The variables included in the model are widely accepted as core explanatory
variables. The log of Real GDP per capita in the year preceding the period (lnGDP0)
is included to capture initial country specific effects or convergence. If initial income
captures convergence the expected sign is negative. However, in a cross-country
regression it may capture country-specific initial conditions, and the sign could be
positive (Mbabazi et al 2001). The coefficient on population growth (POP) is
expected to be negative. The coefficient on investment share of GDP (INV) is
expected to have a positive sign. We employ three alternative measures of trade
policy – average (unweighted) scheduled tariffs (TARIFF), export taxes as a
percentage of exports (XTAX) and import taxes as a percentage of imports (MTAX).
We add to this basic specification other variables believed to be particularly important
in the openness-growth relation. These variables include foreign direct investment
(FDI)3, the interaction effects between trade policy and initial income and a dummy
variable for Sub-Sahara Africa (AFRICA).4

To capture potential contingencies in the relationship between tariffs and growth, we

include in our baseline specification an additional explanatory variable constructed as
the product of log initial income and our individual trade policy variables. The
interaction term is meant to capture the non-linearity in the impact of trade barriers on
economic growth. Evidence of a contingent relationship is provided by a significant
coefficient on the interaction term. This approach is in keeping with the work of
DeJong and Ripoll (2004), which used three approaches to capture potential
contingencies in the relationship between tariffs and growth and found evidence of a
negative relationship only among the relatively rich countries of the world.5

3
There is evidence that FDI contributes to growth (Borensztein et al. 1998). Over the two decades
under study FDI has become the single largest capital flow to developing countries, far surpassing
portfolio equity investment, private loans, and official development assistance. The World Bank (2002)
reported that in 1997 developing countries received 36 percent of total FDI flows.
4
The ‘Africa’ dummy is meant to test if there is an ‘Africa effect’ (to check whether the estimated
coefficient is negative and statistically significant) in our sample. The common belief is that Sub-
Saharan Africa is different from, in the sense that growth is worse than, other regions (Collier and
Gunning, 1999) and much of this difference can be related to trade factors (Mbabazi et al., 2002).
5
In the first approach, they include an additional explanatory variable constructed as the product of the
log of initial income and tariffs. Under the second, they replace this variable with an alternative
interaction term: the product of tariffs and World Bank income rankings. The third approach involves
stratifying the data set into separate sub-samples: one that includes high- and upper-middle income
countries; and one that includes lower-middle and low-income countries.

8
It is useful to examine simple statistics for the measures of openness and growth over
the period under consideration (Tables 1 and 2). Table 1 displays correlations between
per capita GDP growth, trade share and the trade barrier measures. The simple
correlations suggest that while we can expect to find a positive and statistically
significant association between trade volumes (trade share) and growth, the
unconditional relationship between trade barriers and growth is a less clear. There is
evidence of a negative and statistically significant correlation in two cases (export tax
and import tax); for average tariff the correlation appears positive but is not
significantly different from zero. All the trade barrier indicators are negatively and
significantly correlated with trade shares, suggesting that trade barriers do repress
trade. Given the positive relationship between trade share and growth, the negative
correlation between trade barriers and trade share suggests that trade barriers may
have negative effect on growth. As the econometric estimates that follow indicate, the
situation is more complicated than these statistics suggest.

Table 1: Correlation matrix between policy measures

TARIFF MTAX XTAX TRADE (%GDP) GROWTH

TARIFF 1.0000

MTAX 0.5168 1.0000

(0.0000)

XTAX 0.0711 0.1256 1.0000

(0.3399) (0.0693)

TRADE (%GDP) -0.3759 -0.3148 -0.1848 1.0000

(0.0000) (0.0000) (0.0074)

GROWTH 0.0645 -0.1502 -0.1694 0.2384 1.0000

(0.3586) (0.0295) (0.0140) (0.0002)

P-values in parentheses

Table 2 provides information about the means and standard deviations of the main
variables, one aspect being of particular interest in our analysis. With the highest
mean value of all three trade barrier measures SSA remains, on average, the most
closed region.

9
Table 2: Summary Statistics for the Main Variables (1980-99)
Variable Obs Mean Std. Dev. Min Max
East Asia
Average Tariff 40 20.628 12.562 0.200 49.500
Import Tax (% Imports) 38 7.310 4.626 0.268 15.593
Export Tax (% Exports) 38 0.914 1.542 0.000 7.315
Trade (% GDP) 39 98.895 101.483 14.713 407.348
Gross Domestic Investment 40 30.702 7.184 17.650 47.100
Foreign Direct Investment 40 2.661 3.402 -0.004 12.697
Per Capita GDP Growth 40 4.198 3.337 -4.058 11.413
Sub-Sahara Africa
Average Tariff 78 24.200 12.000 6.000 76.400
Import Tax (% Imports) 81 17.400 6.900 3.877 36.280
Export Tax (% Exports) 81 6.030 8.700 0.000 34.575
Trade (% GDP) 99 56.300 28.000 11.390 147.699
Gross Domestic Investment 99 17.300 7.100 4.325 45.525
Foreign Direct Investment 96 0.740 1.400 -6.520 5.715
Per Capita GDP Growth 99 0.007 3.690 -9.910 12.928
Latin America
Average Tariff 51 18.810 10.700 8.000 50.500
Import Tax (% Imports) 52 9.340 3.820 2.039 18.387
Export Tax (% Exports) 52 1.280 2.510 0.000 10.897
Trade (% GDP) 60 62.130 37.884 14.110 168.718
Gross Domestic Investment 60 21.993 4.676 12.250 33.800
Foreign Direct Investment 60 2.420 3.017 -0.087 16.673
Per Capita GDP Growth 60 0.730 3.054 -5.923 10.288
All
Average Tariff 206 25.770 17.403 0.200 99.900
Import Tax (% Imports) 210 14.165 8.056 0.268 46.769
Export Tax (% Exports) 210 3.141 6.223 0.000 34.575
Trade (% GDP) 238 63.957 51.788 11.390 407.348
Gross Domestic Investment 239 21.718 7.680 4.325 47.100
Foreign Direct Investment 236 1.553 2.420 -6.520 16.673
Per Capita GDP Growth 239 1.320 3.553 -9.910 12.928
Calculations are based on all 48 countries in our sample and for the three different regions of interest.
Averages are taken of annual values for 1980-1999.

East Asia is the region with the lowest levels of protection according to import and
export taxes. Sub-Sahara Africa exhibited the lowest average growth in real per capita
GDP over the period 1980 to 1999, whereas East Asia recorded the highest trade
share and rate of average growth.

10
3.2 Empirical Methodology
Drawing from recent modelling techniques in the literature (Greenaway et al. (1998);
Easterly and Levine (2001); DeJong and Ripoll (2004)), we begin with a basic
specification of the form:

y i , t = α y i , t −1 + x i , t ′ β + η i + λ t + ε i , t (1)

where is yi ,t is a country’s per capita growth rate in period t , and xi ,t is the vector of

determinants of economic growth as named above, λt is a time-specific effect, ε i ,t is

the time-varying error term, and i and t represent country and (4-year) time period,

respectively. The term ηi is a permanent but unobservable country-specific effect that

captures the existence of other determinants of an economy’s growth rate that are not

already controlled for by the vector x i ,t . It is time invariant and generally captures
such cross sectional heterogeneity as differences in technology between countries. If
the country-specific parameter were not included, random country-specific
fluctuations would be grouped into the common error term. This would bias the error
term. In a pure cross-sectional regression, the unobserved country-specific effect is

part of the error term. Therefore, a possible correlation between ηi and the
explanatory variables results in biased coefficient estimates.

It can be observed that equation (1) is a dynamic equation with a lagged dependent
variable. Furthermore, the determinants of growth can be classified according to
whether they are strictly exogenous, predetermined or endogenous. The possibility of
endogeneity together with the presence of country specific effects correlated with
some of the explanatory variables implies that estimation methods such as OLS will
not be consistent. A first step in obtaining consistent estimates is to eliminate the
country-specific heterogeneity. One approach is to employ the fixed-effects estimator
by taking deviations with respect to individual country means. However, when the
model includes a lagged dependent variable the dynamic fixed-effects model produces
estimates that are inconsistent if N (number of ‘individuals’, or cross section) is large
relative to T (number of time periods), hence the fixed effects estimator is biased
(Wooldridge, 2002; Baltagi, 1995).

11
In such cases it is appropriate to use an estimation procedure which simultaneously
addresses the issues of correlation and endogeneity. The Generalized Method of
Moments (GMM) estimator proposed by Arellano and Bond (1991) relies on first-
differencing to eliminate unobserved individual-specific effects, and then uses lagged
values of endogenous or predetermined variables as instruments for subsequent first-
differences. Thus, it is able to control for the endogeneity of the lagged dependent
variable as well as the potential endogeneity of other explanatory variables.

Endogeneity is a particular problem in studies that relate growth to openness using

trade outcome measures such as trade share of GDP or its components export/GDP or
imports/GDP. Such openness measures could clearly be endogenous since both the
export and the import share seem likely to vary with income levels. Even direct trade
policy measures, such as average tariffs, are susceptible to potential endogeneity. The
pressure for protection may increase as growth falters, at least in the short run
(Winters, 2004). Trade barriers may present issues of reverse causality if protection
depends on economic growth (O’Rourke, 2000).

To address this and the endogeneity problem, Arellano and Bond (1991) propose
using the lagged values of the explanatory variables in levels as instruments under the
assumptions that there is no serial correlation in the error term and the explanatory
variables. We follow Easterly and Levine (2001) and DeJong and Ripoll (2004) in
addressing the issue of endogeneity by imposing the identifying restriction that the
determinants of growth (variables in the x vector) are predetermined.6 The
assumption is that shocks to economic growth in period t-1 could affect gross
domestic investment, foreign direct investment, population growth, openness or their
interaction terms in period t. Given this assumption, an appropriate instrument for the
difference is the lagged value.

Given the shortcomings of the differenced estimator (Easterly and Levine, 2001), we
use the alternative systems estimator that estimates jointly the regression in

6
This is a testable hypothesis for which the Sargan test of overidentifying restrictions is reported with
all regression results. We relax this assumption later in the sensitivity analysis section to check for the
robustness of our results.

12
differences with the regression in levels, as proposed by Arellano and Bover (1995)
and Blundell and Bond (1998). The consistency of the GMM estimator depends on
the validity of the assumption that the error term does not exhibit serial correlation
and on the validity of the instruments. By construction, the test for the null hypothesis
of no first-order serial correlation should be rejected under the identifying assumption
that the error is not serially correlated; but the test for the null hypothesis of no
second-order serial correlation, should not be rejected. We use two diagnostics tests
proposed by Arellano and Bond (1991) and Blundell and Bond (1998), the Sargan test
of over-identifying restrictions, and whether the differenced residuals are second-
order serially correlated. Failure to reject the null hypotheses of both tests gives
support to our model.

4 RESULTS AND DISCUSSION

Our specification of equation (1) in standard form is:

GROWTH it = δ 0 + δ1GROWTH it −1 + δ 2 ln GDP 0it + δ 3 POPit + δ 4 INVit + δ 5 FDI it

+δ 6 POLICYit + δ 7 GDPOxPOLICYit + ε it (2)

Tables 3a, 3b and 3c report coefficient estimates obtained from the growth regressions
where we measure trade policy by average tariff (TARIFF), import tax (MTAX) and
export tax (XTAX) respectively. Column 1 in Tables 3a, 3b and 3c presents estimates
from a non-linear specification designed to establish whether the relationship between
trade policy and growth is contingent on the level of income by introducing an
interaction term between the respective policy measures and initial income.7 Finally,
we test for the existence of a ‘Sub-Sahara Africa effect’ by introducing a Sub-Saharan
African Dummy (AFRICA) into the specification in column 1 and present the results
in column 2. The baseline specification as in column (1) and its variant as in column
(2) are dynamically specified (with lags of per capita GDP growth) and estimated
using the GMM systems estimator. Various diagnostic tests are reported alongside the
coefficient estimates.

7
We first experimented with a baseline specification under which we model the relationship between
trade barriers and growth as linear (i.e., without interaction between trade policy and income) in all the
regressors. When trade policy alone is introduced into the growth regression it has inconsistent signs.
We do not report the results (available upon request from authors) here for brevity.

13
Table 3a: Trade Policy (TARIFFS) and Growth in Developing Countries (1980-1999)
Dependent variable: Per Capita GDP Growth
Explanatory variables: 1 2
0.319 0.303
GROWTH t−1
(9.92)** (9.59)**
lnGDP0 -0.728 -0.776
(2.68)* (3.03)**
POP -0.236 -0.015
(0.82) (0.05)
INV 0.163 0.144
(7.97)** (6.99)**
FDI 0.274 0.19
(2.43)* (1.8)
TARIFF -0.229 -0.182
(3.73)** (3.28)**
TARIFF x InGDP0 0.044 0.033
(4.41)** (3.61)**
AFRICA -0.87
(2.33)*
Constant 1.081 2.444
(0.51) (1.22)
Sargan Test [0.756] [0.738]
st
1 Order Serial Correlation [0.010] [0.014]
nd
2 Order Serial Correlation [0.694] [0.703]
Wald Test [0.000] [0.000]
Observations 170 170
Number of Countries 48 48

Notes: Absolute t-statistics in parentheses while p-values in brackets, * denotes significant at 10%; **
denotes significant at 5%.
1. Time dummies (not reported) are included to capture the effects of cyclical impacts on growth.
2. The Wald test is for the joint significance of the independent regressors.
3. The Sargan test is for the validity of the set of instruments.
4. The tests for 1st and 2nd - order serial correlation are asymptotically distributed as standard normal
variables (see Arellano and Bond, 1991). The p-values report the probability of rejecting the null
hypothesis of serial correlation, where the first differencing will induce (MA1) serial correlation if the
time-varying component of the error term in levels is a serially uncorrelated disturbance.

14
Table 3b: Trade Policy (MTAX) and Growth in Developing Countries (1980-1999)
Dependent variable: Per Capita GDP Growth
Explanatory variables: 1 2
0.244 0.249
GROWTH t−1
(7.82)** (6.51)**
lnGDP0 -1.126 -1.157
(4.51)** (4.20)**
POP -0.73 -0.61
(3.77)** (2.70)**
INV 0.147 0.1
(6.80)** (3.77)**
FDI 0.279 0.363
(2.69)** (3.23)**
MTAX -0.624 -0.617
(5.40)** (4.37)**
MTAX x InGDP0 0.10 0.101
(5.32)** (4.24)**
AFRICA -0.788
(2.61)*
Constant 6.165 7.144
(3.70)** (4.26)**
Sargan Test [0.843] [0.891]
st
1 Order serial correlation [0.003] [0.003]
nd
2 Order serial correlation [0.380] [0.423]
Wald Test [0.000] [0.000]
Observations 163 163
Number of Countries 48 48
Notes: See Table 3a.

We begin by discussing the relationship observed between trade policy and growth,
given the inclusion of the interaction term, from our baseline specification.8 Trade
policy enters consistently with a negative and statistically significant coefficient, but
the interaction term is significantly positive in all cases, which, according to the
theory, suggests a contingent relationship. This specification reveals a significant
interaction effect under which the marginal impact of trade barriers on growth is

8
We first experimented by estimating the global relationship observed between tariffs and growth
given the exclusion of the interaction term from our baseline specification. We find weak evidence of a
globally positive relationship between growth and trade barriers. When trade policy alone is introduced
into the growth regression it has inconsistent signs. Results available from authors upon request.

15
rising in initial income. These results imply that the impact of trade barriers on growth
is a function both of the level of restriction and of the level of income.

Table 3c: Trade Policy (XTAX) and Growth in Developing Countries (1980-1999)
Dependent variable: Per Capita GDP Growth
Explanatory variables: 1 2
0.245 0.256
GROWTH t−1
(8.10)** (9.90)**
lnGDP0 0.057 -0.033
(0.62) (0.29)
POP -0.733 -0.54
(6.37)** (5.90)**
INV 0.148 0.139
(5.51)** (5.35)**
FDI 0.084 0.065
(1.16) (0.75)
XTAX -0.191 -0.241
(2.41)* (2.54)*
XTAX x lnGDP0 0.03 0.04
(2.09)* (2.51)*
AFRICA -0.659
(3.69)**
Constant -0.403 0.155
(0.82) (0.23)
Sargan Test [0.813] [0.887]
st
1 Order serial correlation [0.003] [0.003]
nd
2 Order serial correlation [0.240] [0.246]
Wald Test [0.000] [0.000] [0.000]
Observations 163 163 163
Number of Countries 48 48 48
Notes: See Table 3a.

To examine the conditional hypotheses, we chose three values for lnGDP0 at which to
compute the marginal effects of trade policy and report the results in Table 4. From
equation (2), the derivative of growth with respect to trade policy (i.e. TARIFF,
MTAX and XTAX) is calculated as

∂GROWTH
= δ 6 + δ 7 ( ln GDP0 )
∂POLICY

16
Table 4: Marginal Effect of Trade Policy on Growth

At lnGDP0 = MIN = 4.40 At lnGDP0 = MEAN = 6.86 At lnGDP0 = MAX = 10.1

TARIFF -0.054 (2.93)** 0.044 (4.0)** 0.174 (4.25)**
MTAX -0.184 (5.32)** 0.066 (3.41)** 0.385 (5.01)**
XTAX -0.059 (3.39)** 0.015 (0.64) 0.303 (4.44)**
Note: t-ratios in parentheses

The interaction model asserts that the effect of a change in POLICY on GROWTH
depends on the value of the conditioning variable log initial GDP (lnGDP0).
Evaluated at mean trade policy level, we find a statistically positive effect for two
(TARIFF and MTAX) of our trade barrier measures. The marginal effect of XTAX is
also positive but not indistinguishable from zero. Similarly, if we compute the
marginal effects at the maximum value of POLICY, we find a statistically positive.
However, evaluated at minimum lnGDP0 the marginal effect of POLICY turns
negative and significant. Thus, the regression indicates that the derivative of growth
with respect to trade policy is an increasing and linear function of the level of initial
income. We know from the fact that the coefficient on the interaction term is positive
that the reductive effect of trade barriers declines as the level of income increases.
Consequently, the potential positive benefits accruing from any given trade policy
reforms will not be universal across different income groups. Thus, for two economies
that belong to different income groups (low and high), similar trade policies will have
different effects on economic growth. Poorer countries stand to benefit more than
richer countries. This finding is particularly interesting for SSA (the world’s poorest
continent) where trade restrictions are still pervasive and where poverty is
widespread. Moreover, we also find that the dummy for SSA countries has a negative
and significant coefficient in all the relevant specifications (Tables 3a, b and c).
Africa does appear to be different – even allowing for the other explanatory
variables,9 SSA countries have a below average growth performance.

9
As expected, the results in all three tables show that both domestic and foreign direct investments are
strong determinants of the growth rate of an economy.

17
We have assumed that the right hand side variables (including trade policy measures
and their interaction terms) are not strictly exogenous. The Sargan Test of the null
hypothesis for over-identifying restrictions (regressors are predetermined) is not
rejected at the 0.05 level of significance. This suggests that the over-identifying
restrictions are valid. The p-values in the test for second order serial correlation
indicates that we cannot reject (at 0.05 level) the null hypothesis that there is no
second order serial correlation in the differenced residuals, suggesting consistency of
our estimates. In addition, we have rejected the null hypothesis that all parameter
estimates are simultaneously zero.

4.1 Evidence on Trade Policy and Growth within Africa

The previous econometric analysis has been based on assessing Africa’s growth
performance in a global perspective. We find evidence that Africa’s slow growth is
partly explicable in a ‘global’ cross-country framework. We recognise the fact that the
‘traditional’ SSA dummy approach adopted in these regressions may obscure
important facts about trade policy and growth within the African continent. If the
region is truly different from other regions, it is appropriate to move away from the
‘traditional’ Sub-Saharan Africa dummy approach and focus on a regression limited
to a sub-sample of African countries. Given the findings from the ‘global’ regression
and the fact that SSA countries are among the poorest countries in our sample10, we
would expect, a priori, to find evidence of a negative effect of trade policy barriers on
growth in SSA.

The vast majority of SSA countries have had restrictive and distortionary trade
policies since independence until the 1980s (at least), typically motivated by some
desire to protect domestic industries, although most have liberalised significantly
since then (Ackah and Morrissey, 2005). Table A2 in the appendix shows how SSA
has lagged behind other regions in trade liberalization. Trade barriers have generally
been higher in Africa; by the end of the 1990s, Africa and South Asia maintained the
highest trade barriers. Latin America and East Asia had pursued intensive trade
reforms during the 1980s such that by the end of the 1990s trade barriers had fallen to
10
In fact, 85% of all SSA countries in the sample are classified as low or lower-middle income by the
World Bank (July, 2005). The only exceptions are Mauritius, South Africa and Botswana who are
classified as upper-middle income countries.

18
relatively low levels of 11 percent and 13 percent respectively. Compared to other
regions, Africa, and especially SSA, has exhibited poor economic performance over at
least the past two decades. While some countries have been exceptions to the trend
and performed very well, the regional performance is cause for concern.

Table 5: Trade Policy (TARIFF, MTAX & XTAX) and Growth in Africa (1980-1999)
Dependent variable: Per Capita GDP Growth
Explanatory variables: 1 2 3
0.229 0.316 0.43
GROWTH t−1
(2.69)* (2.66)* (5.98)**
lnGDP0 0.113 -0.137 -0.219
(0.19) (0.26) (0.36)
POP -1.104 -0.249 -0.226
(1.85) (0.61) (0.59)
INV 0.182 0.114 0.193
(3.14)** (2.08) (2.90)**
FDI 0.218 0.232 0.258
(3122) (0.52) (2.52)*
TARIFF -0.196
(6.14)**
MTAX 0.056
(0.84)
XTAX 0.067
(0.75)
Constant 2.662 -0.761 -0.555
(0.53) (0.21) (0.11)
Sargan Test [0.623] [0.842] [0.398]
st
1 Order serial correlation [0.055] [0.068] [0.067]
nd
2 Order serial correlation [0.249] [0.582] [0.812]
Wald Test [0.000] [0.000] [0.000]
Observations 63 61 61
Number of Countries 20 20 20
Notes: See Table 3a.

Our econometric results for the SSA sample (Table 5) show that trade policy
(TARIFF) is associated with lower growth in SSA countries. The estimated
coefficient, -0.196, on TARIFF is significantly negative, implying that a reduction in
average tariffs by ten percentage points would increase the average growth rate of per

19
capita GDP by about two percentage points annually (ceteris paribus). The results
suggest that African countries with lower tariffs grow faster than countries with
higher tariffs. The evidence in this section and the rest of this paper does not support
the pessimism that trade policy reforms would be growth-retarding in low-income
countries in general and Africa in particular. High levels of trade restriction have been
partly responsible for the slow growth in Africa, and their reduction can be expected
to result in improved economic performance in the region.

Despite the focus on an SSA sample, it is reasonable to believe that average estimates
do not provide enough information about the within-country variance in the impact of
trade policy on growth. For most practical purposes, however, it may be useful to
have some information on individual country experiences. In Table 5 we show the 20
African countries in our regression sample, along with their actual per capita growth
rates and their predicted growth from the regressions. In addition, we estimate how
much of the variations in within-country growth is explained by trade policy.

The evidence suggests that trade policy plays an important role in explaining the
variations in growth within SSA. The table is organized such that the ten observations
for which unexplained growth (residual) is lowest in absolute terms are listed in the
upper panel, and the ten observations with the largest residual in the bottom panel. By
construction, the model explains reasonably well the growth experience of those
countries in the upper panel. The simple mean growth for these countries is about 0.4
per cent whereas TARIFF has a negative impact of almost four per cent on average
(i.e. in many countries, other factors offset the negative impact of tariffs). For those in
the bottom panel, simple mean growth is -1.5 per cent whereas TARIFF has a negative
impact of over five per cent on average. For all the twenty observations in the table
the contribution of TARIFF to growth is negative. The results suggest, unsurprisingly,
that although trade barriers have a negative impact on growth, other factors may offset
this in some countries but exacerbate it in others. A corollary is that trade
liberalisation (reducing barriers) itself has a positive impact but does not ensure
growth – other factors may offset the benefits.

20
Table 6: Actual Growth, Predicted Growth and Tariff contribution
Country Time Period Actual Predicted Unexplained Contribution
Growth Growth Growth of TARIFF
10 lowest absolute values of unexplained GROWTH
Botswana 1992-95 0.009 0.029 0.020
Botswana 1996-99 3.622 3.460 0.162 -2.177
Cameroon 1992-95 -3.922 -3.569 0.353 -3.615
Congo DR 1984-87 0.082 0.017 0.064 -4.484
Ethiopia 1988-91 -3.578 -3.814 0.236 -5.805
Guinea 1996-99 1.943 1.472 0.472 -3.010
Kenya 1996-99 -0.174 0.168 0.342 -3.089
Madagascar 1996-99 0.455 0.445 0.010 -1.399
Mauritius 1988-91 5.415 4.970 0.445 -5.413
Sierra Leone 1984-87 -0.310 -0.565 0.255 -5.060
Simple Average 0.354 0.261 0.236 -3.993

10 highest absolute values of unexplained GROWTH

Burkina 1984-87 0.968 -4.389 5.356 -11.924
Burundi 1992-95 -6.265 0.497 -0.497 -1.451
Congo DR 1988-91 -7.115 -2.188 2.188 -4.452
Congo 1996-99 -1.880 4.600 -4.600 -3.265
Congo Rep 1984-87 -3.058 5.208 -5.208 -6.276
Ethiopia 1992-95 2.938 -2.590 5.528 -4.422
Kenya 1988-91 0.993 -3.430 4.422 -7.692
Rwanda 1992-95 -3.897 0.323 -0.323 -6.668
Uganda 1992-95 4.017 -0.932 4.948 -3.353
Zimbabwe 1984-87 -1.691 4.797 -4.797 -1.726
Simple Average -1.499 0.190 0.702 -5.123
Note: Residuals are from Regressions of Table 8

5 SENSITIVITY ANALYSIS
This section examines the robustness of our results with respect to several
modifications of our model. Simultaneity bias is a particular problem in studies that
relate growth to openness measured by outcome variables (trade/exports/imports
shares). Such openness measures could clearly be endogenous since they are likely to
vary with income levels. Even actual policy-induced barriers, such as average tariffs,
are not totally exonerated from endogeneity issues. This potential problem has been
addressed by the over-identifying assumption that the trade policy measures and their
interaction terms together with other growth determinants are at best predetermined
and not strictly exogenous. This suggests that potential endogeneity has been properly
instrumented with lagged values. We tested for the sensitivity of our results by
relaxing this assumption for the trade barrier indicators and their interaction terms (i.e.

21
we relax the over-identifying restrictions that trade policy is predetermined in favour
of a rather stronger assumption that trade policy is strictly exogenous; hence the trade
policy variables themselves are available as instruments). Further, we experimented
whether our results are sensitive to the exclusion of investment (INV and FDI) in our
models by re-estimating the regressions by dropping INV and FDI from the
regressions. We suspected that the relationship we investigate may be sensitive to the
inclusion of these predictors in the regressions (note however, that these variables
have been assumed predetermined and hence instrumented). In addition, we check
the sensitivity of the results to the exclusion of six countries suspected to be outliers
from the original sample.11 The results were robust to all of these tests (results
available on request).

Finally, we employ an alternative technique to explore the potential contingency

already established by our results in Table 3. To test for the robustness of the
existence of contingency, we follow DeJong and Ripoll (2004) again (but with an
updated version of the World Bank income classification table) by specifying a
regression model under which we interact trade policy with the World Bank’s (July
2005) income-rank index with low-income countries ranked as 1, lower-middle
income countries ranked as 2, upper-middle income ranked 3 and high-income ranked
as 4 (see appendix for exact cut-off values corresponding to the indexes). We then
consider the differential impact of trade policy for high and low – income countries. A
significant coefficient on the interaction term confirms the earlier results of the
existence of contingency. This result suggests that for any two countries classified
differently by the World Bank, the impact of trade restriction on growth will be
dissimilar, given the same tariff level. The results as reported in Table A1 in appendix
confirm our earlier finding of a contingent relationship between trade barriers and
growth. We consider this as a robustness check for our results. In all cases, our results
remain largely unaltered. Both the signs and orders of magnitude of the coefficients are
preserved in most cases. Thus the model parameters are robust in that they show little
sensitivity to changes in the model specification. We still find convincing evidence of

11
It is quite common to omit data points with extreme values of the explanatory variables. Several
standard deviations away from the mean value can define extreme values. The result is robust in a
variety of specifications in which outliers are included or excluded. We arbitrarily set a cut-off point at
five standard deviations from the mean of the variables. By this criterion, Singapore, Malaysia,
Pakistan, India, Bangladesh and St. Lucia are identified as outliers.

22
a negative and statistically significantly relationship between trade barriers and
economic growth which is contingent on income. We also find that the marginal
impact of trade barriers is increasing in income. Overall trade policy openness appears
to be conducive to growth but with differential impact for poor and rich countries.

6 CONCLUSION
In this paper we have investigated several questions regarding the relationship
between trade policy and growth. Our primary question concerned the effect of
protection on economic growth for the poorer countries, SSA countries especially. In
this regard, we investigate the relationship between a variety of trade policy measures
and growth. We find that trade protection has, on average, a robust negative effect on
economic performance for low-income countries in general and SSA countries in
particular. Openness to trade seems to offer the possibility of achieving faster growth
with differing impacts on countries belonging to different income groups. The richer
the country, the smaller are the growth-reducing effects of trade protection and the
poorer the country the more likelihood that trade protection will affect growth
negatively. Thus, for two economies that belong two different income groups (low
and high), similar trade policies will have different effects on economic growth. Our
study demonstrates that the relation between trade barriers and growth performance
can be extremely sensitive to a country’s level of development. This finding seems
consistent with intuition and the commonly held view that poorer, slower growing
countries tend to have much higher tariff rates than richer, faster growing countries.

The perception that low-income countries could increase growth by using tariffs
based on the infant industry argument appears intuitive, but the empirical evidence
supporting that view is not strong and our empirical evidence contradicts this view.
There is no coherent body of evidence that trade restrictions generally stimulate
growth, as even Rodriguez and Rodrik (2001) concede. However, the weight of
evidence suggests that it would be too simplistic to think that trade liberalization per
se is the key to prosperity for all countries. The evidence suggests that the relationship
between openness and growth is likely to be very case-specific. We conclude that the
maintenance of high protection appears to be one of the causes of poor economic
performance in poor countries including SSA countries. If policy makers would like

23
to see the prospects of faster growth for the SSA region, trade liberalization would
definitely be a policy option, although alone it may be insufficient to ensure growth.

REFERENCES

Arellano, M. and Bond, S. R. (1991), “Some Tests of Specification for Panel Data: Monte
Carlo Evidence and an Application to Employment Equations”, Review of Economic
Studies, Vol.58, pp.277–97.
Arellano, M. and Bond, S. R. (1998). ‘Dynamic Panel Data Estimation Using DPD98 for
Gauss: A Guide for Users’ Mimeo.
Barro, R.J., (1997), Determinants of Economic Growth: A Cross-Country Empirical Study.
MIT Press, Cambridge.
Bhagwati, Jagdish, (2000), “Globalization and Appropriate Governance,” Wider Annual
Lecture 4, WIDER, United Nations University.
Bhagwati, Jagdish and T. N. Srinivasan, (2002), “Trade and Poverty,” American Economic
Review. Papers and Proceedings May.
Blundell, R. and S. Bond (1998). “Initial Conditions and Moment Restrictions in Dynamic
Panel Data Models,” Journal of Econometrics, 87, 115-143.
Borensztein, E., De Gregorio, J. and J-W. Lee, (1998). “How does Foreign Direct Investment
affect Economic Growth?” Journal of International Economics 45, 115-135.
Davis, D. (1996), “Trade Liberalization and Income Distribution”, Working Paper No.5693,
NBER.
DeJong, David N. and Marla Ripoll (2004), “Tariffs and Growth: An Empirical Exploration
of Contingent Relationships.” University of Pittsburgh, mimeo.
Dollar, D., (1992), “Outward-oriented developing economies really do grow more rapidly:
Evidence from 95 LDCs, 1976–1985”. Economic Development and Cultural Change 40,
523– 544.
Dollar, David and Aart Kraay (2000), “Growth is Good for the Poor”, World Bank Policy
Research Working Paper 2587.
Dollar, David and Aart Kraay (2001), “Trade, Growth, and Poverty”, World Bank Policy
Research Working Paper 2615.
Easterly, William (2001), “The Lost Decades: Developing Countries’ Stagnation in Spite of
Policy Reform 1980-1998”, Journal of Economic Growth, vol. 6(2), pp. 135-57.
Easterly, W. and R. Levine (2001) “It’s Not Factor Accumulation: Stylized Facts and Growth
Models,” World Bank Economic Review, 15, pp. 177-219.
Edwards, S., (1992), “Trade Orientation, Distortions and Growth in Developing Countries,
Journal of Development Economics, 39, 31– 57.
Edwards, S., (1993), “Openness, Trade Liberalization, and Growth in Developing Countries”,
Journal of Economic Literature 31, 1358–1393.
Edwards, S., (1998), “Openness, productivity and growth: what do we really know?”
Economic Journal, 108, 383– 398.

24
Greenaway, D. Morgan, W. and Wright, P. (2002), “Trade Liberalisation and Growth in
Developing Countries”, Journal of Development Economics, 67, 229-244.
Grossman, G.M., Helpman, E., (1990), “Comparative Advantage and Long-Run Growth”.
American Economic Review 80, 796–815.
Grossman, G.M., Helpman, E., (1991a), “Quality Ladders in the Theory of Growth”, Review
of Economic Studies 58, 43– 61.
Grossman, G.M., Helpman, E., (1991b), “Endogenous Product Cycles”, Economic Journal
101, 1229–1241.
Harrison, A., (1996), “Openness and Growth: a Time Series, Cross-country Analysis for
Developing Countries”, Journal of Development Economics, 48 (2), 419– 47.
Jenkins, R., (2004), “Globalization, Production, Employment and Poverty: Debates and
Evidence”, Journal of International Development, 16, 1-12.
Mbabazi, J., Morrissey, O., and Milner, C. (2002), “The Fragility of the Evidence on
Inequality, Trade Liberalisation, Growth and Poverty”, Credit Research Paper No. 02/19,
University of Nottingham.
Milner, C. and Morrissey, O. (1999), “Measuring Trade Liberalisation”, chapter 3 in M.
McGillivray and O. Morrisey (eds), Evaluating Economic Liberalisation, London:
Macmillan.
Ng, Francis (2001) “Patterns of Trade and Protection: Selected Country Data”,
http://publications.worldbank.org/catalog/content-download?revision_id=1526199.
O’Rourke, K.H., (2000), “Tariffs and Growth in the Late Nineteenth Century”, Economic
Journal 110, 456– 483.
Panagariya, A. (2003), “Miracles and Debacles: Do Free-trade Skeptics Have a Case?”
mimeo, University of Maryland.
Pritchett, L. (1996), “Measuring Outward Orientations in Developing Countries: Can It Be
Done?”, Journal of Development Economics 49(2), 307-35.
Rodrik, Dani and Francisco Rodríguez (2001), “Trade Policy and Economic Growth: A
Skeptics Guide to the Cross-National Evidence”, in Ben Bernanke and Kenneth Rogoff
(eds.), Macroeconomics Annual 2000. pp. 261– 324, Cambridge, MA: MIT Press for
NBER.
Rodrik, D. (1992), “The limits of Trade Policy Reform in Developing Countries”, Journal of
Economic Perspectives, 6(1), 87-105.
Rodrik, D. (1998), “Globalisation, Social Conflict and Economic Growth”, World Economy,
21(2), 143-158,
Rodrik, D. (1999), The New Global Economy and Developing Countries: Making Openness
Work, Washington DC: ODC Policy Essay No. 24.
Rodrik, Dani, (1998), “Trade Policy and Economic Performance in Sub-Saharan Africa”,
NBER Working Paper, No. 6562. http//www.nber.org/papers/w6562.
Sachs, Jeffrey, and Andrew Warner, (1995). “Economic Reform and the Process of Global
Integration,” Brookings Papers on Economic Activity 1, 1-118.
Sachs, Jeffrey, and Andrew Warner, (1997). “Sources of Slow Growth in African
Economies”, Journal of African Economies, 6(3), 335-376.
Sala-i-Martin, X. (1997). “I Just Ran Two Million Regressions”, American Economic Review,
vol. 87 (2), pp. 178–83.

25
Srinivasan, T. N. and Bhagwati, J. (2001). “Outward-orientation and development: are
revisionists right?’, in(D. Lal and R. H. Snape (eds.), Trade, Development and Political
Economy: Essays in Honour of Anne O. Krueger, New York: Palgrave.
Vamvakidis, Athanasios. (2002). “How Robust is the Growth-Openness Connection?
Historical Evidence.” Journal of Economic Growth, 7:57–80.
Wacziarg, R. (2001) “Measuring the Dynamic Gains from Trade”, World Bank Economic
Review, vol. 15 (3), pp. 393–429.
Winters, L. A. (2004), “Trade Liberalisation and Economic Performance: An Overview”, The
Economic Journal, 114 (February), F4–F21.
Yanikkaya, H. (2003), “Trade Openness and Economic Growth: A Cross-Country Empirical
Investigation”, Journal of Development Economics, 72, 57– 89.

26
APPENDIX

Table A1: Trade Barriers and Economic Growth in Developing Countries (1980-1999)
Robustness of Contingency: Interacting Income Classification and Trade Policy.

Dependent variable: Per Capita GDP Growth

Explanatory variables: 13.1 13.2 13.3
0.239 0.205 0.115
Growth t −1
(6.96)** (4.01)** (3.12)**
InGDP0 -0.796 -0.999 -0.023
(4.72)** (7.43)** (0.33)
POP 0.328 -0.429 -0.907
(1.54) (2.25)* (6.60)**
INV 0.112 0.144 0.204
(5.03)** (3.79)** (11.52)**
FDI 0.344 0.256 0.207
(6.83)** (3.33)** (2.58)*
TARIFF -0.049
(3.88)**
TARIFF x Income Classification 0.085
(6.48)**
MTAX -0.206
(4.54)**
MTAX x Income Classification 0.153
(7.07)**
XTAX 0.03
(1.55)
XTAX x Income Classification -0.023
(1.30)
Constant 0.26 5.105 -0.639
(0.23) (3.25)** (0.71)

Sargan Test [0.594] [0.683] [0.753]

1st Order serial correlation [0.020] [0.004] [0.003]
2nd Order serial correlation [0.958] [0.341] [0.231]
Wald Test [0.000] [0.000] [0.000]
Observations 170 163 163
Number of Countries 48 48 48

27
Table A2: Trade Policy and Trade Volume Measures

Country Average Tariff (unweighted) Trade Volume (% GDP)

Pre-Reform 1996-1999 Change Pre-Reform 1996-1999 Change
East Asia % % % % % %
China 50 17 -195 15 40 63
Indonesia 29 11 -159 52 57 8
Korea 24 12 -104 74 74 1
Malaysia 11 8 -28 111 203 46
Philippines 34 12 -182 50 106 53
Thailand 32 19 -74 51 93 46
Average 30 13 -124 59 96 39

Africa
Congo Dem. Rep. 24 18 -34 31 48 36
Cote d'Ivoire 28 20 -39 76 84 10
Egypt 47 28 -72 72 44 -64
Ghana 34 11 -228 11 67 83
Kenya 40 16 -156 58 62 7
Malawi 22 21 -7 55 66 16
Morocco 45 22 -104 51 59 14
Nigeria 33 23 -41 42 71 41
Tunisia 25 30 16 85 86 2
Zimbabwe 10 23 57 43 84 49
Average 31 21 -61 52 67 22

Latin America
Argentina 28 12 -138 14 20 29
Brazil 47 13 -254 19 18 -4
Jamaica 16 10 -64 93 114 19
Mexico 26 12 -110 25 62 59
Paraguay 11 10 -16 34 87 61
Uruguay 47 9 -418 38 43 11
Average 29 11 -167 37 57 35
Source: Author's calculations based on data drawn from Ng (2001) (Average tariff) and Easterly 2001
(Trade volumes). Countries selected based on the availability of tariff and trade data for both pre-
reform and late 1990s.

28
Table A3: LIST OF COUNTRIES
Argentina* Ethiopia Pakistan
Bangladesh Ghana Paraguay
Botswana* Guinea Philippines
Brazil India Rwanda
Burkina Faso Indonesia Sierra Leone
Burundi Jamaica Singapore*
Cameroon Kenya South Africa*
Chile* Korea Rep.* Sri Lanka
China Madagascar St Lucia*
Congo Dem. Rep. Malawi Thailand
Congo Republic Malaysia* Tunisia
Costa Rica* Mauritius* Uganda
Cote d'Ivoire Mexico* Uruguay*
Dominican Rep. Morocco Venezuela*
Ecuador Nepal Zambia
Egypt Nicaragua Zimbabwe
Countries marked with asterisk are classified by the World Bank as high-income (i.e. including ‘Upper
middle income’) countries (with gross GNI per capita of at least $3,256 in 2004. There are 13 ‘rich’
countries (upper-middle and high), 35 ‘poor’ countries (lower-middle and low) and 20 SSA countries
(of which all except 3 are ‘poor’ countries).

Table A4: DEFINITIONS AND SOURCES OF DATA

Variable: Definition Source

GROWTH Real Per Capita GDP growth World Development Indicators (2003)
and Easterly (2001)
lnGDP0 Log Real Per Capita GDP at beginning World Development Indicators (2003)
of each period and Easterly (2001)
TRADE Ratio of total trade (exports + imports) to World Development Indicators (2003)
GDP and Easterly (2001)
INV Gross Domestic Investment World Development Indicators (2003)
and Easterly (2001)
FDI Foreign direct investment (% of GDP) World Development Indicators (2003)
TARIFF Average scheduled tariff (unweighted) Data drawn from World bank
MTAX Import duties as percentage of total World Development Indicators (1999)
imports
XTAX Export duties as percentage of total World Development Indicators (1999)
exports
AFRICA Dummy variable with the value of unity Author’s construction using data from
for countries in Sub-Saharan Africa and Easterly (2001)
zero for all others
POP Country population total; log of World Development Indicators (2003)
population; population growth and Easterly (2001)
Unless stated otherwise, all data series are drawn from the World Development Indicators (WDI, CD-
ROM 2003 and 1999) and Easterly William (2001) data series. The data is available as the Global
Development Network Growth Database at the web site www.worldbank.org/research/growth.
TARIFF data are drawn from http://publications.worldbank.org/catalog/content-
download?revision_id=1526199.

29
30