Transcript Does Trading Partners Matter for Economic Growth? Evidence

GROWTH AND REGIONAL TRADE IN
AFRICA: SOME EMPIRICAL EVIDENCE
Jacob Musila, Athabasca University, Canada
Zelealem Yiheyis, Clark Atlanta University, USA
Abstract
The formation of regional trade agreements among
countries at different levels of economic development
poses the question of whether the composition of
trade, rather than trade itself, is relevant for growth.
This question has not been thoroughly and
conclusively investigated in the case of trade between
developing countries. This paper analyzes the
empirical relationship between growth and intraAfrican export trade as well as exports to other parts of
the world as observed through an inter-country crosssection. A production function model is specified and
estimated using cross-sectional data from Sub
Saharan African countries for the period 1985-2010.
Research Problem/Motivation
• Theoretical models have been constructed to analyze the
relationship between growth and the composition of trade.
Some of these models (e.g., Lucas, 1988) conclude that a
more advanced country would grow more rapidly if it
trades with a less advanced country. Others (e.g.,
Spilimbergo, 2000) demonstrate the possibility of growth
in advanced country slowing down if it trades with a less
developed partner.
• These opposing conclusions about the impact of trade
and/or its composition on growth have not been tested
empirically in the case of SSA countries.
Literature review
The trade-growth nexus:
(1) access to markets leads to exploitation of economies of scale
and comparative advantage, which increases productivity
and growth.
(2) trade can lead to technological transfers.
(3) trade can change the competitive nature of some industries
and, in turn influence the rate of economic growth.
(4) trade can change the sectoral composition of production by
changing the relative prices and supply and demand of
goods, which can influence the overall rate of economic
growth since different goods have different rates of
technological progress.
Literature review (cont.)
Theoretical papers:
• The idea that trade composition can affect economic growth
was first formalized in models by Lewis (1977), Jones (1979),
and Grossman and Helpman (1991).
• The conclusions of the models are dependent on the
assumptions about the nature of technological progress and
the structure of demand function.
• Young (1991) presents a Ricardian model of trade among
countries at different levels of development in which learningby-doing and the spillover effects are same for all advanced
products and the demand function is homothetic. This model
concludes that trade between advanced country and less
advanced country is beneficial to growth in the former.
Literature review (cont.)
• Grossman and Helpman (1991) presents a supply-side model
in which a country that is relatively more endowed with human
capital can suffer a decrease in the rate of growth as a result of
trade with a country that is less endowed with human capital.
• Spilimbergo (2000) formulates a Ricardian model in which a
more advanced country produces both sophisticated and less
sophisticated goods while a less advanced country produces
only a less sophisticated good. The model assumes nonhomothetic demand functions and technological progress
operates through a learning-by-doing process and is countryspecific. The model demonstrates a theoretical possibility that
trade between the two countries can lead to technological
slowdown in the more advanced country.
Literature review (cont.)
Empirical papers:
• Majority of the cross-country studies find positive and
significant correlations between trade and growth. Some
of these studies include Frankel and Romer (1999), Irwin
and Tervio (2002), Dollar and Kraay (2003), Yanikkaya
(2003), Alcalá and Ciccone (2004), and Noguer and
Siscart (2005).
• None of the early empirical studies focuses on analysis of
nature of the relationship between intra- and/or interAfrican trade and growth in SSA.
The growth model
The theoretical model:
• The following production function is used to investigate the
long-run growth.
γyt = ƒ[y0, k0, h0, Z(t)]
where:
γyt
is the growth rate of real GDP per capita in year t
y0
is initial real GDP per capita
k0
is initial physical capital stock per person
h0
is initial human capital per person
Z
is a vector of control and environmental variables
(1)
The growth model (cont.)
Initial conditions
• Telephone lines per person and secondary school gross enrolment
rates in 1985 are used as proxies for initial stock of physical and
human capital, respectively.
• Real GDP per capita in 1985 is used as initial real GDP per capita
and is intended to capture the tendency for SSA countries to
converge or diverge.
Control/environmental variables
• The stock of democracy is constructed and used as a proxy for
political competition.
• The openness of executive recruitment is used as proxy for corporate
openness.
• The export-to-GDP ratios (X/GDP) are used to represent the variable
for export trade. (Exports are disaggregated between Africa, other
LDCs, and high income countries.)
The growth model (cont.)
The econometric model:
Rate of growth = ci + β1 log(real GDP per capita in 1985)i
+ β2 log(Telephone lines per capita in 1985)i
+ β3 (Sec sch. enrolment rates in 1985)i
+ β4 (Stock of democracy)it
+ β5 (Open executive recruitment)it
+ β6 (Trade flow ratio)it + uit
(2)
where uit is the idiosyncratic errors and ci is a random effect
for t = 1,…, T and i = 1,.., N.
The growth model (cont.)
Data sources:
• Data on initial real GDP per capita (GDP per capita in
constant 2000 US$), telephone lines, secondary school
gross enrolment rates, per capita real GDP growth rates,
and shares of merchandize export trade are obtained
from World Development Indicators (December, 2012).
• Data on stock of democracy (Polity2 and durable) and
openness of executive recruitment are obtained from
Polity IV Data Series Version 2010.
Estimated results
Summary Stats: 35 cross-sections (35 SSA countries)
Democracy stock
Open executive recruitment
Exports to within (intra-African trade)
Exports to other LDC
Exports to high income countries
Initial enrol rates, secondary sch = 1985
Initial telephone line per capita= 1985
Initial GDP per capita = 1985
Real GDP per capita growth rates
Maximum Minimum Mean Std. Dev.
399.0 -598.5
-21.1 108.7
4.0
-88.0
-4.3
23.4
27.8
0.0
3.4
4.2
98.9
18.5
76.8
14.9
80.4
1.0
19.8
14.7
64.2
3.3
22.1
15.0
0.1
0.0
0.0
0.0
5113.4 118.1 655.4 974.6
37.1
-47.3
1.0
5.4
Estimated results (cont.)
Table 1: Estimates with aggregate merchandize export trade
a
Dependent variable = per capita real GDP growth
b
Explanatory variable
OLS
OLS
Constant
10.264***
(2.927)
6.903**
(2.179)
10.264***
(2.927)
Log (initial Real GDP per capita)
-0.937**
(-2.258)
-0.722*
(-1.696)
-1.540***
(-3.006)
Log (initial telephone lines per capita)
0.179
(0.816)
0.228
(1.001)
0.294
(1.191)
Secondary school enrolment rates
-0.041**
(-2.055)
-0.040*
(-1.897)
-0.057**
(-2.330)
Stock of democracy
0.007***
(4.444)
0.008***
(4.662)
0.009***
(3.570)
Openness in executive recruitment
0.039***
(4.821)
0.037***
(4.418)
0.036***
(4.319)
Aggregate merchandize exports/GDP
0.060***
(3.432)
0.036**
(1.962)
0.118***
(3.460)
Adjusted R2
F-statistic
DW-statistic
Total panel observations
0.055
9.657
1.847
896
0.047
7.626
1.864
866
0.040
9.430
1.832
825
a
b
The values in parentheses are the t-statistics.
The merchandize export trade ratio variable is lagged.
TSLS
Estimated results (cont.)
Table 2: Estimates with intra-African merchandize export trade
a
Dependent variable = per capita real GDP growth
b
Explanatory variable
OLS
OLS
Constant
5.191*
(1.617)
5.119
(1.580)
5.262
(1.240)
Log (initial Real GDP per capita)
-0.345
(-0.842)
-0.244
(-0.612)
-0.387
(-0.956)
Log (initial telephone lines per capita)
0.157
(0.693)
0.228
(0.984)
0.144
(0.459)
Secondary school enrolment rates
-0.032*
(-1.618)
-0.033*
(-1.721)
-0.033*
(-1.782)
Stock of democracy
0.009***
(5.205)
0.008***
(3.997)
0.012***
(5.081)
Openness in executive recruitment
0.040***
(4.836)
0.037***
(4.408)
0.038***
(4.334)
Intra-African merchandize exports/GDP
-0.038
(-0.736)
-0.043
(-.725)
0.021
(0.086)
Adjusted R2
F-statistic
DW-statistic
Total panel observations
0.041
7.338
1.827
896
0.034
5.752
1.796
802
0.040
7.732
1.810
825
a
b
The values in parentheses are the t-statistics.
The merchandize export trade ratio variable is lagged.
TSLS
Estimated results (cont.)
Table 3: Estimates with merchandize export to other LDCs
a
Dependent variable = per capita real GDP growth
b
Explanatory variable
OLS
OLS
Constant
9.869***
(2.644)
7.332*
(1.900)
17.529
(0.491)
Log (initial Real GDP per capita)
-0.660
(-1.571)
-0.484
(-1.112)
-1.149
(-0.506)
Log (initial telephone lines per capita)
0.136
(0.607)
0.201
(0.862)
0.195
(0.723)
Secondary school enrolment rates
-0.036*
(-1.793)
-0.035*
(-1.681)
-0.047
(-1.005)
Stock of democracy
0.008***
(4.795)
0.009***
(4.896)
0.010*
(1.640)
Openness in executive recruitment
0.040***
(4.900)
0.037***
(4.451)
0.039***
(4.576)
Merchandize exports to other LDCs/GDP
-0.039*
(-1.869)
-0.0014
(-0.670)
-0.092
(-0.339)
Adjusted R2
F-statistic
DW-statistic
Total panel observations
0.044
7.944
1.831
896
0.039
6.863
1.854
866
0.032
6.992
1.820
825
a
b
The values in parentheses are the t-statistics.
The merchandize export trade ratio variable is lagged.
TSLS
Estimated results (cont.)
Table 4: Estimates with merchandize export to HIC
a
Dependent variable = per capita real GDP growth
b
Explanatory variable
OLS
OLS
Constant
6.692**
(2.238)
6.371**
(2.044)
12.159***
(3.711)
Log (initial Real GDP per capita)
-0.723*
(-1.792)
-0.529
(-1.258)
-1.725***
(-3.488)
Log (initial telephone lines per capita)
0.174
(0.819)
0.228
(1.027)
0.357
(1.579)
Secondary school enrolment rates
-0.038*
(-1.932)
-0.037*
(-1.779)
-0.059***
(-2.602)
Stock of democracy
0.008***
(4.933)
0.009***
(5.011)
0.009***
(4.027)
Openness in executive recruitment
0.040***
(4.981)
0.034***
(4.075)
0.042***
(4.862)
Merchandize exports to HICs/GDP
0.044**
(2.184)
0.018
(0.874)
0.154***
(3.533)
Adjusted R2
F-statistic
DW-statistic
Total panel observations
0.047
8.388
1.828
896
0.042
7.265
1.853
866
0.008
10.142
1.755
825
a
b
The values in parentheses are the t-statistics.
The merchandize export trade ratio variable is lagged.
TSLS
Conclusion
• This paper uses the aggregate production function to
investigate the impact of export trade flows on growth in SSA.
• The estimated results show that the trade ratios for aggregate
exports, exports to HIC enter the growth equation positively
and are statistically significant. The positive and significant
coefficients of SSA-HIC export trade support the view that the
SSA growth benefits more from trading with HIC (the North).
• On the other hand, the estimated coefficients of the intra-Africa
exports and other exports to LDCs are negative or insignificant.
This suggests that exports of SSA countries to other SSA or
LDCs have played an insignificant role in the growth of SSA
economies.