The African Financial Review July-August 2014 | Page 62

and do not take note of the individual characteristics of each
entity. While the GMM was used because it defines a class of
estimators that have properties of consistency and asymptotic
normality and is used when the sample moments (mean) are

One strand of growth literature has argued for
the primacy of institutions in economic growth,
and findings from empirical studies have
concluded that institutions are crucial to the
success of economic reforms in developing
countries. This suggests that the failure of
trade reforms to promote trade and growth
in SSA countries may be attributable to the
poor quality of institutions.
used to estimate the population moments (mean). In addition,
since the previous growth level and the current growth level in
the selected countries are integrated, the GMM is a suitable
econometric technique to test this relationship. STATA 11.0
statistical software was used to analyze the data. This is based on
the ability of the software to handle panel data and various test
statistics that the study is interested in.

Model specification

The model for this study is adapted from the works of Kagochi
et al., (2007); Baliamoune-Lutz and Ndikumana, (2007) and
Bhattacharyya, (2011). The Solow and endogenous growth
theories constitute the theoretical base of the model specified
in this study. For the purpose of this study, the growth model is
specified as:

Grgdp = f(gkap, Lab, Hkap, Reprisk, Polrig, Ethsion, Open, Nare,
Taxes) (3.4)
Recall that the model has some conventional variables found
in the Solow growth model, and it is assumed that a non-linear
relationship exists between the variables based on the CobbDouglas production function assertion. Hence, equation (3.4)
stated in Cobb-Douglas form gives:
Grgdp = AGkapα1 Labα2 Hkapα3 Repriskα4 Polrigα5 Ethsionα6
Openα7 Nareα8 Taxesα9ε
(3.5)
where; A is the total factor productivity – a measure of
productivity.
Equation (3.5) cannot be estimated directly using the OLS
technique of estimation since it is non-linear. Therefore, it
would be necessary to transform it into linear form that allows
the use of the OLS technique. In doing this, the double logtransformation rule is applied on the equation. The essence of this
is that it provides estimated parameters that can be interpreted
directly as elasticities that is, the sensitivity of a change in the
Grgdp following a change in the variables included in the model.
Consequently, equation (3.6) becomes:
lGrgdpt = α0i + α1i lGkapt + α2ilLabt +α3ilHkapt + α4iRepriskt +

(+)
(+)
(+)
(+)
(-)
α5iPolrigt + α6iEthsiont + α7ilOpent + α8ilNaret + α9ilTaxest + εt

(-) (+) (-) (+) (3.6)
where; α0 is the intercept. The αis, for i = 1, 2, 3, 7, 8, 9, being
elasticities such that αi><1; the signs below the variables in brackets
indicate the apriori expectations.

where; Grgdp: growth rate of real GDP;

Gkap: gross fixed capital formation (proxy for capital or
investment);

Lab: employment to population ratio (proxy for labour);

INST: a vector of institutional variables;

TLIB: trade liberalization variable.

Recall that this study made use of thirty (30) SSA countries;
which means that we have both time series and cross-sectional
data. The Ordinary Least Squares (OLS) technique cannot be
used to estimate combined time series and cross-sectional data.
Therefore, there is a need to use an appropriate technique that
takes care of panel data, hence the usage of the LSDV technique.
Consequently, equation (3.6) expressed in panel data form
becomes:

Trade liberalization and institutional variables are made up of a
combination of variables which are specified in equations (3.2)
and (3.3):

lGrgdpit = α0i +α1ilGkapit +α2ilLabit +α3ilHkapit + α4iRepriskit+
α5iPolrigit + α6iEthsionit + α7ilOpenit + α8ilNareit + α9ilTaxesit +εit
(3.7)

INST = f (Reprisk, Polrig, Ethsion, Hkap, Taxes, Nare) (3.2)

where; i = 1, 2... 30 (countries); t = 1, 2... 28 (years). i =1,...,N , t
= 2,...,T;ε is the error term,i is ith
country and t is the time period for the variables we defined
above. The intercept term carrying a subscript i suggests that
the intercepts of the selected countries may be different. The
coefficients α1... α3 and α7... α9 are elasticities because they
measure the rate of change. α0 is the intercept.

However, the limitations of the LSDV includes; (i) there is
the degrees of freedom problem arising from introducing too many
dummy variables; (ii) the problem of multicollinearity arising
from too many variables, both individual and multiplicative, this
makes precise estimation of one or more parameters difficult;
and (iii) the LSDV may not be able to identify the impact of
time invariant variables. Due to these limitations, this study
introduced the concept of dynamic panel data. As a result of

Grgdp = f (Gkap, Lab, INST, TLIB)

TLIB = f (Open)

(3.1)

(3.3)

where; Open: degree of openness (measure of trade liberalization);

Taxes: proxied by tax revenue on natural resources;

Hkap: human capital (proxied by primary and secondary

school enrolments); Nare: natural resource endowment

(proxied by the share of fuel in total export); Reprisk:

repudiation risk (proxy for contracting institutions);

Polrig: political Rights (proxy for political institutions);

Ethsion: ethnic tensions (proxy for cultural institutions).
Putting equations (3.1), (3.2) and (3.3) together in one equation
gives us the growth model used in this study:

62 | The African Financial Review