Chapter 3

Research Methodology
3.0 Introduction
In this chapter, I outlined the research methodology employed to investigate the relationship
between trade liberalization and economic growth in developing countries in Africa. The
methodology encompasses data collection, sources and method of specification, variable
measurement and statistical techniques utilized to analyze the research questions and objectives.
3.1 Sources of Data
The data used for the study is sourced from secondary sources. Specifically data from World
Bank Development Indicators (WDI), of the World Bank, International Monetary Funds and
United Nations Comtrade Database. However, where the data from the aforementioned sources
were not available for a given country, National Statistical Institution would be used for the
country to ensure accuracy and consistency.
3.2 Variable Measurement
The study would use GDP per capita as the dependent variable and Trade liberalization measured
as Export + import as a ratio of GDP as used by Moyo et al., (2017) with slight modification to
his model, inflation rate, interest rate and investment rate would be used as control variables.
Dependent Variable: Economic growth is measured by annual growth rate of GDP per capita
(gross domestic product divided by midyear population). According to world development
indicator of the World Bank, GDP is the sum of gross value added by all resident producers in
the economy plus any product taxes and minus any subsidies not included in the value of the
products. It is calculated without making deductions for depreciation of fabricated assets or for
depletion and degradation of natural resources. Data presented in this study are in United States
Dollar (World Bank, 2024).
Independent Variable: Trade openness, measured as the ratio of total trade (exports plus
imports) to GDP. Export of goods and services according to World Bank represent the value of
all goods and their market services provided to the rest of the world. They include the value of
merchandise, freight, insurance, transport, travel, royalties, license fees and other services, such
as communication, construction, financial, information, business, personal and government
services. They exclude compensation of employees and investment income (formally called
factor services) and transfer payments. While import of goods and services represent the value of
all goods and market services received from the rest of the world. They include the value of
merchandise, freight, insurance, transport, travel, royalties, license fees and other services, such
as communication, construction, financial, information, business, personal and government
services. They exclude compensation of employees and investment income (formally called
factor services) and transfer payments. Data are in current U.S dollars (World band, 2024).
Control Variables: Various control variables will be included to account for factors that can
influence economic growth, including inflation rates, investment rates, and interest rate.

1|Page
Inflation rate: as measured by the consumer price index reflect the annual percentage change in
the cost to the average consumer of acquiring a basket of goods and services that may be fixed or
changed at specified intervals, such as yearly. The Laspeyres formula is generally used
Investment rate: Gross fixed capital formation formally known as gross domestic fixed
investment will be used as an indicator for measuring investment rate. It include land
improvements (fences, ditches, drains and so on); plant, machinery, and equipment purchases;
and the construction of roads, railways, and the like, including schools, offices, hospitals, private
residential dwellings, and commercial and industrial buildings. According to the 1993 SNA, net
acquisitions of valuables are also considered capital formation. Data are in current U.S dollars
(World Bank, 2023)
Interest rate spread: this is the interest rate charged by the banks on loans to private sector
customers minus the interest rate paid by commercial or similar banks for demand, time, or
savings deposits. The terms and conditions attached to these rates differ by country, however,
limiting their comparability.
3.3 Sample size and Techniques
The data for this study covers a period from 2012 to 2022, which varies depending on data
availability for each country. To account for cross-country variations and time-series dynamics,
this study employs panel data analysis. Panel data models allow for a more robust analysis by
considering both cross-sectional and time-series variations. The study apply a simple random
sampling technique in selecting the countries to be used in the study. However, each selected
country is based on their affiliation to its regional organization like the ECOWAS, AU etc. as
used by Guei (2019).

3.4 Model specification

Following the relevant studies reviewed in chapter 2 (Moyo et al, 2017) and (Onafoworo &
Owoye, 1998). The study adopts their model with slight modification.
GDP = f (TRL, INV, INT, INF)…………………………………….. (1)
The equation above is specified in the econometrics form as follows;
GDPit = TRLit + INVit + INTit + INFit + it..................... (2)
The model is further expressed in linear logarithmic form as follows
LogGDPit = log TRLit + log INVit + INTit + INFit + it...... (3)
Where:
GDP = Gross Domestic Product per capita current (U.S dollars)

2|Page
TRL: Trade Liberalization = Export of goods and services + import of goods and services current
(U.S dollars)
INV: Investment rate = Gross fixed capital formation current (U.S dollars)
INF = Inflation rate
INT = Interest rate
Log = logarithms
it = panel data for cross-country variation and time series variation from (2012-2022).
and are parameters to be estimated
Error term
3.5 Method of Data Analysis
3.5.1 Descriptive Statistics
Descriptive, or deductive, statistics is the phase of statistics that seeks only to describe and
analyze a given group without drawing any conclusions or inferences about the larger group.
(Spiegel, 2011). This provides essential information about the variables involved, like their
average values, variability, and range. These statistics help understand the data’s distribution and
potential relationships between variables.

3.5.2 Research Techniques

The research models and techniques used in this study includes Generalized Method of Moments
(GMM), Arellano-Bond estimators, Sargen-Hansen test as well as the use of panel data to
account for cross-country variation and time series variation.
The generalized method of moments (GMM) estimator and a test of the over identifying
restriction were programmed in TSP. the specific formulas for the estimator and the test are
briefly stated below with the results of the tests of the over identifying restrictions for the
equations of the model. Each equation of the model can be written in the form
y = Z
y = (T x 1) vector of observations;
Z = (T x K) matrix of observations;
(K x 1) vector of parameters to be estimated;
= (T x 1) vector of disturbances.
The GMM estimator of  is given by

And an estimate of the covariance matrix of if given by

3|Page
Where
W = (T x q) matrix of instrumental variables,
(q x q) consistent estimate of the optimal weighting matrix.
Actual computation of  andrequires calculation of
Hansen (1982) showed that the optimal GMM estimator of (the consistent estimate
with the smallest asymptotic variance) is obtained when the weighting matrix is proportional to
the inverse of the variance-covariance matrix

Dynamic Panel Data Models: To address potential endogeneity issues and time lags in the
relationship, Arellano-Bond estimators, will be employed.
A static panel data model takes the form
Yit = Xit+ vi + it, (1)
where vi can be random or fixed effect for unit i.
A dynamic panel data model allows past realizations of the dependent variable
to affect its current level. Consider the model

where vi can be random or fixed effect for unit i.

Two new terms are introducted: wit is a vector of pre-determined covariates, which can be lags

of Xit’s. are lagged dependent variables, up to  lags.

wit does not have a big impact on how to estimate the model. Given the
condition that they are predetermined, a regular panel data estimator (for example, a fixed effect

estimator) is still consistent. However, the introduction of makes a regular

panel data estimator inconsistent. To see why it is so, let’s consider a fixed effect estimator
(within estimator). Suppose we have

which means, we only have yi,t−1 on the right hand side.

A within estimator regresses and other “demeaned”
regressors, with an erorr term . The within estimator gets rid of vi.

4|Page
However, yi,t correlates with it, so yi,t−1 correlates with i,t−1. Therefore yi,t−1 correlates with .
So correlates with , which makes the within estimator inconsistent.
This inconsistency goes away if is very small comparing to , which can occur in a long
panel.
Let’s consider a simpler situation, with p = 2; that is, only two lags of yi
are included in the regressors. Also let’s omit wit for now. Consider first differencing equation 2,

As we discussed, OLS estimator is inconsistent, since yi,t−1 is correlated with

i,t−1.
Anderson and Hsiao (1981) proposed using instrumental variables estimator
with yi,t−2 as an instrument for (yi,t−1 − yi,t−2).
Arellano and Bond (1991) expanded the idea by using additional lags of the dependent variable
as instruments. For example, both yi,t−2 and yi,t−3 can be uased as instruments. In fact, as t
increases, the number of instruments avialable also increases. In period 3 only yi,1 is available.
In period 4 yi,1 and yi,2 are available. In period 5 yi,1 and yi,2 and yi,3 are availabe, and so on.

5|Page