The African Financial Review July-August 2014 | Page 63

this, the study assumed study that there is a connection between
the level of growth experienced in a country in the preceding year
with that of the current level, that is, the level of growth achieved
in the previous year has a link with the level of growth that the
country would attain in the current year. In other words, there is
integrated growth in the country. This is particularly necessary
because the economy is assumed not to exist in isolation; there
are interconnections among the various sectors in the economy,
hence, the economic activities in the preceding year have a bearing
with current economic activities. This is why the dynamic panel
data is used in this study to estimate this link. Thus the linear
dynamic panel data model is expressed as:
Grgdpit = α 1Grgdpi,t-1 + α 2INST it + α 3TLIB it + vi + eit

(3.8)

where; Grgdpt-1: one period lag of growth rate of real GDP;
INST is a vector that comprises of strictly institutional exogenous
covariates (ones dependent on neither current nor past eit ); TLIB
is the trade liberalization exogenous covariate.Thus, expressing
equation (3.8) in dynamic panel data form putting all the variables
in equation (3.6) gives equation (3.9):
lGrgdpit = α0i + α1ilGrgdpt-1 + α2ilGkapit + α3ilLabit + α4ilHkapit
+ α5iRepriskit + α6iPolrigit + α7iEthsionit + α8ilOpenit + α9ilNareit +
α 10ilTaxesit + εit (3.9)
Equation (3.9) was estimated using the Generalized Methods
of Moments (GMM) technique.

Data analyses and discussion

In this section, data analysis and discussion of results for this
study are made. The unit root test is used to test the nature of time
series to determine whether they are stationary or non-stationary.
If a time series is stationary, it means that its mean, variance and
auto covariance are the same at the very point they are measured.
That is, they are time invariant. But if the mean, variance and
auto covariance of a time series are not the same at any point they
are measured, the time series is non stationary. This is a unit root

problem. This implies that the study of the behaviour of that time
series is only possible for the time period under consideration. It
cannot be generalized to other time periods. Such time series may
be of little value for forecasting. The stationarity of the time series
is important because correlation could persist in non stationary
time series even if the sample is very large and may result in what
is called spurious or nonsense regression (Yule, 1989; Wei, 2006).
Thus, in order not to have spurious results, this study carried out
panel unit root tests. The panel unit test can be carried out on a
pooled data when two conditions are met; first, the time series and
cross-sectional observations must be more than fifteen years each
and second, the panel must be balanced, that is, there should not
be any missing data. These two conditions are met by this study.
There are thirty countries selected and the time period is twentyeight years; while the data used is a balanced one. Panel unit root
test is the panel data (both time series and cross- sectional data)
version of the time-series unit root test. The null and alternative
hypotheses are formulated as:
H0: All panels contain unit roots.
H1: At least one panel is stationary.
The rule of thumb for decision making under panel unit root
test involves the rejection of the null hypothesis at the 1 percent
statistical significance level, this implies that all panel series in
the panel data set do not contain a unit root; therefore, at least
one panel is stationary. This automatically implies the acceptance
of the alternative hypothesis which means that at least one panel
is stationary.

The results presented in Table 1 are the panel unit root tests
of the variables. It reveals that all the variables used in the growth
model are statistically significant at 1 percent and are stationary at
levels. Therefore, we reject the null hypothesis that states that all
panels contain unit roots. This means that there are no unit roots
in the panels of this study, therefore, this implies that at least one
panel is stationary. The implication of this is that the variables are
stationary which means that the results obtained from this study

Table 1: Augmented Dickey Fuller (ADF) unit root test results at levels
Variables

Chi-squared Statistic Remark

Lngrgdp

206.02*** (0.0000)

I(1) Stationary

Lnssenr

132.43*** (0.0086)

I(1) Stationary

Lngkap

Lnpsenr
Lnopen

Ethsion
Reprisk
Polrig

Lntaxes
Lnnare

Number of panels 30

142.09*** (0.0034)
123.02*** (0.0000)

181.09 *** (0.0002)
244.47*** (0.0000)
128.87*** (0.0012)
89.61*** (0.0084)
88.23*** (0.0074)

166.12*** (0.0000)

I(1) Stationary
I(1) Stationary
I(1) Stationary
I(1) Stationary
I(1) Stationary
I(1) Stationary
I(1) Stationary
I(1) Stationary

Number of periods 26
Source: Estimated by the Author. Probability values are displayed in parentheses beside the chi-squared coefficients.
Note: *** - significant at 1 percent, ** - significant at 5 percent.

The African Financial Review | 63