4 Data and Empirical Results of Growth Enhancing Variables

The study of growth is principally about the demand and supply sides and generally about the medium or long-run. It revolves around issues about accumulation of physical capital and intensity, population growth, knowledge and innovation, utilization and combination of factors and so forth (Stern 1991). In this section of the study, we employ annual data range spanning 1961–2012 depending on the time series for each of the four countries studied. Following the literature, we use growth rate of total GDP for the analysis. Openness to trade is measured as ratio of real exports and imports to real GDP. Private and government investments are expressed as ratios of GDP. Official Development Assistance is expressed as a ratio of GDP. The data set used in the analyses is obtained from the 2013 World Development Report and the 2013 African Development Report.

4.1 Empirical Results and Economic Implications

4.1.1 Stationarity Tests

As a first step, the time series properties of the variables tested include the stationarity test by conducting the Augmented Dickey-Fuller (ADF) and DickeyFuller GLS (DF-GLS) unit root tests. The results presented in Table 3 offer strong evidence of almost all the variables to be integrated of order one; that is I(1). Variables that are integrated at levels are the output growth variable (gdp) for all the sampled countries (The Gambia, Ghana, Nigeria and Sierra Leone); foreign direct investment (fdi) for The Gambia and Sierra Leone; government consumption expenditure (gc) for Sierra Leone; foreign aid or official development assistance (oda); for Nigeria; trade openness (trad) for Sierra Leone and inflation variable (inf) for The Gambia, Ghana and Nigeria. Consequent upon this, we applied the Dynamic Ordinary Least Squares (DOLS) to estimate a single cointegration vector that characterizes the long-run relationship among the variables in each country's economic growth function. The DOLS “simply regress one of the variables onto contemporaneous levels of the remaining variables, leads and lags of their differences, and a constant using ordinary least squares” Stock and Watson (1993,

p. 784). The DOLS model a la Stock and Watson is specified as:

(11)

Yi =dependent variable

X = matrix of explanatory variables

Table 3 Results of unit root test

Variable

The Gambia 1978–2011

Ghana 1965–2012

Nigeria 1981–2012

Sierra Leone 1965–2012

ADF

DF-GLS

ADF

DF-GLS

ADF

DF-GLS

ADF

DF-GLS

fdi

-2.83

-2.89

-9.21*

-7.74*

-2.99***

-1.15

-4.89*

-4.92*

Δfdi

-7.77*

-6.97*

–

–

-2.93***

-10.03*

–

–

lgc

-1.12

-3.13

-2.58

-2.51

-2.11

-1.92

-4.01**

-4.01*

Δlgc

-7.46*

-7.17*

-6.78*

-6.79

-4.75*

-4.92*

–

–

gdp

-6.53*

-6.05*

-5.42*

-5.50*

-5.87*

-4.31*

-6.19*

-6.02*

lm2

-0.92

-0.98

-1.70

1.69

-2.85

-2.48

-1.96

-1.85

Δlm2

-5.76*

-5.83*

-6.75*

-6.89*

-4.81*

-4.80*

-7.20*

-7.35*

ldcp

-0.85

-1.00

-0.83

-0.91

-3.19

-3.07***

-1.71

-1.86

Δldcp

-5.13*

-5.25*

-6.29

-5.71*

-7.36*

-7.34*

-8.01*

-8.10*

loda

-1.48

-1.59

-1.70

-1.69

-4.70**

-4.60*

-2.97

-2.60

Δloda

-6.01*

-6.14*

-6.75*

6.89*

–

–

-5.17*

-7.23*

ltrad

-1.85

-1.95

-2.25

-2.11

-2.95

-3.03**

-3.78**

-3.82*

Δltrad

-6.36*

-6.53*

-5.00*

-4.63*

-3.42**

-7.60*

–

–

infla

-3.12**

-3.18*

-3.39**

-3.41*

-3.71**

-3.80*

-2.79

-2.82

Δinfla

–

–

–

–

–

–

-8.09*

-10.52*

*, **, *** are the significant levels of 1% 3.610, 5% 3.055 and 10% 2.762 levels of significance

ᾱ= cointegrating vector; i.e., represents the long-run cumulative multipliers or, alternatively, the long-run effect of a change in X on Y

p = lag length, and

q = lead length

DOLS regression includes the lag and lead terms in order to make its stochastic regressors (Camacho-Gutierrez 2010). The method involves augmenting the cointegration regression so that the resulting cointegration equation error term and addition of the p lags and q leads of the differenced regressors soaks-up the long-run correlation between μ1t and μ2t and least squares estimates of θ=( ᾱ, γ^{i})^{i} .

Finally, stationarity test are performed on the residuals of the estimated DOLS regression. This is to determine whether the estimated results are spurious. The EViews 7 is applied in the estimation processes.

4.1.2 Estimated Results

The DOLS estimated results for the countries are presented in Table 4. The number of leads and lags are selected according to the Akaike information criterion. In order to preserve space, only the best and plausible results are reported. This is subsequently followed by the post-estimation diagnostics.

The estimated results for The Gambia show that the coefficient on the share of investment in GDP is significant at 1 % level and high. This implies a gross rate of return on investment of 15 % per year. The negative but significant coefficient on official development assistance (oda) might be that it is less productive at the margin and probably mis-directed according to political, rent-seeking objectives. The democratic variable has a positive effect on the growth of The Gambia and statistically significant. The openness and inflation coefficients are not statistically significant and as such, may not have impacted much on the growth path of the country.

Column 3 of Table 4 reports the DOLS estimated results of Ghana. The growth enhancing variables which drive the economy are investment (gfc), openness (trad), democratic institution (democ) and foreign direct investment (fdi). Of significant note is the official development assistance (oda) variable which is negatively signed. This may not be surprising as Osei et al. (2005) study underscores the 'fungibility' of foreign aid in the case of Ghana.

The DOLS results of Nigeria are, however, not as robust as those of Ghana. The foreign direct investment (fdi) coefficient appears to be the most plausible of the growth enhancing variables that entered the Nigerian growth function. Such investments may have occurred more in the oil and gas sector; however, the communication sector in last decade has equally witnessed unprecedented foreign investments. Government consumption represents government in the model. Its attached coefficient is negative and significant at approximately 8.5 % while private investment is not significant statistically (p-value ¼ 0.573). This may perhaps to some extent imply 'crowding out' situation. In explaining government consumption expenditure negatively correlating with growth, Barro (1990) observed that government activity is taken to be a productive input into private sector production, albeit with decreasing effect on private sector marginal productivity. One reason for this is that for government to finance its activities, it requires distorting taxation.

Table 4 DOLS estimates for the four WAMZ countries

Variables

The Gambia

Ghana

Nigeria

Sierra Leone

(a)

(b)

Intercept

2.186

9.146

3.422

-3.513

-5.424

(2.201)

(2.263)

(2.293)

(-2.341

(-2.733)

[0.080]

[0.066]

[0.070]

[0.063]

[0.043]

GFC

5.890

8.836

(4.594)

(3.777)

[0.000]

[0.003]

ODA

-8.114

-6.599

-2.759

-7.821

-2.311

(-4.696)

(-3.906)

(-2.518)

(-2.379)

(-2.269)

[0.000]

[0.003]

[0.063]

[0.083]

[0.038]

TRAD

4.070

3.763

-3.598

8.010

(1.602)

(2.105)

(-2.325)

(4.698)

[0.122]

[0.059]

[0.068]

[0.000]

DEMOC

1.644

0.999

-8.858

0.282

(3.766)

(4.200)

(-2.857)

(1.138)

[0.001]

[0.001]

[0.065]

[0.272]

INFLA

0.001

0.208

-2.338

-1.792

(0.025)

(1.416)

(-2.579)

(-2.430)

[0.980]

[0.216]

[0.082]

[0.027]

GC

7.001

-0.009

6.620

(3.815)

(-2.531)

(5.309)

[0.012]

[0.085]

[0.000]

FDI

-3.016

-3.730

(-3.102)

(-3.412)

[0.027]

[0.042]

M2

5.010

9.04

(0.379)

(5.813)

[0.720]

[0.000]

K

1.387

(0.631)

[0.573]

DCP

-7.442

(-2.741)

[0.014]

ECM

-0.431

-0.663

-0.252

-0.688

-0.517

(-2.435)

(-2.628)

(-2.48)

(-2.386)

(-2.413)

[0.025]

[0.042]

[0.049]

[0.054]

[0.053]

Notes: Dependent variable is growth of output t-statistics in ()

Probability in []

DOLS Dynamic ordinary least squares

Tax drives a wedge between private returns to investment and social returns, thus reduces private agents interest to invest and consequently reduces the long term growth rate of the economy (Dowrick, 1995). Like in the other economies, Official Development Assistance (ODA) negatively and significantly impact on the growth of the Nigerian economy. The fungibility of ODA in Nigeria, particularly in capital expenditure and as a disincentive to non-export domestic tax has been established elsewhere (Omotor 2010). Fungibility also inhibits diversification and competition.

The DOLS results for Sierra Leone follow the same path as those of Ghana and others. The significant difference of the empirical results relative to other WAMZ selected economies is the striking level of statistical non-significance of the democratic institution coefficient. Sierra Leone, after all, can be aptly described as a post-conflict country.

4.1.3 Post-estimation Analysis

The results of the various post estimation tests are reported in Table 5. The diagnostic tests show that the models are correctly specified as attested to by the Eagle-Granger stationarity of the residual tests. Second, the stability tests of the estimated vector process using the maximum eigenvalue for all the model is also established as all the values stayed within a unit circle. Third, there is absence of autocorrelation in the residuals as attested to by the LM tests. In summary, the empirical analyses are not spurious.

Table 5 Post estimation test

The Gambia

Ghana

Nigeria

Sierra Leone

(a)

(b)

EG residual test

-4.064**

-4.814*

-5.012*

-4.211**

-5.261*

BPG

(p-value)

0.216

0.463

0.408

0.621

0.611

LM test (1)

(p-value)

0.521

0.536

0.711

0.314

0.406

LM test (2)

(p-value)

0.428

0.501

0.488

0.269

0.387

JB

(p-value)

0.146

0.769

0.308

0.117

0.121

Notes: EG Engle–Granger t-test for cointegration, BPG Breusch–Pagan–Godfrey Heteroskedasticity test, LM test Breusch–Godfrey serial correlation LM test, JB Jarque–Bera Normality test

*, **, *** are the significant levels of 1% 4.94, 5% 4.35 and 10% 4.02 levels of significance

Found a mistake? Please highlight the word and press Shift + Enter