The survey explores the cointegration relationship between authorities outgo and economic growing for five ECOWAS states. A panel cointegration attack is used for the dataset for the first clip. From the consequences, Government outgo and economic growing are non cointegrated in both their common and idiosyncratic constituents. Therefore farther analysis on the causalities among the variables could non be assessed. The consequences collaborate recent findings on the failing in utilizing financial policy to explicate economic growing behavior in these development and resource forced states.

Keywords: authorities outgo, economic growing, panel cointegration, unit root, Ghana

JEL Codes: C23, E21, E32, G15

1. Introduction

Though there has been considerable figure of research into the relationship between authorities outgo and economic growing, the inquiry still remains relevant in developing states in the aftermath of the planetary fiscal crisis. Harmonizing to the World Bank ( 2008 ) , at the highs of the crisis, developing states faced a funding deficit of $ 270-700 billion in 2008, as private sector creditors shun emerging markets, and merely one one-fourth of the most vulnerable states have the resources to forestall a rise in poorness. Many of the universe ‘s poorest states have seen their financial grosss decline because of the crisis. This has left many of these states going evermore more dependent on development assistance and exports. Many economic experts have outlined the impact of the crisis on developing states among others, financial shortages and decelerate down in GDP growing. In resource scarce developing states, the function of authorities outgo in work outing market failure is really important in forcing productiveness. In most West African states, authoritiess are the largest Spenders and therefore a freezing on disbursement can hold really terrible effects on economic growing. For case, existent GDP growing in the states of West Africa stagnated at 5.4 % in 2008 as it was in 2007 and has been projected to decelerate down by more one per centum point to about 4.2 % ( African Economic Outlook, 2009 ) .

The function of authorities outgo in exciting economic growing has been widely accepted in the macroeconomics literature since the publications of Wagner in 1890 and Keynes in 1936 severally. The importance of the causal relationship between authorities outgo and economic growing has long been recognized, though empirical estimations have non yielded important and conclusive consequences for developing states. Several surveies in developing states such as Ansari et Al ( 1997 ) , Bagdigen et Al, ( 2003 ) , Dogun and Tang, ( 2006 ) , among others have through empirical observation studied the causal relationship by using conventional unit root and cointegration testing processs for single states. These surveies nevertheless, found small or no long tally causal relationship between steps of authorities outgo and economic growing in most underdeveloped states. With regard to ECOWAS states, old surveies by Iyare et Al ( 2004 ) and Oteng-Abayie and Frimpong ( 2009 ) have found no long tally causal relationship between growing and authorities outgo.

The double aims of this survey is to revisit the issue on the causal relationship between authorities outgo and economic growing utilizing an expanded information set covering five ECOWAS member states and besides to set up the cointegration belongings in more precise footings by using a more robust panel cointegration techniques.

The remainder of the paper is organised as follows. The following subdivision nowadayss trends in the relevant economic information. The econometric methodological analysis of the survey and informations are described in subdivision 3. The empirical consequences follow in subdivision 4. Section 5 concludes the paper with suggestions of policy deductions.


2.1 Model Specification

There are assorted preparation of the theoretical relationship between authorities outgo and economic growing. Based on Mann ( 1980 ) , besides cited by Iyare et Al ( 2004 ) among others, we formulate a simple log-linear theoretical account to prove the long tally relationship between outgo and growing theoretical account as follows:

ln ( GEX ) = A0 + A1ln ( PCGDP ) it + eitaˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦ . [ 1 ]

where GEX is log of authorities outgo as a per centum of GDP and PCGDP is log of existent per capita income.

2.2 Datas

Annual clip series informations on the log of authorities ingestion expenditures as per centum of GDP ( GEX ) measured and log of per capita Gross domestic merchandise ( PCGDP ) for The Gambia, Ghana Guinea, Nigeria and Sierra Leone from the African Development Indexs published by the World Bank were used. The information for is for the period 1965 to 2007. The tendency secret plans ( Figure 1 and 2 ) of the two variables of involvement show variable tendencies over clip in the assorted states. Real GDP Growth rates for the selected states show a tendency towards convergence about 5 % with the exclusion of Guinea which is diverging. The financial index which is the balance between outgo and gross in general besides shows a inclination towards long tally convergence. With respect to the individual currency policy being pursued by these states as West African Monetary Zone, this is a really satisfactory image.

Figure 1: Real GDP growing ( % )

Beginning: African Development Indicators, 2009

Figure 2: Fiscal Balance Indicators ( % of GDP )

Beginning: African Development Indicators, 2009

2.3 Econometric Methodology

The late developed panel econometric techniques has become popular for several grounds. First, it is designed to command for cross subdivision dependences and secondly to better on the power of the appraisal by working the commonalties of the states under probe. In peculiar, a long tally equilibrium between authorities outgo and economic growing may happen due to the being of regional or national tendencies, or both.

To look into these issues, each variable is separated into common and idiosyncratic constituents, as suggested by Bai and Ng ( 2004 ) . Cointegration between the series can happen merely if the common factors of the variables cointegrate. If both the common factors and idiosyncratic constituents are I ( 1 ) , cointegration is examined individually for the common and the idiosyncratic constituents. If the common factors are I ( 1 ) , but the idiosyncratic constituents are I ( 0 ) , the nonstationarity in the panel could be wholly driven by a decreased figure of regional stochastic tendencies. This would be the instance of cross member cointegration. Suppose that the series Y and X have a individual I ( 1 ) common factor, i.e.

aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦.. ( 2 )

aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦ . ( 3 )

where F denote the common factors and E the idiosyncratic elements of the several variables. A panel cointegrating relationship between Y and X

aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦aˆ¦ ( 4 )

requires that the nothing of no cointegration is rejected for both the common and the idiosyncratic constituents. Cointegration between common factors can be examined by the usual clip series trials such as the Johansen reduced rank attack. As the idiosyncratic constituents are independent by building, their analysis is done by standard panel trials such as those of Pedroni ( 1999, 2004 ) . It should be noted, nevertheless, that the being of cointegrating relationships that wipe out both the common and idiosyncratic tendencies is really improbable, see equation ( 4 ) .

As with clip series cointegration, estimates from a cointegrated panel are robust to a assortment of jobs that frequently plague empirical work, including endogeneity, omitted variables, and measurement mistake ( Phillips and Moon, 2000 ; and Baltagi and Kao, 2000 ) . Furthermore, panel cointegration techniques can be implemented with shorter informations spans than their clip series opposite numbers and statistical illation is simplified because restricting distributions are standard normal. Evans ( 1998 ) refers to panel cointegration as a stable long-term relation between the dependant variable and the independent variables in states in a panel. This construct is distinguishable from cross-member cointegration whereby the dependant variable in one state is cointegrated with the dependent variable of another state in the panel.

We make usage of both the clip series ( old ages ) and the cross sectional ( state degree ) information of the information set. Econometric testing is performed in two stairss. Before gauging the long-term cointegratiion relationships, the relevant variables are pre-tested for unit roots and cointegration. Recent literature suggests that panel-based unit root trials have higher power than traditional ADF and Philips-Perron unit root trials based on single clip series. Though these trials are normally termed “ panel unit root ” trials, theoretically, they are merely multiple-series unit root trials that have been applied to panel informations constructions ( Eviews 6 User ‘s Guide, 2007 ) . First, We used the panel unit root trials as proposed by Levin et Al ( 2002 ) and Im et Al ( 2003 ) . The Levin et Al ( 2002 ) and Im et Al ( 2003 ) panel unit trial consequences are presented in Table 1.

Second, we tested for cointegration. A figure of panel cointegration trials have been proposed in the econometrics literature, correspondent to clip series cointegration trials ( Pedroni, 2004 ) . Panel cointegration trials, can be done as pooled ( within dimension ) trials and as group mean trials, matching to the panel parallels of the time-series rho-statistic, t-statistic, and ADF trials ( Phillips and Perron, 1988, Phillips and Ouliaris, 1990, and Dickey and Fuller, 1979 ) . These are reported in Table 2.


3.1 Panel Unit Roots Test

Panel unit root trials are applied to prove for stationarity of the clip series. From the Levin, Lin & A ; Chu t group unit root and Philips-Perron single unit root process, the hypothesis of a unit root in the degrees of all variables for all states could non be rejected ( Table 1 ) . The hypothesis is nevertheless rejected after first differencing, bespeaking that all variables are stationary in their first log foremost differences. These consequences are robust to the trial method and the exogenic variable specification.

Table 1: Panel Unit Root Test


First difference


Prob value**


Prob value**

Per Capita GDP ( PCGDP )

Levin, Lin, and Chu t-test*





Im, Pesaran, and Shin W-test





ADF-Fisher chi-square trial





PP – Fisher Chi-square





Government Expenditure as % of GDP ( GEX )

Levin, Lin, and Chu t-test*





Im, Pesaran, and Shin W-testi‚?





ADF-Fisher chi-square testi‚?





PP – Fisher Chi-squarei‚?





*Null: Unit of measurement root ( assumes common unit root procedure ) i‚? Null: Unit of measurement root ( assumes single unit root procedure ) . **Probabilities for Fisher trials are computed utilizing an asymptotic Chi-square distribution. All other trials assume asymptotic normalcy.

3.2 Panel Cointegration

Following from the consequences of the panel unit root trials, we were able to prove for the cointegration relationship between the variables among the five pooled states utilizing the panel cointegration technique discussed earlier. A lag length of 2 was chosen for the trial based on SBC information. From Table 2, the consequences suggests that a stable long tally cointegration relationship between authorities outgo and economic growing can non non be estimated. The assorted cointegration trials did non bring forth consentaneous important consequences. For case, Johansen Fisher panel cointegration trials ( hint and maximal Eigen ) were non important at the traditional 5 per cent degrees. The Kao residuary ADF cointegration trials statistics was besides non important. With regard to the Pedroni residuary cointegration trials, merely the panel ADF and PP – statistics were important. However, the group ADF and PP statistics were non important.

These consequences supports earlier findings from surveies by Ansari et Al ( 1997 ) , Iyare et Al ( 2004 ) , and Oteng-Abayie and Frimpong ( 2009 ) . Thus the hardiness of the techniques adopted in this survey does non demo any important difference in the consequences obtained in the old surveies. To this point the survey did non happen it necessary to gauge a long tally dynamic panel arrested development for the theoretical account in conformity with Baltagi. .

Table 2: Panel Cointegration

Test statistic


Pedroni Residual Cointegration Trials

Panel v-Statistic


A 0.3989

Panel rho-Statistic

A 1.867458

A 0.0698

Panel PP-Statistic

A 2.585236

A 0.0141*

Panel ADF-Statistic

A 2.078238

A 0.0460*

Group rho-Statistic

A 1.881744

A 0.0679

Group PP-Statistic

A 1.838432

A 0.0736

Group ADF-Statistic


A 0.3219

Kao Residual Cointegration Trials



A 0.1711

Johansen Fisher Panel Cointegration Test

Fisher stats ( hint )

A 17.28

A 0.0684

Fisher stat ( max-eigen )

A 17.28

A 0.0684

Probabilities are computed utilizing asymptotic Chi-square distribution

Therefore the grounds attributed to the decisions that have been arrived at in Ansari et Al ( 1997 ) and Oteng-Abayie and Frimpong ( 2009 ) could be sustained.

4. Decision

This paper through empirical observation analyzed the cointegration relationship between authorities outgos and economic sciences growing utilizing the advanced method of panel unit root and panel cointegration. A nonstationary panel informations analysis is an appropriate method of analysis in parts, such as Sub-Saharan Africa, where it is hard to roll up a sufficiently a big sample size. The empirical consequences revealed that there is no long tally cointegration relationship between authorities outgos and economic growing among the five ECOWAS states consisting The Gambia, Ghana Guinea, Sierra Leone, and Nigeria over the period from 1986 and 2004. Further analysis on causalities were hence non necessary. The findings could bespeak that authorities expenditures do non play a important function in advancing economic growing in the ECOWAS states in our survey.

The ability of authoritiess in the states studied to increase revenue enhancements beyond a certain degree prevent authorities are challenged. Government financial policy through disbursement cuts may non hold the necessary direct impact on economic growing in these states. For case, in Ghana, cuts in authorities disbursement are expected to decelerate growing in productiveness. Therefore in ECOWAS states, financial policy is improbable to present growing impact without important external budgetary support. Most significantly, the findings of this survey are possible if non-economic factors are more of import in explicating the growing of authorities outgo than economic factors.