GOVERNMENT CONSUMPTION EXPENDITURE S AND THE CURRENT [614738]

GOVERNMENT CONSUMPTION EXPENDITURE S AND THE CURRENT
ACCOU NT IN NIGERIA

Abstract

This study contributes to the body of knowledge by examining the effect of government
consumption expenditure s on the current account balance in Nigeria , covering the p eriods
from 1970 to 2018. The study answers the question: Does government consumption
expenditure s affect the current account balance in Nigeria? Employing the Autoregressive
Distributed Lag (ARDL) bound cointegration testing approach for evaluation, the u nit root
test results shows a mix level of stationarity. The ARDL results showed that the government
consumption expenditure s and total debt burden contributed significantly to the current
account deficit in Nigeria . The error -correction coefficient of the ECM( -1) term is negative as
expected , indicating that the gap between long run equilibrium value and the actual value of
the dependent variable is corrected with speed of adjustment equal to 1.06 percent annually.
The study recommends reduction in Govern ment borrowing in order to reduce debt servicing
affecting the country’s current account.

Keywords: Government consumption, Current account, Bound cointegration, ARDL , Nigeria
Jel Classification: B22, C13, F32, H50, N47

1. Introduction

There has been an extensive research conducted without a conclusive evidence , on how
changes in government consumption expenditure s affect the current account balance of
countries. The views of the neoclass ical economists are that government spending
expend itures on goods and services affects both output and investments positively , while
spending based on worked hours negatively affect the two variables (see Baxter and king,
1993). For example, Baxter (1995) asserts that as the final consumption expenditures
increases by a 1 %, it results to a 0.5% decline of the current account balance . While
evaluating the United Kingdom’s economy, Ahmed (1987) noted a negative relationship
between increases in expenditures of the government and the balance of trade in the 1 8th and
19th centuries.
Also , Yi (1993) found that purchase of the government contributed immensely to the
decline of the economy of the United States ( US) in the 1970s and 1980s . According to
Abbas, Bougha -Hagbe, Fatas, Moun & Valleso (2011) , the questio n most economies have
asked overtime are, to what extent through which adjustment of fiscal policy would result to
resolving problems relating to external balances. Oseni &Onakoya (2013) were of the view
that issuance of bonds by the government, raises the rate of interest, leading to higher
consumption expenditure as a result of wealth effect. High interest rate leads to currency
appreciation, and as loss in competitiveness sets in , coupled with higher marginal propensity
to consume, leads to deterioration of the current account (Oseni & Onakoya, 2013) .
Government consumption expenditure is t he values of both goods and services utilised
by various agencies, departments and institutions of g overnment s, in the provision of goods
and services to the populatio n (Cavallo, 2005) . This type of goods and services which include,

national defence, administrative activities are not easily provided by private business owners.
On the other hand, current account balances of countries are the differences between an
econ omy’s export and imports of goods and services, as well as income. It also includes the
differences between an economy’s national income and domestic investments and
consumption. A greater amount of local expenditure especially the in advanced economies are
mainly government consumption expenditure s (Cavallo, 2005) .
In Nigeria, during in 1970, expenditure of the government was estimated at a billion
dollars ; this placed the country 11th in the global rankings . Also, the country’s share to the
world’s total on government consumption expenditures was 1.2 %. After about a decade, in
1980 the expenditure of the rose to 68.8 billion dollars, placing the country 8th in the global
rankings . This brings a share of the consumption expenditure of the government to 3. 1% in
the world. In 1991 , consumption expenditure of the government was 4.83% of the Gross
Domestic Product ( GDP ) and in 1994, the consumption expenditure rose to 17.94 % of the
GDP. The country spent a total o f 2.9 billion dollars in 1999, placing the coun try 68th in the
world. The government final consumption expenditure in 2003 was -23.926% , while in
2004 , it reached to 565.59% .
In 2013, the Nigeria n government spent a total of 41.6 billion dollars , thus placing the
country 38th in the global rankings , and a share to the global consumption expenditure s to
0.31% . The final consumption expenditure s in 2015 and 2016 was -11.897% and –
15.116% respectively . The final consumption expenditures as percentage of the GDP in
2017, was 4.6% of the GDP and 22.3 billion US dollars, in 2018, about 33.16% of the
GDP (The World Bank, 2019) . The figure 1 below shows the Nigerian government
consum ption expenditure s from 1970 to 2018.

Figure 1: The consumption expenditures of Nigeria from 19 70 -2018

10000 15000 20000 25000 30000 35000 CEXP
1970 1980 1990 2000 2010 2020
YEAR
Source: Extracted from the Central Bank of Nigeria (CBN) statistical bulletin 2018

From the fig ure, it could be seen that consumption expenditure s of Nigeria was on a
constant increment from 19 70 to 1980 ; it however decreased between 1987 and 1990 . From
the figur e, the consumption expenditure s of the government continued to in crease on higher
scale from 2004 to 2018.
Since the study conducted by Cavallo (2005) on the effect of government consumption
expenditure s for the US economy, fewer studies had focused on this area of research. As
studies on the effect of fiscal policy on current account balance continued to attract debates
among researchers without conclusive evidence, theoretical debates on current account
balances explaining variations in fiscal measur es, had centered on macroeconomic models .
Among the models include, the Mundell –Fleming and Ricardian Equivalent models. In the
Mundell –Fleming approach, fiscal deficit causes current account deficit , while the Ricardian
equivalent hypothesis posit it that deficit financing , either through reduced taxes , or by
issuing bonds does not change the present value wealth of individual households, since both
temporarily reduced taxes and issuance of bonds represent future tax liabilities (Kim,1995 ;

Oseni & Onakoya, 2013). Following the twin -deficit hypothesis, it is expected that recorded fiscal
deficit is accompanied by huge current account deficit (Kim & Roubini, 2004).
On the other hand, the Keynesian principle suggest that increase s in fiscal de ficit as a
result of expenditure increment, households will be compelled to utilize s additio nal income
to boost consumption , thereby causing national saving, both public and private to decline
(Ekundayo, 2011). The implication of this, is that the country would resort to borrowing, unless
the local investment activities decreases enough to offset the short fall in the saving rate. Nonetheless,
the Keynesian approach centred on the fact that , there is a positive interaction effect between fiscal
deficits and private cons umption, implying that higher deficit results to a higher private consumption
expenditures. This view point however may not be realistic according to Briotti (2005) , based on the
literature of expansionary effect of fiscal consolidation
It therefore beco mes crucial to understand the current account dynamics especially in the
Sub-Saharan African economies, as the performance of the economies of the region in the recent past
has been discouraging. For instance, since Nigeria n government embarked on th e Stru ctural
Adjustment Program me (SAP), geared towards liberalization of the economy, it has in practice led to a
decline of the economy, and further worsen ed the current account im balances. Thus understanding
current account dynamics, might lead to a mitigatio n of these effects ; though it is still unclear what the
current account determinants are , and also, not yet if the imbalances are sustainable. Equally, the
burden of debt service repayments , has also contributed to the huge c urrent account deficits of
the country . As an import dependent economy, consumption expenditure s which constituted
greater part of the total expenditures , and a greater share of the total output, has resulted to a
huge trade deficit for the country.
The questions this study tends to a nswer is therefore stated as follows: Does
government consumption expenditure s affect the current account balance in Nigeria? What
are the other variables that affect current account balance in Nigeria? The overall objective of

this research is therefore to evaluate the effect of government consumption expenditure s on
the cur rent account balance in Nigeria

2. Literature Review
In the literature, it is argued that that there are opposite effects of the consumption
expenditures on goods and hourly worked o n a key macroeconomic variables which include,
private investment and the total output. These variables are the key determinants of current
account (Cavallo, 2005). The model of the Neo-Classical model economy posits that there is
positive effect between ex penditures on goods, output and investment; and a negative effect
between hourly worked and the variables (Baxter & king, 1993 ; Finn, 1998 ). However,
Baxter (1995) noted that expenditures on hourly worked , are smaller in magnitude than the
goods expenditu res on current account. According to the author, a 1% increases in the hourly
worked expenditure, decreases the current account to 0.5% of the GDP.
Equally, increases in government spending on goods is accompanied by an increases
in imports, while an incr eases in hourly worked is accompanied by an increase in domestic
supply of labour. Increases in the rate of import leads to trade decline and current account
deficit, while increases in the domestic labour supply have little impact on services, income
and net exports (Baxter, 1995). Since hourly worked is a non -traded goods, an increase in
hourly worked have no direct impact on local expenditures goods, and therefore does not
impact directly on current but rather, an indirect effect. Thus, it is important t o firstly evaluate
consumption expenditure changes while assessing its effect on current account (Cavallo,
2005) .

2.1. Fiscal policy effects on current account balance

Studies on fiscal policy effects on current account balance according to Abbas et.al.
(2011) , includes effect on consumption , tradable goods demand and investment. Large part of
the local demand for goods are always accounted for the government, thus depending on
propensity of import, a change in the demand for import by the government lead s to a
movement in the balance of trade. Also, changes in the rate of exchange affects current
account by altering the relative prices of non -tradable goods (Abbas et.al. 2011) . An increased
government spending on non -tradable goods like real estate sector s or services, can lead to
exchange rate appreciation , thereby tilting forward private consumption and production away
from tradable goods . Other channel through which fiscal policy affects current account
include, country ’s risk premia and rate of interes t (Abbas et.al, 2011

2.2. Government expenditures as a measure of size of government
In the literature, there exist several ways through which the size of government is
measured. According to Alesina, Glaeser & Sacerdote ( 2001), these measures are typi cally
based on, either total expenditure of the g overnment , or general consumption expenditure s.
The authors were of the view that the relative measure of the government size is the
expenditure ratio of government and that depicts how the government utiliz es local resources.
However, Leonard (1986) argued that budget might not actually state clearly , the true size of
since it may not account for the government ‘quiet size’ activities. Government role in an
economy therefore goes beyond collection of revenue s and spending (Leonard, 1986) .
2.3. A cross country studies on government consumption expenditure
Cross country studies on government consumption expenditure shows that in 2013, the
Nigerian government spent a total sum of 41.6 billion dollars on consumpti on expenditure s.
This amount bring s the country on par with other countries like Greece 41.6 billion dollars ,
the Czech Republic 41 billion dollars , Ireland 40.6 billion dollars , Algeria 39.7 billion dollars ,

Malaysia 42.3 billion dollars and Portugal 4 3.1 billion dollars respectively. Also, the
country’s consumption expenditures far exceeded that of the neighbouring countries . For
instance, consumption expenditures for counties like Camero un in the same year 2013, was
3.4 billion dollars , Chad 2.6 billi on dollars, Republic of Niger , 1 billion dollars, and the
Republic of Benin 0.93 billion dollars respectively. As a percentage of the GDP, t he Nigeria’s
consumption expenditures in 2015 and 2017 was 5.94% and 4.6% respectively. The
consumption expenditur es as a percentage of the GDP in 2017 for Ghana was 8.8 %, Niger
15.8%, and Cameroun 11.5% respectively (The World Bank , 2019) .

2.4. Empirical literature
Bhavesh & Prabheesh (2017) investigate macroeconomic and external factors that
affect current account balance in India from 1997 to 2012. The study show that twin deficit
hypothesis that fiscal deficit reduction amelior ate the current account deficit applies. Ișık,
Yılmaz & Kılınç . (2017) evaluate the relationship between current account balance and credit
to firms in 26 Organization of Economic Cooperation and Development (OECD) countries
from 2005 to 2015. The study results indicate that credit to households and firms have
negative effect on current account balance in the short run while credit to firms and
government have positive effect in the long run .
Klemm & Iakova(2013) investigate growth following investment and consumption
driven current account crises for industrialized countries. The results show that a deficit
reversal that follows investment boom are identified with better growth than those following
consumption booms. The study find that large current account deficits are as a result of
consumption or non-productive investment booms. Also, Oseni & Onakoya (2013) examine
fiscal policy effect o n current account balance in Nigeria from 1980 to 2010. The study
employ Structural Vector Auto Regression (VAR) model approach. The result show that

expansionary fiscal policy has a positive effect on the output and exchange rate and
negatively impact on the current account balance
Nurudeen & Usman (2010) evaluate the effect of government expenditure and the
growth of the Nigerian economy from 1970 to 2008 using cointegration and error correction
model techniques. The study results show that government cap ital expenditure, recurrent and
education expenditures indicate a negative effect on economic growth. Nickel and
Vansteenkiste (2008) investigate the relationship between fiscal policy variables and current
account balance of 22 advanced economies . The panel data results show that countries with
debt to GDP ratio of 90% indicate a positive relation between fiscal policy variables and
current account balance. The result also indicate that for 11 largest Euro area economies , a
statistically insignificant result exists between fiscal policy variables and current account
balance .

3. Methodology of Research
3.1. Data analysis method
The study adopts an Auto Regressive Distributive Lag (ARDL ) model approach for
evaluation. The current account is the dependent variable, while government final
consumption expenditure s, total debt , exchange rate , interest rate and trade openness, are the
independent variables. The model for the effect of government consumption expenditure s on
current account in Nigeria is therefor e specified as follows:
  F , , , , CAB CEXP TDEBT EXCH INTR OPEN
……………………………… …….(1)
where
CAB = The current account balance
CEXP = The Government consumption expenditure
TDEBT = Total government debt

EXCH = Excha nge Rate
INTR = Interest rate
OPEN= Trade openness

In order to establish a linear relationship between the dependent variable and the
independent variables , an econometrics model is t herefore specified as follows:
0 1 2 3 4 5 t = + + + + + + CAB CEXP TDEBT EXCH INTR OPEN      
…………….……(2)
where
t
= Error term
0
= the constant term

’s = the parameters to be estimated

3.2. Method of result evaluation
This study utilizes the Augmented D ickey Fuller (ADF) test for stationarity test. This
is because, the Auto Regressive Distributed Lag (ARDL) bound test procedure is conducted
based on the assumption that , the variables under consideration are either integrated of order
zero or order one. H ence an order of integration was determined applying the ADF test
procedure s. The null hypothesis that guided the test is stated as follows: H 0: β=0 (β has a unit
root); H 1: β<0 (for alternative hypothesis). According to Pesaran, Shin & Smith (2001), the
variables under consideration cannot be integrated of order two to avoid a spurious regression .
In conducting an ARDL model procedure , a choice for an appropriate lag length is provide d.
Pesaran & Shin (1999 ) were of the view that Akaike info criterion (AIC) and Schwarz
criterion (SC) are appropriate in a small sample, though SC is superior to AIC : The ARDL
model is written as follow:

n n n n n n
t 0 1 t-1 2i 1t-1 3i 2t-1 4i 3t-1 5i 4t-1 6i 5t-1
i=1 i=0 i=0 i=0 i=0 i=0
n
7i t-1 8 t-1 9 t-1 10 t-1 11
i=0CAB = + CAB CEXP + TDEBT + EXCH + INTR + OPEN
+ CAB + CEXP TDEBT EXC Hi
INT      
           
       
 t-1 12 t-1 OPENt R…… (3)
where
Δ = Difference operator
𝜀𝑡 = Stochastic term
Conducting ARDL bound test, an Ordinary Least Square (OLS) is estimated firstly in
order to establish if there exists a long run relationship between the variable under
consideration. The test is based on an F-Statistic for the joint statistically signif icance of the
lagged variables.
The null hypothesis of no cointegration (
0 7 8 9 10 11 12:0H           ) is
evaluated using Pesaran et.al. (2001) procedure. The underlining assumption is therefore
stated as follow s: if the F -statistic is > upper critical bound, H 0 is reje cted and concludes that
the variables under consideration are cointegrated, otherwise it is not accepted. However, if F –
statistic ≥ lower critical bound ≤ upper critical bound, then decision becomes inclusive. If the
null hypothesis of n o cointegration is rejected, a vector error -correction model (VECM) is
therefore estimated (Narayan & Narayan, 2006). The VECM model is therefore specified as
follows:
n n n n
t 0 1 t-1 2i 1t-1 3i 2t-1 4i 3t-1
i=1 i=0 i=0 i=0
nn
5i 4t-1 6i 5t-1 t
i=0 i=0CAB = + CAB CEXP + TDEBT + EXCH +
+ + ECM( -1)+i
INTR OPEN    
       
   

.(4)
where:
ECM( -1) = The error correction term
𝜆 = the error c oefficient

4. Presentation and Analysis
The trend of the current account balance in Nigeria from 19 70 to 2018 is therefore
presented in the figure 2 below:
Figure 2: The trend of the current account balance (CAB) in Nigeria from 19 70- 2018

-10
010 20CAB
1970 1980 1990 2000 2010 2020
YEAR
Source: Extracted from the Central Bank of Nigeria (CBN) statistical bulletin 2018

A look at the figure of the c ountry’s current account showed a record low in 19 78 as
the figures are close to negative. There was an upward spike from 1981 to 1996, w ith a
surplus for Nigeria in 1990s . The figure showed current account balance to be fluctuating in
the period from 2000 to 201 0; also, t he country recorded a deficit in 2015 and a surplus in
2016 and 201 8 respectively .
4.1. Descriptive statistics
The descr iptive statistics for the variables under consideration are therefore presented in table
1 below:
Table 1: The Descriptive statistics
CAB CEXP TDEBT EXCH INTR OPEN
Mean 2.064350 23921.83 1725203. 72.68884 9.945877 62.90421
Median 1.716712 25726.02 1037296. 21.88610 10.46250 74.60000
Maximum 20.73932 36012.05 6260595. 365.0090 26.90000 97.30000
Minimum -10.27790 11976.64 1252.900 0.544500 2.500000 19.60000
Std. Dev. 5.589969 5988.234 1862432. 94.84836 5.263218 21.760 45
Skewness 0.936028 -0.360192 0.661268 1.457820 0.793408 -0.540000
Kurtosis 4.783077 2.310211 2.148752 4.787226 3.715457 1.814041
Jarque -Bera 13.64641 2.030975 5.050525 23.87757 6.185975 5.253008
Probability 0.001088 0.362226 0.080037 0.000007 0.045366 0.072331
Sum 101.1531 1172169. 84534963 3561.753 487.3480 3082.306
Sum Sq. Dev. 1499.892 1.72E+09 1.66E+14 431818.1 1329.670 22728.83
Observations 49 49 49 49 49 49
Source: Authors’ compilation from the Eviews results

From the table above , the descriptive statistics indicated that from 19 70 to 201 8, all
the variables under study showed an averaged positive mean values with 49 observations. The
standard deviation indicated that the highest standard deviat ion is recorded by the variable
CAB, while the least standard deviation is recorded by variable INTR . The Jarque -Bera (JB)
test of normality for the variables und er consideration revealed that five of the variables are
significant at 5% level.
4.2. Unit root test
The results of the Augmented Dickey Fuller (ADF) unit root test obtained are
presented in table 2 below as follows:
Table 2: The ADF Unit root test
Variable Level
difference Prob. First difference Prob. Order of
integration
CAB -3.500394 ** 0.0122 I(0)
CEXP -3.014496** 0.0406 I(0)
TDEBT -1.498503 0.5257 -4.996533** 0.0002 I(1)
EXCH 1.717017 0.9995 -4.041700** 0.0028 I(1)
INTR -2.493807 0.1233 -8.479398** 0.0000 I(1)
OPEN -1.913846 0.3234 -8.479398** 0.0000 I(1)
Note: * , ** and *** denote 1% ,5% and 10% critical values respectively
Source: Authors’ compilation from the Eviews results

From the table above , the results showed that two of the variables , CAB and CEXP
are stationary at level at 5% significant level s. The other variables , TDEBT, EXCH , INTR
and OPEN are stationary at first difference I(1) at 5% significant levels in the ADF test
procedure respectively. Therefore, s ince there exist a mix order of stationarity among the
variables , an ARDL model Procedure is therefore conducted . Oteng -Abaiye & Frimpong
(2006) stated that instead of utilising Johanson cointegration approach which requires that all
variables are integrated o f order one I(1), an ARDL that incorporates variables that are I(0)
and I(1) are therefor e utilised.
4.3. Lag Length selection
To determine the lag order of the estimated model, we evaluated several order
selection criteria. From estimated lag length criteria results (see table 4 below) , the Akaike

Information Criterion (AIC), the Schwarz Cr iterion (SC) and Hannan -Quinn information
criterion (HQ) indicated 4 lag. Therefore we utilized the AIC and SC test results and
estimate d the ARDL model with a (4) lags.
Table 3: lag length selection for the model
Lag LogL LR FPE AIC SC HQ
0 -1383.021 NA 2.03e+30 86.81383 87.08866 86.90493
1 -1215.202 262.2172 5.61e+26 78.57514 80.49892 79.21282
2 -1181.830 39.62922 8.53e+26 78.73940 82.31213 79.92366
3 -1099.125 67.19788* 1.00e+26 75.82033 81.04201 77.55117
4 -1021.558 33.93570 6.29e+25* 73.22237* 80.09301* 75.49979*
* indicates lag order selected by the criterion
Source: Authors’ compilation from the Eviews results

4.4. Auto regressive distributed lag (ARDL), bound testing approach
The study tries to establish whe ther there exist a long run relationship among the
variables under study by conducting bound cointegration test procedure. An F -statistic to test
the long -run relationship is therefore computed. A maximum lag length of 4 is adopted. The
result of ARDL bou nd cointegration test is presented in table 4 below:

Table 4: The ARDL Bound test result
ARDL Bounds Test
Null Hypothesis: No long -run relationships exist
Test Statistic Value k
F-statistic 6.081641 5
Critical Value Bounds
Significance I0 Bound I1 Bound
10% 2.26 3.35
5% 2.62 3.79
2.5% 2.96 4.18
1% 3.41 4.68
Source: Authors’ compilation from the Eviews results

The ARDL bound test result from the table 4 above indicated that the F – Statistic
value is greater than the critical value bounds at both 1%, 5% and 10% significant levels
respectively. We can see that the value of 6.081641 is higher than 2. 62 and 3.79 critical
values at 5% significant levels . The hypothesis of no long run relationship is therefore
rejected. Thus it showed that ARDL model estimate d, is proven to determine the long -run
slope -estimated coefficients and the short -run dynamic -estimated coefficients. Hence there
exists a long run relationship between CAB, CEXP, TDEBT, EXCH, INTR and OPEN

4.5. The estimated short run ARDL Regre ssion model
The ARDL (1, 4) is selected based on Akaike info criterion.

Table 5: The short run error correction dynamics

Source: Authors’ compilation from the Eviews results

From the result table 5 above, the coefficient of the variables, CAB ( -1) is positively
signed and significant statistically. It indicates that the lag value of the dependent variable
contributes positively to the growth of th e current account balance in Nigerian . The
coefficient of the variables CEXP and TDEBT , indicate neg ative signs and are significant
statistically. The negat ive coefficient of the variable CEXP implied that the consumption
expenditure contributed significantly to the decrease in the current account balance of Nigeria
in the short run . The finding conform to the study by Cavallo (2005) that consumption
expenditures of the governments overestimates its role while accounting for current account
movement . Also, t he result of TDEBT conforms to the findings of Ramanan (2017) that
current account deficit leads to fall in demand, lower output and income and this translates
into huge government deficit and in turn affect the public debt.
The coefficient of the variable INTR indicate a positive signs and is significant
statistically at 5% significant level . A rise i n interest rate as the Central bank buys domestic
currency , leads to a liquidity squeeze , thereby contributing to high interest rate. The result
conform to the findings by Spiro (1997) , that testing fiscal policy and interest rate in Canada
have been uns uccessful as the interaction between fiscal policy and current account seems to Vari able Coefficient Std. Error t-Statistic Prob.
D(CAB( -1)) 0.519540 0.182469 2.847272 0.0103
D(CEXP) -0.000320 0.000164 -1.955314 0.0654
D(TDEBT) -0.000004 0.000002 -2.169548 0.0429
D(EXCH) 0.019264 0.041366 0.465705 0.6467
D(INTR) 0.541786 0.229532 2.360392 0.0291
D(OPEN) 0.131376 0.139391 0.942499 0.3578
ECM( -1) -1.063659 0.242426 -4.387560 0.0003
R-squared = 0.798349
Adjusted R -squared = 0.533018
F-statistic = 3.008879
Prob(F -statistic) = 0.008291
Durbin -Watson stat= 2.034288

be indirect and thus are influenced by monetary policy. The coefficient of EXCH shows a
positive sign and i nsignificant statistically. Heera (2014) noted that once there is a b reech in
exchange rate comfort level, the apex bank intervenes by selling foreign exchange in order to
curb local currency depreciation.
However, Pettinger (2017) stated that the Marshall Lerner condition asserts that
exchange rate depreciation leads t o current account improvement. It also leads to an increase
cost of import purchase, thus leading to low demand fo r imported goods and services, thereby
reducing current account deficit. Also, the coefficient of the variable OPEN indicates a
positive sign and significant statistically at 5% level. The result shows that the level of trade
openness of the economy leads to a current account balance improvement. The result
conforms to the finding of Khan, Jaffri & Abbas (2017) that tariff reduction improves
balance of trade in the short run while non -tariff barriers leads to deterioration of balance of
trade in the long run.
The error -correction term, ECM( -1) from the ARDL model estimate is negative as
expected with a value of -1.063659 and is statistically significant. This implies that there is a
possibility for cointegration and therefore the existence of a long run relationship among the
variables. We can therefore conclude that the gap between long run equilibrium value and the
actual value of the depende nt variable CAB is corrected with speed of adjustment equal to –
1.06 percent annually. It also showed that there exists a ba ckward move towards equilibrium.
The coefficient of determination
2R indicates a value of 0.798349 and the ad justed
2R
with a value of 0.533018. This shows that 53% of variatio ns in the dependent variable (C AB)
are exp lained by independent variables . The F -statistics results indicate that the overall model
is significant with a value of Prob(F-statistic) = 0.008291; while the Durbin -Watson (DW)
statistics value of 2.034288 indicates absence of serial correlation in the model under
consideration.

4.6 Testing for the long run relationship between government consumption expenditure
and curr ent account balance in Nigeria
The long run model for the relationship between consumption expenditure and current
account balance is therefore presented as follows:

Table 6: The long run relationship result
Long Run Coefficients
Variable Coefficient Std. Error t-Statistic Prob.
CEXP -0.000254 0.000226 -1.123377 0.2753
TDEBT 0.000006 0.000001 4.623973 0.0002
EXCH -0.132058 0.037141 -3.555569 0.0021
INTR -0.548237 0.312123 -1.756476 0.0951
OPEN 0.087151 0.088138 0.988812 0.3352
C 5.492530 7.6547 91 0.717528 0.4818
Source: Authors’ compilation from the Eviews results

From the results, it showed that the estimated coefficient s of the long -run relationship
are significant for TDEBT, EXCH and INTR , and insignificant for CEXP and OPEN. The
estimated coefficients are positive for TDEBT, and OPEN, and negative for CEXP , EXCH
and INTR. This implies that the country’s debt and degree of openness of the economy
positively affected the current account balance , while consumption expenditure, exchange rate
and interest rate indicate a negative relationship with the current account in the long run
during the period under review.
4.7. Diagnostic test
Under the diagnostic tests conducted, the Normalit y test results is presented in figure 3
below, shows t hat th e average mean value is -4.00e-15 and a median value of 0.364506 .

Figure 3: The normality test result

0246810
-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5Series: Residuals
Sample 1974 2018
Observations 45
Mean -4.00e-15
Median 0.364506
Maximum 4.741840
Minimum -6.158739
Std. Dev. 2.288164
Skewness -0.358698
Kurtosis 2.999733
Jarque-Bera 0.964980
Probability 0.617244
Source: Authors’ compilation from the Eviews results

The maximum value is 4.741840 and the minimu m value -6.158739 . Evidence from
the Jarque – Bera (JB) test indicates that the null hypothesis is therefore rejected and the error
terms of model are normally distributed. Equally, the serial correlation and Heteroskedasticity
tests results as presented in the table 7 below, shows that the null hypotheses of no serial
correlation is accepted and the Heteroskedasticity test results indicate that the error term of
the variables are homoscedastic.
Table 7: The diagnostic tests results
Breusch -Godfrey Seri al Correlation LM Test:
F-statistic 0.068464 Prob. F(1,18) 0.7966
Obs*R -squared 0.170511 Prob. Chi -Square(1) 0.6797
Heteroskedasticity Test: Breusch -Pagan -Godfrey
F-statistic 0.963562 Prob. F(25,19) 0.5418
Obs*R -squared 25.15737 Prob. Chi-Square(25) 0.4536
Scaled explained SS 4.484245 Prob. Chi -Square(25) 1.0000
Source: Authors’ compilation from the Eviews results

4.8. Stability test
The stability test procedure is conducted using the Cumulative Sum of Recursive
Residuals (CU SUM) and the Cumulative Sum of Squares of Recursive Residuals
(CUSUMSQ) in line with Durbin, Brown and Evans (1975). To ascertain if the ARDL
estimated results are stable, the CUSUM and CUSUMSQ test results are presented in figures
4 and 5 below:

Figure 4: The CUSUM test result

-15-10-5051015
2000 2002 2004 2006 2008 2010 2012 2014 2016 2018
CUSUM 5% Significance
Source: Authors’ compilation from the Eviews results

Figure 5: The CUSUM of squares test result

-0.40.00.40.81.21.6
2000 2002 2004 2006 2008 2010 2012 2014 2016 2018
CUSUM of Squares 5% Significance
Source: Authors’ compilation from the Eviews results

From t he figures 4 and 5 above , the critical bounds within the straight line s are
significant at 5% level. The CUSUM and CUSUMSQ tests with a blue lines are equally
within the area restricted by the lines; thus, the null hypothesi s for CUSUM and CUSUMSQ
are rejected and the conclusion is that the estimated model s are effective with stable recursive
residuals.

5. Conclus ion and policy recommendations
This study in contributes to the body of knowledge by examining the relationship
existing between government consumption e xpenditure and current account in Nigeria
covering the periods from 1970 to 2018. It is noted that Consumption expenditure s, such as
goods and services used by agencies and departments of the government cannot be
economically produced by business enterprises. Examples are public administration and
defence. This study particularly evaluated government consumption expenditure s on goods

and services to ascertain its effect on the current account in Nigeria. Thus , the s tudy answered
the question, does government consumption expenditure af fect the current account in Nigeria?
This study utilized the ARDL model procedure and the results showed that the
government consumption expenditure contributed significantly to the cu rrent account deficit
in Nigeria . The result equally indicated that the country’s total debt burden contributed to the
current account deficit of the Nigeria. Based on the findings of the study, the policy
recommendations are as follows: The Federal Govern ment of Nigeria should reduce the level
of borrowing as this would reduce debt service payment affecting the country’s current
account balance. There should be more funds allocated to sectors that create employment
opportunities for the teeming population.

References

Abbas, s. et al., 2011. Fiscal policy and the current account , Washington: IMF working
paper No.WP/10/121.
Ahmed, S., 1987. Government spending,the balance of trade and the terms of trade in British
history. Journal of Monetary Economics , Volume 20, pp. 195 -220.
Alesina, A., Glaeser, E. & Sacerdote, B., 2001. Why Doesn't the US Have a European -Style
Welfare System?, Cambridge : NBER.
Bank, T. W., 2019. General government final consumption expenditure. [Online]
Available at: https://data.wo rldbank.org/indicator/NE.CON.GOVLCD
Bartoli, G., 1989. Fiscal Expansion and External Current Account Imbalances. In: M. I. Blejer
& K. Chu, eds. Fiscal policy, Stabilisation and Growth in Developing Countries . Washington
: International Monetary Fund .
Baxter, M. & King Rich, G., 1993. Fiscal Policy in General Equilibrium. American Economic
Review, 83(3), pp. 315 -334.
Bhavesh, G. & Prabbeesh, K., 2017. Drivers of India's current account deficits, with
implication for ameliorating them. Journal of Asian Eco nomics.
Brennam, G. & Pincus, J., 1996. A minimalist model of federal granmts and flypaper effects.
Journal of Public Economics, 61(2), pp. 229 -246.
Briotti, G., 2005. Economic Reactions to Public Finance Consolidation: A Survey of the
Literature . ECB Occa sional Paper No. 38.

Cavallo, M., 2005. Government Consumption Expenditures and the Current Account, San
Francisco: Federal Reserve Bank of San Francisco Working paper No.2005 -03.
De Mello, L., 1999. Fiscal Federalism and Government Size in Transition Eco nomies,
Washington: IMF.
Ekundayo, B. I., 2011. Fiscal policy and dynamics of current accouint in Sub -Saharan Africa.
A Ph.D proposal presented at the African Economic Research Consortium (AERC) Binnual
Wrokshop. Nairobi, Kenya, AERC.
Heera, G., 2014. Why does a current account leads to a rise in interest rate? Econ -friendly.
[Online] Available at: https://www.quora.com/profile/Gurmeet -Heera -1
Isik, N. S. & Kilinc, E. C., 2017. The Relationship between Current Account Balance and
Types of Credits: An Applic ation on selelcted OECD Counries. Journal of the Faculty of
Economics, 7(2), pp. 105 -126.
Khan, M. A., Jaffri, A. A. & Abbas, F., 2017. Does Trade Libralisation Improve Trade
Balancein Pakistan?. South Asia Economic Journal, p. Available online at:
https:/ /journals.sagepub.com/home/sae.
Klemm, A. & Iakova, D., 2013. Growth followinginvestment and consumption -Driven current
account crises, Washington: IMF Working Paper WP/13/217.
Mundell, R. A., 1968. International Economics. New York: Macmillan.
Narayan, N. P. & Narayan, S., 2006. Savings behaviour in Fiji:An empirical assessment
ujsing ARDL approach to cointegratioin. International journal of Social Economics, 33(7),
pp. 468 -480.
Nickel, C. & Vansteenkiste, I., 2008. Fiscal policies, the current account and Ricardian
Equivalence, s.l.: European Central Bank Working Paper No.935/September.
Nurudeen, A. & Usman, A., 2010. Government Expenditure and Economic Growth in Nigeria
1970 -2008: A Disaggregated Analysis. Business and Economics journal, 4(1).
Oseni, I. & Onukoya, A. B., 2013. Empirical Analyisis of Fiscal Policy Shocks and Current
Account Dynamics in Nigeria. African research review , 7(1).
Oteng -Abayie, E. & Frimpong, J., 2006. Bounds Testing Approach to Cointegration:An
Examination of Foreign Direct Investment,Trade and Growth nrelationships. American
journal of Applied Sciences, 3(11), pp. 2079 -2085.
Perasan, M. H., Shin, Y. & Smith, R. J. , 2001. Bound testing approaches to the analysis of
levelrelationships. Journal of Applied Econometrics, 16(3), pp. 289 -326.
Pettinger, T., 2017. Exchange rate and Cuurent account. [Online]
Available at: https://www.economichelp.org/blog/2089/economics/ex change -rate-and-
cuurent -account/
Pevicin, P., 2004. Determinants of the size of governement and their implications for public
sector reform in Slovenia. A Doctoral Dissertation. Ljubljana, Faculty of Economics,
Ljubljana.

Posner, R., 1971. Regulation as Ta xation. The Bell Journal, 2(1), pp. 22 -50.
Ramanan, V., 2017. Public Debt and Current Account Deficits. [Online]
Available at: https://www.concertedaction .com/201`7/07/01`/public -debt-andcurrent -account –
deficits/
Spiro, P. S., 1997. The effect of the current account balance on interest rate. Munich:
University of Munich, Germany.
Yi, Y., 1993. The Determinants of Consumer Satisfaction :The moderating role of Ambiguity.
In: M. Leigh, L. Michael & U. T. Provo, eds. Advances in consumer researh Volum 20. pp.
502-506. Association of consumer Research