Page 1
AFRICA DEVELOPMENT AND RESOURCES RESEARCH INSTITUTE (ADRRI) JOURNAL
ADRRI JOURNAL (www.adrri.org)
pISSN: 2343-6662 ISSN-L: 2343-6662 VOL. 3 ,No.3, pp 1-19,December, 2013
1
AFRICA DEVELOPMENT AND RESOURCES RESEARCH INSTITUTE (ADRRI) JOURNAL
ADRRI JOURNAL (www.adrri.org)
pISSN: 2343-6662 ISSN-L: 2343-6662 VOL. 3 ,No.3, pp 1-19,December, 2013
The Impact of Trade Liberalisation on Tax Revenue in Ghana: A Co-Integration
Analysis.
Mustapha Immurana1, Abdul Mumin Abdul Rahman
2 and Abdul-Aziz Iddrisu
3
1Department of Accounting Studies Education, University of Education Winneba, Kumasi-
Ghana. Tel: +233-246110693 Email: [email protected]
2 Department of International Development, London School of Economics and Political
Science, E-mail:[email protected]
3Lecturer, Department of Accountancy and Accounting Information Systems, Kumasi
Polytechnic, Kumasi-Ghana. E-mail: [email protected]
Correspondence: Mustapha Immurana, Department of Accounting Studies Education,
University of Education Winneba, Kumasi-Ghana. Tel: +233-246110693
Email: [email protected]
Received: 4th
November, 2013 Accepted: 30th
November, 2013 Published Online: 1st December, 2013
URL: http://www.adrri.org/journal
Abstract
There is ambiguity in the literature on the impact of trade liberalization on tax revenue and this tends to be one
of the major concerns of developing countries in opening up their economies. This paper therefore explored how
the trade liberalisation policy agenda is consistent with meeting the revenue objective of developing countries
using Ghana as a case study. It used the Autoregressive distributed lag (ARDL) model to find the impact of
trade liberalisation on total tax revenue in Ghana using time series data from 1980 to 2010. It found that trade
liberalisation had positive significant impact on tax revenue in Ghana. The study concluded that trade
Page 2
AFRICA DEVELOPMENT AND RESOURCES RESEARCH INSTITUTE (ADRRI) JOURNAL
ADRRI JOURNAL (www.adrri.org)
pISSN: 2343-6662 ISSN-L: 2343-6662 VOL. 3 ,No.3, pp 1-19,December, 2013
2
liberalization increased tax revenue and that developing countries should open their economies without any fear
of losing tax revenue.
Keywords: Ghana, Tax revenue, Trade liberalization, Cointegration, ARDL
INTRODUCTION
Domestic resource mobilisation has been an important factor in the development strategies of
low income countries. The volatility of foreign assistance and the downward trends of aid,
make domestic resource and taxation necessary for sustainable funding in social programs,
improving institutional capacity and enhancing public investment for sustainable growth and
development (Di John, 2005). The West African Economic and Monetary Union (WAMU)
for instance, is committed to raise tax revenue to 17 per cent of GDP among its member
states as convergence strategy (Keen and Baunsgaard, 2005, 3). Achieving the Millennium
development goals requires 4 percentage points increment in the Tax-GDP ratio of low –
income countries (UN 2005; IMF 2011).
Another important development policy initiative for developing countries is trade
liberalisation. Developing countries, under Structural Adjustment Policies by the IMF, are
demanded to liberalise their market to enhance domestic competition, facilitate technology
diffusion, increase efficiency and ultimately foster economic growth and development
(Dornbush, 1992). The implementation of these policies is sometimes conditioned in
International Financial Institution’s loans to developing countries. Trade liberalisation often
entails the reduction and unification of tariffs and the relaxation and elimination of
quantitative barriers, and may be accompanied by currency devaluation and domestic tax
reform (Agbeyegbe et al 2004;3). It is generally believed that increased in market access for
agriculture and labour-intensive exports, would provide the basis for growth and poverty
reduction for developing countries. According to the World Bank (2002), trade liberalisation
accelerates growth; enhance productivity through specialisation, and leads to high
employment and poverty reduction (Bhasin and Obeng 2007; 2).
However, the effect of trade liberalisation on growth and development both in theory and
empirical evidence has attracted debates in the academic literature, but less attention has been
given to its effect on tax revenue. One major concern of developing countries’ further
Page 3
AFRICA DEVELOPMENT AND RESOURCES RESEARCH INSTITUTE (ADRRI) JOURNAL
ADRRI JOURNAL (www.adrri.org)
pISSN: 2343-6662 ISSN-L: 2343-6662 VOL. 3 ,No.3, pp 1-19,December, 2013
3
liberalisation of trade through regional, bilateral and multilateral agreements by tariff
reduction under the current DOHA round, is the potential adverse fiscal consequence (Keen
and Baunsgaard 2005; 3). There is an ambiguous relationship between trade liberalisation and
trade tax revenue in both theory and empirical evidence (Gabriel and Obeng, 2008, 1). Trade
liberalisation in the form of converting quantitative restrictions to tariffs can initially lead to
an increase in trade tax revenue. Further liberalisation in the form of tariffs cut can cause
trade tax revenue loss on one hand, but can also amount to an increase in the volume of
imports, and hence the tax base and revenue. The net effect of trade liberalisation depends on
many factors, including the structure of liberalisation and the elasticity of demand for imports
(Gabriel and Obeng 2008; 1).
The fear of the effect of trade liberalisation on international trade tax is even substantial in
low income countries because it constitutes an average of 26 per cent total tax revenue
(Pelzman 2004 cited in Pupongsak 2009; 130). The ambiguity of the relationship between
trade liberalisation and total tax revenue in the literature creates a knowledge gap, which is
worth probing empirically.
This paper looked at the effect of trade liberalisation on total tax revenue using Ghana as a
case study. Ghana has been described as a country that has gone deep structural adjustment
reform from the 1980s. Trade liberalisation has formed one of the key reform programs in
Ghana since its implementation of Structural Adjustment policies in 1983. Its liberalisation
policies were not gradual from relaxation and elimination of quantitative restrictions to tariff
cuts but proceeded with simultaneous elimination of quotas and reduction in the range and
level of tariffs at the same time (Gabriel and Obeng 2008; 1). The aim of this paper was to
look at whether liberalisation policies advanced by the International Financial Institutions are
consistent with effective state formation and the achievement of the Millennium
Development Goals through the meeting of the revenue objectives since its actual effect on
tax revenue is ambiguous in the literature. This is vital for policy makers in the design of
policies for further liberalisation and subsequent tax reforms.
The remainder of this paper was structured as follows: Section 2 dealt with the literature
review, section 3 covered data sources and methodology and section 4 covered the analysis of
results and discussion and the final section (section 5) dealt with conclusion and policy
implication.
Page 4
AFRICA DEVELOPMENT AND RESOURCES RESEARCH INSTITUTE (ADRRI) JOURNAL
ADRRI JOURNAL (www.adrri.org)
pISSN: 2343-6662 ISSN-L: 2343-6662 VOL. 3 ,No.3, pp 1-19,December, 2013
4
LITERATURE REVIEW
Brief theoretical literature
In theory the direction of changes in revenue to trade liberalisation is ambiguous as it
depends on the tax structure and the revenue productivity (Suliman 2005; 6). Theoretically,
trade liberalisation in the form of tariff cuts and unification can lead to loss in direct trade tax
revenue on one hand, but on the other hand, can also result to increase in volume of imports
and hence the tax base and revenue (Gabriel and Obeng 2008; 1). The net effect on revenue
will therefore depend on the form of liberalisation as well as the elasticity of demand for
imports. Considering trade liberalisation as stages, the first stage of trade policy reform-
elimination of prohibitive tariffs and quantitative restrictions, unification of tariffs,
elimination of exemptions- may lead to increase in trade tax revenue. Further liberalisation
towards a free trade or freer trade in tariff cuts can however have negative fiscal consequence
(Keen and Baunsgaard 2005; 3).
Furthermore, trade liberalisation exerts indirect effect on profit and corporate income tax.
Privatisation, restructuration and automation accompanying trade liberalisation could lead to
job loss, wage cuts resulting in the contraction in the base of personal income tax as well as
decline in personal income tax revenue (Pupongsak 2009; 8). This effect is however
dependent on the growth effect of trade liberalisation, country specific factors and conditions
and some other underlining contingent factors (Pupongsak 2009; 8). Real exchange rate
depreciation associated with trade liberalisation also affects the profitability of corporate
firms through changes in the relative prices of imported inputs as well as exports (Pupongsak
2009; 9). This affects the corporate tax base and revenue, but the net effect on corporate tax
revenue is ambiguous and depends on the growth effect of trade liberalisation as argued by
Agbeyegbe et al (2004).
Empirical literature
The empirical evidence on the relationship between the share of tax revenue and trade
liberalisation is largely based on cross-country analysis. Ebrill et al (1999), based on a panel
of 27 countries from Africa, Asia and Western Hemisphere for the period 1980-1992 and a
panel of 105 countries for 1980-1995, examined two complementary models of the
determinants of import and international trade tax revenue. They concluded that using a
Page 5
AFRICA DEVELOPMENT AND RESOURCES RESEARCH INSTITUTE (ADRRI) JOURNAL
ADRRI JOURNAL (www.adrri.org)
pISSN: 2343-6662 ISSN-L: 2343-6662 VOL. 3 ,No.3, pp 1-19,December, 2013
5
fixed-effect and instrumental regression framework, tariff reform does not necessarily lead to
loss in trade tax revenue. In contrast, Khattry and Mohan Rao (2002), using a similar fixed-
effect regression framework for a panel of 84 countries from developing and industrialised
countries covering 1970-1998, found trade liberalisation in developing countries particularly
low and upper middle-income countries, to be associated with declining tax revenues as a
result of falling income and trade tax revenue. This findings support Rao (1999) study on the
effect of changes in openness and changes in the tax base on overall tax revenue, which
concluded that there is a trade-off for Sub-Saharan African countries between reduced
protection and reduced revenues from liberalisation.
Adam et al (2001), using a difference General Method of Moments (GMM), examined the
relationship between tax revenues, exchange rate and trade liberalisation. They found that
trade liberalisation with an openness variable as a proxy, raises tax revenue in CFA Franc
countries but little effect on non-CFA franc countries. Longoni (2009), using General Method
of Moment Regression on panel-data of large number of African countries from the period
1970-2000, taking into account macroeconomic features of African economies as well as
their political capacity in undertaking trade policy reforms, found a large trade-off between
the degree of openness to international trade and the revenue from import and export
taxation. The paper also detected the existence of ‘Laffer effect’, in which further tariff
reduction will lead to further reduction in trade tax revenue in these African countries.
In a study by Pupongsak (2009), using a traditional tax effort model with a panel data of 134
countries covering the period of 1980-2003 and employing the two way fixed-effect
approach, found trade liberalisation (with share of trade to GDP, average tariff rates and the
number of free trade agreements as proxies) to have positive effect on both international trade
tax and domestic taxes for 30 low income countries. The result is highly inconsistent with
Khattry and Rao (2002) findings on low income countries.
Keen and Baunsgaard (2005), using a panel data of 111 countries over 25 years, studied the
extent to which countries have recovered revenue loss resulting from trade liberalisation.
They found advanced countries to have successfully overcome the fiscal impacts of trade
liberalisation through other domestic taxation. Middle income countries recovery has been in
the order of 45-60cents for each dollar loss from trade tax revenue. Low income countries, on
the other hand, were far from revenue recovery. Their recovery is at best 30 cents for each
Page 6
AFRICA DEVELOPMENT AND RESOURCES RESEARCH INSTITUTE (ADRRI) JOURNAL
ADRRI JOURNAL (www.adrri.org)
pISSN: 2343-6662 ISSN-L: 2343-6662 VOL. 3 ,No.3, pp 1-19,December, 2013
6
dollar lost. These findings are consistent with that of Khattry and Rao (2002), but both
studies ignored the possible effect on other domestic taxes of trade liberalisation. Suliman
(2000) examined the buoyancy and elasticity of the Sudanese tax system with particular
attention to the impact of trade liberalisation reforms of 1992 on revenue mobilisation and the
stabilisation role of the fiscal sector over 1970-2002. The study found an improvement in tax
yield from import duties over the liberalisation period. Zafar (2005) study of the revenue and
fiscal impact of trade liberalisation in Niger from 1983-2003, found that, tariff reductions
during the 1980s and 1990s in the context of Structural Adjustment and West African
Regional integration initiatives, had negative effect on trade tax revenue. It however revealed
that, the increase in the volume of imports after 1994 following trade liberalisation succeeded
in partially offsetting the revenue losses.
Oduro (2000) based on descriptive analysis of tax revenue variations, finds trade
liberalisation to be fiscally incompatible in Ghana in the 1990s. This method does not
however provide an exact impact of trade liberalisation on total tax revenue (Gabriel and
Obeng, 2008, 1). A similar study on Ghana, by Gabriel and Obeng (2008), used regression
analysis to examine the impact of import liberalisation on tariff revenue. Comparing post-
liberalisation period (post-1983) with pre-reform period (pre-1983), the study found trade
liberalisation to be in conflict with revenue objectives of economic reforms. This study
however differed from the previous two papers on Ghana. It looked at how trade
liberalisation has affected total tax revenue in general. It recognises, as in the study of
Agbeyegbe et al (2004), the direct and indirect effect on other domestic taxes of trade
liberalisation.
METHODOLOGY
Data
The empirical investigation of this paper was based on annual time-series data on Ghana
taken over the period 1980-2010. 1983 marked the year for the implementation of Structural
Adjustment policies with trade liberalisation as one of the main policy initiatives. Data was
drawn from the Ghana Revenue Authority (GRA), Ministry of Finance and Economic
Planning (MoFEP), the World Development Indicators (WDI, World Bank 2011), African
Development Indicators (ADI, World Bank 2011) and the International financial Statistics
(IFS, IMF 2012)).
Page 7
AFRICA DEVELOPMENT AND RESOURCES RESEARCH INSTITUTE (ADRRI) JOURNAL
ADRRI JOURNAL (www.adrri.org)
pISSN: 2343-6662 ISSN-L: 2343-6662 VOL. 3 ,No.3, pp 1-19,December, 2013
7
Method
The explanatory variables widely used as the determinants of tax revenue include the level of
real income per capital, sectoral shares in GDP, real effective exchange rate, terms of trade
and indices of trade liberalisation (Obeng et al 2011). These variables were used in Adam et
al (2001), Khattry and Rao (2002), Agbeyegbe et al (2004), Suliman (2005), among others.
Other variables used in some of these papers included the net aid, and quality of governance
indicators.
Based on the literature, data availability, cointegration, diagnostics and stability test, the
study derived a simple tax revenue model for Ghana as follows:
lnTR/GDP t =β1 + β
2ln(OPEN)t + β
3ln(TRADE)t + β
4lnAGRICt + β
5lnINDt + β
6ln(SER)t +
β7ln(RER)t + β
8ln(ODA)t + Ԑt...........................(1)
Where; TR/GDP is tax revenue as a share of GDP, Log (OPEN) is an index for openness
implying the collected tariff rate (the ratio of import duties to the value of imports); Log
(TRADE) is a measure of trade liberalization which is import + exports/ GDP, AGRIC is
agriculture share to GDP, IND is Industry share to GDP, SER is services’ share of GDP, RER
is the real exchange rate, ODA is the net transfers of aid and Ԑ is the error term and t is time
in years.
Tax Revenue as a percentage of GDP (TR/GDP) was the Dependent variable and data was
calculated using Tax revenue on Ghana from the MoFEP & Ghana Revenue Authority and
GDP data from the WDI (2011).
Trade Liberalization proxies: According to Rodrik and Rodrigues (1999), the effective rate
of trade taxation and the collected tariff rate are the two main indices for trade liberalisation
used in empirical works (cited in Longoni 2009; 8). The effective rate of taxation is the ratio
of international trade tax to the volume of international trade. This measure has been used by
among other authors by Khattry and Rao (2002) and Longoni (2009), where a drop in the
ratio indicates greater liberalisation, while a rise in the index shows a country is becoming
more restrictive or closed relative to its trade volumes. It has been argued that this index
ignores the Laffer effect on revenue by extremely tax rate and ignores tax evasion through
smuggling and other practices (Kattry and Rao, 2002). But it gives an idea of ‘’realised’’
Page 8
AFRICA DEVELOPMENT AND RESOURCES RESEARCH INSTITUTE (ADRRI) JOURNAL
ADRRI JOURNAL (www.adrri.org)
pISSN: 2343-6662 ISSN-L: 2343-6662 VOL. 3 ,No.3, pp 1-19,December, 2013
8
tariff as it is based on actual tariff collected (Longoni 2009; 8). The collected tariff rate
measure is the ratio of import duties to the value of imports, where a decline in the ratio
represents greater liberalisation (Agbeyegbe et al 2004; 12). Ebrill (1999) and Agbeyegbe et
al (2004) employed this measure.
This study therefore employed the share of trade to GDP (TRADE) and the collected tariff
rate (OPEN) as the measures of trade liberalization. The unavailability of data does not
permit the use of the effective rate of trade taxation, and given the small share of export tax in
trade tax, it is not likely to affect the results. Data on the share of trade to GDP (TRADE) was
calculated from the WDI (2011) and data on the collected tariff rate (OPEN) was calculated
using import duties data from the MoFEP & Ghana Revenue Authority and data on the value
of imports from the WDI (2011). The impact of trade liberalization as has been stated already
is not certain thus it can be negative or positive.
Sectors (Agriculture, Industry and Service) share to GDP; A higher agriculture share to
GDP is expected to be associated with lower revenue because of difficulty in taxing
agricultural activities as it is less monetized and largely informal and hence Agriculture as a
share of GDP is expected to have a negative impact on tax revenue. A larger share of industry
and service to GDP is expected to be positively related to tax revenue. Data on Sectors
(Agriculture, Industry and Service) share to GDP were obtained from the WDI (2011).
Real exchange rate; trade reforms are usually accompanied by exchange rate reforms which
include domestic currency devaluation. In the short-run, the aggregate elasticity of import
demand might be quite inelastic, but could lead to increase in revenue if the valuation effect
(the effect due to higher domestic currency value of imports) dominates the volume effect
(the negative effect on trade revenue due to the fall in imports) (Longoni 2009; 10). Data
calculated from the IFS (May, 2012, IMF).
Net aid; the relationship between net aid and tax revenue is also uncertain. There is some
evidence that net aid reduces tax effort and administrative capacity and efficiency in tax
mobilisation, but these results might depend on the nature of aid and other factors
(Agbeyegbe et al 2004; 21). So the impact of Net aid could be either positive or negative.
Data was obtained from the ADI (2011).
Page 9
AFRICA DEVELOPMENT AND RESOURCES RESEARCH INSTITUTE (ADRRI) JOURNAL
ADRRI JOURNAL (www.adrri.org)
pISSN: 2343-6662 ISSN-L: 2343-6662 VOL. 3 ,No.3, pp 1-19,December, 2013
9
Empirical methodology
Unit Root Tests
It was important to verify the stationarity properties of the variables used for the study in
order to avoid the risk of spurious regression since literature has shown that most time series
variables have stochastic trends, thus their variances and unconditional means are non-
stationary. This study adopted the standard test for unit root developed by Dickey-Fuller
(1979) known as the Augmented Dickey-Fuller (ADF) test to examine the order of
integration of the variables used in the model.
Co-integration test
The Ordinary Least Square (OLS) estimator becomes inappropriate since literature has shown
that most time series variables are mostly non-stationary. Therefore to curb the issue of
spurious regression associated with non-stationary variables; this paper adopted the co-
integration technique. Co-integration simply means although it is possible that single series or
variables may not be stationary, their linear combinations with other variables may generate
stationarity. Thus if co-integration is established, it means a stationary co-integrating
relationship is established between the variables and hence the problem of spurious
regression can be avoided. This study used the Autoregressive Distributed Lag (ARDL)
model approach to co-integration to estimate the impact of trade liberalisation on total tax
revenue. The ARDL model was used as it is suitable for small sample size as it is in the case
of this study. It also allows for co-integration to be performed curbing the problem of
spurious correlation from non-stationarity of variables associated with time-series data.
The ARDL model
Following Pesaran et al (2001), the study therefore adopted the ARDL model which involved
three steps.
In step 1, equation (1) was estimated by the ordinary least square (OLS) in order to test for
the existence of a level or a long-run relationship among the variables. This was done by
conducting an F-test for the joint significance of the coefficients of lagged levels of the
variables.
Page 10
AFRICA DEVELOPMENT AND RESOURCES RESEARCH INSTITUTE (ADRRI) JOURNAL
ADRRI JOURNAL (www.adrri.org)
pISSN: 2343-6662 ISSN-L: 2343-6662 VOL. 3 ,No.3, pp 1-19,December, 2013
10
The hypothesis would be:
H0: β
1= β
2 = β
3 = β
4 = β
5= β
6 = β
7 = β
8 = 0 (No long run relationship or co-integration)
H1: β
1= β
2 = β
3 = β
4 = β
5= β
6 = β
7 = β
8 ≠ 0 (There is a long run relationship or co-integration)
So if the F-statistic is above the upper critical value, the null hypothesis of no long-run
relationship is rejected regardless of the orders of the variables. Also, if the F-statistic, falls
below the lower critical value, the null hypothesis is accepted, implying that there is no long-
run relationship among the series. Lastly, if the F-statistic falls between the lower and the
upper critical values, the result is inconclusive. When the existence of co-integration (long
run relationship) was established, the study proceeded to the second step to run the long-run
model for . Finally, the third step in the ARDL bounds approach involved
estimating an Error Correction Model (ECM) to capture the short-run dynamics of the model.
The ECM provides the means of reconciling the short-run behaviour of an economic variable
with its long-run behaviour.
ANALYSIS AND DISCUSSION
Here the Unit root test was done using EVIEWS 7 and the ARDL model was run using
Microfit 5.0.
Analysis of Time Series Properties
Results for Unit Root Test
In order to find out the impact of trade liberalization (measured by the collected tariff rate and
trade) on Tax revenue as a percentage of GDP in Ghana, the stationarity status of all the
variables in the model specified for the study was tested. This was done to ensure that the
variables were not integrated of order two (I(2)) so as to avoid spurious results.
Page 11
AFRICA DEVELOPMENT AND RESOURCES RESEARCH INSTITUTE (ADRRI) JOURNAL
ADRRI JOURNAL (www.adrri.org)
pISSN: 2343-6662 ISSN-L: 2343-6662 VOL. 3 ,No.3, pp 1-19,December, 2013
11
Table 1.0 Results of the Unit Root Test
Variables Augmented Dickey Fuller – Test
lnTR/GDP -1.827429 -5.270035***
LnOPEN -2.476902 -8.122838***
LnTRADE -1.454554 -4.852030***
LnAGRIC -1.363277 -5.479209***
LnIND -1.443409 -4.613640**
LnSER -1.012922 -5.685564***
LnRER -1.615355 -4.415467***
LnODA -2.080984 -5.917258***
Legend: *** (**) denotes the rejection of the null hypothesis of unit root or no-stationarity at
1% (5%).
As can be seen from Table 1.0, when the regression was estimated at the log level (with
intercept), none of the variables becomes stationary. But using the first difference, all the
variables become stationary. Hence, the null hypothesis of non-stationarity is rejected and the
alternative hypothesis of stationarity accepted. Thus, the first differences of the variables
were stationary.
Results for Cointegration test
After establishing stationarity among the variables, the study now proceeded with the co-
integration test and the results can be seen in Table 1.1 below:
Table 1.1 Bounds Test for Long Run Relationship or Co-integration
Critical Values
F-statistics 5.1306
Lower Bound I(0) Upper Bound I(1)
5% 2.9476 4.5130
10% 2.4002 3.7804
Page 12
AFRICA DEVELOPMENT AND RESOURCES RESEARCH INSTITUTE (ADRRI) JOURNAL
ADRRI JOURNAL (www.adrri.org)
pISSN: 2343-6662 ISSN-L: 2343-6662 VOL. 3 ,No.3, pp 1-19,December, 2013
12
Since the F-statistic (5.1306) was above the upper bounds (4.5130), it means the null
hypothesis of no level relationship is rejected and hence there is co-integration. Based on this,
we therefore run our ARDL model.
Diagnostic and Stability Test
The diagnostic and stability test was done to find out the robustness of the model. The result
as seen in appendix I showed that, the null hypotheses of no normal distribution, the
existence of autocorrelation, incorrect functional form and heteroscedasticity in the model
could not be accepted at the 5% level. The stability test was also done using the Cumulative
sum and the Cumulative sum of squares of recursive residuals and it showed the non-
existence of structural break and the results can be found in graph 1.0 and 1.1 below.
Graph 1.0
Graph 1.1
Long Run Impact of Trade Liberalisation on Total Tax revenue
-20
-10
0
10
20
1981 1988 1995 2002 2009
The straight lines represent critical bounds at 5% signif icance level
Plot of Cumulative Sum of Recursive Residuals
-0 .4
-0 .2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1981 1988 1995 2002 2009
The straight lines represent critical bounds at 5% signif icance level
Plot of Cumulative Sum of Squares of Recursive Residuals
Page 13
AFRICA DEVELOPMENT AND RESOURCES RESEARCH INSTITUTE (ADRRI) JOURNAL
ADRRI JOURNAL (www.adrri.org)
pISSN: 2343-6662 ISSN-L: 2343-6662 VOL. 3 ,No.3, pp 1-19,December, 2013
13
The long run impact on the share of tax revenue to GDP estimated from the ARDL (1, 1, 0, 1,
0, 0, 0, 0) model based on the Schwarz Bayesian Criterion (SBC) can be seen in Table 1.2
below:
Table 1.2 Estimated Long run model for the selected ARDL (1, 1, 0, 1, 0, 0, 0, 0) based
on Schwarz Bayesian Criterion. Dependent Variable: lnTR/GDP
Variable Coefficient Std. Error T-Statistic Prob.
LnOPEN .47732 .098406 4.8505 .000***
LnTRADE .39032 .11128 3.5074 .003 ***
LnAGRIC -.65698 .47974 -1.3695 .188
LnIND .78143 .19825 3.9416 .001***
LnSER .26361 .37798 .69742 .494
LnRER -.26999 .10423 -2.5903 .018**
LnODA .31526 .080257 3.9282 .001***
C .65919 3.5934 .18344 .856
*** (**)* significant at 1% (5%) 10%
From Table 1.2 above, both measures of trade liberalisation had positive and significant
impacts on tax revenue in the long run, with tax revenue being much sensitive to the collected
tariff proxy (lnOPEN) than the share of trade to GDP index (lnTRADE). A percentage
increase in each of the indices led to an approximate 0.4 – 0.5 per cent increase in tax
revenue. These findings contrast that of Khattry and Mao (2002) and Agbeyegbe et al (2004)
but consistent with that of Pupongsak (2009). The result could suggest a high elasticity of
demand for imports where the increase in the volume of imports resulting from openness
exceed the magnitude of tariff cuts resulting in positive fiscal impact. It can also be argued
that, the tax reforms in Ghana following trade liberalisation such as the introduction of the
VAT in 1998; have succeeded in replacing reliance on international trade tax by domestic
taxation. The positive effect of trade liberalisation on total tax revenue can also be as a result
of increase in corporate profit as a result changes in the profitability of import and exports for
companies.
The share of industry had a positive and significant effect on tax revenue as argued in
literature. The positive and significance effect of the share of industry also suggests a higher
tax base for domestic taxes such as corporate and profit tax, and income tax. Surprisingly, the
share of Agriculture, although had a negative impact on tax revenue as predicted by literature,
Page 14
AFRICA DEVELOPMENT AND RESOURCES RESEARCH INSTITUTE (ADRRI) JOURNAL
ADRRI JOURNAL (www.adrri.org)
pISSN: 2343-6662 ISSN-L: 2343-6662 VOL. 3 ,No.3, pp 1-19,December, 2013
14
but was statistically insignificant. The share of Service also had a positive but statistically
insignificant effect on tax revenue.
Furthermore, real exchange rate had a negative impact on total tax revenue at a 5 per cent
level of significance. This result is consistent with Agbeyegbe et al (2004) findings and also
confirms Tanzi’s (1989) hypotheses that, there is an inverse relationship between a country’s
real exchange rate and its tax revenue. Thus in Ghana, the high demand for foreign exchange
has kept the real exchange rate in upward trend resulting in negative effect on tax revenue as
the result shows.
Also, official aid had a positive and significant impact on tax revenue, indicating that the
receipt of official external assistance in Ghana is not as a result of its weak domestic revenue
generating capacity as aid is not used as a substitute for domestic revenue mobilisation.
Short run Impact of Trade Liberalisation on Total Tax Revenue- the Error Correction
Model or the Short run model.
Since co-integration was established in the Tax revenue as a percentage of GDP model, it
means that from the ARDL model, there is an error correction model that captures the short
run dynamics of the Tax revenue as a percentage of GDP model. The error correction model
enables us to estimate coefficients that tell us the speed at which the variables adjust to their
long run equilibria following any shock in the short run. The results of the short run dynamics
or the error correction model from the ARDL (1, 1, 0, 1, 0, 0, 0, 0) model based on the
Schwarz Bayesian Criterion (SBC) are given in table 1.3 below:
Table 1.3 Error Correction Representation for the selected ARDL (1, 1, 0, 1, 0, 0, 0,0)
based on Schwarz Bayesian Criterion. Dependent Variable: dlnTR/GDP.
Variable Coefficient Std. Error T-Statistic Prob.
DlnOPEN .22922 .063163 3.6291 .002***
DlnTRADE .32701 .10342 3.1620 .005 ***
DlnAGRIC .059816 .45267 .13214 .896
DlnIND .65468 .15538 4.2134 .000***
DlnSER .22085 .31514 .70082 .491
DlnRER -.22620 .093806 -2.4113 .026**
DlnODA .26413 .077111 3.4253 .003***
Ecm(-1) -.83780 .073672 -11.3720 .000
*** (**)* significant at 1% (5%) 10%
Page 15
AFRICA DEVELOPMENT AND RESOURCES RESEARCH INSTITUTE (ADRRI) JOURNAL
ADRRI JOURNAL (www.adrri.org)
pISSN: 2343-6662 ISSN-L: 2343-6662 VOL. 3 ,No.3, pp 1-19,December, 2013
15
From Table 1.3 above, the two indices of trade liberalisation had positive and significant
impact on tax revenue. That is, the positive effect on total tax revenue by trade liberalisation
is not sensitive on the proxy being used or on the time period of analysis. Given the nature of
trade liberalisation in Ghana, which proceeded with simultaneous elimination of quantitative
restrictions and reduction and unification of tariff levels, one would expect it to have adverse
fiscal consequence at least in the short run before any tax reforms, but the results showed
otherwise. The results suggest a high elasticity of demand for imports resulting in higher
demand for imports for any marginal reduction in tariff levels. That is, the increase in tax
revenue from a rise in the volume of imports as a result of any tariff cut more than offset the
possible revenue loss from decrease in the value of imports.
Also, in the short run, the ratio of agriculture to GDP (AGRIC) had a positive coefficient
which was different from its long run negative coefficient but was still not significant as it
was in the long run. The share of industry to GDP, the share of service to GDP and Official
foreign aid, all had positive and significant impact on Tax revenue in the short run as in the
long run. Real exchange rate also had a negative significant impact on total tax revenue in the
short run as in the long run.
Additionally, the error correction (ECM) term had a negative coefficient of (-.83780) with a
p-value of .000 which means that it is highly significant at 1% level. This result met the
expectation of econometric theory. The coefficient of the ECM measures the alacrity with
which the Tax revenue as a percentage of GDP model adjusts to changes in the independent
variables in the model. Thus the ECM coefficient of -.83780 shows a high speed of
adjustment to equilibrium after a shock and means approximately more than 84% of
disequilibria from the previous year’s shock converges back to the long-run equilibrium in
the current year.
Combining the short run and long run effect on tax revenue of trade liberalisation in Ghana,
this paper argues that, the positive effect of trade liberalisation indices on tax revenue can be
explained by both the high elasticity of demand for imports as well as the subsequent tax
reforms especially the introduction and implementation of the VAT in the late 1990s.
CONCLUSION AND RECOMMENDATION
The relationship between trade liberalisation and total tax revenue is ambiguous. Policy
makers in devising trade policies are challenged by ascertaining their fiscal impacts.
Page 16
AFRICA DEVELOPMENT AND RESOURCES RESEARCH INSTITUTE (ADRRI) JOURNAL
ADRRI JOURNAL (www.adrri.org)
pISSN: 2343-6662 ISSN-L: 2343-6662 VOL. 3 ,No.3, pp 1-19,December, 2013
16
The endeavour to unravel the ambiguity in the relationship between trade liberalisation and
total tax revenue was undertaken by this paper, using the ARDL regression model based on
country case study of Ghana for the period 1980-2010. The results showed trade liberalisation
to have a positive and significant effect on total tax revenue both in the short run and in the
long run. The paper can conclude that, the observed positive effect on total tax revenue by
trade liberalisation in the short run is due to the high elasticity of demand for import in the
Ghanaian economy, and this high elasticity is complemented by the subsequent tax reforms
for the observed relationship to persist in the long run. This result has a serious policy
implication as it suggests the reluctance of developing countries in further opening up their
national economies on the basis of diminishing tax revenue has no basis and the paper would
recommend that they should open up their economies and institute important domestic taxes
in order to raise tax revenue.
This paper addresses the effect on total tax (in aggregate) of trade liberalisation. It does not
decompose the components of total tax revenue to analyse the impact on each component of
trade liberalisation. The paper does not also estimate the exact elasticity of import demand
following trade liberalisation that it argues to be one of the factors driving the results. These
limitations raise queries for further research.
REFERENCE
Adam, C., David, B., & Gerrard, C. (2001). Exchange Rate Regimes and Revenue
Performance in Sub-Saharan Africa. Journal of Development Economics, 64, 173-
213.
Agbeyegbe, T., Stotsky, J., & Woldemariam, A. (2004). Trade Liberalisation, Exchange
Rates Changes and Tax revenue in Sub-Saharan Africa. IMF Working Paper,
WP/04/178.
Bhasin, K., Vijay, & Obeng, C.K. (2007). Trade Liberalisation, Foreign Aid, Poverty and
Income Distribution of Households in Ghana. [Online] Available:
http://www.africametrics.org/documents/conference07/Day%202/Session%206/Bhasi
n%20Trade%20Liberalization%20and%20Foreign%20AID.pdf
Dickey, D., Fuller, W.A., (1979). Distribution of the Estimates for Autoregressive Time
Series with Unit Root. Journal of the American Statistical Association, 74, 427-31.
Page 17
AFRICA DEVELOPMENT AND RESOURCES RESEARCH INSTITUTE (ADRRI) JOURNAL
ADRRI JOURNAL (www.adrri.org)
pISSN: 2343-6662 ISSN-L: 2343-6662 VOL. 3 ,No.3, pp 1-19,December, 2013
17
Di John, Jonathan. (2005). The Political Economy of Taxation and Resource mobilisation in
Sub-Saharan Africa. In Vishun Padazachee (Ed.), The Political Economy of Africa.
London: Routledge.
Dornbusch, R. (1992). The Case for Trade Liberalisation in Developing Economies. Journal of
Economic Perspectives, 6, No.1, 69-85.
Ebrill, L., J., Stotsky, & Reint, G. (1999). Revenue Implications of Trade Liberalisation. IMF
Occasional Paper, 99/80.
Gabriel, W., Brafo-Insaidoo, & Obeng, C. K. (2008). Effect of Trade Liberalisation on Tariff
Revenue in Ghana. African Economic and Research Consortium Research Paper, 180.
International Monetary Fund. (2011). Revenue Mobilisation in Developing Countries. New
York: IMF.
Keen, M., & Baunsgaard, T. (2005). Tax Revenue and (or?) Trade Liberalisation. IMF
Working Paper, WP/05/112.
Khattry, B., & J., Mohan, Rao. (2002). Fiscal Faux Pas?: An Analysis of The Revenue
Implications of Trade Liberalisation. World Development, 30, No. 8,1431-1444.
Krueger, A. (1986). Problems of Liberalisation. In A., Choksi & Papageorgiopu (Eds).,
Economic Liberalisation in Developing Countries. Oxford; Blackwel.
Longoni, E. (2009). Trade Liberalisation and Trade Tax Revenues in African Countries.
Working Paper Series, 158.
Oduro, A.D. (2000). Performance of the External Trade Sector since 1970. In Aryeetey, E.,
J. Harigan & M. Nissanke (Eds.), In Economic Reforms in Ghana. New Jersey:
African World Press.
Pelzman, Joseph. (2004). Trade Liberalization and Fiscal Reform: Evidence from Two Case
Studies-Morocco and Jamaica and a General Cross Country Econometric Analysis.
Fiscal Reform in Support of Trade Liberalization Project, Development Alternatives
Inc. Working Paper fr1004.
Pesaran, M. H., Y., Shin, & R. J., Smith. (2001). Bounds Testing Approaches to the
Analysis of Level Relationships. Journal of Applied Econometrics, 16, 289-326.
Rao, J.M. (1999). Globalisation and Fiscal Autonomy of the State. Human Development
Report Background Papers 1999, Vol. 1. New York: UNDP.
Page 18
AFRICA DEVELOPMENT AND RESOURCES RESEARCH INSTITUTE (ADRRI) JOURNAL
ADRRI JOURNAL (www.adrri.org)
pISSN: 2343-6662 ISSN-L: 2343-6662 VOL. 3 ,No.3, pp 1-19,December, 2013
18
Rodriguez, F., & Rodrick, D. (1999). Trade Policy and Economic growth: A Sceptics Guide
to Cross-national Evidence. NBER Working Paper, 7081.
Pupongsak, S. (2009). The Effect of Trade Liberalisation on Taxation and Government
Revenue. PhD diss., University of Birmingham Research Archive.
Suliman, Kabashi, M. (2005). The Impact of Trade Liberalisation on Revenue Mobilisation
and Stability in Sudan. International Research Project on Macroeconomic Policy
Challenges of Low Income Countries, GDN. [Online] Available: http://www.ieo-
imf.org/external/np/res/seminars/2005/macro/pdf/sulima.pdf.
Tanzi, Vito. (1989). The Impact of Macroeconomic Policies on the Level of Taxation and the
Fiscal Balance in Developing Countries. IMF Staff Papers, 36, 633-656.
United Nations. ( 2005). Investing in Development. New York: United Nations.
World Bank. (1981). Accelerated Development in Sub-Saharan Africa: an agenda for action
(Known as the Berg Report). Washington D.C.: World Bank.
World Bank. (2011). World Development Indicators. Washington D.C.: World Bank.
World Trade Organisation. (2001). Trade Policy Review: Ghana. PRESS/TPRB/157.
[Online] Available: http://www.wto.org/english/tratop_e/tpr_e/tp157_e.htm
World Trade Organisation. (2008). Trade Policy Review: Ghana. PRESS/TPRB/ 294.
[Online] Available: http://www.wto.org/english/tratop_e/tpr_e/tp294_e.htm
Zafar, A. (2005). Revenue and the Fiscal Impact of Trade Liberalisation: The Case of Niger.
World Bank Research Working Paper, 3500.
ACRONYMS
ARDL- Autoregressive Distributed Lag VAT-Value Added Tax
ADF- Augmented Dickey-Fuller WAMU-West African Monetary Union
ADI- African Development Indicators WB- World Bank
CEPS- Custom Excise and Preventive Service WDI- World Development Indicators
CPIA- Country Policy and Institutional Assessment
ECM- Error Correction Model
EOWAS- Economic Community of West African States
Page 19
AFRICA DEVELOPMENT AND RESOURCES RESEARCH INSTITUTE (ADRRI) JOURNAL
ADRRI JOURNAL (www.adrri.org)
pISSN: 2343-6662 ISSN-L: 2343-6662 VOL. 3 ,No.3, pp 1-19,December, 2013
19
ERP- Economic Recovery Programme
GDP- Gross Domestic Product
GMM- General Method of Moments
GRA- Ghana Revenue Authority
IFS- International Financial Statistics
IMF- International Monetary Fund
MoFEP- Ministry of Finance and Economic Planning
OECD- Organisation of Economic Co-operation and Development
OLS- Ordinary Least Squares
SAP- Structural Adjustment Program
UN- United Nations
This academic research paper was published by the Africa Development and Resources Research Institute’s
Journal (ADRRI JOURNAL). ADRRI JOURNAL is a double blinded peer review, open access international
journal that aims to inspire Africa development through quality applied research.
For more information about ADRRI JOURNAL homepage, follow: www.adrri.org/journal.
CALL FOR PAPERS
ADRRI JOURNAL calls on all prospective authors to submit their research papers for publication. Research
papers are accepted all yearly round. You can download the submission guide on the following page:
www.adrri.org/journal
ADRRI JOURNAL reviewers are working round the clock to get your research paper publishes on time and
therefore, you are guaranteed of prompt response. All published papers are available online to all readers world
over without any financial or any form of barriers and readers are advice to acknowledge ADRRI JOURNAL. All
authors can apply for one printed version of the volume on which their manuscript(s) appeared.