Page 1 of 12
Journal for Studies in Management and Planning
Available at http://internationaljournalofresearch.org/index.php/JSMaP
e-ISSN: 2395-0463
Volume 01 Issue 09
October 2015
Available online: http://internationaljournalofresearch.org/ P a g e | 1
Exports, Gdp, Imports and Exchange Rate in Ghana:
Evidence from Cointegrated Analysis and Vector
Error Correction Model
Emmanuel Amoah MPhil Statistics1, Isaac Zinkure MSc. Biometry2
1Department of Statistics, University of Ghana, Legon
Email: emmanuelamoah27@gmail.com
2Department of Statistics, University for Development Studies, Navrongo
Email: zingure11@yahoo.com
ABSTRACT
This study used co-integrated analysis and
vector error correction model (VECM) in
modelling exports of Ghana with other three
selected macroeconomic such as GDP,
imports and exchange rate for the period of
1980 to 2013. Data were taken from the World
Bank’s World Development Indicators and
Bank of Ghana. The objectives of the study are
to examine both long-run relationships and
direction of causality between the Exports and
the macroeconomic variables. The time series
properties of the data were, first, analysed
using the Augmented Dickey-Fuller (ADF)
test, PP test and KPSS test. The empirical
results derived indicate that all the variables
were stationary after their first differencing;
i.e. variables are integrated of order one, I
(1). The study further established that there is
co-integration between the selected
macroeconomic variables and Exports growth
in Ghana indicating long run relationship.
The above long term relation indicates that
imports and exchange rate have a negative
effect on Exports growth whiles GDP growth
showed a positive effect on Exports growth.
The study further investigated the causal
relationship using the Granger Causality
analysis, which indicates a unidirectional
causality between Exports and GDP growth
rate. This finding implies that Exports granger
cause GDP growth rate in Ghana. Also there
is unidirectional causality between imports
and Exports, which implies that imports
granger cause Exports in Ghana. Also there is
unidirectional causality running from imports
and GDP growth rate, Exchange rate and
GDP growth, imports and Exchange rate but
there is no directional causality from Exports
and Exchange rate and visa versa in Ghana
for the study period at 5%.
Key world:
Exports, ADF test, PP test, KPSS test, co- integration, VECM, granger causality test.
Introduction
From the early 1960, policy makers and
researchers had a great interest in relationship
between export and economic growth. Their
main reason and motivation is that they want
to know if a country should increase its export
Page 2 of 12
Journal for Studies in Management and Planning
Available at http://internationaljournalofresearch.org/index.php/JSMaP
e-ISSN: 2395-0463
Volume 01 Issue 09
October 2015
Available online: http://internationaljournalofresearch.org/ P a g e | 2
to lead to a more economic growth or they
should stimulate economic growth from the
outset to lead to more export.
The relationship between exports and imports
has also attracted the attention from both
researchers and policy makers. The fact that
exports and imports constitute a major part of
balance of payment in a country,
unsustainable trade deficit depicts a violation
of international budget constraints over time.
The existence of long-run relationship
between exports and imports is desirable to
nations and was the subject of intensive
research in developed and developing
economies.
Annan and Acquah (2011) examined the long- run relationship between exports and imports
in Ghana for the period from 1948-2010.
Engels-Granger’s method of co-integration
was used, and variables were found to be co- integrated. The paper adopted Husted (1992)
and Arize (2002) models for testing the
hypothesis. The study concluded that long-run
relationship between nominal exports and
imports in Ghana exist.
Uddin (2009) in a paper attempted at studying
time series behaviour of total exports and total
imports in Bangladesh. Johansen co- integration method was applied to data, and
revealed existence of long-run equilibrium
relationship between the two variables. Long
term causality was also investigated and found
to be bidirectional between exports and
imports in Bangladesh.
Also, various empirical studies investigated
the relationship between openness of business
and economic growth and different results are
obtained. Frankel and Romer (1999) indicate
that despite many attempts which are done for
studying this issue, there is little convincing
evidence that strongly confirm the trade
impact on growth. Hence, there are different
degrees of uncertainty about this issue. In
other words, the issue that trade has always a
positive impact on countries ‘economic
growth is not an imminent matter.
Empirically, apart from Klaassen (2011) who
established that exchange rate has
insignificant effect on exports, several studies
have found significant impact of exchange
rate on export (see, Erdal et al. 2012, Dincer
and Kandil 2011 and Granskog and Wisdom
2003). These impact findings are however
very mixed with some establishing positive
while others getting negative impacts (see Hsu
and Chiang, 2011).
Hence this paper aims at using co-integration
test and Vector error correction model to find
out the long-run and short-run relationships
between exports, GDP, imports exchange rate
in Ghana for the period 1980-2013 using time
series data.
METHODOLOGY
Source of Data and Data Collection
Procedure
The study employs secondary data. Time
series annual data on exports, imports, real
exchange and real GDP growth from Ghana
over the study period 1980 to 2013 are used
for the study, which gives thirty three (33)
data points which is statistically large to be
used for the study. Data was obtained from
Bank of Ghana and WDI, World Bank.
Analysis Plan
Page 3 of 12
Journal for Studies in Management and Planning
Available at http://internationaljournalofresearch.org/index.php/JSMaP
e-ISSN: 2395-0463
Volume 01 Issue 09
October 2015
Available online: http://internationaljournalofresearch.org/ P a g e | 3
In analysing the dataset the following tests are
expected to be employed: Unit root test for
stationarity, ADF (Augmented Dickey Fuller)
test, PP (Phillips-Perron) test and KPSS
(Kwiatkowski, Phillips, Schmidt and Shin,
1992) test, Co-integration test, Vector error
correction model (VECM), etc. We rely on
Eviews 8 statistical computing software to
implement the time series methods that will be
discussed and all statistical tests were carried
out at 0.05 level of significance.
Unit Root Tests
Unit root tests are applied to check whether
the time series is stationary. Many unit root
tests are proposed in the literature. In this
paper we used three tests namely ADF
(Augmented Dickey Fuller) test, PP (Phillips- Perron) test and KPSS (Kwiatkowski, Phillips,
Schmidt and Shin, 1992) test.
The Dickey-Fuller test is based on testing the
hypothesis that series contains unit root
against the series is stationary under the
assumption that errors are white noise. The
test may be carried out using a conventional t
statistic. However, the tests do not follow the
standard student t distribution so the critical
values for the test are obtained by simulation.
An extension which will accommodate some
forms of serial correlation is the augmented
Dickey–Fuller test.
The Phillips–Perron test on the other hand test
that a variable has a unit root. The null
hypothesis is that the variable contains a unit
root, and the alternative is that the variable
was generated by a stationary process.
KPSS proposed an LM test for testing trend
and/or level stationarity. That is, now under
the null hypothesis the series is assumed
stationary, whereas in the former tests it was a
unit root process. Taking the null hypothesis
as a stationary process and the unit root as an
alternative is in accordance with a
conservative testing strategy. One should
always seek tests that place the hypothesis we
are interested in as the alternative one. Hence,
if we then reject the null hypothesis, we can
be confident that the series indeed has a unit
root.
Co-integration Test
Two broad approaches for co-integration have
been developed. These are Engle and Granger
(1987) method and the Johansen approach,
due to Johansen (1998), based on vector
autoregressive model (VAR) (Green, 2003).
There are two types of tests for Johansen co- integration approach called trace test and the
maximum eigenvalue test. The test statistics
for the trace test and maximum eigenvalue test
are shown in the following equations [1] and
[2] respectively.
Jtrace= -T ∑ ln(1 − λ̂
i)
n
i=r+1
[1]
Jmax= -T Ɩn (1- λ̂r+1) [2]
Where T is the sample size, n is number of
endogenous variables and λ̂
i
is thelargest
eigenvalue. The trace test tests the null
hypothesis of r co-integrating vectors against
the alternative hypothesis of n co-integrating
vectors. The maximum eigenvalue test, on the
other hand, tests the null hypothesis of r co- integrating vectors against the alternative
hypothesis of r +1 co-integrating vectors
where r =1,2,.....n.
The variance decomposition was used to
analyse the proportion of the unanticipated
change of a variable that is attributable to its
own innovations and shocks to other variables
in the system.
