Transcript Current Account Dynamics and the Adjustment to

Chuku Chuku 1
Johnson Atan2
Felix Obioesio2
&
Kenneth Onye2
1Centre
for Growth and Business Cycle Research, University of
Manchester, U.K & Department of Economics, University of Uyo
Uyo, Nigeria,
2Department
of Economics, University of Uyo, Nigeria
1. BACKGROUND
Current Account (C.A, hence forth) imbalance and the
process of adjustment to desirable and sustainable level
that is consistent with regional integration are two major
aspects of regional economic integration that have received
scanty attention from the research and policy communities.
This problem is compounded by limited knowledge of what
the structural determinants of CA are, especially, for West
African (W.A) countries. But there is more knowledge of this
for Europe (see e.g Gosse & Serranito, 2014) and some
industrial and developing countries (see e.g., Chin and
Prasad, 2003).
It is important to understand the factors that drive CA
fluctuations because they indicate the health of the
external sector of an economy. The understanding is also
important because unsustainable CA has been indicted
severally as a major factor responsible for macro instability
in Europe and East-Asian regional economic blocks
(Obstfeld and Rogoff, 2009).
Further, a perceptive stylized characterization of the W.A region reveals
large C.A deficit in recent years, poor growth, heavy reliance on import
and foreign aid. Therefore, given these realities, knowledge of C.A level
that may be considered to be sustainable based on the macroeconomic,
political and demographic characterization is needed to achieve better
regional economic integration in W.A.
OBJECTIVES
 1.. To examine the S.R and L.R determinant
of CA imbalance in West African Economies


2. To identify the “sustainable level of CA”
or “path of adjustment of CA” that is
consistent with regional integration in WA.
3. To analyze the process of adjustment to
“sustainable level” of CA
SYNOPSIS OF RESULT




(i). The effect of the structural determinants of C.A
tend to switch (especially in sign) from the S.R to the
L.R.
(ii). Real effective exchange rate (REER), TRADE
Openness (OPN), investment (INV), Income and Fiscal
Balance (FB) are the key determinants of C.A in the S.R
and L.R
(iii).
The adjustment path from disequilibrium to
equilibrium (SR or LR) is relatively fast.
(iv).
The deviation of C.A of many W.A countries
vary considerably from the “regionally sustainable
equilibrium path”.
TWO (2) MAJOR COMTRIBUTION OF THE PAPER:
 The study made use of second generation panel unit root
and cointegration test that improves on the power and
size of the first generation panel unit root and
cointegration testing schemes.

It calculates the country-specific CA adjustment path and
compares it with the regionally “sustainable level” of CA
using recent methods that are due to Gosse and Serranito
(2014) and Lane and Milesi-Ferreti (2012).
THE BALANCE OF THE PAPER IS AS FOLLOWS:





Section 2: Stylized Characterization of C.A data and its
hypothesized determinants –
including their densities, unconditional correlations, &
non-parametric regressions.
NOTE: Literature (Empirical and Theoretical) is contained
in the full paper earlier circulated. We not only separate
theoretical from empirical literature but also discriminate
contributions from industrial countries from their
developing countries counterparts.
(see full paper for up-date survey of literature).
Section 3: Empirical Strategy
Section 4: Results and Discussions
Section 5: Conclusion
SECTION 2: STYLIZED CHARACTERIZATION OF
C.A & ITS HYPOTHESIZED DETERMINANTS:
An inspection of the charts in Figure 1 shows that many countries (Benin,
Burkina Faso, Ghana, Guinea Bissau, Niger, Senegal Sierra Leone and Togo) have
had persistent and large C.A deficit since the 1980s until recently.
Figure 1: TREND OF C.A IN W.A COUNTRIES
A possible reason is the prolonged TOT deterioration. For Nigeria, however, the
CA position has been switching almost evenly from surplus to deficit position
over the same period. This is understandable given the crude oil component of
Nigeria’s export which is likely to be driven by booms and burst in the
international price of crude oil.
2.1 CROSS-SECTIONAL SCATTER CHARTS
Next we plot cross-sectional scatter charts of C.A and growth to check the
relationship – namely, to check if CA can tell us something about growth
performance in the region. The scatter charts shows that CA surplus or
deficit does not readily say much about growth as countries like Ghana and
Nigeria both has impressive growth performances but yet with divergence
CA account positions; Ghana has C.A deficit while Nigeria has CA surplus;
hence, there is need to further probe the nexus.
Figure 2: cross-sectional scatter Chart of C.A versus Growth in W.A (2011)
FIG 3: NON-PARAMETRIC REGRESSION LINE, SCATTER CHART,
HISTOGRAM & UNCONDITIONAL CORRELATION COEFFICIENT OF C.A
IN W.A
The result from Fig. 3 appears to validate those of cross sectional
scatter chart (Fig 2) which seems to suggest that CA deficit does not
readily say much about growth in WA.
3. EMPIRICAL STRATEGY





The empirical methodology we use is very similar to
Gosse and Serranito (2014), Lane and Milesi-Ferretti
(2012) and Dreger (2013).
The approach entails a three-stage strategy, thus:
Stage 1: we use a second generation type panel cointegration estimation framework due to Westerlund
(2007) to identify the L.R determinants of CA in W.A
Stage 2: we use dynamic OLS (DOLS ) to calculate the
equilibrium LR targets for CA in the region which we call
the “regionally sustainable level of CA” for WA.
Stage 3: We account for S.R dynamics by estimating
equilibrium error correction versions of the L.R
relationship.
3. Empirical Strategy (CONTD.1)
We set out with a specification of a reduced-form version of the
determinants of C.A in what follows:
𝑀
𝑗
𝐶𝐴𝑖,𝑡 = 𝛿𝑖 +
𝑋 + 𝜇𝑖,𝑡 … … … (1)
𝑗=1
𝑖 = 1, 2, … , 𝑁; 𝑡 = 1, … , 𝑇
where CA is the current account to GDP ratio, 𝛿𝑖 are the country fixed effects, 𝑋𝑗 are the
theoretically and empirically suggested structural determinants of the current account,
highlighted in the literature (see full paper), and M are the number of regressors.
In line with Lane and Milesi-Ferretti (2012), and Goss 𝑒 and Serranito (2014), a measure of
current account imbalances can be computed as the error term from the long-run reduced
form equation in (1). In other words, the difference between the fitted values from (1) and the
actual values can be treated as a measure of current account imbalances, while the
fitted values serve as the long-run equilibrium relationship.
3. Empirical Strategy (CONTD. 2)
If we find at least one cointegrating equation from the L.R reducedform equation, then it is possible to account for S.R dynamics by
estimating the ECM which will enable us to capture the speed of
adjustment of any deviations from the LR path of CA.
The estimable equation for the S.R adjustment path (the ECM) is given
as:

𝑀
𝑗
∆𝐶𝐴𝑖,𝑡 = 𝛿𝑖 +
𝜃𝑖 ∆𝑋 + 𝛾𝐸𝐶𝑇𝑖,𝑡 + 𝜇𝑖,𝑡 … … (2)
𝑗=1
where as before 𝛿𝑖 is a country fixed-effect, ∆ is a difference operator,
and ECT is the error-correction term which is the measure of CA
imbalances retrieved as the error-term from the long-run
cointegrating relationship in (1). In other words, the error-correction
term is given as;
𝛾𝐸𝐶𝑇𝑖,𝑡 = 𝐶𝐴𝑖,𝑡 − 𝛿𝑖 +
𝑗
𝑀
𝑗=1 𝑋
𝛽 ---------3)
The estimated value 𝛾 is then interpreted as the adjustment
coefficient which measures the speed of convergence, and is used to
calculate the half-period required for full equilibrium to be restored
after a current account imbalance. (see Full paper)
4.



RESULTS & DISCUSSIONS
Due to known size and power limitations of the first
generation panel unit root tests (see e.g Harris and
Travalis, 1999; Hadri,2003; Levin, Lin & Chu, 2002, LLC;
Im, Pesaran & Shin, 2003, IPS; Breitung and Das, 2005),
we conduct a battery of panel unit root tests and rely
mainly on the conclusions of the second generation test,
namely, Pesaran (2007). An up-to-date survey of the
limitations of the first generation tests can be found in
Battagi (2008).
Our panel unit root test, thus, include the cross-sectional
augmented IPS test (CIPS), simply referred to as the
Pesaran (2007) test in order to account for crosssectional dependence and heterogeneity in the panels.
This second generation test is also preferred to the first
as it not only account to cross sectional dependence but
is also not sensitive to the specification of the
deterministic term in the model
4.
RESULTS & DISCUSSIONS (CONTD.): TABLE 3-PANEL UNIT ROOT RESULTS:
KEYS: LLC IS FOR THE LEVIN ET AL. (2002) TEST; BREITUNG IS FOR THE BREITUNG AND DAS (2005) TEST; IPS IS FOR THE
IM ET AL. (2003) TEST; HADRI IS FOR THE HADRI (2000) TEST AND PESARAN IS FOR THE PESARAN (2007) TEST, WHICH
ACCOUNTS FOR CROSS SECTIONAL DEPENDENCE AND HETEROGENEITY. THE KEYS FOR REJECTION OF THE NULL
HYPOTHESIS ARE *, ** AND *** FOR THE 10, 5 AND 1 PERCENT CONDENCE LEVELS RESPECTIVELY.
Panel unit root results
Variable
CA/GDP
1st-generation tests
IPS
-4.50***(0.00)
Hadri
5.87***(0.00)
Pesaran
-1.85(0.37)
1.65(0.95)
-6.29***(0.00)
3.76***(0.00)
-2.11(0.37)
-12.85***(0.00)
-3.56***(0.00)
-12.96***(0.00)
2.52***(0.00)
-3.58***(0.00)
-6.45***(0.00)
6.78(0.99)
-2.28**(0.01)
5.92(0.99)
-12.94***(0.00)
5.77(0.99)
5.59***(0.00)
11.59***(0.00)
-3.58***(0.00)
-1.34(0.96)
Specification
Constant
LLC
-1.71**(0.04)
Breitung
-1.49*(0.06)
Constant & trend
0.06(0.52)
$\Delta$CA/GDP
Constant
GDPpc
Constant & trend
Constant
2nd-generation test
Constant & trend
1.74(0.95)
2.54(0.99)
-1.24(0.11)
9.01***(0.00)
-2.18(0.74)
$\Delta$GDPpc
Constant
-35.15***(0.00)
-5.17***(0.00)
-10.36***(0.00)
6.23***(0.00)
-3.14***(0.00)
Openness
Constant & trend
Constant
3.03(0.99)
-3.98***(0.00)
-7.02***(0.00)
-2.61***(0.00)
-11.40***(0.00)
-2.79***(0.00)
2.91***(0.00)
5.50***(0.00)
-3.14***(0.00)
-2.11(0.16)
Constant & trend
0.37(0.64)
-0.96(0.16)
-3.26***(0.00)
5.59***(0.00)
-2.09(0.74)
$\Delta$Openness
Constant
-17.05***(0.00)
-8.76***(0.00)
-13.00***(0.00)
-0.30(0.62)
-2.88***(0.00)
Investments
Constant & trend
Constant
-13.27***(0.00)
-2.95***(0.00)
-7.59***(0.00)
-1.77**(0.03)
-13.13***(0.00)
-2.57***(0.00)
3.09***(0.00)
6.61***(0.00)
-2.95***(0.00)
-2.21**(0.03)
Constant & trend
1.67(0.95)
-0.01(0.49)
-3.69***(0.00)
6.60***(0.00)
-2.91***(0.00)
$\Delta$Investments
Constant
-13.32***(0.00)
-6.58***(0.00)
-12.71***(0.00)
2.14**(0.01)
-3.29***(0.00)
Fiscal Balance
Constant & trend
Constant
-1.54*(0.06)
-4.69***(0.00)
-6.59***(0.00)
-0.75(0.22)
-13.19***(0.00)
-3.51***(0.00)
3.89***(0.00)
7.51***(0.00)
-3.41***(0.00)
-2.12*(0.07)
Constant & trend
5.74(0.99)
-1.88**(0.02)
-4.52***(0.00)
6.71***(0.00)
-2.29(0.58)
$\Delta$Fiscal Balance
Constant
-0.45(0.32)
-3.38***(0.00)
-12.72***(0.00)
1.21(0.11)
-3.28***(0.00)
Fin Openness
Constant & trend
Constant
6.84(0.99)
-1.31*(0.09)
-4.87***(0.00)
-0.21(0.41)
-13.05***(0.00)
-0.67(0.24)
3.16***(0.00)
9.51***(0.00)
-3.26***(0.00)
-0.98(0.99)
Constant & trend
-1.55*(0.06)
-1.84**(0.03)
-3.52***(0.00)
4.12***(0.00)
-1.54(0.99)
$\Delta$Fin Openness
Constant
-14.04***(0.00)
-12.15***(0.00)
-12.68***(0.00)
-0.65(0.74)
-1.71(0.58)
REER
Constant & trend
Constant
-10.06***(0.00)
2.23(0.98)
-11.42***(0.00)
3.99(0.99)
-12.78***(0.00)
14.71(0.99)
3.61***(0.00)
10.72***(0.00)
-2.02(0.92)
-1.35(0.99)
$\Delta$REER
Constant & trend
Constant
1.72(0.95)
-9.54***(0.00)
2.35(0.99)
-4.35***(0.00)
3.31(0.99)
-9.03***(0.00)
9.01***(0.00)
9.48***(0.00)
-1.37(0.99)
-2.34***(0.00)
TOT
Constant & trend
Constant
-8.24***(0.00)
9.41(0.99)
-3.61***(0.00)
-0.79(0.21)
-10.62***(0.00)
-2.37***(0.00)
6.79***(0.00)
4.39***(0.00)
-3.42***(0.00)
-1.53(0.82)
Constant & trend
9.94(0.99)
-0.45(0.32)
-3.59***(0.00)
4.42***(0.00)
-2.19(0.73)
Constant
-7.96***(0.00)
-5.79***(0.00)
-11.97***(0.00)
0.18(0.42)
-2.85***(0.00)
Constant & trend
-3.41***(0.00)
-5.91***(0.00)
-12.97***(0.00)
2.28**(0.01)
-3.41***(0.00)
$\Delta$TOT
From the panel unit result presented in Table 3, the result of the
1st generation tests are mixed.



4. RESULTS & DISCUSSIONS (CONTD.),
UNIT ROOT
From Table 3, the test result seems to depend mainly on
whether we include only a “constant” or a “constant and
trend”. For instance, using the first generation test, CA,
OPN,INV and FB were I(0) when we included only a
constant but the variables turned out to be I(1) when we
included a constant and trend.
Thus, given the limitations and non-dependability of the
1st generation tests, we relied mainly on the conclusions
from Pesaran (2007) test.
Overall, the panel unit root result indicates that all
variables are I(1) except financial openness whose result
was inconclusive
as it returned different results at
different tests and specifications used. Next, we present
the
Westerlund
(2007)’s
error
correction-based
cointegration test results.



PANEL (AND GROUP-MEAN) COINTEGRATION RESULT
In testing for panel co-integrating relationship, we depart from
the norm of using residual based panel cointegration test (
see e.g Pedroni (2004); Ouliaries (1990); Engle and Granger
(1987))
which
although
are
able
to
accommodate
heterogeneous
dynamics,
endogenous
regressors
and
individual-specific constants and trends, have been shown to
have low power (Ho, 2002) mainly because they require the
L.R cointegrating vector of the level-variables to be equal to
the S.R adjustment process of the differenced- variables.
As this requirement which is popularly called “common
factor restriction” often do not hold, panel cointegration
tests that are based on residual dynamics (rather than
structural dynamics of Westerlund (2007)-type) are often
bereaved of high power and size accuracy.
We, therefore, rely on Weterlund (2007) panel cointegration
test which is based on testing the null of no cointegration in
the structural (rather than residual) dynamics and thus
requires no common factor restriction.
TABLE 4: PANEL (AND GROUP-MEAN) CO-INTEGRATION RESULT
(CONTD.1)
s/no
Structural specification
Westerlund test statistic
P_tau
p-val
P_alpha
p-val
1
CA/GDP, TOT, OPN, REER, Finop, GDPpc, FSB
-5.27
(0.99)
-5.66
(0.99)
2
CA/GDP, TOT, OPN, REER, Finop, FSB
-14.7
(0.00)
-9.27
(0.82)
3
CA/GDP, OPN, REER, GDPpc, FSB, INV
-16.03
(0.00)
-20.35
(0.00)
4
CA/GDP, TOT, OPN, REER, GDPpc, FSB, INV
-6.07
(0.99)
-5.78
(0.99)
5
CA/GDP, OPN, REER, GDPpc, FSB
-15.47
(0.00)
-21.03
(0.00)
6
CA/GDP, TOT, OPN, REER, Finop, GDPpc, INV
-4.63
(0.99)
-2.51
(0.99)
7
CA/GDP, OPN, REER, FSB, GDPpc, Finop
-14.25
(0.00)
-19.34
(0.00)
8
CA/GDP, OPN, REER, FSB, Finop
-15.99
(0.00)
-21.65
(0.00)
Note: The test statistic P is based on Westerlund (2007) asymptotic results . P is the test statistic after normalization by
cross-sectional averages of the effective number of observations per country. The keys for rejection of the null hypothesis of
no cointegration are *, ** and *** for the 10, 5 and 1 percent confidence levels respectively. The probability values of the test
statistics are based on 399 bootstrap resampling, and are reported in parenthesis.
The Westerlund (2007) panel cointegration result presented in
Table 4 indicates that once TOT is included into any of the
specifications (row 1 through 8), we are bound to accept the Ho
of no cointegration.
PANEL (AND GROUP-MEAN) COINTEGRATION RESULT (CONTD.2)



This may be explained by the well-known result in the
literature (see e.g Gosse and Serranito, 2014) that EXR
rather than TOT is more likely to be a fundamental
determinant of CA as it represents a better proxy for
price competitiveness in W.A.
The result in Table 4, thus, shows that there are four
different combination of CA and its structural
determinants for which there is a cointegrating
relationship. These are the specifications: 3, 5, 7 and 8.
The result also implies that there are at least four
different combinations of equilibrium targets for the CA
that would be consistent with regional integration in WA.
Next, we estimate cointegrating vectors for the
significant relationship (specifications 3,5, 7 and 8) using
Panel DOLS.
TABLE 5: ESTIMATE OF THE COINTEGRATING VECTORS BY DOLS:
DEPENDENT VARIABLE CA/GDP
Structural Specifications
Model 3
0.013(-0.024)
Model 5
Model 7
Model 8
-0.021 (-0.028)
-0.0208 (-0.029)
0.013 (-0.024)
REER
-0.0016** (-0.001)
-0.378 (-0.573)
-0.361 (-0.576)
-0.444 (-0.467)
LGDPpc
-0.521 (-2.469)
0.637 (-2.951)
0.676 (-2.958)
FSB
-0.162** (-0.07)
-0.337***(-0.09)
-0.330*** (-0.09)
INV
-0.359***(-0.05)
Openness
-0.379*** (-0.047)
0.413 (-3.548)
FINOP
-0.154** (-0.078)
-2.14 (-2.865)
ote: The long-run co-integrating vectors are estimated by dynamic OLS for cointegrated panel data with
omogeneous long-run covariance structure across cross-sectional units, as in; Kao and Chiang (2000).
tandard errors of the estimates are in parenthesis, and the keys; *, ** and ***, are for the 10, 5, and 1
ercent confidence levels respectively. The model numbers correspond to the numbering of the structural
pecifications in Table 4
.
Our choice of Panel DOLS over the commonly used FM OLS is
motivated primarily by the finite sample property in Kao and
Chiang (2000) which shows that the DOLS may be more
promising than the OLS or FMOLS estimators in estimating
cointegrating panel regressions.
ESTIMATES OF THE COINTEGRATING VECTORS BY DOLS:
DEPENDENT VARIABLE CA/GDP (CONTD.1)



Table 5 shows the L.R equilibrium relationship (estimates of
cointegrating vectors) which serves as targets that could be
consistent with regional integration in W.A.
From Table 5, model 3 is selected as the best-performing model
based on both the number of statistically significant variables and
the consistency of sign of the coefficients with the theoretical
expectations.
Model 3 show that REER and INV are both negative and
statistically significant as expected. The only puzzling result from
TABLE 5 is that the sign of FB is negative rather than positive.
Nonetheless, there is no consensus about the theoretical
expectations of FB even though many past studies have found it to
be positively related to CA (See e.g Gose and Serranito, 2014;
Chinn and Ho, 2008;and Lane and Milesi-Ferretti, 2012). We now
turn to the estimate of S.R relationship.
NEXT WE PRESENT RESULT OF S.R ADJUSTMENT (VECM)
TABLE 6: DISEQUILIBRIUM ADJUSTED ESTIMATES: DEPENDENT
VARIABLE CA/GDP
∆Openness
∆REER
∆lnGDPpc
∆Fiscal Balance
∆Fiscal Balance_{t-1}
∆Investments_{t-1}
(1)
POLS
-0.0437**
(0.023)
1.2
(1.098)
8.739***
(-3.295)
-0.170**
(0.074)
-0.0745
(0.071)
-0.387***
(0.043)
(2)
FE
-0.0477**
(0.023)
2.202*
(1.251)
8.574***
(3.475)
-0.163**
(0.076)
-0.0728
(0.073)
-0.384***
(0.044)
(3)
RE
-0.0437**
-0.023
1.2
-1.098
8.739***
-3.295
-0.170**
-0.074
-0.0745
-0.071
-0.387***
-0.043
-0.423***
-0.043
-0.102
-0.231
(4)
POLS
-0.0448**
-0.023
2.134*
-1.225
9.297***
-3.307
-0.173***
-0.074
-0.0871
-0.072
-0.386***
-0.043
-0.0278*
-0.016
0.306
-3.036
-0.424***
-0.043
0.0672
-0.251
(5)
FE
-0.0492**
(0.023)
3.109**
(1.336)
9.089***
(3.478)
-0.166**
(0.076)
-0.088
(0.074)
-0.381***
(0.044)
-0.0366**
(0.019)
0.873
(3.104)
-0.417***
(0.043)
0.0632
(0.272)
(6)
RE
-0.0448**
-0.023
2.134*
-1.225
9.297***
-3.307
-0.173***
-0.074
-0.0871
-0.072
-0.386***
-0.043
-0.0278*
-0.016
0.306
-3.036
-0.424***
-0.043
0.0672
-0.251
-0.423***
(0.043)
-0.102
(0.231)
-0.417***
(0.043)
-0.19
(0.239)
No
No
Yes
Yes
4.27
(0.748)
1.28
Yes
No
No
No
Yes
Yes
6.17
(0.723)
1.28
Yes
No
Inflation$_{t-1}
∆Fin. Openness$_{t-1}
ECT_{t-1}
Constant
R^{2}
Country effects
Time effects
Hausman test: FE vs. RE
Half-life of adjustment
1.26
1.26
1.25
1.26
Note:
The error correction
term, ECT, is computed
from the
structural long-run
equation of Model 3 in Table 5. POLS, FE, and RE are abbreviations for the
pooled OLS, fixed effect, and random effect estimators respectively. The
estimated half-life of the adjustment (in years) back to equilibrium is computed
as; ln(0:5)= / ln (1− 𝛾). Standard errors of the estimates are in parenthesis, and
the keys; *, ** and ***, are for the 10, 5, and 1 percent confidence levels
respectively.
DISEQUILIBRIUM ADJUSTED ESTIMATES: DEPENDENT
VARIABLE CA/GDP (CONTD.1)





Making use of the most plausible L.R equilibrium relationship (Model 3 of
Table 5), we now focus on how disequilibrium in the established L.R
relationship are corrected in the S.R using panel VECM.
In
implementing
the VECM, we relied on three different panel data
estimators, namely; Pooled OLS, Fixed Effect (FE), and Random Effect (RE)
models based on Hausman’s specification tests.
Column 1, 2 and 3 of Table 6 report results of S.R determinants of CA when
we used exactly the same variables from the best-performing L.R model
(model 3 in TABLE 5). In column 4, 5 and 6, we report results of S.R
determinants when we include two additional S.R determinats (inflation and
financial openness) that are meant to improve the fitness of the SR model.
Because results do not significantly change with different estimators, we
concentrate on discussing the results from FE estimators especially because
the Hausman’s test does not discriminate against it.
One of the fascinating results from Table 6 is that the coefficient of
REER seems to be switching from positive sign in the S.R (see
columns 2 & 5 in Table 6) to negative sign in the L.R (see column 1
in Table 5).
REER
In the S.R, a depreciation of REER improves the C.A status but worsens in in
the L.R. This is comparable to the result in the OECD countries for which Gosse
and Serranito (2014) found that the short-run enhancing effect of REER
depreciation is lower in W.A when compared to OECD countries.

DISEQUILIBRIUM ADJUSTED ESTIMATES: DEPENDENT
VARIABLE CA/GDP (CONTD.2)
Trade openness (OPN)

Expectedly, trade openness worsened CA position in the S.R while the L.R effect can be
considered to be ambiguous. The worsening effect in the S.R can be explained by the
fact that the region export mainly primary commodities with little or no technological
embodiment and imports tertiary goods, manufactured goods and services with high net
value, hence the worsening of the CA .

Income (GDP)

The S.R effect of GDP on CA which is positive is difficult to rationalize unlike the L.R
effect which is negative (in line with the prediction of real business theory – namely, that
increase in AD raises income by increasing INV more than saving which therefore leads
to CA deficit.

Investment (INV)

The negative effect of INV on CA is in line with the prediction of equilibrium open
economy business cycle theory that rising investment would lead to worsening of C.A
position.

Fiscal Balance: Worsened C.A position both in the S.R and L.R

Error Correction Term (ECTt-1)

The effect of the error correction term is negative and significant for all the models
(with an average of around -0.42). This coefficient implies that every year about 42% of
disequilibrium in C.A dynamics of previous year is corrected. This speed of adjustment is
relatively fast compared to that of OECD countries (obtained by Gosse and
Serranito,2014); it, therefore, point to the potential to achieve equilibrium in W.A in the
S.R to medium term period. But the speed of adjustment diminishes with time and as the
gap reduces.
Next we examine the trend of L.R adjustment path (LAP) of CA vis-à-vis the actual path
(AP)and estimated adjustment path (EAP) of CA so as to check positions (how far or near)
W.A economies are from the regionally sustainable path of CA.

FIGURE 4: ACTUAL, EQUILIBRIUM AND ESTIMATED ADJUSTMENT PATH OF THE CURRENT
ACCOUNT IN WEST AFRICAN COUNTRIES
Note: The keys; AP, EAP, and LRP are for the Actual Path, Estimated Adjustment Path, and
the Long-run path of current account in West Africa, respectively. Estimates of the long-run
path are based on the co-integrating vector of Model 3 in Table 5 (best performing L.R
equation), while the estimated adjustment path is based on the fixed effect equilibrium
correction estimates of Column 5 in Table 6 (well performing S.R equation).
WHAT THEN ARE THE POSITIONS OF W.A COUNTRIES W.R.T
REGIONALLY SUSTAINABLE EQUILIBRIUM C.A PATH (L.R TARGET
OF C.A)?
Figure 4 plots the actual path of the C.A (AP, blue solid lines) the
estimated adjustment path (EAP, black dot-dash lines) and the L.R
structural equilibrium path (LRP, red dashed lines) for all the
countries in the sample. Figure 4 shows that there is
considerable variation by countries in the nature of responses.
In particular, 6 countries; Burkina Faso, Ghana, Guinea, Mali and
Togo are relatively far from the regional structural equilibrium
path while 6 other countries; Cape Verde, Guinea Bissau,
Mauritania, Niger, Sierra Leone and the Gambia have adjustment
path (EAP) that are relatively closer to the regional L.R structural
equilibrium path. Cote D’Ivoire and Benin may be considered to be
at intermediate range
The result for Nigeria is slightly different from others because
although the AP and EAP are close to the LAP, the level of
volatility is pretty much the highest in the region. We recognize
that this may have been driven by volatility in oil prices coupled
with macroeconomic instability. Thus, for Nigeria, what is needed
is an effective stabilization mechanism such as a carefully
implemented Sovereign Wealth Fund (SWF) to help mitigate the
actual account imbalances in Nigeria.
5. CONCLUSIONS
This study committed itself to providing answers to three basic questions: (i) what
are the S.R and L.R determinants of CA in West Africa; (ii) is there any sustainable
C.A path that is consistent with regional integration in W.A; (iii) what is the
process of adjustment towards that path.
The key points from the results are:
(i)The effect of the structural determinants of C.A tend to switch
(especially in sign) from the S.R to the L.R. For instance, for REER and
income, the effects were positive in the S.R but switched to negative in
the L.R
(ii)There are five major determinants of C.A for the region, namely: trade
openness, REER, income, fiscal balance and investment
(iii)The speed of adjustment back to equilibrium is relatively fast (about
42%) compared to 15 % obtained for OECD countries by Gosse and
Serranito (2014).
(iv)Finally, there is considerable variation in the deviation of W.A
countries from the regionally sustainable path.
Six countries are relatively far from the regionally sustainable path; six
other countries as relatively closer; two countries fall into a sort of
intermediate range; one country (Nigeria) stands out as its volatility is
pretty much the highest in the region even though its adjustment
path is quite close to the regionally sustainable level.
THANKS
THANKS
THANKS
THANK YOU FOR
YOUR KIND
ATTENTION