Exchange rate volatility and manufacturing exports in South Africa

Abstract

The primary objective of this study is to investigate the impact of exchange rate volatility on South Africa's manufacturing exports to the United States for the period 1990Q1 to 2014Q1. The study employs the EGARCH model to measure exchange rate volatility, and the ARDL bounds tests as developed by Pesaran, Shin and Smith to determine the longrun and short-run effects of exchange rate volatility on the country's manufacturing exports. The study also carries out a Granger causality test between real exchange rates and exports of manufactured products. The study results show that an increase in exchange rate volatility has a significant positive effect on manufacturing exports in the long run. However, the results are insignificant in the short run. It is also found that real exchange rates Granger cause manufacturing exports. Manufacturing exports, however, do not Granger cause real exchange rates.

Full text

“Exchange rate volatility and manufacturing exports in South Africa”
AUTH ORS Arnauld Ishimwe
Harold Ngalawa http://orcid.org/0000-0002-1946-3983
ARTICLE INFO Arnauld Ishimwe and Harold Ngalawa (2015). Exchange rate volatility and
manufacturing exports in South Africa. Banks and Bank Systems, 10(3), 29-38
JOURNAL "Banks and Bank Systems"
FOUNDER LLC “Consulting Publishing Company “Business Perspectives”
NUMBER OF REFERENCES
0
NUMBER OF FIGURES
0
NUMBER OF TABLES
0
© The author(s) 2019. This publication is an open access article.
businessperspectives.org

Banks and Bank Systems, Volume 10, Issue 3, 2015
29
Arnauld Ishimwe (South Africa), Harold Ngalawa (South Africa)
Exchange rate volatility and manufacturing exports in South Africa
Abstract
The primary objective of this study is to investigate the impact of exchange rate volatility on South Africa’s manufac-
turing exports to the United States for the period 1990Q1 to 2014Q1. The study employs the EGARCH model to meas-
ure exchange rate volatility, and the ARDL bounds tests as developed by Pesaran, Shin and Smith to determine the long-
run and short-run effects of exchange rate volatility on the country’s manufacturing exports. The study also carries out a
Granger causality test between real exchange rates and exports of manufactured products. The study results show that an
increase in exchange rate volatility has a significant positive effect on manufacturing exports in the long run. However, the
results are insignificant in the short run. It is also found that real exchange rates Granger cause manufacturing exports. Manu-
facturing exports, however, do not Granger cause real exchange rates.
Keywords: exchange rate volatility, EGARCH, ARDL bounds tests.
JEL Classification: F10, F31, F41.
Introduction¤
The relationship between exchange rate fluctuations
and international trade has been a subject of debate for
many years (Obi et al., 2013). It has generally been
argued that a rise in exchange rate risk causes eco-
nomic agents to invest in less risky assets and raises
the level of risk to trade, which in turn reduces the
level of trading activity (Ndung’u, 1999; Walters and
de Beer, 1999; Bah and Amusa, 2003). The literature,
however, presents confusing and contradictory theo-
retical and empirical outcomes on this issue (see Se-
kantsi, 2011), which prompted Klaassen (1999) to
argue for the necessity of more empirical studies, es-
pecially in less developed countries where deficient
time series have been blamed for the lack of adequate
studies (Vergil, 2002; Takaendesa et al., 2006;
Sekantsi, 2011).
Different studies on the relationship between ex-
change rate volatility and manufacturing exports have
obtained different results (see, for example, Hook and
Boon, 2000; Kumar and Dhawans, 1991; Arize et al.,
2000; De Vita and Abbott, 2004; Morgenroth, 2000).
In the case of South Africa, there are a few studies that
have been carried out on the subject. Using ARCH
and GARCH frameworks, Bah and Amusa (2003)
found a statistically significant inverse relationship
between exchange rate fluctuations and exports from
South Africa to the United States of America (hereaf-
ter the US) during the period 1990Q1 to 2000Q4.
Takaendesa et al. (2006) extended this study by using
quarterly data from 1992 to 2004. Employing the ex-
ponential general autoregressive conditional hete-
roskedasticity (EGARCH) approach, they found simi-
lar results to those of Bah and Amusa (2003). Sekantsi
¤ Arnauld Ishimwe, Harold Ngalawa, 2015.
Arnauld Ishimwe, MCOM (Economics), School of Accounting, Eco-
nomics & Finance, University of KwaZulu-Natal, Westville Campus,
Private Bag X54001, Durban, South Africa.
Harold Ngalawa, Ph.D., Senior Lecturer, School of Accounting, Eco-
nomics & Finance, University of KwaZulu-Natal, Westville Campus,
Private Bag X54001, Durban, South Africa.
(2011) used higher frequency (monthly) data for the
period January 1995 to February 2007, to analyze the
same relationship. Using the autoregressive distributed
lag (ARDL) bounds test approach, Sekantsi (2011)
found that real exchange rate volatility influences
exports significantly and negatively, consistent with
Bah and Amusa (2003) and Takaendesa et al. (2006).
However, Todani and Munyama (2005) found posi-
tive relationships in some instances and insignificant
results in others when they investigated the impact of
exchange rate volatility on aggregate exports (goods,
services and gold exports) of South Africa for the
period 1984 to 2004. The conflicting findings under-
score the absence of consensus on the relationship
between exchange rate fluctuations and manufacturing
exports in South Africa. This study, therefore, contri-
butes to the literature by attempting to provide more
updated empirical research to the debate. The study
employs the EGARCH model to measure exchange
rate volatility, and the Pesaran et al. (2001) ARDL
bounds tests to determine the long-run and short-run
effects of exchange rate volatility on the country’s
manufacturing exports to the US for the period
1990Q1 to 2014Q1. The study also carries out a Gran-
ger causality test between real exchange rate instabili-
ty and South Africa’s manufactured exports to the US.
Following this introduction, the rest of the paper is
structured in five sections. Section 1 is a discussion of
South Africa’s exchange rate fluctuations and trading
with the US. A review of the literature of exchange
rate volatility and export performance of manufac-
tured goods is carried out in Section 2. Section 3
presents the methodology. Estimation results are dis-
cussed in Section 4 followed by a summary and con-
clusion in final Section.
1. Exchange rate behavior and manufacturing
export performance in South Africa: 1990-2014
In order to cope with economic and political crises
the country faced after the end of apartheid in 1994,
South Africa dedicated its efforts to stabilization

Banks and Bank Systems, Volume 10, Issue 3, 2015
30
measures in the domestic foreign exchange market
(Van der Merwe, 1996). This was done through
many changes in the exchange rate regime. In the
post-apartheid era, three main exchange rate re-
gimes have been adopted by South Africa, as pre-
sented in Table 1.
Table 1. South Africa’s exchange rate regimes since 1985
Episode Period Exchange rate regime
I Sept 1985
–
Feb 1995 Two-tier system is re-established, with commercial and financial Rand.
II Mar 1995
–
Jan 2000 Unitary exchange rate: managed float in Rand.
III Feb 2000 – present Unitary exchange rate: free floating Rand with inflation targeting framework of
monetary policy.
Source: Mtonga (2011).
The financial sanctions imposed on apartheid South
Africa in the 1980s and 1990s forced the South
African Reserve Bank (SARB) to enter the foreign
exchange market as an active participant with di-
rect control measures to regulate capital flows and
mo-netary reserves (Van der Merwe, 1996). Van
der Merwe (1996) affirms that South Africa took
steps in the development of a forward market in the
first two years of the post-apartheid era (1994-1995)
as part of financial reforms. This happened with
progressive relaxation of foreign exchange controls
and a decline in SARB’s involvement in the foreign
exchange market.
From March 1995 to September 2000, South Africa
adopted a unitary exchange rate under a managed
floating Rand. This development occurred follo-
wing South Africa’s political reconciliation in 1994
that not only heralded a political transition from the
apartheid regime to inclusive democracy, but also
ended the country’s economic isolation. Aron et al.
(2000) argue that this change of regime was a huge
step toward the liberalization of South Africa,
which reinstated the country into the global econo-
my. The financial liberalization resulted in the re-
moval of exchange rate control regulations (Mton-
ga, 2011). Under this regime, SARB neither set the
fixed rate to be quoted by banks nor predetermined
its own rate of buying and selling dollars. Nattrass
et al. (2002) state that the administered float allows
the currency to fluctuate and the Reserve Bank to
intervene and diminish the market’s short-run fluc-
tuations.
In February 2000, South Africa adopted inflation
targeting, which was followed by implementation of
a free floating exchange rate. The current monetary
policy framework is employed to allow market
forces to determine the exchange rate without in-
terfering in the market (South African Reserve
Bank (SARB), 2012). The Reserve Bank, howe-
ver, still has authority over the foreign exchange
rate by participating in the market through pur-
chases and sales of foreign currency, even though
it stopped the control of foreign exchange rate directly
(Mtonga, 2011).
Figure 1 in the Appendix shows trends of the real
exchange rate of the South African Rand per US
Dollar from 1990Q1 to 2014Q1. The figure shows a
peak in 2002Q1. According to the Myburgh Com-
mission of Inquiry, the 2002 depreciation of the
Rand against the US Dollar was caused by (1) a
continuous slowdown in global economic activity;
(2) contagion from events in Argentina; (3) a wor-
sening in the current account of the balance of pay-
ments in 2001Q3; and (4) a complete shift from a
surplus position in the financial account of the ba-
lance of payments in 2001Q3 to a deficit in 2001Q4
(Bhundia and Gottschalk, 2003).
Since 1994, there has been a significant increase in
exports of South African goods to its major trading
partners such as Germany, US, China, and Japan
(see Figure 2 in the Appendix). Figure 2 shows that
the US has been the largest importer of South Afri-
can manufactured exports, followed by Japan and
Germany, in that order. It is also observed that there
has been a considerable increase in exports to Chi-
na, especially since 2002.
Figure 3 in the Appendix shows that there is an appar-
ent inverse relationship between manufacturing ex-
ports and volatility. The periods of low volatility in the
exchange rate tend to be followed by an increase in
South Africa’s exports to the US. However, we also
observe that higher exchange rate volatility (i.e., in
1998, 2001 and 2002, 2008) appears to be positively
related to an increase in exports to the US. This
represents the ambiguity regarding the relation be-
tween exchange rate fluctuations and manufactured
exports.
2. Literature review
Exchange rate fluctuations are largely explained by
macroeconomic variables. A high demand for South
African exports relative to its imports, for instance,
may increase the country’s terms of trade, which in
turn may lead to an appreciation of the South Afri-
can Rand against other currencies. Conversely, if
the price of South African exports increase com-
pared to its imports, the Rand may depreciate rela-
tive to foreign currency, as it is expected to be on
high demand compared to other trading currencies

Banks and Bank Systems, Volume 10, Issue 3, 2015
31
(Coudert et al., 2008). The rate of interest is another
determinant of exchange rate variation, and the two
are positively correlated (see Hnatkovska et al.,
2008). This means, for example, that an increase in
South Africa’s interest rates will attract capital in-
flows, which may lead to an appreciation of the
Rand. Money supply changes operating through
interest rates also influence exchange rate move-
ments. A rise in money supply is expected to put
downward pressure on interest rates, consequently
leading to a decrease in the rate of return on domes-
tic financial assets and a fall in the value of the do-
mestic currency (see Krugman and Obstfeld, 2006).
According to Kandil and Mirzaie (2003), a devalua-
tion of the domestic currency has a positive effect
on the demand for domestic goods by foreigners.
This occurs because a depreciation of the domestic
currency increases the value of foreign currency
which allows foreigners to buy more of domestic
goods since they are cheaper (in foreign currency)
compared to foreign goods.
Several studies have been carried out on the rela-
tionship between exchange rates and export perfor-
mance. Many of these have found an inverse rela-
tionship between exchange rate fluctuations and
trade (see, for example, Hook and Boon, 2000;
Kumar and Dhawans, 1991; Arize et al., 2000);
several have found an insignificant relationship
between the two variables (see, for example, De
Vita and Abbott, 2004; Morgenroth, 2000); some
have shown that an increase in exchange rate uncer-
tainty increases trade flows (see, for example, To-
dani and Munyama, 2005); and others show that the
correlation between exchange rate fluctuations and
trade is poor (see, for example, Musonda, 2001;
Adubi and Okumadewa, 1999).
There are four studies on the effect of exchange rate
fluctuation on trade in South Africa that we are
aware of. These are Bah and Amusa (2003), Todani
and Munyama (2005), Takaendesa et al. (2006), and
Sekantsi (2011). Bah and Amusa (2003) investi-
gated the effect of exchange rate volatility on South
African exports to the US. Using quarterly data
from 1990Q1 to 2000Q4 in ARCH and GARCH
models, they found that both in the long run and the
short run, real exchange rate fluctuations tend to
have a negative and statistically signi-
ficant effect on exports. Takaendesa et al. (2006)
extended the Bah and Amusa (2003) study to 2004Q4,
although the starting period is also moved forward to
1992Q1. Using the EGARCH model of Nelson (1990)
to measure exchange rate volatility, they found similar
results to Bah and Amusa (2003).
Todani and Munyama (2005) investigated the im-
pact of exchange rate fluctuations on aggregate
exports of South Africa to the rest of the world
using quarterly data from 1984 to 2004 using the
ARDL bounds testing model. To measure exchange
rate volatility, they used the GARCH (1, 1) and the
moving average standard deviation. Todani and
Munyama (2005) found a positive but insignificant
correlation between exchange rate fluctuations and
the exports of manufactured products from South
Africa using different measures of exchange rate
variability. Sekantsi (2011) examined the impact of
real exchange rate volatility on South Africa’s ex-
ports to the US for the period 1995 to 2007. Using
the GARCH model to measure exchange rate vola-
tility, the study also estimated long-run coefficients
using the ARDL model. Sekantsi (2011) found that
exchange rate fluctuations are significantly and
inversely related to exports.
3. Methodology
3.1. Model specification. Following Savvides
(1992), Todani and Munyama (2005) and Sekantsi
(2011), the estimated exports equation is given by:
01 2 3 4tttttt
exp = Į+Įgdp +Įrer + Įdummy +Įvol + İ,
(1)
where expt represents South Africa’s manufactur-
ing exports to the US, gdpt is real income of the
foreign country (the US), rert is the exchange rate
of the South African Rand to the US Dollar, volt is
exchange rate volatility, dummyt is a dummy variable
representing the African Growth and Opportunity
Act (AGOA) bilateral trade agreement with the US
signed in 2000, ܽ଴ is a constant and
H
t is a white
noise error term. The variables expt, gdpt and rert
are expressed in natural logarithms.
3.2. Definition of variables, data and sources of
data. The study uses quarterly data covering the
period 1990Q1 to 2014Q1. Data on South Afri-
ca’s EXPORTS to the US were collected from the US
Census Bureau and are expressed in US Dollars. In
order to generate real exports, we follow Vergil (2002)
and Takaendesa et al. (2006) who deflated the nomin-
al value of South Africa’s exports to the US with the
consumer price index of the US. Though demand
theory proposes that the volume rather than the value
of manufacturing exports be used, this study uses
manufacturing exports values for easy comparability
with previous studies (see, for example, Bay and
Amusa, 2003; Todani and Munyama, 2005; Ta-
keandesa et al., 2008; and Sekantsi, 2011).

Banks and Bank Systems, Volume 10, Issue 3, 2015
32
Real income data of the foreign country, which is
proxied by the real GDP of the US, was obtained
from International Financial Statistics (IFS), a data-
base of the International Monetary Fund (IMF). The
bilateral real exchange rate was computed using the
formula:
,
US
SA
NER* CPI
rer = CPI
(2)
where CPIUS is the consumer price index for the
US, CPISA is the consumer price index for South
Africa and ܰܧܴ is the nominal exchange rate in
Rands per US Dollar. Data for the nominal ex-
change rate were obtained from the South African
Reserve Bank (SARB). The CPI data of both countries
were obtained from the Organization for Economic
Cooperation and Development (OECD) statistics.
3.3. Measuring exchange rate volatility. The autore-
gressive conditional heteroscedasticity (ARCH) mod-
el, the general ARCH (GARCH) model, and the ex-
ponential GARCH (EGARCH) model are some
common measures of exchange rate volatility (see
Engel, 1982; Bollerslev, 1986; Todani and Munyama,
2005; Takaendesa et al., 2006; Sekantsi, 2011). This
study adopts the EGARCH method of Nelson (1990),
consistent with Takaendesa et al. (2006) and Su
(2010). The variance specification of the EGARCH
model can be presented as:
  
22
11
01 1 22
11
t- t-
tt-
t- t-
ȦȦ
ln ı=IJ+șln ı+Ȗ+ Æ ,
ıı
(3)
where
W
0 is the intercept term and
T
1
J
,׎ are parame-
ters to be estimated. The
T
1 parameter measures per-
sistence in conditional volatility in the economy; the
term
Z
t-1 symbolizes the ARCH term and measures
fluctuations in the previous period; and ߪ௧ିଵ
ଶ is the
GARCH term which represents the variance of the
previous period estimate.
One of the advantages of the EGARCH specification
is that even though the parameters are negative, ߪ௧
ଶ
would be positive. Accordingly, there would be no
violation of the positive variance conditions. In addi-
tion, unlike the GARCH specification, the
J
parameter
measures the leverage effect or the asymmetric order
(see Brooks, 2002; Su, 2010; Takaendesa et al., 2006).
If
J
= 0, the EGARCH model is symmetric because
the ׎ parameter denotes the symmetric or magnitude
effect of the model. If
J
< 0, there are positive shocks
in the economy that produce less fluctuations than
negative shocks. However, when
J
> 0, the opposite
applies, meaning positive shocks are more threatening
than negative shocks (Su, 2010).
3.4. Autoregressive distributed lag bounds testing
approach.This paper adopts the autoregressive distri-
buted lag (ARDL) bounds test approach proposed by
Pesaran et al. (2001). Among its many advantages,
this approach permits examining the presence of coin-
tegration without the need to recognize whether the
variables are stationary in levels, integrated of order
one or mutually cointegrated (Todani and Munyama,
2005). In addition, this procedure has small-sample
properties which are better than what is obtained in
Johansen (1991, 1995) and Engle and Granger (1987)
approaches (see Sekantsi, 2011). Following Todani
and Munyama (2005) and Pesaran et al. (2001), equa-
tion (1) can be rewritten as:
01 12 13141
51
10 0
1
00
gp
,
d
'    
-'' '
 ''
¦¦ ¦
¦¦
tt-t-t-t-
p
nm
t- i t-i k t-k k t -k
i= k= k
qv
k t -k k t-k t
k= k=
exp ș+ștɉexp ɉɉrer ɉvol
ɉdummy exp Ȗgdp įrer
ijvol dummy ȝ Ĭ

(4)
where ߠ଴ and ߠଵݐ are the constant and trend com-
ponents, respectively,
P
t is a white noise error term
and the remaining variables are similar to the va-
riables in equation (1). De Vita and Abbott (2004)
and Sekantsi (2011) argue that the nonexistence of
serial correlation in the residuals is explained by the
formation of the first difference explanatory va-
riables. The parameters ɉଶ, ɉଷ, ɉସ, and ɉହ
represent long-run coefficients that affect manufac-
turing exports.
In order to estimate equation (4), the starting point
is to determine the lag length that specifies the final
ARDL by using the general-to-specific approach
(Shin and Yu, 2006). The next step is to test if ma-
nufactured export products and the explanatory va-
riables are cointegrated. This is done by carrying out
a joint test for cointegration. After the model is found
to be cointegrated, the normalized long-run relation-
ship resulting from equation (4) is given by (see Pe-
saran and Shin, 1999):

Banks and Bank Systems, Volume 10, Issue 3, 2015
33
12 3 4 5 6 ,
tttttt
exp = + t + gdp + rer + dummy + vol + ȟ
E
EE E E E
(5)
where
E
1 = -T0/ɉ1; 2 = -T1/ɉ1, 3 = - ɉ2/ɉ1, 4 = - ɉ3/ɉ1, 5 =
- ɉ5/ɉ1,6 = - ɉ5/ɉ1 and ߦ௧ is an error term which is
assumed to be white noise. These long-run coeffi-
cients (3, 4, 5 and 6) correspond to the estimated
coefficients (
D
1,
D
2,
D
3 and
D
4) in equation (1) re-
spectively.
3.5. Granger causality. This study uses the Granger
causality test to determine the direction of causality
between manufacturing exports and the real ex-
change rate. According to Mousavi and Leelavathi
(2013), to carry out the Granger causality test, the
variables of interest must be stationary. Akaike and
Schwarz Information Criteria are used to find the op-
timal number of lags.
4. Estimation results
4.1. Exchange rate volatility. The Lagrange Multip-
lier (LM) – ARCH test shows that the value of the test
statistic (R2 * (number of observations = 3.75998)) is
greater than the probability of the chi-squared value
(0.1526), revealing the presence of ARCH effect in
the real exchange rate series. This prompts us to use
the EGARCH model. The anti-cipated conditional
variance of the exchange rate volatility in the
EGARCH model is summarized as follows:
  

22
-1 -1
-1 22
-1 -1
= -0.2877 + 0.908 + 0.1503 -0.3068
0.0016 0.0000 0.0000 0.0021
tt
tt
tt
P-va
ȦȦ
ln ıln ııı
lue .
(6)
In equation (6), all coefficients are statically signifi-
cant at 1%. The coefficient of the measure of persis-
tence ሾ݈݊ሺߪ௧ିଵ
ଶሻሿ shows that exchange rate volatility
does not die instantaneously following a shock. The
asymmetric parameter (0.1503) reveals that positive
shocks are more threatening than negative shocks
(see Su, 2010). To ascertain the robustness of the
exchange rate volatility estimates, we check if there
is any ARCH effect remaining in the EGARCH
residuals. A summary of the ARCH effect test is
presented in Table 2.
Table 2. Heteroskedasticity test: ARCH
F-statistic 0.035321 Prob. F (1.93) 0.8513
Obs*R-squared 0.036067
Prob. chi-square
(1) 0.8494
The Table shows that the F-statistic and Chi-square
statistic are insignificant. We, therefore, conclude
that there is no ARCH effect remaining in the ex-
change rate fluctuation. Next we carried out unit
root tests of the data using the Augmented Dickey-
Fuller (ADF) and the Phillips Peron (PP) tests. The
results show that all variables are integrated of the
first order, except for exchange rate volatility,
which is stationary in levels for both ADF and PP
tests. These findings provide further justification for
using the ARDL bounds test approach (see Table
A1 in the Appendix).
4.2. Estimation of the ARDL bounds test for
cointegration approach. In the estimated equation
(4), the optimal lag length was chosen by examining
the sequential modified likelihood ratio test statistic
(LR), Akaike information criterion (AIC), Schwarz
information criterion (SC), final prediction error
(FPE), and Hannan-Quinn information criterion
(HQ). FPE, AIC and HQ show seven lags and LR
and SC show six and three lags, respectively, as the
optimal lag lengths (Table of results available on
request from authors). Following Shin and Yu
(2006), we use the general-to-specific approach,
beginning from seven lags (max n = max m = max p
= max q= max v = 7), then removing all the va-
riables that are insignificant. The long-run estima-
tion results of the model are presented in Table 3.
Table 3. Results of the long-run cointegration equation
Manufacturing exports Constant Trend
Foreign
income Real exchange rate Exchange rate
volatility
Dummy
variable
Coefficients 1.000 -70.0406 -0.0198 2.9297 -0.9656 44.9146 0.2517
Std. error -0.0840 9.5818 0.0035 0.3493 0.1472 12.6412 0.0509
t-statistic -6.8843 -4.2287 -3.2873 4.8515 -3.7978 2.0559 2.8625
Prob. 0.000*** 0.0001*** 0.0015*** 0.000*** 0.0003*** 0.043** 0.0053***
Note: *, **, *** represent 10%, 5%, and 1% level of significance, respectively.
Using the critical values tabulated by Pesaran et al.
(2001), we observe that the F-statistic (10.2807) and
chi-square (51.4035) are statistically significant at 5
percent. We observe that the F-statistic (10.2807) is
greater than the upper-bound critical values of Pesaran
et al. (2001) at all levels of significance, indicating that

Banks and Bank Systems, Volume 10, Issue 3, 2015
34
there is cointegration between manufacturing exports
and the explanatory variables in equation (5).
Table 3 shows that exchange rate volatility is statisti-
cally significant and positively related to manufactu-
ring exports. This finding is consistent with De
Grauwe (1988), Arize et al. (2003), Todani and Mu-
nyama (2005), and Obi et al. (2013). The positive
relationship between exchange rate volatility and
manufacturing exports might be due to income effects
exceeding substitution effects. This relationship may
also be a result of the openness of the South African
economy (Todani and Munyama, 2005). This might
be a situation where exporters are aware that all excess
supply may not be consumed by the domestic market
in case trading becomes risky as exchange rate volatil-
ity increases. Therefore, exporters increase manufac-
turing exports as exchange rate volatility increases
with the intention of avoiding a fall in revenues and an
exchange rate risk exposure.
We also observe that the real exchange rate is nega-
tively related to manufacturing exports and statistical-
ly significant at 1 percent, which may suggest that the
negative effect of a devaluation on net exports in the J-
curve is persistent. This is also in line with a structu-
ralist view, which states that the depreciation of a
currency might have a negative effect on job creation
and production, which in turns negatively affects ex-
ports (Acar, 2000). Acar (2000) maintains that this
view works mostly for developing countries, where a
depreciation raises both costs of domestic productions
and imports. The foreign income coefficient is posi-
tive, higher than unity and statistically significant.
Even though this elasticity is high, it is consistent with
other studies (see, for example, Arize et al., 2000; Bah
and Amusa, 2003). Coefficients of foreign income
mostly range between 2.0 and 4.0 for both developed
and developing countries (Riedel, 1988). Riedel
(1988, 1989) argues that a high income elasticity
shows lack of action from the supply side of exports.
Arize (1990), however, states that increased export
penetration would result in high income elasticity.
Since the AGOA bilateral trade agreement was signed
in 2000, South Africa’s exports of manufactured pro-
ducts to the US market have increased (Bah and Amu-
sa, 2003). The coefficient of the AGOA bilateral trade
agreement was found to be statistically significant and
consistent with a priori theoretical expectations.
4.3. Error correction model. An error correction
model (ECM) was estimated to present the short-term
dynamics that exist between South Africa’s manufac-
turing exports to the US and its main determinants
(see Table 4 for the estimation results).
Table 4. Error correction model
Variables Coefficients Standard error t-statistic Probability value
Constant -0.069193 0.032028 -2.160412 0.0336**
Trend 0.000493 0.000414 1.190674 0.2371
D (exp (-4)) 0.267280 0.081954 3.261323 0.0016***
D (gdp) 4.186642 1.318292 3.175808 0.0021***
D (vol) 40.29242 14.94883 1.695356 0.1985
D (vol (-1)) -46.07119 14.61612 -1.152081 0.2022
ECM (-1) -0.483196 0.082928 -5.826681 0.0000***
The speed of adjustment towards long-run equilibrium
is found to be negative (-0.4832) and statistically sig-
nificant, showing that almost 48 percent of the dise-
quilibrium in the previous quarter are adjusted to their
long-run equilibrium in the current quarter. All short-
run explanatory variables were found to be statistically
significant, except exchange rate volatility.
4.4. Diagnostic tests. The model was tested for
normality, serial correlation, autoregressive condi-
tional heteroscedasticity and stability. Results of the
diagnostic tests show that there is no serial correla-
tion in the model, the error terms have equal va-
riance and the residuals are normally distributed
(see Table 5).
Table 5. Residual diagnostic test results
Test Null hypothesis t-statistic Probability value
Breusch-Godfrey LM-test No serial correlation 0.3561 0.7015
White test (Chi-sq) No conditional heteroscedasticity 0.7395 0.8419
Jarque-Bera(JB) There is a normal distribution 1.9574 0.3758
4.5. Granger causality test.The test of causality
between manufacturing exports and the real ex-
change rate is carried out using the Granger cau-
sality test. Table 6 presents the results of the test.

Banks and Bank Systems, Volume 10, Issue 3, 2015
35
Table 6. Granger causality test results
Dependent variable: manufacturing exports
Excluded Chi-sq Df Prob.
Real exchange rate 8.658541 2 0.0132
A
ll 8.658541 2 0.0132
Dependent variable: real exchange rate
Excluded Chi-sq Df Prob.
Manufacturing exports 4.129443 2 0.1269
A
ll 4.129443 2 0.1269
The Table shows a probability value of 0.0132 in a
test of whether the exchange rate Granger causes
manufacturing exports. We, therefore, conclude that
the real exchange rate Granger causes manufacturing
exports at 5%. Testing if manufacturing exports
Granger cause real exchange rates, a probability
value of 0.1269 reveals that we cannot reject the null
hypothesis. We, therefore, conclude that manufactu-
ring exports do not Granger cause exchange rates.
Summary conclusions and policy implications
The primary objective of this study was to examine
the long-run and short-run relationship between
exchange rate volatility and manufacturing exports
in South Africa, covering the period of 1990Q1 to
2014Q1. The study also analyzes the causality be-
tween exchange rates and manufacturing exports
from South Africa to the US. Using the EGARCH
model to determine the volatility of exchange rates
and the ARDL bounds test approach for cointegra-
tion, it has been found that exchange rate volatility
and manufacturing exports are positively related in
the long run. However, the real exchange rate and
export of manufactured products are observed to be
negatively associated. In the short run, the results of
exchange rate volatility and foreign income were
found to be insignificant. It is also established that
the speed of adjustment to equilibrium is nearly 50
percent and statistically significant. Thus, the study
finds no evidence that exchange rate volatility ad-
versely affects manufacturing exports in South
Africa. Therefore, central bank intervention in the
South African foreign exchange market to smoothen
exchange rate movements cannot be justified on the
basis that it encourages manufacturing exports.
References
1. Acar, M. (2000). Devaluation in Developing Countries: Expansionary or Contractionary? Journal of Economic
and Social Research, 2 (1), pp. 59-83.
2. Adubi, A. and Okumadewa, F. (1999). Price, Exchange Rate Volatility and Nigeria’s Agricultural Trade Flows: A
Dynamic Analysis, African Economic Research Consortium Research Paper No. 87.
3. Arize, A. (1990). An Econometric Investigation of Export Behavior in Seven Asian Developing Economies,
Applied Economics, 22, pp. 891-904.
4. Arize, A.C., Osang, T. and Slottje, J.D. (2000). Exchange Rate Volatility and Foreign Trade: Evidence from
Thirteen LDC’s, Journal of Business and Economic Statistics, 18 (1), pp. 10-17.
5. Arize, A.C., Osang, T. and Slottje, J.D. (2003). Exchange Rate Volatility in Latin America and its Impact on
Foreign Trade. Retrieved from: http://faculty.smu.edu/tosang/pdf/latin.pdf.
6. Aron, J., Elbadawi, I., and Kahn, B. (2000). Determinants of the Real Exchange Rate in South Africa. Centre for
the Study of African Economics. Working paper number WPS/97-16.
7. Bah, I. and Amusa, A. (2003). Real Exchange Rate Volatility and Foreign Trade: Evidence from South Africa’s
Export to United State, African Finance Journal, 5 (2), pp. 1-20.
8. Bhundia, A. and Gottschalk, J. (2003). Source of Nominal Exchange Rate Fluctuation in South Africa. IMF
Working Paper No. WP/03/252.
9. Bollerslev, T. (1986). Generalised Autoregressive Conditional Heteroscedasticity, Journal of Econometrics, 31,
pp. 307-327.
10. Brooks, C. (2002). Introductory Econometrics for Finance. Cambridge: Cambridge University Press.
11. Coudert, V., Couharde, C., and Mignon, V. (2008). Do Terms of Trade Drive Real Exchange Rates? Comparing Oil and
Commodity Currencies? Centre d’Etude Prospectives et d’Inforamtion International Working Paper No. 32.
12. De Grauwe, P. (1988). Exchange Rate Volatility and the Slowdown in Growth of International Trade, IMF Staff
Papers, 35 (1), pp. 63-84.
13. De Vita, G. and Abbott, A. (2004). The Impact of Exchange Rate Volatility on UK Exports to EU Countries,
Scottish Journal of Political Economy, 51 (1), pp. 63-81.
14. Engel, R. (1982). Autoregressive Conditional Heteroscedasticity with Estimates of the Variance of the United
Kingdom Inflation, Econometrica, 50, pp. 987-1007.
15. Engle, R.F. and Granger, C.W.J. (1987). Co-integration and Error Correction: Representation, Estimation, and
Testing, Econometrica, 2 (55), pp. 251-276.

Banks and Bank Systems, Volume 10, Issue 3, 2015
36
16. Hnatkovska, V., Lahiri, A., Vegh, C.A. (2008). Interest Rate and the Exchange Rate: A Non-Monotonic Tale.
National Bureau of Economic Research Working Paper No. 13925.
17. Hook, L.S. and Boon, T.H. (2000). Real Exchange Rate Volatility and Malaysian Exports to its Major Trading
Partners. University Putra Malaysia Working Paper No. 6.
18. Johansen, S. (1991). Estimation and Hypothesis Testing of Cointegration Vectors in Gaussian Vector
Autoregressive Models, Econometrica, 59 (6), pp. 1551-1580.
19. Johansen, S. (1995). Likelihood-Based Inference in Cointegration Vector Autoregressive Models. New York:
Oxford University Press.
20. Kandil, M. and Mirzaie, I.A. (2003). The Effect of Exchange Rate Fluctuations on Output and Prices: Evidence
from Developing Countries, International Monetary Fund Working Paper.
21. Klaassen, F. (1999). Why is it so Difficult to Find an Effect of Exchange Rate Risk on Trade? Retrieved from
http://greywww.kub.nl:2080/greyfiles/center/.
22. Krugman, P.R. and Obstfeld, M. (2006). International Economics: Theory and Policy (7th Ed.). Boston: Pearson
Addison-Wesley.
23. Kumar, R. and Dhawan, R. (1991). Exchange Rate Volatility and Pakistan’s Exports to the Developed World:
1974-1985, World Development, 19, pp. 1225-1240.
24. Morgenroth, L.W. (2000). Exchange Rate and Trade: The Case of Irish Exports to Britain, Applied Economics, 32,
pp. 107-110.
25. Mousavi, S., and Leelavathi, D.S. (2013). Agricultural Export and Exchange Rates in India: The Granger Causality
Approach, International Journal of Scientific and Research Publications, 3 (2), pp. 1-8.
26. Mtonga, E. (2011). Did it matter? Monetary Policy Regime Change and Exchange Rate Dynamics in South Africa,
paper presented at the Centre for the Study of African Economies, Oxford.
27. Musonda, A. (2001). Exchange Rate Volatility and Non-Traditional Exports Performance: Zambia (1965-1999),
Paper Presented at the AERC Biannual Workshop, Nairobi, May (2001)
28. Nattrass, N., Wakeford, J. and Muradzikwa, S. (2002). Macroeconomics Theory and Policy in South Africa (3rd
ed.). Cape Town: David Philip.
29. Ndung’u, N. (1999). Monetary and Exchange Rate Policy in Kenya. AERC Research Paper No. p. 94.
30. Nelson, D.B. (1990). Stationarity and Persistence in the GARCH (1, 1) Model, Econometric Theory, 6 (3),
pp. 318-334.
31. Obi, A., Ndou, P. and Peter, B. (2013). Assessing the Impact of Exchange Rate Volatility on the Competitiveness
of South Africa’s Agricultural Exports, Journal of Agricultural Science, 5 (10), pp. 227-250.
32. Pesaran, M. and Shin, Y. (1999). An Autoregressive Distributed Lag Modeling Approach to Cointegration
Analysis, in Strom, S. (Ed.), Econometrics and Economic Theory in the 20th Century: The Ragnar Frisch
Centennial Symposium. Cambridge: Cambridge University Press.
33. Pesaran, M.H., Shin, Y., and Smith, R.J. (2001). Bounds Testing Approaches to the Analysis of Level
Relationships, Journal of Applied Econometrics, 16, pp. 289-326.
34. Riedel, J. (1988). The Demand for LDC Exports of Manufactures: Estimates from Hong Kong, Economic Journal,
98, pp. 138-148.
35. Riedel, J. (1989). The Demand for LDC Exports of Manufactures: Estimates from Hong Kong: A Rejoinder,
Economic Journal, 99, pp. 467-70.
36. South African Reserve Bank (SARB) (2012). Inflation Targeting Framework: South African Reserve Bank.
Retrieved from: https://www.resbank.co.za/MonetaryPolicy/DecisionMaking/Pages/default.aspx>.
37. Savvides, A. (1992). Unanticipated Exchange Rate Variability and the Growth of International Trade,
Welwirtschaftliches Archives, 128, pp. 446-463.
38. Sekantsi, L. (2011). The Impact of Real Exchange Rate Volatility on South African Exports to the United States
(U.S.): A Bounds Test Approach, Review of Economic and Business Studies, 8, pp. 119-139.
39. Shin, Y. and Yu, B. (2006). An ADRL Approach to an Analysis of Asymmetric Long-Run Co-Integrating
Relationships, Mimeo. Leeds University Business School.
40. Su, C. (2010). Application of EGARCH Model to Estimate Financial Volatility of Daily Returns: The Empirical
Case of China (Unpublished Master’s Degree in Finance Project) University of Gothenburg.
41. Takaendesa, P., Tsheole, T. and Aziakpono, M. (2006). Real Exchange Rate Volatility and its Effect on Trade
Flows: New Evidence from South Africa, Studies in Economics and Econometrics, 30 (3), pp. 79-97.
42. Todani, K.R. and Munyama, T.V. (2005). Exchange Rate Volatility and Exports in South Africa. Retrieved from:
http://www.tips.org.za/files/773.pdf.
43. Van der Merwe, E.J. (1996). Exchange Rate Management Policies in South Africa: Recent Experience and
Prospects, South African Reserve Bank Occasional Paper No. 9.
44. Vergil, H. (2002). Exchange Rate Volatility in Turkey and its Effect on Trade Flows, Journal of Economic and
Social Research, 4 (1), pp. 83-99.
45. Walters, S. and De Beer, B. (1999). An Indicator of South Africa’s External Competitiveness, South African
Quarterly Bulletin, 213, September, pp. 54-67.

Banks and Bank Systems, Volume 10, Issue 3, 2015
37
Appendix
Source: South Africa Reserve Bank.
Fig. 1. Trend of the real exchange rate
Source: Department of Trade and Industry, South Africa.
Fig. 2. Export products by country of destination 1994-2010
Source: International Monetary Fund, International Financial Statistics.
Fig. 3. Manufacturing exports and exchange rate volatility in South Africa
Table A1. Summary of the augmented Dickey-Fuller and the Phillips Peron test results
Null hypothesis: exp,gdp,re
r
and vol have unit roots
Exogenous: constant and trend
ADF test statistics
A
DF test critical
values
PP test
statistics PP test critical values Order of integration
Manufacturing exports (exp) -10.38926 -2.589531*** -12.94891 -4.05753*** I (1)

Banks and Bank Systems, Volume 10, Issue 3, 2015
38
Table A1 (cont.). Summary of the augmented Dickey-Fuller and the Phillips Peron test results
Null hypothesis: exp,gdp,re
r
and vol have unit roots
Exogenous: constant and trend
Real exchange rate (re
r
) -7.791884 -3.500669*** -7.74961 -4.05753*** I (1)
Foreign income (inc) -5.898253 -3.500669*** -6.69778 -4.05753*** I (1)
Volatility (vol) -7.58564 -4.057528*** -8.5786 -4.25789*** I (0)
Notes: the LS method was used in the ADF test; maximum number of lags was set to 11; *, **, *** represent 10%, 5%, and 1%
level of significance, respectively.