Effects of Exchange Rate Volatility on Trade in Some Selected Sub-Saharan African Countries

doi:10.4236/me.2011.24059

Paper Menu >>

Journal Menu >>

Modern Economy, 2011, 2, 538-545

doi:10.4236/me.2011.24059 Published Online September 2011 (http://www.SciRP.org/journal/me)

Effects of Exchange Rate Volatility on Trade in Some

Selected Sub-Saharan African Countries

David Olayungbo, Olalekan Yinusa, Anthony Akinlo

Department of economics, obafemi Awolow o Uni versi t y , Ile-Ife, Nigeria

E-mail: [email protected]

Received April 2, 2011; revised June 2, 2011; accepted June 15, 2011

Abstract

The paper investigates the impact of exchange rate volatility on trade in 40 selected sub-Saharan African

countries for the period 1986-2005. The study employs a gravity model with pooled ordinary least square

(POLS) allowing for fixed effect and panel Generalized Method of Moments (GMM) techniques. The results

of the analysis show that the net effect of exchange rate volatility on aggregate trade was positive using the

two approaches. In the way the results show that there is not much difference between the impact of ex-

change rate volatility on primary and manufactured trade as well as between ECOWAS and non-ECOWAS

countries. However, the results should be interpreted with caution as the history of exchange rate volatility is

still relatively young compared with the developed countries.

Keywords: Exchange Rate Volatility, Trade, Sub-Saharan Africa

1. Introduction

Foreign exchange rate for sub-Saharan African countries

have been highly volatile following introduction of the

structural adjustment reforms since early 1980s. A cen-

tral question has been the effect of such high exchange

rate volatility on the growth of foreign trade. In the de-

veloped and other industrialized economies, several

studies have provided empirical evidence on the rela-

tionship between exchange rate volatility and trade. In

general, most of these studies have concluded that ex-

change rate volatility deters the growth of foreign trade.

However, little is known about the extent this conclusion

may be true for sub-Saharan African countries consider-

ing their peculiar characteristics including low exports

volume and dominance of primary commodities in the

aggregate exports. Essentially, studies of the experience

with sub-Saharan African countries have been very few,

due mainly to the unavailability of sufficient time series

data1.

This article examines the impact of exchange rate

volatility on trade for forty sub-Saharan African coun-

tries. The paper adds to existing literature in many ways.

One, the paper focuses specifically on sub-Saharan Afri-

can countries as against many others that merely only

included few African countries as part of larger sample.

Moreover, this study covers more Sub-Saharan African

countries compared with few existing studies in Sub Sa-

haran countries. Also, we examine the differential impact

of exchange of rate volatility on both primary and manu-

factured exports in the sub region2. Finally, we examine

the impact of regional grouping on the relationship be-

tween exchange rate volatility and trade in the region.

Essentially, this study is important for two main rea-

sons. One, the effect of the exchange rate volatility on

trade has significant impact on the reforms in the

sub-region. If exchange rate volatility adversely affects

trade, the export expansion programme will be jeopard-

ized. Moreover, the intended effect of the current trade

liberalization policy being implemented in the sub region

may be dammed thereby precipitating a balance of pay-

ment crisis.

The paper is organized as follows. In Section 2, we

discuss an overview of exchange rate management and

trade development in the region, in Section 3, we equally

examine the specification of the model. Data sources and

variable definitions are described in Section 4. In Section

1Few recent studies that focus exclusively on African Countries data

include [1], [2] and [3]. Out of the forty five countries that make up the

sub-Saharan African countries only five countries such as Eritrea,

Somalia, United Republic of Tanzania, Guinea Bissau and Comoros

are excluded from our sample due to unavailability of data.

2Indeed, it has been observed in the literature that total trade aggregate

tends to hide substantial variation across sectors. There is the need to

look at some of the components of the aggregate trade.

D. OLAYUNGBO ET AL539

5, we discuss the empirical results for the forty selected

sub Saharan African countries. Conclusions are drawn in

the last section.

2. An Overview of Exchange Rate

Management and Trade Development in

Sub-Saharan Africa

The trade policy of most sub-Saharan countries in the

late 1960 s to early 1970 s had been export promotion

policy. The export of sub-Saharan African countries has

been basically primary products and raw materials such

as vegetable oils, palm oil, palm nut, kernels and ground-

nuts. During this time, the growth of export of sub-Sa-

haran countries started from 3.1% in the late 1960s

(UNCTAD 2004). The export performance of the region

started declining in 1970 from 3.9 percent to 3.4 percent

in 1979. The oil-price shocks, the slow growth in the

world trade in primary commodities, institutional weak-

ness, political instability, civil war, trade restriction, tar-

iff barriers and persistent rise in price of imported manu-

factured goods were factors identified in the economic

literature responsible for low export in sub-Saharan Af-

rica [4]. During this period, the exchange rate policy of

these countries were fixed and pegged to the U.S. dollars

which is a fixed exchange rate system. By 1980, the

share of non-fuel exports had reduced from 18 percent in

1970 to about 9 percent, while the growth of import ex-

panded by 5.8 percent (World Development Indicator,

2006).

As a result of this import dependent tendency, coupled

with overvaluation of the exchange rates, most of the

countries in the region in the 1980s had to shift from

export-promotion policy to import-substitution strategy

to bring their economies back to the growth path. In or-

der to further correct the distortions in the economy,

most of the countries in sub-Saharan region adopted lib-

eralized policy of exchange rates after massive loss of

world market share. Nigeria for instance in 1986 adopted

Structural Adjustment Programme (SAP) with liberalized

exchange rate. The South African Reserve Bank’s

(SARB) flexible exchange rate regime resulted in vola-

tility of the Rand in 1997 [5]. Moreover, between 1987

and 1998, the average quarterly depreciation of the Gha-

naian cedi was 6.59%. The real effective appreciation of

the naira also in the 1980s eroded Nigeria’s competi-

tiveness, and growth of trade slowed remarkably during

those periods. By 1990, sub- Saharan share of world

trade had fallen to 1.2 percent compared to Asian world

share of 19.81 percent in the same year (UNCTAD

2004).

With the pursuit of trade liberalization in the 1990 s,

the U.S., through the Uruguay Round Agreement Act

(1994), devised measures to improve trade relation with

sub-Saharan Africa. In 2000, the U.S imports from

sub-Saharan region were petroleum products, followed

by non-ferrous metal, apparel and clothing and iron and

steel. While major U.S exports to the region were aircraft

and parts, mining machinery, wheat, general industrial

machinery and road vehicles. Likewise, sub-Saharan

export of primary products to Europe improved to 44

percent, while that of manufactured and energy products

was 22 and 34 percent respectively. The trade liberaliza-

tion also opens way for Asian-African trade relation with

sub-Saharan countries’ total trade as a percentage of

GDP reaching 71.75 percent in 2006. It equally paid off

with Mauritius’ manufactured exports of 19.13 percent

expressed as a percentage of GDP in 2006 [6]. However,

the overall world trade share of African countries was on

the decline (see Table 1).

3. Model Specification

In specifying the model, the study adopts the gravity

approach as employed by [7]3. In a gravity model, the

volume of trade between two countries increases with the

product of their gross domestic products (GDP) and de-

creases with their geographical distance. This implies

that high-income countries trade more than low-income

countries. Also, more proximate countries have lower

real exchange rate volatility and trade more than distant

countries. The gravity model has been widely used in

empirical work in international economics. The theoreti-

cal foundation of the gravity model assumes monopolis-

tic competition, identical and homothetic preferences

across countries. It relies heavily on the concept of in-

tra-industry trade as postulated by the new trade theories

such as the product-differentiation model and the tech-

nological-gap models.

Besides the distance, the empirical specification of the

gravity model often includes a number of dummy vari-

ables to control for different factors augmenting or re-

ducing trade; such as land areas, similarity, language,

Table 1. Sub-Saharan Africa’s share in world trade (%)

1970-2006.

Period Figure

1970-1976 3.1 %

1970-1979 5.2%

1980-1989 3.2%

1990-1999 1.1%

2000-2006 1.2%

Source: Inte rnational Financial Standard (2006).

3To conserve space, no theoretical discussions on the relationship be-

tween higher exchange rate volatility and foreign trade are presented

here. Several studies in the developed and industrialized countries have

rovided detailed discussions on the theoretical and empirical evidence

of the relationship between the two variables. See [8] and [9] among

others.

D. OLAYUNGBO ET AL

540

geographical position, historical links, and preferential

trading arrangements.

The gravity model is of the following form:

ijtit ijt

TRADEAY D



 (1)

where 10



0, 2



T, ijt is a measure of bi-

lateral total trade between sub-Saharan African countries

i and trading countries j at time t, A is the constant term,

RADE



and 2



are coefficients, ijt is the distance be-

tween sub-Saharan African countries i and trading coun-

tries j at time t and it is the income of sub-Saharan

African countries i at time t.

Taking logs of the gravity model, we derive an equa-

tion for country i and j at time t as:

12ijtitijt it

trade agdpd



  (2)

where ijt is a measure of bilateral trade between

sub-Saharan African countries i and trading countries j at

time t, a becomes the intercept of the gravity model,

trade

dp d

, a proxy for income it

Y, is the gross domestic

product of sub-Saharan African countries i, ijt is the

distance between sub-Saharan African countries i and

trading countries j, 1



and 2



are the coefficients, it

is the error term of sub-Saharan African countries i at

time t.



To capture the effect of population as a determinant of

trade the model becomes:

123ijtitijtit it

trade agdpdpop



  (3)

where it represents population of sub-Saharan Af-

rican countries i at time t.

pop

Incorporating exchange rate volatility and real ex-

change rate to capture exchange rate risk, the model be-

comes:

123ijtit ijtit

trade agdpdpop

45it it it

reer



 

 

  (4)

The variable it



and it are both the measure of

exchange rate volatility and the real exchange rate of

sub-Saharan African countries i respectively at time t.

reer

Assuming fixed-effect to account for time-varying

factors the model becomes:

12ijtiit ijt

trade agdpd

5ititit it

pop reer





 







 (5)

where i



stands for time-varying effects such as coun-

try’s size, economic power etc. However, the result is not

reported due to space constraint. A priori, the signs of the

coefficients are as follows: 10



2

, 0



, 30



,

4or 0



 5or0,



.

We disaggregate Equation (5) to incorporate the like-

lihood differential effect of exchange rate volatility on

primary and manufactured products, the model therefore

becomes:

ijtimitm ijt

Magdp d

 

 

mit mit mitit

pop reer







 

(6)

ijtipitp ijt

345

itp itpitit

pop reer





P agdpd

 

 



 

(7)

Equations (5), (6) and (7) are the estimated equations.

Where ijt

is a measure of manufacturing trade be-

tween sub-Saharan African countries i and trading coun-

tries j at time t, ijt is a measure of primary product

trade between sub-Saharan African countries i and trad-

ing countries j at time t, and

a are constant terms,

15mm





and 15





are coefficients, while m



and



are time-varying effects for the manufacturing

and primary product trade respectively, we expect the

sign of the coefficients to be as before, and errors across

equations are assumed independent,























. In estimating the models, we used the

Pooled ordinary least Square technique (POLS). How-

ever, for robustness check we equally used the General-

ized Method of Moments (GMM) method of estimation.

The results of our estimations are presented in the section

5 of the paper.

4. Data Sources and Variable Measurements

4.1. Data Source

Data for the study were obtained from World Bank,

World Bank Development Indicator (WDI) CD-ROM

2007) and Commodity Trade Statistics (COMTRADE)

database available at www.comtrade.org.

4.2. Variable Measurements

Operational definition and measurement of variables is

as presented below:

Trade (trade): Trade is the volume of aggregate sum

of import and export series sourced from World Devel-

opment Indicator (WDI 2007) published by World Bank,

prmtrade denotes primary trade also sourced from

Commodity Trade Stastistics Database (COMTRADE)

available at www.comtrade.org published by United Na-

tions (UN). The primary products are classified based on

SITC rev.3 (Standard International Trade Classification

revision 3) commodity code, such as Food and Live

Animals, Beverages and Tobacco, Cotton and Textiles,

Crude Materials, Inedible, Minerals fuels, Ores and Met-

als, Animal and Vegetable oils. The mantrade variable is

manufactured trade also available at COMTRADE with

SITC rev.3 commodity classification as Equipment,

Utensils, Appliances and Machines. Trade variable de-

D. OLAYUNGBO ET AL541

notes the volume of aggregate bilateral trade among se-

lected trading sub-Saharan African Countries. The vol-

ume of the aggregate bilateral trade is constructed as

bilateral trade value deflated by the relative price index

(import and export prices of trading countries).

Gross Domestic Product (gdp): This is the produc-

tive capacity of an economy. The real domestic product

is the nominal value of the GDP deflated by the con-

sumer price index.

Real Effective Exchange Rate (REER): Exchange

rate is a relative price that measures the worth of a do-

mestic currency in terms of another currency. It relates

the purchasing power of a domestic currency, in terms of

the goods and services it can purchase, vis-à-vis a trading

partners’ currency over a given period.





niiiJ

BREERE WPIWPI







where n= Bilateral trade weighted real exchange

rate = Nominal exchange rate.

BREER

= Wholesale price index for importing country i.

WPI

WPI = Wholesale price index for exporting country j.

Exchange rate volat ility (exvol): Exchange rate vola-

tility is a measure that intends to capture the uncertainty

faced by both exporters and importers due to unpredict-

able fluctuations in the exchange rates. Clearly, this is an

unobservable variable and thus its measure is a matter of

serious contention. This study follows recent literature

and uses the measures derived from the GARCH (1, 1)

model as measures of exchange rate volatility. Following

[10] and [11], the conditional volatility of exchange rate

was extracted and modeled via a state space representa-

tion of the form:

;

zht

tt t





 (8)



0,1iid

1π

ttt



 NID, (9)



0,/ /1

 



t is the exchange rate. The term z



is a scale factor

and subsumes the effect of a constant in the regression of

t, π, is a parameter, t



is a disturbance term that is

uncorrelated with t



is an iid (0, 1) are random distur-

bances symmetrically distributed about zero. The ht

equation is a transition equation in autoregressive form

where the absolute value of π is less than unity to ensure

that the process in Equation (8) is stationary [10]. These

equations generate the conditional volatility of exchange

rate.

Population (pop): This is a measure of a country size.

This is another determinant of trade. It is expected that

countries with higher population trade more. Therefore,

positive relationship should exist between population and

trade.

Distance (d): This is a measure of distance between

trading countries. In the literature some studies used

transport cost as a proxy while some represented distance

by air distances between capital cities. Tariffs, import

and export taxes, and taxes on international trade can

also be used. Taxes on international trade include import

duties, export duties, profits of export or import monopo-

lies, exchange profits, and exchange taxes, World De-

velopment Indicator (2007). This study makes use of

taxes on international trade as proxy for distance due to

data availability and the bilateral model adopted. All

variables are expressed in log-form except exchange rate

volatility (exvol).

5. Empirical Results

The first step in our analysis is to perform a panel unit

root tests to overcome the heterogeneity biases that are

common characteristics of panel data analysis. Specifi-

cally, we used Levin, Lin & Chu, Im, Pesaran and Shin

W-stat, ADF-Fisher Chi-square and PP-Fisher Chi-

square tests. These tests assume individual unit root

process to allow for heterogeneity across cross-sectional

units. As a check, Hadri Z-stat test is reported as well,

which imposes the same unit root process across coun-

tries. The full sample exhibits stationarity for all the

variables at first difference. The results are contained in

Table 2.

From the panel unit root tests (see Table 2), taxes and

mantrade variable are stationary at levels, while other

variables i.e pop, gdp, exvol, prmtrade and trade are sta-

tionary at first difference. Given the unit root properties

of the variables, we proceed to conduct our estimation

using both POLS and GMM techniques.

Table 3 presents the results for the pooled OLS with

fixed effects for total trade over the period 1986-2005

(see column 1). The model performs well empirically,

yielding precise and generally reasonable estimates. The

R2, which measures the goodness of fit is relatively high

and the F- statistics is significant. The results for total

trade as shown in column 1 of Table 3 show that ex-

change rate volatility positively related to trade. The co-

efficient is only significant at 10 percent. This simply

suggests that volatility of the exchange rate enhances

aggregate trade in the sub-Saharan African countries.

This possibly suggests that traders are risk takers who

see increase in volatility as opportunity for profit. This

finding is consistent with [12]. The coefficient of tax is

negative but barely significant at 20 percent. The results

show that a 10 percent increase in tax would reduce trade

by 0.8 percent. This means that higher tax tend to in-

crease trade costs, which depresses exports. The coeffi-

cients of population and gross domestic product have

positive signs and they are both significant. Which

means that trade responds positively with increase in

population and GDP.

D. OLAYUNGBO ET AL

542

Table 2. Panel unit root tests-individual effects, full sample.

Variables Level LLC p-v IPS p-v ADF p-v PP p-v Hadri Zp-v

Trade 0 4.62 0.96 5.86* 0.01 64.95* 0.89 56.3 0.98 8.57* 0.00

1 –17.78* 0.00 –16.63* 0.00 422.83* 0.00 741.15* 0.00 6.22* 0.00

Exvol 0 –3.99* 0.00 1.41 0.92 65.06 0.89 103.88** 0.04 15.98*0.00

1 –13.63* 0.00 –12.98* 0.00 317.96* 0.00 366.56* 0.00 5.93* 0.01

Prmtrade 0 –1.27*** 0.1 0.31 0.62 99.26*** 0.07 100.53*** 0.06 –1.7 0.96

1 –19.57* 0.00 –16.88* 0.00 425.28* 0.00 916.37* 0.00 14.01*0.00

Mantrade 0 –3.24* 0.00 –3.24* 0.00 131.64* 0.00 104.03** 0.04 12.01*0.00

Gdp 0 –0.94 0.17 1.27 0.9 106.24** 0.03 92.13 0.17 14.28*0.00

1 –8.54* 0.00 –9.66* 0.00 272.93* 0.00 331.37* 0.00 5.71* 0.00

Taxes 0 –5.78* 0.00 –7.51* 0.00 211.36* 0.00 222.57* 0.00 5.60* 0.00

Pop 0 0.76 0.78 12.14 0.75 53.73* 0.01 50.32* 0.01 12.02*0.00

1 5.32* 0.01 –2.58* 0.01 164.94* 0.00 406.05* 0.00 6.52* 0.00

Notes: The null hypothesis (Ho) is that there is no unit root, (H1) some do not have a unit process. Significance levels are denoted by *: 1%, **: 5%, ***: 10%:

and indicate rejection of the null hypothesis. 0 and 1 represent level and first difference respectively. Probabilities for Fisher tests are computed using an as-

ymptotic Chi-square distribution. All other tests assume asymptotic normality. LLC denotes Levin, Lin and Chin, IPS denotes Im Pesaran Shin W-Stat, ADF

indicates Augmented Dickey Fuller test, PP denotes Phillip Peron, Hadri Z Stat is also reported, and P-V indicates Probability Value.

Table 3. POOL-OLS with Fixed Effects Full- sample period

1986-2005.

Dependent Variables: Trade

(1)

Prmtrade

(2)

Mantrade

(3)

Constant 11.3982 4.1469 3.1322

(5.38) (2.24) (2.46)

Tax –0.0841 0.053668 0.0085

(–1.69) (1.21) (0.29)

Population (LOG(POP) 0.3246 0.0642 0.0537

(2.58) 0.59 (0.72)

Gross Domestic Product

(LOG (GDP) 0.7507 –0.1329 –0.0276

(5.34) (–1.13) (–0.28)

Real Effective Exchange Rate

(EXCH) 0.0E+01 –0.0001 0.0E+04

(0.96) (–5.44) (–0.29)

Exchange Rate Volatility

(EXVOL) 0.0397 0.0190 0.1017

(1.90) (1.18) (6.69)

Summary Statistics

R-Square 0.7789 0.5416 0.6623

Durbin-Watson Statistic 1.85 1.13 1.17

F-Statistic 60.29 20.17 33.52

Prob(F-statistic) 0.0000 0.0000 0.0000

AIC 2.2924 1.2439 1.0083

SC 2.5564 1.5085 1.2726

Cross sections included 40 40 40

Obsevations 798 796 797

N.B.: t-statistics in parenthesis.

In order to get a better picture of the relationships be-

tween exchange rate volatility and trade we look at the

two of the components of the aggregate trade namely

primary trade and manufactured trade. The results are as

shown in columns (2) and (3) of Table 3 . Essentially, the

results obtained for these two categories of trade are not

significantly different from those of aggregate trade.

However, tax variable has positive sign for both primary

and manufactured trade as against the negative sign ob-

tained for aggregate trade. But the coefficient of taxes is

not significant in both cases. Hence, firm conclusion

cannot be based on it. The real effective exchange rate

variable has the expected negative sign on both primary

and manufactured trade but significant only in the case of

the latter. The coefficient of gross domestic product is

negative for both categories of trade. However, the coef-

ficient is not significant. The exchange rate volatility

variable has positive effect on both primary and manu-

factured trade but only significant for the former. The

results show that a 10 percent increase in exchange rate

volatility would increase manufactured trade by 1.0 per-

cent. The corresponding figure for primary trade is 0.2

percent.

Next we address the question that do members of trade

union affect the impact of exchange rate volatility on

trade? To address this question, we divide the selected

sub-Saharan African countries into ECOWAS and

non-ECOWAS countries4. The results for ECOWAS

countries are shown in Table 4. Columns 1, 2 and 3 are

results for aggregate trade, primary and manufactured

trade respectively. For aggregate trade, the exchange rate

volatility coefficient is positive and barely significant

with t-statistics of 1.64. 10 percent rise in exchange rate

volatility would increase aggregate trade by 0.9 percent.

Tax variable is negative as expected but not significant.

The coefficient gross domestic product (GDP) is positive

and significant at 1 percent.

With respect to primary and manufactured trade in the

ECOWAS countries, the results show that exchange rate

volatility has significant positive effect on the two cate-

gories of trade. For manufactured trade, a 10 percent

increase in exchange rate volatility will lead to 0.2 per-

cent increase in manufactured trade in the ECOWAS sub

region. The corresponding figure for primary trade is 1.8

percent. The results is in Table 4 show that GDP is sig-

4This exercise is important for two main reasons. One, it will help to

ascertain the assertion that members of the same regional grouping

tend to trade more even in the face of high exchange rate volatility than

non members. Two, such division tends to reinforce the homogenous

nature of these countries thereby obviating the problem often associ-

ated with pooling countries with different underlying time series prop-

erty. 16 countries actually make up Ecowas, but 15 countries are used

in this paper while the remaining 25 countries represent non-Ecowas.

D. OLAYUNGBO ET AL543

nificantly positively related to manufactured trade. The

reverse is the case with primary trade though the coeffi-

cient is not significant. Real effective exchange rate has a

significant negative effect on primary trade while the

coefficient is positive for manufactured trade though not

significant.

In the case of non-ECOWAS countries, the results for

aggregate, primary and manufactured trade are as shown

in columns 1, 2 and 3 of Table 4 respectively. The re-

sults in Table 4 for aggregate trade show that the ex-

change rate volatility variable is significantly positively

related to aggregate trade for non-ECOWAS countries.

The results indicate that a 10 percent increase in ex-

change rate volatility will lead to 0.3 percent increase in

aggregate trade in non-ECOWAS sub region.

The coefficient of tax is negative and significant. This

is means that an increase in taxes will lead to reduction

in aggregate trade in non ECOWAS sub region. Popula-

tion and gross domestic product both have significant

positive impact on aggregate trade in the non-ECOWAS

sub region.

With respect to primary and manufactured trade, the

results show that tax variable has positive impact on both

primary and manufactured trade though the coefficient is

only significant in the case of primary trade. In the same

way, population variable is positively related to both

primary and manufactured trade but only significant in

the latter. The coefficient of GDP is negative and sig-

nificant for both manufactured and primary trade. Real

effective exchange rate variable has negative impact on

the two categories of trade but only significant for pri-

mary trade. Finally, the coefficient of exchange rate is

positive for both primary and manufactured trade. The

variable is only significant in the case of manufactured

trade.

Further Consideration

The basic assumption behind Pooled Ordinary least

Square (POLS) results presented above is the exogeneity

of explanatory variables. However, when this assumption

is relaxed, the POLS breaks down. Therefore, relaxing

the assumption requires that we use another approach

capable of correcting biases introduced by including the

lagged dependent variable on the right hand side of the

equation. Therefore, a Generalized Method of Moments

(GMM) estimator in [13] approach was used to obtain

consistent estimates. Such panel techniques allow one to

control for endogeneity or simultaneity of some of the

explanatory variable in particular GMM estimators, as

well as for potential biases due to correlation between the

explanatory variables and the regression residual. More-

over, the use of GMM estimation technique provides the

robustness check for for the results obtained through the

pooled OLS technique. The panel GMM with fixed ef-

fects is performed on aggregate trade, primary product

and manufacturing product trade5. The results are pre-

sented in Table 5.

Columns 1, 2 and 3 of Table 5 show the GMM results

for aggregate trade, primary and manufactured trade re-

spectively. Overall, the results from Generalized Method

of Moments (GMM) perform better considering the

j-statistics, instrument rank, significant t-statistics, and

the coefficients. With respect to aggregate trade from

Table 5 column 1, the coefficient of exchange rate vola-

tility is positive and significant. The results show that a

10 percent increase in exchange rate volatility would

increase trade by 0.6 percent. In the same way, the coef-

ficients of population and gross domestic product are

positive and significant. A 10 percent increase in GDP

would lead to 6 percent increase in aggregate trade. Tax

variable is negative and significant as expected. The re-

sults indicate that a 10 percent increase in taxes would

reduce aggregate trade in sub-Saharan Africa by 2 per-

cent.

As regards primary and manufactured trade, the results

show that exchange rate volatility has significant nega-

tive effect on primary trade while it has significant posi-

tive effect on manufactured trade. The results indicate

that increase in population would lead to increase in pri-

mary trade. The reverse is the case with manufactured

trade though the coefficient is not significant. The coef-

ficient of gross domestic product is negative and signifi-

cant for both primary and manufactured trade. The coef-

ficient of tax is positive and significant for both primary

and manufactured trade. A similar panel study carried

out by [3] between 1972-1987 on sub-Saharan Africa

reported a negative effects of exchange rate volatility on

trade. However, the estimation period was a period of

fixed exchange rate regime and this might have biased

the result. A Study conducted also by [2] analyzed the

effects of bilateral exchange rate movements in terms of

real effective exchange rate misalignment and volatility

on the growth of non-oil exports in Nigeria over the

1960-1990 periods. The findings of the study showed

that exporters in Nigeria are less risk averse and would

readily substitute other activities for exporting should

adverse movement in real exchange rate occur. Apart

from a single country study, the conclusion may be as a

5However, the reliability of the GMM estimator depends very much on

the reliability of the instruments. The validity of the instrument was

evaluated using the popular Sargan test [14]. The Sargan test is a test

on over-identifying restrictions by comparing both the j-statistic and

instrument rank. It is asymptotically distributed as χ2 and tests the null

hypothesis of validity of the (over-identifying) instruments. P-values

report the probability of incorrectly rejecting the null hypothesis, so

that a P-value above 0.05 implies that the probability of incorrectly

rejecting the null hypothesis above 0.05. In which case, a higher

P-value makes it more likely that the instruments are invalid. Our

P-values are generally lower than 5% with the value of 0.03, which

means that instruments used are valid.

D. OLAYUNGBO ET AL

544

Table 4. POOL-OLS with Fixed Effects Full-sample period

1986-2005 (NON ECOWAS COUNTRIES) 25 countri es .

Dependent Variables: trade

Prmtrade

Mantrade

Constant 10.9604 4.6812 1.9147

(3.65) (1.95) (1.48)

Tax –0.1010 0.0817 0.0034

(–3.40) (1.53) (0.11)

Population( LOG (POP) 0.4526 0.0907 0.1943

(3.34) (0.61) (2.37)

Gross Domestic Product (LOG

(GDP) 0.5232 –0.2951 –0.1681

(3.49) (–3.67) (–1.98)

Real Effective Exchange Rate

(EXCH) 0.0E+06 –0.0E+09 –0.0E+01

(2.70) (–6.28) (–0.10)

Exchange Rate Volatility

(EXVOL) 0.0338 0.0079 0.1044

(8.22) (1.08) (2.73)

Summary Statistics

R-Square 0.9472 0.5068 0.6571

Durbin-Watson Statistic 0.64 1.14 1.24

F-Statistic 311.62 18.74 34.15

Prob (F-statistic) 0.0000 0.0000 0.0000

AIC 0.5950 1.3901 1.3404

SC 0.8485 1.6441 1.5940

Cross sections included 25 25 25

Observations 520 519 520

Table 5. Panel generalized method of moments fixed effects (first

difference) 1988-2005 (A ll coun tries).

Variables: Trade

Prmtrade

Mantrade

LOG (TRADE (–1)),

LOG (PRMTRADE (–1)),

LOG (MANTRADE(–1))

0.0520 0.2135 0.1343

(1.07) (53.3) (71.3)

EXVOL 0.0568 –0.0341 0.1570

(3.75) (–5.73) (128.85)

LOG (POP) 0.4270 0.0880 –0.0274

(2.83) (2.75) (–0.41)

LOG (GDP) 0.5890 –0.6537 –0.8362

(12.4) (-34.9) (–24.04)

LOG (TAX) –0.2257 0.0930 0.0789

(–7.04) (26.4) (17.4)

SUMMARY STATISTICS

j-statistic 35.70211 35.84797 38.02933

Instrument rank 42.00000 41.00000 40.00000

result of the data span not extending beyond 1990.

6. Concluding Remarks

The purpose of this study is to examine the impact of

exchange rate volatility on trade in sub-Saharan African

countries. Therefore, we begin by specifying gravity

model that incorporates exchange rate volatility as argu-

ment. We then estimate the model using pooled OLS and

GMM techniques for the period 1986-2005. Essentially,

the results for aggregate trade show that exchange rate

volatility tends to enhance trade in the sub-Saharan Af-

rican region. This suggests that traders in the sub-region

perceive increase in volatility as opportunity for profit

making and thus ready to export more in the face of in-

creased exchange rate volatility. The evidence reported

here suggests that there is not much difference between

the impact of exchange rate volatility on primary and

manufactured trade as well as between ECOWAS and

non-ECOWAS countries.

However, the results should be interpreted with cau-

tion because the history of exchange rate volatility is still

very short in Sub-Saharan African countries compared to

the developed countries. Therefore, its impact on the

macro variables in these economies might not yet be

substantial.

7. References

[1] G. O. Akpokodje, “Exchange Rate Volatility and Exter-

nal Trade Performance of Selected African Countries,

1973-2003,” A Ph. D Thesis Submitted to University of

Ibadan, Nigeria, 2007.

[2] O. Ogun, “Real Exchange Rate Movements Export

Growth in Nigeria, 1960-1990,” AERC Research Paper

82, Nairobi, 1998.

[3] D. Ghura and T. Grennes, “The Real Exchange Rate and

Macroeconomic Performance in Sub-Saharan Africa,”

Journal of Development Economic, Vol. 42, No. 2, 1993,

pp. 155-174. doi:10.1016/0304-3878(93)90077-Z

[4] A. Subramanian, “The Mauritian Success Story and its

Lessons,” Wider Research Paper 2009/36, UNU-WIDER,

Helsinki, 2009.

[5] I. Bah and H. A. Amusa, “Real Exchange Rate Volatility

and Foreign Trade: Evidence from South Africa’s Export

to the United States,” The African Finance Journal, Vol.

5, No. 2, 2003, pp. 1-20.

[6] World Bank, “World Development Indicators,” World

Bank, Washinton DC, 2006.

[7] E. Helpman, “Imperfect Competition and Inter National

Trade: Evidence from Fourteen Indus Trial Countries,” In:

M. Spence and H. Hazard, Eds., International Competi-

tiveness, Cambridge, Ballinger, 1987. pp. 197-220.

[8] M. D. Mckenzie, “The Impact of Exchange Rate Volatil-

ity on International Trade Flows,” Journal of Economic

Surveys, Vol. 13, No. 1, 1999, pp. 71-106.

doi:10.1111/1467-6419.00075

[9] R. Anderton and F. Skudelny, “Exchange Rate Volatility

and Euro Area Imports,” European Central Bank Work-

ing Paper No. 64, 2001.

[10] N. S. Ndung’u, “Liberalization of the Foreign Exchange

Market and Short Term Capital Flows Problem,” AERC

Research Paper 109, Nairobi, 2001.

[11] D.O. Yinusa, “Exchange Rate Variability, Currency Sub-

stitution and Monetary Policy in Nigeria (1986-2001),”

An Unpublished Ph.D Thesis, Depart Economics,

Obafemi Awolowo University, Ile-Ife, Nigeria, 2005.

[12] A. Asseery and D. A. Peel, “The Effect of Exchange Rate

Volatility on Export-Some New Estimates,” Economic

D. OLAYUNGBO ET AL

545

Letters, Vol. 37, 1991, pp. 173-177.

doi:10.1016/0165-1765(91)90127-7

[13] M. Arellano and S. R. Bond, “Some Tests of Specifica-

tion for Panel Data: Monte Carlo Evidence and an Appli-

cation to Employment Equations,” Review of Economic

Studies, Vol. 58, No. 2, 1991, pp. 277-297.

doi:10.2307/2297968

[14] J. D. Sargan, “The Estimation of Economic Relationships

Using Instrumental Variables,” Econometrica, Vol. 26,

No. 3, 1958, pp. 393-415.

doi:10.2307/1907619