This article was downloaded by: [The UC Irvine Libraries]On: 19 December 2014, At: 11:23Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

Applied EconomicsPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/raec20

How does exchange rate policy affect manufacturedexports in MENA countries?Mustapha Kamel Nabli & Marie-Ange Véganzonès-Varoudakis ba World Bank , Washington, DC, USAb CERDI , CNRS, Université d’Auvergne , Clermont Ferrand, Francec CERDI , CNRS, Université d’Auvergne , Clermont Ferrand, France E-mail:Published online: 02 Feb 2007.

To cite this article: Mustapha Kamel Nabli & Marie-Ange Véganzonès-Varoudakis (2004) How does exchange rate policy affectmanufactured exports in MENA countries?, Applied Economics, 36:19, 2209-2219, DOI: 10.1080/0003684042000271373

To link to this article: http://dx.doi.org/10.1080/0003684042000271373

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information. Taylor and Francis shall not be liable forany losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

How does exchange rate policy affect

manufactured exports in MENA

countries?

MUSTAPHA KAMEL NABLI andMARIE-ANGE VEGANZONES-VAROUDAKISz*

World Bank, Washington, DC, USA and zCERDI, CNRS, Universite d’Auvergne,Clermont Ferrand, France

This paper shows that, during the 1970s and 1980s, MENA economies were char-acterized by a significant overvaluation of their currency. This overvaluation hashad a cost in terms of competitiveness. To determine the degree of overvaluation ofthe MENA currencies, an indicator of misalignment was developed based on theestimation of an equilibrium exchange rate (Edwards, Exchange Rate Misalignmentin Developing Countries, The Johns Hopkins University Press, Baltimore, 1988). Theempirical work was based on a panel of 53 developing countries, ten of which areMENA economies. Although overvaluation decreased in the 1990s, probably due toflexibilization of the exchange rate regime in some MENA countries and to bettermacroeconomic management in others, misalignment remained higher than in otherregions. This may be explained by the MENA countries’ delay in adopting moreflexible exchange rates, as well as in reforming their economies. In terms of competi-tiveness, the estimation of an export equation has shown that manufactured exportshave been significantly affected by the overvaluation of the MENA currencies.Countries that already had a more diversified economy benefited more from thedecreased overvaluation in the 1990s. These countries also saw a continuous risein diversification of their manufactured exports, resulting from the significant declinein exchange rate misalignment.

I . INTRODUCTION

Recent assessments of economic policies and performance

in developing countries have underlined the crucial issue of

the management of the real exchange rate (RER). It has

been shown that best performers are countries that have

maintained an ‘appropriate’ RER, i.e., one close to the

equilibrium real exchange rate (ERER) (Williamson,

1985; Harberger, 1986; Razin and Collins, 1997). In parti-

cular, all countries that have been successful in promoting

manufactured exports have avoided real exchange rate

overvaluation.1 In fact, RER misalignment – especially

overvaluation – is damaging to economic performance

because it decreases the profitability of production and

the export of tradable goods. In this way, RER misalign-

ment leads to a reduction in economic efficiency and a

misallocation of resources. In addition, by increasing

uncertainty and raising the risk of macroeconomic col-

lapse, RER misalignment can also hinder growth by dete-

riorating domestic and foreign investment and contributing

to capital flight. These negative effects of misalignment

on growth and export performance have been shown by

*Corresponding author: E-mail: veganzones@aol.comNote: The views expressed in this paper are those of the authors, not of their institutions.1 See, for example, Balassa (1990) and Reinhardt (1995) for empirical evidence in both developed and developing countries.

Applied Economics ISSN 0003–6846 print/ISSN 1466–4283 online # 2004 Taylor & Francis Ltd 2209

http://www.tandf.co.uk/journalsDOI: 10.1080/0003684042000271373

Applied Economics, 2004, 36, 2209–2219

Dow

nloa

ded

by [

The

UC

Irv

ine

Lib

rari

es]

at 1

1:23

19

Dec

embe

r 20

14

Edwards (1988), Cottani et al. (1990), and Ghura and

Grennes (1993) for different groups of developing

countries.

In addition to misalignment, inconsistencies among

macroeconomic, trade, and exchange rate policies increase

the variability of the RER – which in turn can affect

growth. Higher RER volatility sends confusing signals to

economic agents. It raises the uncertainty of long-run

investments and of the profitability of producing tradable

goods. The sensitivity of export performance to RER

volatility has been highlighted in the case of various econo-

mies by Ghura and Grennes (1993), Grobar (1993),

Cushman (1993), and Gagnon (1993).

The harmful effect of RER misalignment on exports of

the MENA economies is well confirmed by the present

study. It is shown that, during the past three decades,

MENA countries experienced substantial RER misalign-

ment, with a net tendency to overvaluation of their RER.

This had a negative significant impact on the export growth

of manufactured products, though the effect was less sig-

nificant when total exports are considered. This appears to

have resulted in slower economic growth, as manufactured

products exports have become a major factor of economic

growth in developing economies, a large number of which

have now successfully entered world markets.2

The findings bring new empirical evidence on the subject

of misalignment and of export growth in the case of

MENA countries, on which little previous work has been

undertaken. Our results were obtained through the estima-

tion of an export equation on a panel of 53 developing

countries (see Appendix A), among which ten are MENA

economies. The calculations cover the period 1970–1980 to

1999, during which tremendous changes in trade and

exchange rate policies have been observed.

The first step has been to provide an accurate measure of

the gap (or misalignment) between RER and its equili-

brium level (ERER). The estimates of ERER and of

RER misalignment are based on a reduced form approach.

RER behaviour is modelled using an equation that

includes both the role of fundamental factors in the

medium-to-long term (e.g., terms of trade, investment,

capital flows, trade openness), and the less persistent

impact of short-term variables (e.g., macroeconomic

policies, nominal devaluations). The ERER is then

computed using this equation, by eliminating the effect of

transitory variables and using estimates of sustainable or

long-term values of the fundamental variables. This

approach, initiated by Edwards (1989), has been extended

by Elbadawi (1994) and Baffes et al. (1997).

The use of this approach also represents a new con-

tribution to the study of exchange rate policy in MENA

economies, since the few previous studies were generally

based on a time series approach (Mongardini, 1998;

Domac and Shabsigh, 1999; Sorsa, 1999; Sundararajan,

et al., 1999; Achy, 2001).3 In addition to the interest

of using panel data estimations techniques compared to

times series econometrics,4 our calculations allow some

comparative analysis among the different regions, as well

as among the MENA countries themselves.

The paper is organized as follows. In the second section,

panel data calculations of the RER’s long-run equilibrium

are presented. In the third section, the misalignment and

volatility of the MENA countries’ RER are discussed,

compared to that of other regions. The fourth section

presents estimation of the impact of RER misalignment

and of RER volatility on the export performance of the

economies. It is illustrated that the misalignment has had a

negative influence on the MENA countries’ manufactured

exports. The fifth section concludes.

II . MODELLING THE LONG-RUNEQUILIBRIUM OF THE RER

The long-run equation explaining RER behaviour is based

on Edwards (1994), who has developed a dynamic model

of RER determination for a small, open economy with a

single nominal exchange rate system. The model allows

for both real and nominal factors to play a role in the

short run. In the long run, only real factors – the funda-

mentals – influence the ERER. In the present case, the

2 In fact, export diversification – through promotion of manufactured exports – is an important factor of sustained growth for differentreasons. First, income elasticity of demand is higher for manufactured goods than for primary products. In this way, growth in foreignincome is expected to increase the growth prospects of country’s manufactured exports. Second, both price elasticity of demand andsupply are presumed to be higher for manufactured goods than for primary commodities. This implies a stabilizing effect on the terms oftrade and a more stable growth of exports over time. Third, development of the manufacturing sector involves substantial prospects fordynamic productivity gains through economies of scale, learning effects, and externalities among firms and industries. See Nishimizu andRobinson (1986) for cross-country evidence at a two-digit industry level of positive correlation between export growth and total factorproductivity (TFP) changes.3 See Sekkat and Varoudakis (2002) for a panel data approach to assessing the misalignment of North African countries.4 The comparative advantage of panel data regressions compared to time series estimations can be seen, first, in the double dimension ofthe sample (time series-cross section), which improves estimates by adding information; second, in the country dummy variables, whichgenerally ask for an important degree of freedom, and which improve the results of the estimations.

2210 M. K. Nabli and M.-A. Veganzones-Varoudakis

Dow

nloa

ded

by [

The

UC

Irv

ine

Lib

rari

es]

at 1

1:23

19

Dec

embe

r 20

14

long-run relationship is specified as follows:5

lnðei, tÞ ¼ cþ a1 � lnðInvi, tÞ þ a2 � lnðOpeni, tÞ þ a3 � lnðTOTi, tÞ

þ a4 � Capinfi, t þ a5 � lnðDebtServi, tÞ þ "i, t ð1Þ

where:e ¼ bilateral RER between the country con-

cerned and the USA, measured as the ratioof the consumption price index in the coun-try (PD) to the wholesale price index in theUSA (Pw), multiplied by the nominalexchange rate in local currency/US$ (E ).These price indices are used, respectively,as proxies of the price of non-tradablegoods (PD) and the price of tradable goods(Pw �E );

RERi, t¼ðPi,DtÞ=ðPi,wt � Ei, tÞ

Inv ¼ investment ratio to GDP;Open ¼ indicator of trade openness, measured as the

sum of imports and exports divided byGDP.As improved measure of trade openness, ithas been substituted to Open a policy-induced trade openness indicator (TP),which consists in adjusting Open for the‘natural trade openness’ of the economyconstituted by the size of the country andthe distance from markets (see Frankel andRomer, 1999). This is Equation 10.

TOT ¼ external terms of trade, measured as theratio of export to import prices (in dollars);

Capinf ¼ capital inflows, calculated as the net changein reserves minus the trade balance scaled byGDP;6

DebtServ ¼ debt service to total exports;c ¼ intercept, a1 to a4¼ parameters, c¼ country

index, t¼ time index, and "t¼ error term.

Following Edwards (1989), it is assumed that, in the longterm, an increase in the investment rate (Invt) results inan increase in the demand for and relative price of non-tradable products – thus appreciating the real exchangerate. This assumption implies that as the investmentrate grows, investment is increasingly constituted of non-tradable products (e.g., services and construction), and

relatively less of tradable goods (e.g., equipment).

This effect can also be due to the multiplier effect of the

investment, which raises the aggregated demand mainly for

non-tradable products.

The RER is positively affected by trade restrictions,

which implies a negative sign on the coefficient on the

proxy for trade openness, measured as the ratio of imports

plus exports to GDP (Open). The same negative sign is

expected for the improved measure of policy-induced

trade openness (TP).

The impact of the terms of trade (TOT ) on the RER is

more ambiguous, since there are two opposite effects. That

is, an increase in the relative price of exported goods to

imported goods leads to an appreciation of the RER

only if the income effect (which results in higher demand

for non-tradables) dominates the substitution effect (which

is associated with a decline in the relative cost of imported

intermediate goods used in the production process of

non-tradables).

An increase in capital inflows (Capinf ) involves stronger

demand for both tradable and non-tradable goods.

Increased inflows, therefore, lead to a higher relative

price of non-tradables, and conversely, appreciate the

RER – as is necessary for domestic resources to be diverted

towards production in the non-tradable sector to meet

increased demand. On the other hand, a rise in the debt

service (DebtServ) – which captures the important impact

of debt relief in many MENA countries – contributes to

depreciating the RER.

The existence of this long-term relationship implies that

the variables of Equations 1 and 10 are cointegrated. It is

therefore necessary to determine the order of integration

of the series. Table B1 in Appendix B provides the results

of the Augmented Dickey–Fuller (ADF) tests of the data

for the sample of 53 developing countries7 over the time

period. The Im et al. (1997) methodology is used – which

provides critical values of ADF tests in the case of hetero-

geneous panel data. The results indicate that the series

are stationary at either the 1% or 5% level, which allows

Equations 1 and 10 to be run.

Hence, Equations 1 and 10 describe the long-run relation-

ship between RER and a number of fundamental variables.

The equations were estimated on the panel of 53 develop-

ing countries of which ten are MENA economies.8 The

5 The short-run dynamic of the RER has also been estimated through an error correction model (Equation (1) in Appendix C. Results areshown in Table C3.6An increase in net capital inflows may result from: (a) an autonomous augmentation in foreign aid, foreign voluntary lending, or foreigndirect investment (FDI); (b) an increase in borrowing due to the removal of domestic capital controls; (c) a fall in world interest rates; or(d) an increase in public borrowing to finance the fiscal deficit.7 Of which 19 are African countries (eight CFA and 11 non-CFA), 13 are Latin American countries, ten are Asian countries, and 11 areMENA countries. (CFA is for ‘Communaute Francophone d’Afrique’.)8 The countries were selected based on level of income per capita. To preserve a kind of coherence of the sample, intermediate-incomecountries were generally chosen so they could be compared to countries in the MENA region.

Exchange rate policy and manufactured exports 2211

Dow

nloa

ded

by [

The

UC

Irv

ine

Lib

rari

es]

at 1

1:23

19

Dec

embe

r 20

14

results of the regressions – using the White estimator tocorrect for the heteroscedasticity bias – are presented inTable 1. The equations were estimated using the fixed-effect methodology.9 The estimated regressions explain afairly large amount of the observed variation of the RER.

Estimated relationships between RER and its fundamen-tals are consistent with theory (Edwards, 1989): an increasein investment and in capital income, or an improvement ofthe terms of trade, result in a RER appreciation – whichindicates, in the latter case, that the income effectdominates the substitution effect. Conversely, the openingof the economy and the increase in the debt service leadto a RER depreciation.

III . RER MISALIGNMENT

The misalignment (MIS) of the real exchange rate (RER)is measured as the per cent difference between the RERand its equilibrium value (ERER):

MIS ¼ ðRER=ERERÞ � 1

The estimations of the long-term relationship between theRER and its fundamental determinants have been usedto compute the ERER based on Equation 10. For thispurpose, the sustainable or equilibrium values of thefundamental variables had to be assessed. The idea isthat the deviation of the fundamental variables fromtheir equilibrium – in addition of variations in short-termeconomic policy variables (see the estimation of the errorcorrection model through Equation A3-1 in Appendix C) –leads to a misalignment of the RER. The permanent valuesof the five fundamental variables – Invt, Opent, TOTt,Capinft,, DebtServt – were computed using moving averagesof the series over a three-year period. This simple methodwas possible because the series are stationary.10

Following this methodology, an excessive trade protec-tion, an unexpected appreciation of the terms of trade, an

increase in investment and in capital flows, or a reductionof the debt service – in comparison to the normal or long-term trend in the economy – leads to an overvaluation ofthe RER. It can also be shown from the estimation ofthe error correction model (Table C1 in Appendix C)that, in the short run, nominal devaluations (Dev), blackmarket premiums (BMP), and inflation (Inf l ) explain thedeviations of the RER from the ERER.

The results confirm that, during the past three decades,MENA countries have experienced substantial overvalua-tion of their RER – 29% a year in average from the mid-1970s to the mid-1980s, and 22% from the mid-1980sto 1999 (see Table 2). In general, the extent of overvalua-tion has not significantly decreased – contrary to Latin

9 The use of the fixed-effect methodology is supported by the data, as shown by the Fischer test of equality of intercepts across countries,and is preferable to the random effect methodology, as revealed by the value of the Haussmann test (see Table 1).10 Other attempts have consistence in an ‘economic’ determination of these ‘sustainable’ levels (inspired by Edwards, 1988). As asustainable value for openness, the average of the three higher values of the variable were taken, and in the case of capital inflows,zero if the rate of growth of the economy was inferior to the international interest rate – which means that borrowing was notsustainable. Those calculations are not shown here because they did not give better results as far as misalignment is concerned.

Calculation of misalignment has been adjusted according to a base year, where the RER could be considered close to its equilibriumlevel. This has been the case especially in periods following devaluation and structural adjustment, when balance of payments were alsoclose to the equilibrium. For example it has been considered that RER was close to the equilibrium in 1989 in Morocco;1991 and 1994–1995 in Algeria; 1993–1994 in Egypt; 1995 in Iran; 1992 in Jordan; and 1980, 1994, and 1997 in Tunisia. The method used to determinethe probability of such an event of the RER being in equilibrium has been to identify a period of time when the difference between theobserved and the sustainable value of the fundamental variables was very small.

Some more sophisticated calculations consist – when a variable has a unit root – in using time series techniques introduced byBeveridge and Nelson (1981), where variables are decomposed into a random walk with a drift and a stationary component. Thistechnique – unlike the trend stationary model-based decomposition – allows the steady-state growth path of the series to shift over time.Fluctuations around the shifting permanent path reflect cyclical effects.

Table 1. Estimation results of the cointegrating Equations 1 and 10;dependant variable: ln(e)

Variable Eq 1 Eq 10

ln(Inv) 0.09 0.11(2.0) (2.3)

ln(Open) �0.71 –(14.4)

ln(TP ) – �0.32(6.7)

ln(TOT ) 0.23 0.24(4.9) (4.8)

Capinf 0.45 0.5(4.5) (4.7)

ln(DebtServ) �0.18 �0.14(9.9) (7.5)

Adjusted R2 0.63 0.55Fischer test 25.9 19.1Haussmann test 20 18.8

Note: Student t statistics are within parentheses. The number ofobservations used in Equation 1 and 10 are, respectively, 1092 and1080. Data have been compiled from World Development Indi-cators, Global Development Finance, Global Development Net-work, and Live Data Base (World Bank)Source: Authors’ estimations.

2212 M. K. Nabli and M.-A. Veganzones-Varoudakis

Dow

nloa

ded

by [

The

UC

Irv

ine

Lib

rari

es]

at 1

1:23

19

Dec

embe

r 20

14

America, Africa, or Asia. Overvaluation remains higherin MENA than in the other regions, except CFA Africa(see next section for the experience of individual MENAcountries).

On the other hand, exchange rate volatility has generallybeen lower in the MENA economies than in the otherregions of the sample (see Table 2). This can surely beexplained by the less flexible exchange rate regimes ofthe MENA countries. This conclusion should, however,be nuanced. In particular during the second subperiod(1985–1999), the volatility of the exchange rate in theMENA region is not very different from that in LatinAmerica, and is higher than in Asia.

IV. RER MANAGEMENT ANDMANUFACTURED EXPORTPERFORMANCE

Manufactured exports in the MENA countries

Table 3 presents data on the performance of some MENAcountries in terms of manufactured exports. Over the lastthree decades, the success of these countries varied widelyin increasing exports and in diversifying their economies.

Tunisia and Jordan have been the most successful in

increasing their export of manufactures. Tunisian manu-factured exports rose, on average, from 24.5% of total

exports in the 1970s to 75% in the 1990s (4.6% to 21.2%of GDP). If Jordan’s performance seems less impressive

than Tunisia’s, its increase in manufactured exports asa percentage of GDP is, in fact, comparable to Tunisia’s

(although Jordan’s level of exports to GDP remains lower).Morocco also significantly increased its exports during the

1970s and 1980s, but these gains slowed in the 1990s.In Egypt, manufactured exports increased slowly

throughout the period, growing from 27.1% of totalexports in the 1970s to only 36.6% in the 1990s (and, in

fact, decreasing from 3.1 to 2.4% of GDP). The two majoroil-exporting countries, Algeria and Iran, showed the most

dismal performance, with manufactured exports remainingnegligible throughout the period.

Modelling exports of manufactured products

Overvaluation should have had a cost for the MENA

countries that can be quantified. As seen previously,manufactured exports should have suffered from RER

misalignment and volatility. The following model is usedto test for these effects:

lnðXi, tÞ ¼ cþ b1 � GDPgrTPi, t þ b2 � lnðTOTni, tÞ

þ b3 � lnðInvi, tÞ þ b4 � lnðRoadsi, tÞ þ b5 � lnðH1i, tÞ

þ b6 � RERVoli, t þ b7 � lnðRERMisi, tÞ þ "i, t ð2Þ

The model explains exports to GDP in logarithmic form

by:

(a) the GDP growth rate of the trade partners(GDPgrTP), which can have a ‘pulling’ role in

exports;(b) the logarithm of the terms of trade ln(TOTn), the

improvement of which increases the profitabilityof production for export;

(c) the logarithm of the ratio of investment to GDP[ln(Inv)], which is conducive to an increase in

Table 3. Average manufactured exports of selected MENA countries

Algeria Egypt Iran Jordan Morocco Tunisia

%X %GDP %X %GDP %X %GDP %X %GDP %X %GDP %X %GDP

1970–1979 3.0 0.6 27.1 3.1 2.9 0.6 25.8 1.9 16.0 2.1 24.5 4.61980–1989 1.5 0.3 19.2 1.5 4.0 0.3 42.7 5.4 39.4 6.0 49.4 11.71990–1999 3.3 0.8 36.6 2.4 6.6 1.5 48.9 9.5 52.9 7.5 74.9 21.2

Note: For the first subperiod, four values were missing for Iran (1970, 1971, 1972, and 1973). For the third subperiod, two values weremissing for Iran (1991 and 1992) and one for Jordan (1996).Source: Authors’ calculations.

Table 2. Average misalignment and volatility

1975–1980 to 1984 (% per year) Misalignment Volatility

MENA 29 7.9Latin America 20 11.2Africa (CFA) 61 12.7Africa (non-CFA) 29 11.3South Asia 43 13South East Asia 10 5.4

1985 to 1999 (% per year)

MENA 22 12.4Latin America 10 12.9Africa (CFA) 28 14.5Africa (non CFA) 13 16South Asia 15 8.3South East Asia 5 8.6

Source: Authors’ calculations.

Exchange rate policy and manufactured exports 2213

Dow

nloa

ded

by [

The

UC

Irv

ine

Lib

rari

es]

at 1

1:23

19

Dec

embe

r 20

14

overall production capacity, and thus to an increasein export capacity;

(d) the availability of core infrastructure, measured bythe logarithm of the length of roads ln(Roads) in kmper km2, as well as the availability of human capital,approximated by the logarithm of the average num-ber of years of primary schooling of the adult popu-lation [ln(H1)];

(e) the volatility in relative prices, approximated bythe volatility of the RER (RERVol ) and calculatedas the coefficient of variation of the RER over a five-year period.11 RER volatility increases uncertaintyregarding the profitability of producing tradablegoods;

(f) the distortions in relative prices, as measured by theRER misalignment (RERMis), where overvaluationhampers competitiveness and diverts investment outof the more productive tradable goods sectors. RERmisalignment can also disrupt exports by increasingRER uncertainty.

In addition, the heterogeneity of the sample is controlledfor by considering country dummy variables. These vari-ables reflect differences in the quality of institutions ordifferent endowments in natural resources – which can bethe origin of large discrepancies in the natural propensityto export. The hypothesis of country dummy variablesis supported by the data for manufactured exports12

(see Table 4). A time dummy variable is also introducedfor the years 1974–1975, corresponding to the first oil shock.

Econometric results

Equation 2 was estimated on the panel of 53 developingcountries from 1970–1980 to 1999, for both total (Xtot)and manufactured exports (Xmanuf ). Manufacturedexports are more sensitive to competitiveness problems.They should consequently be more negatively influencedby RER overvaluation. Because of missing data for somevariables, the model was finally estimated on two unba-lanced panels of, respectively, 943 and 837 observations.13

Results are shown in Table 4.The estimations confirm the negative impact of exchange

rate misalignment on total and manufactured export. Thecoefficient is rather strong in the case of manufactured

exports (�0.72). It remains significant for total exports(�0.10). The weaker elasticity in the latter case can be

explained by the fact that total exports include productsthat are less sensitive to competitiveness – such as oilproducts and other primary goods, which are often

owned and managed by governments.For the MENA region as a whole, exchange rate

policy helps explaining losses in competitiveness and inmanufactured exports. During the whole period, RER

overvaluation reduced the ratio of manufactured exportsto GDP by 18% a year on average. Manufactured exports,

which averaged 4.4% of GDP from 1970 to 1999, could

11 To compute this indicator, some economists use more or less sophisticated regression techniques, such as the variance of the residual ofthe regression of the RER on a time trend, or an ARCH modellization RER behaviour. However, from an empirical point of view, allthese measures are highly correlated, and the standard deviation or the coefficient of variation measures perform as well as moresophisticated ones (see Grobar, 1993; Kenen and Rodrik, 1986).12 As shown by the value of the Fischer test of equality of intercepts across countries, and by the value of the Haussmann test as far as therandom effect method is concerned (Table 4).13 Before proceeding to the estimation of Equation 2, the degree of integration of the series entering into the regression was tested, andthe existence of a long-term relationship among them. The results of the ADF tests of the variables of Equation 2 – using Im et al. (1997)critical values – are shown in Table B2, Appendix B.

Table 4. Estimation results of the exports equations; dependentvariables: ln(Xmanuf) and ln(Xtot)

VariableManufactured exportsln(Xmanuf )

Total exportsln(Xtot)

GDPgrTP 2.83 1.48(1.9) (2.5)

ln(TOTn) �1.4 0.1(0.8) (2.15)

ln(Inv) 0.87 0.30(5.8) (8.17)

ln(Roads) 0.08 0.10(1.4) (3.15)

ln(H1) 1.92 0.26(11.1) (5.17)

RERVol �0.27 �0.1(0.8) (1.2)

ln(RERMis) �0.72 �0.10(5.7) (2.7)

Year1974 0.25(1.6) –

Year 1975 0.34(1.7) –

Intercept – �1.14(9.0)

Adjusted R2 0.81 0.13Fischer test 31.7 78.3Haussmann test 12.4 0.20

Note: Student t-statistics are within brackets. The number ofobservations used in the regressions are, respectively, 816and 964. Data have been compiled from World DevelopmentIndicators, Global Development Finance, Global DevelopmentNetwork, and Live Data Base (World Bank).Source: Authors’ estimations.

2214 M. K. Nabli and M.-A. Veganzones-Varoudakis

Dow

nloa

ded

by [

The

UC

Irv

ine

Lib

rari

es]

at 1

1:23

19

Dec

embe

r 20

14

Table 5. Cost of misalignment on manufactured exports; selected MENA countries

Algeria Egypt Iran Jordan Morocco Tunisia

ExpM* Mis Cost** ExpM* Mis Cost** ExpM* Mis Cost** ExpM* Mis Cost** ExpM* Mis Cost** ExpM* Mis Cost**

1970–1979 3 1.79 �1.7 27 1.15 �2.9 3 1.42 �0.9 26 1.57 �10.5 16 1.49 �5.7 251980–1989 1.5 1.59 �0.6 19 1.22 �3.0 4 1.24 �0.7 43 1.31 �9.4 39 1.08 �2.4 49 1.03 �1.01990–1999 3.3 1.08 �0.2 37 1.09 �2.4 7 1.84 �4.0 49 1.09 �3.1 53 1.10 �3.7 75 1.16 �8.7

1970–1999 2.6 1.49 �0.8 27.6 1.15 �2.7 4.5 1.49 �1.8 39.1 1.25 �7.7 36.1 1.21 �3.9 49.6 1.09 �4.8

Note: *ExpM ¼ manufactured exports as percent of total exports.**Cost ¼ cost of overvaluation as percent of total exports.

Exchangerate

policy

andmanufactu

redexports

2215

Dow

nloa

ded

by [

The

UC

Irv

ine

Lib

rari

es]

at 1

1:23

19

Dec

embe

r 20

14

have reached 5.2% of GDP if no overvaluation had

taken place. These losses were more concentrated in the

1970s and 1980s than in the 1990s – due to the higher over-

valuation of the currencies during those two subperiods.

Some countries with a more diversified export base –

such as Jordan and Morocco (see Table 5) – faced high

losses during the 1970s and 1980s. Because of its high

level of manufactured exports, Tunisia still incurred a

large loss during the 1990s despite a relatively low level

of misalignment. But in these countries, RER misalignment

either declined significantly or remained low during

the 1990s, as the countries saw a continuous rise in

diversification of their manufactured exports.

In the major oil-exporting countries – Iran and Algeria –

the large overvaluation of the currency has certainly

contributed to the low diversification of their exports

away from oil. But the losses, as measured here, appear

small given the low initial level of manufactured exports,

which can be explained by the structure of their economies.

The estimations fail, however, to show a significant

impact of RER volatility on manufactured as well as on

total exports of the countries. This finding does not confirm

the empirical results of several studies of different groups

of economies (see in particular Ghura and Grennes, 1993;

Grobar, 1993; Cushman, 1993; Gagnon 1993).

The results also highlight that total, as well as manufac-

tured exports are positively influenced by the GDP growth

rate of the trade partners, the ratio of investment to GDP,

and the physical and human infrastructure (measured

respectively by the length of the road network and by the

level of primary education of the population).14

The pulling effect of the trade partners’ GDP growth rate

is particularly strong in the case of manufactured exports

(elasticity of 2.8 against 1.5 for total exports). This result

goes in the direction expected. The income elasticity is

higher for manufactured products than for other products

in the economy. The same conclusions can be drawn

for human capital which improves the profitability of

investment and the competitiveness of manufactured

exports much more than in other sectors of the economy.

The effect of primary education is particularly strong

(elasticity of 1.9, compared to 0.26 for total exports).

This makes of education a key variable for the competi-

tiveness of the manufacturing sector in the developing

world. Manufactured exports, however, are not sensitive

to improvement of terms of trade, which were supposed

to provide an incentive to produce for the export sector.

This can reflect the fact that the terms of trade measures

include prices of exports of agriculture and mining

products, which are not included in manufacturing.

IV. CONCLUSION

In this paper, it is shown that – during the 1970s and 1980s– MENA countries are characterized by a significant over-valuation of their currency. Overvaluation decreases in the1990s, probably due to some degree of flexibilization ofthe exchange rate regime or to better macroeconomic man-agement. Misalignment remains, however, higher than inother developing countries (but CFA Africa). This may beexplained by the delay in adopting more flexible exchangerates and in reforming the economy.

In fact, although many economies have progressivelyadopted more flexible exchange rate regimes – leading toa better management of their RER – most MENA coun-tries are still implementing fixed or adjustable-pegexchange rate policies. In addition, if the shift towards amore open economy has begun in several countries of theregion, this process needs to be deepened, since the currentsituation reduces manufactured competitiveness and weak-ens the incentive for exporters to increase their penetrationof foreign markets. This is partly the case for oil exportingcountries, which have failed to address the volatility oftheir economies and which diversification of exports isstill very low. But this lack of trade openness also explainsthe low diversification of other MENA countries in the1970s and 1980s.

The study also illustrates that overvaluation has hada cost for the region in term of competitiveness.Manufactured exports, in particular, have been affectedby the overvaluation of the exchange rate. These findingsconfirm recent assessments of economic policies andperformance in developing countries, which underline thecrucial issue of the management of the real effectiveexchange rate. The results corroborate the findings ofEdwards (1988); Balassa (1990) and Cottani et al. (1990)for different groups of developing countries.

REFERENCES

Achy, L. (2001) Equilibrium exchange rate and misalignment inselected MENA countries, EUI Working Papers,Mediterranean Programme Series, RSC no. 2001/42.

Baffes, J., Elbadawi, I. A. and O’Connell S. A. (1997) Single-equation estimation of the equilibrium real exchange rate,mimeo, Swarthmore College, May.

Balassa, B. (1990) Incentive policies and export performance inSub-Saharan Africa, World Development, 18(3), 383–91.

Beveridge, S. and Nelson C. R. (1981) A new approach todecomposition of economic time series into permanentand transitory components with particular attention tomeasurement of the business cycle, Journal of MonetaryEconomics, 7(2), 151–270.

14 Surprisingly, in the case of roads, the elasticity for manufactured exports is weakly significant. This may be due to the fact that, inseveral MENA countries, oil exports represent an important percentage of total exports (as well as of GDP). In this case, it can beassumed that oil exports have led to the construct of good physical infrastructure.

2216 M. K. Nabli and M.-A. Veganzones-Varoudakis

Dow

nloa

ded

by [

The

UC

Irv

ine

Lib

rari

es]

at 1

1:23

19

Dec

embe

r 20

14

Cottani, J. A., Cavallo, D. F. and Khan, M. S. (1990) Realexchange rate behavior and economic performance inLDCs, Economic Development and Cultural Change, 39(1),61–77.

Cushman, D. O. (1993) The effects of real exchange rate risk oninternational trade, Journal of International Economics, 15(2),45–64.

Domac, I. and Shabsigh, G. (1999) Real exchange rate behaviorand economic growth: evidence from Egypt, Jordan,Morocco and Tunisia, IMF Working Paper, WP/99/40,March.

Edwards, S. (1988) Exchange Rate Misalignment in DevelopingCountries, The Johns Hopkins University Press, Baltimore.

Edwards, S. (1989) Real Exchange Rate, Devaluation andAdjustment: Exchange Rate Policy in Developing Countries,MIT Press, Cambridge, MA.

Edwards, S. (1994) Real and monetary determinants ofreal exchange rate behavior: theory and evidence from devel-oping countries, in Estimating Equilibrium Exchange Rates, J.Williamson (Ed.) Institute of International Economics,Washington DC.

Elbadawi, I. (1994) Estimating long-run equilibrium realexchange rates, in Estimating Equilibrium Exchange Rates,J. Williamson (Ed.) Institute of International Economics,Washington DC.

Frankel, J. A. and Romer, D. (1999) Does trade cause growth?,The American Economic Review, 89(3), 399–400.

Gagnon, J. E. (1993) Exchange rate variability and the levelof international trade, Journal of International Economics,34(3–4), 269–88.

Ghura, D. and Grennes, T. J. (1993) The real exchange rate andmacroeconomic performances in Sub-Saharan Africa,Journal of Development Economics, 42(1), 155–75.

Grobar, L. M. (1993) The effect of real exchange rate uncertaintyon LDC manufactured exports, Journal of DevelopmentEconomics, 14(2), 367–75.

Harberger, A. (1986) Economic adjustment and the real exchangerate, in Economic Adjustment Exchange Rates in DevelopingCountries (Eds) S. Edwards and L. Ahamed, Universityof Chicago Press, Chicago.

Im, K. S., Pesaran, M. H. and Shin, Y. (1997) Testing for unitroots in heterogeneous panels, DAE Working PaperAmalgamated Series, no. 9526, Cambridge University.

Kenen, P. B. and Rodrik, R. (1986) Measuring and analysingthe effects of short-term volatility in real exchange rates,Review of Economics and Statistics, 68(2), 311–5.

Mongardini, J. (1998) Estimating Egypt equilibrium realexchange rates, IMF Working Paper, WP/98/5, January.

Nishimizu, M. and Robinson, S. (1986) Productivity growthin manufacturing, in Industrialization and Growth: AComparative Study (Eds) H. Chenery, S. Robinson, andM. Syaruin, World Bank Research Report, OxfordUniversity Press, Oxford.

Razin, O. and Collins, S. M. A. (1997) Real exchange ratemisalignments and growth, NBER Working Paper, 6174,September.

Reinhardt, C. (1995) Devaluation, relative prices, and inter-national trade, IMF Staff Paper, No. 42.

Sekkat, K. and Varoudakis, A. (2002) The impact of exchangeand trade policy reforms on manufactured exports in NorthAfrica, Development Policy Review, 20(2), 177–89.

Sorsa, P. (1999) Algeria: the real exchange rate, export diversifi-cation, and trade protection, IMF Working Paper, no. 99/49,April.

Sundararajan, V., Lazare, M. and Williams, S. (1999)Exchange rate unification, the equilibrium real exchangerate, and choices of exchange rate regime: the case ofIslamic Republic of Iran, IMF Working Paper, WP/99/15,January.

Williamson, J. (1985) The exchange rate system, Policy Analysesin International Economics, No. 5. Institute for InternationalEconomics, Washington, DC.

Exchange rate policy and manufactured exports 2217

Dow

nloa

ded

by [

The

UC

Irv

ine

Lib

rari

es]

at 1

1:23

19

Dec

embe

r 20

14

APPENDIX A:

Table A1. List of countries in the sample

Africa Asia

MENA CFA NonCFA South East Asia Latin America

Bahrain (BHR) Burkina Faso (BFA) Botswana (BWA) Indonesia (IDN) Argentina (ARG)Algeria (DZA) Cote d’Ivoire (CIV) Gambia, The (GMB) Korea, Rep. (KOR) Bolivia (BOL)Egypt, Arab Rep. (EGY) Gabon (GAB) Kenya (KEN) Malaysia (MYS) Brazil (BRA)Iran, Islamic Rep. (IRN) Cameroon (CMR) Madagascar (MDG) Philippines (PHL) Chile (CHL)Jordan (JOR) Ghana (GHA) Mozambique (MOZ) Thailand (THA) Colombia (COL)Kuwait (KWT) Niger (NER) Mauritius (MUS) Costa Rica (CRI)Malta (MLT) Senegal (SEN) Malawi (MWI) South Asia Ecuador (ECU)Morocco (MAR) Togo (TGO) Nigeria (NGA) Guatemala (GTM)Syrian Arab Republic (SYR) Tanzania (TZA) Bangladesh (BGD) Mexico (MEX)Tunisia (TUN) India (IND) Peru (PER)

China (CHN) Paraguay (PRY)Other countries Sri Lanka (LKA) Uruguay (URY)Israel (ISR) Pakistan (PAK) Venezuela, RB (VEN)

APPENDIX B:

Table B1. Augmented Dickey–Fuller (ADF) unit root tests;Equations 1 and 10

VariableADFstatistic k (1)

Criticalvalue (2) ADF test

RERln(e) �1.73 1 �1.69** I(0)

Fundamentalsln(Inv) �1.92 1 �1.82* I(0)ln(Open) �1.69 1 �1.69** I(0)ln(TP) �3.77 1 �1.82* I(0)ln(TOT ) �2.15 1 �1.82* I(0)Capinf �2.79 1 �1.82* I(0)DebtSev

Other variablesDef �2.43 1 �1.82* I(0)P �2.76 1 �1.82* I(0)Depr �3.07 1 �1.82* I(0)BMP �2.69 1 �1.82* I(0)

Notes:(1) k is the number of lags in the ADF test.(2) Im et al (1997) critical values (respectively, *1 and **5% level).Data have been compiled from World DevelopmentIndicators, Global Development Finance, Global DevelopmentNetwork, and Live Data Base (World Bank).Source: Authors’ calculations.

Table B2. Augmented Dickey–Fuller (ADF) unit root tests;Equation 2

VariableADFstatistic k (1)

Criticalvalue (2) ADF test

ln(Xmanuf ) �1.76 �1.69** –GDPgrTP �3.69 1 �1.82* I(0)ln(TOTn) �2.15 1 �1.82* I(0)ln(Inv) �1.92 1 �1.82* I(0)ln(Roads) �3.65 1 �1.82* I(0)ln(H1) �1.86 1 �1.82* I(0)RERVol �2.83 1 �1.82* I(0)ln(RERMis) �2.24 1 �1.82* I(0)

Notes:(1) k is the number of lags in the ADF test.(2) Im et al. (1997) critical values (respectively, *1 and **5%level).Data have been compiled from World Development Indicators,Global Development Finance, Global Development Network,and Live Data Base (World Bank)Source: Authors’ calculations.

2218 M. K. Nabli and M.-A. Veganzones-Varoudakis

Dow

nloa

ded

by [

The

UC

Irv

ine

Lib

rari

es]

at 1

1:23

19

Dec

embe

r 20

14

APPENDIX C: SHORT-TERM DYNAMICS OFTHE RER

Since the variables are cointegrated, the short-term

dynamic adjustment of the RER towards its equilibrium

level can be estimated through an error correction model.

The estimated equation is as follows:

�lnðei, tÞ ¼ �a½lnðei, t�1Þ�lnðe�i, t�1Þ�þa0�lnðei, t�1Þ

þ b1 ��lnðInvi, tÞþb2 ��lnðOpeni, tÞ

þ b3 ��lnðTOTi, tÞþb4 ��lnðCapinfi, tÞ

þ b5 ��lnðDebtSevi, tÞ

þ c1 ��lnðInvi, t�1Þþc2 ��lnðOpeni, t�1Þ

þ c3 ��lnðTOTi, t�1Þþc4 ��lnðCapinfi, t�1Þ

þ c5 ��lnðDebtSevi, t�1Þ

þ d1 �Depri, tþd2 �Depri, t�1

þ e1 � Infli, tþe2 � Infi, t�1

þ f1 �Def i, tþf2 �Defi, t�1

þ g1 � BMPtþg2 � BMPi, t�1þe2t ðCÞ

In addition to the error correction term, i.e., the

lagged error term of the cointegrating Equation [ln (et-1)�

ln (e*t-1)], and lagged variables of Equations (1) and (10) in

first differences, indicators of fiscal policy (fiscal deficit as

percentage of GDP, Def ) and of exchange rate policy

(nominal depreciation, Depr, and black market premium,

BMP), as well as inflation (Inf l) are included. The assump-

tion is that the adjustment path of the RER towards its

equilibrium level may be affected (accelerated or slowed

down) by short-term economic policies, including capital

controls (for which BMP is a proxy), nominal exchange

rate depreciation, and fiscal policy, of which inflation can

be a consequence. Table C1 shows the estimates of the

error correction model.

Nominal devaluations show a short-run impact on the

RER, which is in the expected direction and significant.

The change in the official nominal exchange rate (NER)

hence captures the strong temporary effect that devaluation

may produce on the RER due to price rigidities.

In addition, these estimations highlight the role of othershort-term economic policies through the black marketpremium (BMP) and inflation (Inf l ). These variables (Inf l,BMP), by leading to a rise in the price of non-tradablegoods, appreciate the RER and lead to its overvaluation.Although public deficit does not show a significant effect,it can be captured by the inflation variable, the effect ofwhich is strong and which is also supposed to be a proxyfor some other inappropriate policies.

Table C1. Estimates of the error correction model; dependantvariable: �ln(et)

Eq (1) Eq (10)Variable Elasticity Student Elasticity Student

"1t�1 �0.13 (7.3) �0.2 (9.7)�ln(Invt) 0.04 (1.4) 0.2 (0.8)�ln(Opent) �0.27 (7.0) �0.5 (14.5)�ln(TOTt) 0.1 (2.7) 0.15 (4.8)�(Capinf ) 0.006 (1.3) 0.25 (3.8)�ln(DebtSevt) – – 0.02 (1.81)�ln(Invt�1) 0.01 (0.3) 0.03 (1.2)�ln(openvt�1) 0.06 (1.6) 0.02 (0.5)�ln(TOTt�1) 0.02 (0.7) 0.04 (1.4)�(Capinft�1) 0.78 (1.8) �0.33 (5.1)�ln(DebtSevt�1) – – 0.04 (2.1)�ln(et�1) 0.06 (1.6) 0.16 (4.9)Depr �0. 22 (18.0) �0.04 (10.9)Deprt�1 �0.05 (8.0) 0.006 (1.4)Inf lt 0.19 (17.8) 0.04 (10.4)Inf lt�1 0.05 (7.9) �0.007 (1.6)Deft�1 0.05 (0.4) – –Deft 0.05 (0.4) – –BMPt 0.006 (2.5) 0.12 (5.5)BMPt�1 0.21 (0.9) �0.003 (1.47)

D–W 1.74 2.03

Note: Student t statistics are within brackets. The sample includes,respectively, 640 and 828 observations over the 1970–1999 period.*"1t�1 is the lagged error term of the cointegrating Equation 1.Data have been compiled from World Development Indicators,Global Development Finance, Global Development Network,and Live Data Base (World Bank).Source: Authors’ estimations.

Exchange rate policy and manufactured exports 2219

Dow

nloa

ded

by [

The

UC

Irv

ine

Lib

rari

es]

at 1

1:23

19

Dec

embe

r 20

14