EXPORTS AND ECONOMIC GROWTH IN ETHIOPIA:

AN EMPIRICAL INVESTIGATION

DEBEL GEMECHU

A thesis Submitted to the School of Graduate Studies of Addis Ababa

University in partial fulfillment of the requirements for the Degree of
Master of Science in Economics (Economic Policy Analysis)

JUNE, 2002

ADDIS ABABA
Declaration

The thesis is my original work, has not been presented for a degree in any other university
and that all sources of material used for the thesis have been duly acknowledged.

Declared by:

Name: Debel Gemechu Guta

Signature: _________________

Date: June, 2002

Place: Addis Ababa, Ethiopia

This thesis has been submitted for examination with my approval as an M.Sc. thesis
supervisor.

Name: Dr. Girma Estiphanos

Signature: __________________

Date: __________________
ACKNOWLEDGEMENTS

With a heart full of praises and adorations will I exalt you Lord for what you have done for

me. Without your help this thesis wouldn't have been possible.

Appreciation and gratitude are extended to all who have assisted me during the course of my

graduate studies and in undertaking this thesis. My special thanks goes to my advisor Dr.

Girma Estifanos for his guidance and comments throughout the development of this study. I

also express my gratitude to Dr. Haile Kebret for his useful comments and provision of

working materials.

I am grateful to the Department of Economics for providing me the opportunity to participate

in the Masters Program and the African Economic Research Consortium (AERC) for funding

my studies at the Addis Ababa University.

I owe more than I can say to my parents, especially my father (ababi) and my mother

(emmami) for their unreserved help, care and encouragement throughout my study. I will

never forget the friendship and companionship of my sisters Lelisie and Hawi Gemechu.

Finally, My heart felt thanks goes to all my friends, especially Wondimagegnehu Negera and

Miraf Tamene for their prayer and moral support throughout my study.
 TABLE OF CONTENTS

PAGES

Acknowledgements

List of Tables

List of Figures

Abstract

1. INTRODUCTION......…………………………………………..................... 1

Background ................................................................................................... … 1

Statement of the Problem ............................................................................. … 3

Objectives of the Study................................................................................. … 7

Significance of the Study ............................................................................ … 7

Organization of the Study ............................................................................. … 8

2. THE ROLE AND PERFORMANCE OF THE EXPORT SECTOR

IN THE ETHIOPIAN ECONOMY………………………………………….9

2.1 The Role of the Export Sector .............................................................…....9

2.2 Trends in Export Performance ................................................................….12

2.3 Commodity and Sectoral Structure of Export .......................................…...14

2.4 Export Promotion Efforts in Ethiopia ..........................................................22

2.4.1 Pre 1991/92 ……......................................................…...22

2.4.2 Post 1991/92 ……....................................................…...26

3. REVIEW OF LITERATURE………………………………………………...30

3.1 Theoretical Literature.............................................................................. .….30

3.2 Trade Strategies for Development: Export Promotion Versus

Import Substitution.....................…….....................................................…..34

3.3 Export and Economic Growth ...............................................................…....41

3.4 Empirical Literature ………….......................................................................45

3.4.1 Cross-sectional Studies ...................................................…45

3.4.2 Time Series Studies ........................................................…50

4. DATA SOURCES, METHODOLOGY AND MODEL

SPECIFICATION ............................................................................................59

4.1 Econometric Methods ....................................................................................59

4.1.1 Stationary and Non-Stationary Series …………………….59

4.1.2 Tests for Unit Roots ............................................................60

4.1.3 Cointegration and the Error Correction

Model (ECM) ......................................................................61

4.1.4 Granger Causality Test .......................................................63

4.2 Model Specification ......................................................................................64

4.3 Data Description and sources ........................................................................69

5. EMPIRICALANALYSIS…...............................................................................71

5.1 Result of Unit Roots Tests .............................................................................71

5.2 Estimation the Long Run and Error Correction Models ……………………72
 6. CONCLUSION AND POLICY IMPLICATIONS………................................83

6.1 Conclusions......................................................................................................83

6.2 Policy Implications.........................................….............................................87

BIBILOGRAPHY…………………………………………………………………..90

ANNEXES

\
 LISTS OF TABLES AND FIGURES

TABLES PAGES
2.1 Share of Exports in the different sectors of the economy……..…………...……9

2.2 Average Annual Growth Rate of Export ………………………………………12

2.3a Commodity Structure of Exports ………………………………………...……..14

2.3b Sectoral Structure of Exports ……..…………………………………………….20

2.3c Relative Performance of Ethiopia's manufacturing Sector (1998) ………….….21

2.4 Effective exchange rate (Birr/Dollar) and anti-export-bias

(EERm/EERx) ……………………………………………………………………29

5.1 Results of Unit Root Tests for Order of

Integration of the variables ………………………………………………….….70

5.2a Result of the Test for the number of cointegrating vectors …………………….72

5.2b Estimation Result of the long-run Model ………………………………………73

5.2c Estimation Result of the Error-Correction Model ……………………………...74

5.2d Results of the DF and ADF test for the Residuals of

cointegration equations …………………………………………………………78

5.2e Results of the Error correction Model for the dependent

variable LGDPt ………………………………………………………………….79

5.2f Results of the Error correction Model for the dependent

variable LEXPt ……………………………………………………………...…..80

FIGURES

2.1a Output (GDP) and Exports …………………………………………………..…..10

2.1b Exports and Imports …………………………………………………………..…10

2.1c Total Revenue and Export Tax …...………………………….……………….….11

 ABSTRACT

This study investigated the effect of exports on economic growth in Ethiopia for the period

1960/61-2000/01. The study aimed to review the policies undertaken by the different regimes

in relation to export policies, and to empirically test the relationship between exports and

economic growth using different techniques. In addition, attempts were also made to examine

the supply (structural) constraints to export growth in Ethiopia.

To test the export-economic growth relationship, in addition to using the frameworks that

have been followed by most of the previous studies, an extension to the previous studies was

made by introducing cointegration and error correction approaches in the regression analysis.

Furthermore, a simultaneous equation model and the Granger causality test were conducted

to examine the indirect effects of export on economic growth and to address a possible

simultaneity problem that may arise because of the correlation between export and economic

growth.

The results from the cointegration and error correction models revealed that export

significantly affected economic growth in the short run. In addition to its direct effect, export

is also found to indirectly affect economic growth as evidenced from the simultaneous

equation models. Furthermore, the causality test conducted indicated that causality runs from

exports to economic growth. The key finding in this study is that export growth positively

and significantly affected economic growth and the result is not sensitive to the

methodologiesused.
1.INTRODUCTION

1.1 Background

The issue of accelerated economic growth has been the main agenda in economic policy

formulation for most of the Sub-Saharan Africa (SSA) and other developing countries of the

world since the early 1970’s. It is given paramount importance precisely because the

economic growth of these countries has been showing a deteriorating trend from time to time.

The per capita income of most of these countries has registered a negative growth rate during

the past three decades owing to the sluggish economic performance coupled with the startling

population growth. The records of the economic performance of most SSA counties exhibit

that they had been performing better before their colonial independence than today [Rodney,

1982]. Hence, considerable attention has been paid by a number of development economists

and government policy makers to review the experiences of these countries in order to

promote economic growth and improve their living standards.

One area that has been given much focus in order to promote the economic performance of

these countries is external trade. Following the traditional trade argument, trade is viewed as

an "engine" if not as a "handmaiden" of growth playing a supportive role in the economic

growth of the least developed countries (LDCs).

The economic growth of the present day developed nations like the United States, Canada,

Australia, New Zealand (referred as regions of recent settlement) that were once developing

nations is largely attributed to international trade [Salvatore, 1990]. Hence international trade
1
has been given much importance in the policy formulation of many LDCs, viewing it as a

vehicle to transform the economic performance of these countries.

International trade has also played a crucial role in the historical development of the third

world. In the second half of the 20th century, the tremendous economic performance of the

"four tigers"- South Korea, Taiwan, Hong Kong and Singapore has been largely attributed to

the performance of the external sector where the export sector was given a greater emphasis.

Strong political commitment towards export promotion and the application of appropriate

policies together with efficient institutional mechanisms helped these countries attain a

higher growth rate of exports and hence of the overall economy.

The success of these East Asian countries coupled with failure of the import substitution (IS)

strategy, which once was held by many LDCs as the appropriate policy during the 1950s and

1960s, led most LDCs to give due attention to the export promotion trade strategy. It is

argued that this strategy would help primary commodity exporting countries achieve optimal

scale and enable them tap foreign technology which is deemed to be important for the growth

process of these countries.

Many studies have been conducted on LDCs to check whether exports do contribute to

economic growth or not. The results of these studies are very important to forward relevant

policy recommendations that would enable these countries enjoy the benefits of economic

growth.

2
Ethiopia, like many other developing countries has actively pursued the import- substitution

industrialization strategy during the Imperial and Derge regimes. The World Bank (1987)

classified Ethiopia as one of the strongly inward oriented countries during the periods of

1963-73 and 1973-85, which coincides with the Imperial and part of the Derge regimes,

respectively. However, the IS trade strategy hadn't performed well, where the import

competing industries remained infant and were at their rudimentary stage despite the tariff

and non-tariff protection. With the fall of the Dergue regime, however, the current regime

initiated trade liberalization in which export promotion is the major component of the

program.

Hence, a closer look into the policies that were once followed by these governments and an

empirical investigation to find out the contribution of exports to economic growth is very

essential in order to help the country experience a sustainable economic growth.

1.2 Statement of the Problem

Owing to both internal and external factors, the growth performance of the Ethiopian

economy has been less than satisfactory during the past four decades. Externally, the

quadrupling of oil prices that took place during the period 1973-74 and the export short fall

associated with the world recession of 1974-75 resulted in chronic balance of deficit problem

that greatly affected the economic growth of many LDCs and the country. In addition, the

period of the1980s was also characterized by the collapse of commodity prices that resulted

in deterioration of terms of trade of primary-commodity exporting countries, which further

worsened the economic performance of the country.

3
Internally, recurrent drought and the prolonged war that took place till the demise of the

Derge regime have greatly affected the growth of the economy. Furthermore, the shifting of

policies from one regime to the other has negative implication on the overall economic

performance.

As the country is not self sufficient in generating the saving that is essential to realize a

sustainable economic growth, the external sector becomes very crucial for the growth

performance of the economy. During the past four decades, the country is becoming more

and more open increasing its outward linkages with the rest of the world. For instance the

openness of the economy (measured as the ratio of export plus import to GDP) which was

about 23 percent on average during the imperial regime reached 26.1 and 36.2 percents

during the Derge and the present government, respectively1.

The export sector can play a crucial role in the growth performance of the country, as can be

evident from its contribution to the different sectors of the economy. During the past four

decades, for instance, export has contributed on average about 11.3 percent to GDP. In

addition, it generates the much needed foreign exchange earning that is essentially used to

finance the imports of the country. Together with foreign aid and grants, the country uses the

foreign exchange generated from the export of primary agricultural products to import almost

all of its intermediate inputs, fuel and capital goods, which are believed to be essential for the

economic growth of the country.

__________________
1
Data from MEDaC

4
During the Imperial regime for instance, proceeds from exports used to cover on average

about 91 percent of the import bill, which has declined to 54.4 percent in the current regime.

Furthermore, taxes from foreign trade are important sources of the state revenue. During the

three regimes all together foreign trade tax accounted for more than 28.7 percent of the state

revenue of which 6.5 percent comes from export tax.

Despite its contribution to the overall economy, however, the performance of the export

sector has been less satisfactory. The country's exports are highly concentrated in agricultural

commodities, while the share of non-agricultural products in total merchandise exports is

almost insignificant. Coffee, pulses & oilseeds, hides & skins and chat, in that order, have

contributed on average about 54.4, 13.1, 11.5 and 4.1 percent of the total export earnings

during the last four decades2. During the same period, the commodity concentration index for

these commodities together with gold and petroleum averaged about 0.56. The country's

heavy reliance on these few export commodities, which are highly subjected to price

fluctuations, is one reason for the poor performance of the export sector. In addition to this,

there is geographical concentration of exports that makes the country vulnerable to the

economic conditions (demand) of its trading partners. Germany, Japan, United States,

Djibouti and Italy are the five major trading partners of the country, which altogether absorb

about 73.3 percent of the country's export3. This concentration on few trading partners

resulted in demand constraint for the nation's primary exports and could be one reason for the

poor performance of the sector and hence of the economy.

____________________________
2
Data from MEDaC
3
Data from NBE

5
Failure of the different government policies to diversify and promote exports is also one

problem that greatly reduced the competitiveness and performance of the export sector. Until

the demise of the Derge regime, the country has been recognized as one of the strongly

inward-looking countries. Anchored by high level of protection and overvalued exchange

rate, the policy of inward-looking has weakened the export sector. The policies that were

followed by the then governments had a strong anti-export bias that greatly reduced the

competitiveness of the sector.

The Transitional Government of Ethiopia (TGE) who came to power in 1991/92 launched a

new economic policy where the role of exports to economic growth was given due

importance in the development strategy of the country. However, the export supply response

to the policy change is not as anticipated. The incentives provided by the new policy to

promote exports could not totally eliminate the anti-export-bias incentive structure that

originated from heavy protection of the domestic industries. As a result the export supply

response was weak and the recovery since 1991 in export earning mainly resulted from

coffee price boom and institutional reforms than the effect of trade and exchange reform. The

factors behind the sluggish performance of the sector could be due to a combination of the

structural problems existing in the whole economy or the insufficiency of policy measures

taken to totally nullify the anti-export bias that prevailed during the previous regimes

[MEDaC, 1997]. Therefore, a closer look at the policies followed by the different regimes,

examining the contribution of structural constraints and assessing the contribution of exports

to economic growth are issues that will be examined in this study.

6
 1.3 Objectives of the Study

In light of the problems stated above, the specific objectives of this study are:

• To review the different policies undertaken to diversify and promote exports by the present

and the previous regimes,

• To investigate the relationship between exports growth and economic growth using

different methodologies,

• To examine the supply/structural constraints to export growth,

• To suggest, on the basis of empirical evidence, policies that would help the country

experience economic growth.

1.4 Significance of the Study

Understanding the contribution of export to the different sectors of the economy would help

the country fully exploit the benefits of the sector that is essential for sustainable economic

growth. In this regard, a clear picture of the domestic policies that once have been followed

and other supply constraints become very important in order to remove impediments that

deter the performance of the sector. The outcomes of this study may provide useful inputs in

the formulation of development plans and policies. For instance, it provides an empirical

magnitude of the contribution of export to economic growth, which could be important to

understand the proportion of the overall economic growth that is attributed to the

performance of the export sector. In addition it shades light on domestic policies and related

supply constraints that hinder the performance of the export sector. Understanding these

would help policy makers and related bodies take appropriate measures to remove the

impediments and be able to fully utilize the benefits of the sector.

7
 1.5 Organization of the Study

The rest of the thesis is organized as follows. The next chapter reviews the role and the

performance of the export sector and the trade policies and efforts undertaken to promote and

diversify exports. Chapter three reviews the relevant theoretical and empirical literatures. The

methodology, model specification and the data are dealt within chapter four. Chapter five is

devoted to the empirical examination of the issues using different econometric techniques.

Finally, conclusions and policy recommendations are provided in chapter six.

8
2. THE ROLE AND PERFORMANCE OF THE EXPORT
SECTOR IN THE ETHIOPIAN ECONOMY

2.1 The Role of the Export Sector

Like in most of the Sub-Sahara African and other developing countries, exports can play a

significant role in the growth performance of the Ethiopian economy. This is evident from

looking at the figures of export contribution to the different sectors of the economy (see

table2.1).

Table 2.1: Share of Exports in the different sectors of the economy

Period Average Share of Average Share of Average Share of

Exports in GDP Exports Covering Export Tax from
Imports Bill the State Revenue
(%) (%) (%)
1960/61-1973/74 10.9 91.1 7.7

1974/75-1990/91 10.0 62.7 10.1

1991/92-2000/01 12.8 54.4 1.6

1960/61-2000/01 11.0 70.4 7.3

Source: Author’s Calculation using the data from MEDaC, Excise Authority& Ministry
of Finance

It can be seen that on average exports contributed about 11.0 percent to GDP during the past

four decades. The share of exports in GDP was the highest during the present government.

9
 Export

Export
GDP

Import
00 00
20 20
98 98
19 19
96 96
19 19
94 94
19 19
92 92
Fig. 2.1a Output (GDP) and Exports

19 Fig. 2.1b Exports and Imports 19

90 90
19 19
88 88
19 19
86 86
19 19
84 84
19 19

10
82 82
19 19
80 80

Year
19 19
78
19
78 Year 19
76 76
19 19
74 74
19 19
72 72
19 19
70 70
19 19
68 68
19 19
66 66
19 19
64 64
19
19
62 62
19
19
60 60
19
19

18000
16000
14000
12000
10000
8000
6000
4000
2000
0
60000

50000

40000

30000

20000

10000

0
000'Birr
000'Birr
 Fig. 2.1c Total Revenue and Export Tax

12000

10000

8000
Tot.Rev
000'Birr

6000 Exp.Tax

4000

2000

0
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
20
60
62
64
66
68
70
72
74
76
78
80
82
84
86
88
90
92
94
96
98
00
Year

The highest share was recorded in the year 1996/1997 which was about 16.2 percent of GDP,

the lowest being 4.5 percent at the end of the Derge regime in 1991/92 (Fig. 2.1a).

The revenue from exports made the import of inputs possible that are crucial for development

purposes thereby playing as an engine of growth to other sectors. During the period 1960/61-

2000/01 proceeds from exports covered more than 70.4 percent of the import bill of the

country. In some years during the imperial regime the proceeds from export was able to

cover the total imports bill and even register a surplus (see fig. 2.1b). Hence expanding

exports enables the country reduce the foreign exchange constraint that acts as a bottleneck

for the growth of the economy. Due to administrative inconveniences or high cost to raise

government revenue, tax on foreign trade constitute a major part of the state budget. During

the three regimes all together foreign trade tax accounted for more than 27.9 percent of the

state revenue of which 7.3 percent comes from export tax.

11
 2.2 Trends in Export Performance.

The value of goods and services exported has been growing at an average annual rate of 10.2

percent for the last four decades 1960/61-2000/01. During the period under consideration, the

revenue from exports of pulses and oilseeds has been growing at an average annual rate of 59.4

percent followed by chat whose rate was 59.1 percent.

The revenue from the export of coffee and hides and skins has been growing at average annual

rate of 10.3 and 7.1 percent, respectively (See Table 2.2).

Table 2.2 Average Annual Growth Rate of Export

Period Growth Rates in Total and Major Components of Export
Total Coffee Hides& Pulses& Chat
Export Skins Oilseeds
1960/61- 8.2 2.7 9.1 13.1 0.8
1973/74
1974/75- 4.7 7.1 5.6 2.4 69.8
1990/91
1991/92- 22.5 27.4 14.1 221.0 122.3
2000/01
1960/61- 10.2 10.3 7.1 59.4 59.1
2000/01

Source: Author’s calculation using data from MEDaC, Excise Authority& Ministry
of Finance

Prior to 1974(the imperial era), there has been a modest growth in the total value of export

earning. The average annual growth rate of real value of exports was 8.2 percent. During the

12
period, the receipt from the export of pulses and oilseeds has been growing at an average

annual rate of 13.1 percent. Coffee, which is the largest export commodity, has been at an

average annual rate of 2.7 percent.

During the Derg regime (1974/75 – 1990/91), the growth rate of real exports was lower from

the rate recorded in the previous regime. The growth rate of total value of export earning

registered more than 42 percent decline from the growth rate recorded for the imperial

regime. This can be largely attributed to the poor performance in the export of pulses and

oilseeds. The proceeds from the export of the commodity, which was growing at an average

annual rate of 13.1 percent during the imperial regime, declined to 2.4 percent during the

Derge regime. During this regime, export revenue from chat has demonstrated an average

annual growth of 69.8 percent compared to 0.8 percent during the imperial regime.

Under the period of Ethiopian Peoples’ Revolutionary Democratic Front (EPDRF), that is

1991/92-2000/01,the growth rate in the real value of total exports has shown a significant

increase. In real value, total exports grew by 22.5 percent. Owing to different policy

measures under taken by the government to promote exports, revenue from the various

export commodities has shown a remarkable increase. Proceeds from oilseed & pulses, which

was growing at an average annual rate of 2.4 percent during the previous regime started to

grow at a rate of 221.0 percent. And the revenue from Chat export has been growing at an

average annual rate of 122.3 percent. Such an increase in the value of export of these

commodities is mainly attributed to a very significant growth rate recorded during the period

1991/92-1992/93. Owing to the favorable environment created after the prolonged war that

prevailed in the country, the proceeds from these commodities escalated very sharply from

the period 1991/92 to 1992/93.

13
 2.3 Commodity and Sectoral Structure of Export

Commodity structure of the Ethiopian export sub-sector is a mirror reflection of the country’s

over all economic structure at large. The nation’s output and exports are highly concentrated

in agricultural commodities, while the share of non-agricultural products in total merchandise

exports is almost insignificant. For the past four decades, primary agricultural products

accounted for 80-90 percent of the merchandise export earnings of Ethiopia.

Table 2.3a Commodity Structure of Exports (%of total)

Commodity 1970/71- 1974/75- 1991/92- 1970/71-
1973/74 1990/91 2000/01 2000/01
Coffee 42.6 61.6 59.1 54.4
Oilseeds & Pulses 25.4 6.6 7.2 13.1
Hides&Skin 10.6 12.7 11.3 11.5
Chat 0.8 2.2 9.4 4.1
Gold - 1.2 4.9 2.0
Petroleum 0.6 5.0 1.4 2.3
Fruit&vegatable 1.7 1.0 0.8 1.2
Live animals 1.2 1.8 0.2 1.1
Sugar &Molasses 2.1 1.6 0.5 1.4
Textile 0.1 0.4 0.5 0.3
Meat products 3.5 0.7 0.6 1.6
Spices 0.6 0.4 0.6 0.5
Natural Gum - 0.3 0.4 0.2
Cotton 1.3 1.0 0.5 0.9
Others 9.5 3.5 2.8 5.3
Total 100 100 100 100
Source: Author’s calculation using data from MEDaC, Excise Authority& Ministry of
Finance

14
Among the major export products, as shown in Table 3.2 above, coffee accounts for the

lion’s share of primary exports and of total merchandise exports as well. From 1970/71-

2000/01 coffee alone accounted for 54.4 percent of the total export proceeds. The average

percentage share of coffee in the total merchandise exports was 42.6, 61.6 and 59.1 percent

for the imperial, Derge and the present government, respectively. The smallest share of

coffee in the total export was 25.9 percent in 1974/75, which was due to the change in the

government and political instability. The largest share was 80.7 percent in 1977/78 due to the

then government’s development campaign efforts.

All these figures illustrate the fact that the Ethiopian merchandise export sub-sector is largely

dependent on coffee export for its badly needed foreign exchange earning. However,

Ethiopia’s share of the world coffee market has been stable at less than 2 percent during the

last twenty years and coffee exports have declined since 1997/98 along with the decline in

the world prices [World Bank, 2001]. In addition to the fluctuation in the international price

of coffee, which acts as a demand side constraint, there are supply side factors (government

policies and Institutional problems) that inhibit the performance of the commodity’s export.

According to the World Bank (2001), four factors can be mentioned.

a. In order to prevent undermining the reputation of Ethiopian coffee on the world

market, all coffee is classified as export quality or as domestic quality. However, this

has unintended consequences. When domestic price are higher than international

price, farmers are likely to illegally divert more of their production into the domestic

market. Estimates suggest that this could amount to 25 percent of production and half

of exports.
15
b. The auction system, which has been widely practiced but is now being relaxed, has a

number of drawbacks. First, the system prevents direct trading between processors

and exporters, which might result in the development of coffee with special

characteristics. Second, it prevents vertical integration; exporters involved in

processing and washing activates cannot integrate these activities because they may

not be able to re-acquire the coffee they supplied at an auction. Third, it inhabits

exporters from making long-term contracts with importers since they cannot be

assured of buying at an auction the type of coffee they contracted to supply.

c. The inability of buyers to inspect and test coffee is another constraint that further

reduces confidence in quality. However, exporters report cases where the certified

coffee was letter rejected as unfit for export when submitted for final inspection. A

solution for this problem would be to hold coffee in storage for 2-3 days before

auction and to allow private sector testing which would require additional storage

facilities at the action.

d. The auction system is weak and is being hindered by a variety of factors. For

instance, some inexperienced traders default on the payment of minimal entry

requirements for traders (200 Birr license) that was envisaged to promote competition

among coffee buyers at the auction. Others have disrupted the auction by working for

coffee sellers to bid-up the auction prices, thus raising the price to exporters. This will

widen the trade margin (as bid-up prices are not transmitted back to producers) and

makes Ethiopian coffee less competition with world market.

16
Hence in order to reap the benefits of this single export commodity that the country heavily

relies on, the institutional problems mentioned above need to be addressed and government

regulations and polices should be revised.

Following coffee, oilseeds and pulses together rank second in their share in the total

merchandise export during the period 1970/71-2000/01.During this period, their share in the

total export proceeds was about 13.1percent. Though their share was the highest during the

pre-revolution Period (which was about 25.4 percent), it declined significantly to 6.6 percent.

Although affected by declining would market prices, oilseed exports surged in recent years.

In 1999 Ethiopia ranked 25th among the top oilseed exporters, up from 84th place in1990

(Ibid.). The most rapid growth has been sesame seeds, where production has increased five-

fold during the last five years, from practically zero during the early 1990s. Export earnings

from pulses have declined in the late 1990s due to the down turn in international price.

Despite this, export prospects are reasonably good and the country can benefit a lot by

increasing the production of the commodity in response to price.

During the period 1970/71-2000/01, hides and skins accounted, on average, for about 11.5

percent of the total export proceeds. Its share during the three regimes has been somewhat

constant ranging between 10-13 percent. Ethiopia has traditionally been a major exporter of

hides and skins because of its large livestock population. However, in recent years it has lost

market shares in both hides and skins and finished leather products. The country ranked 70th

among the top leather and leather manufactures exporters, down from 49th and 48th place in

1980 and 1990, respectively (Ibid.).

17
One of the major reasons for the decline in its share is the outbreak of parasitic disease that

has reduced the value of hides and skins by causing blemishes in the finished leather. In an

effort to derive the benefits of processing, the government banned exports of raw hides and

skins. The ban has benefited the tanneries, which are now the sole suppliers of semi-

processed hides and skins to the international market. But live stock producers receive lower

prices for their hides and skins because the tanneries are the only buyers. Domestically, the

tanneries are also protected by high import tariffs. The rapid expansions of tanneries since the

export ban likely reflects the opportunity to profit in protected market.

The export commodity whose share increased significantly during the last four decades is

Chat. Its share in the total export earning was 0.8,2.2 and 9.4 percent during the imperial,

Derge and the present government, respectively. Recorded exporters of Chat increased

significantly in 1999 following the liberalization of export to Somalia. Chat is also legally

exported to Djibouti and the U.K. In addition, substantial unrecorded illegal trade occurs,

which is estimated to be equal in size to the recorded exports. Chat exports are expected to

grow even without government interventional because of strong demand in neighboring

countries. While such exports are not encouraged, because of the negative health effects, they

are likely to occur (Ibid.). The above export commodities including gold and petroleum

products, on average, account for more than 88 percent of the country’s export earning. The

rest 12 percent is distributed among fruits and vegetables, live animals, sugar and molasses,

meat and meat products, cotton spices, textiles, natural gum and others.

18
Furthermore, commodity concentration index3 (Hirschman Concentration index) for the first

five products (coffee, hides and skins, chat, oilseeds and pulses, gold and petroleum

products) during the period 1970/71-2000/01 averaged 0.56. This indicates that the export

sector is concentrated on few products. Only two export commodities have been added to the

list of the country’s export items since 1970/71(natural gum in1980/81 and gold in1985/86).

This exposes the export sector to undesirable consequences in the event of a fall in price of

the commodities in question or recession in the major markets.

As any Sub-Sahara African (SSA) countries, the sectoral structure of Ethiopia's exports is

classified as very much concentrated on primary agricultural products. Despite the fact that

there is an associated higher degree of risks and uncertainties with the export of primary

agriculture products, the country heavily relies on these products on average for about 80%

of its total export earnings.

As shown on Table 2.4, the Ethiopian export is sectorally concentrated on agriculture. For the

period 1991/92-2000/01, raw agricultural products export on average accounted for about

78.8 percent of the total merchandise exports. The rest 11.2, 8.6, and 1.4 percent on average

came from manufacturing, mining and unspecified sectors, respectively. This shows that the

Ethiopian export sub-sector in particular and the whole economy in general is agriculture

dominated which in turn is rain fall dependent and back ward in technology utilization.

_______________________
3
Hirschman Concentration Index ( H j = [∑ (X ij ]
/ X j ) where X ij is the value of a particular
2

commodity and X j total export revenue. The indices range from 0 to 1, higher value reflecting export
concentration.

19
Table 2.3b Sectoral Structure of Exports (% share in total)
SECTOR 1991/92 1992/93 1993/94 1994/95 1995/96 1996/97 1997/98 1998/99 1999/00 2000/01
Raw
Agricultural 54.7 69.0 71.5 81.5 81.0 71.9 90.9 87.6 72.3 83.6
Product
Exports
Industrially
Processed 20.8 11.4 8.0 11.2 10.1 12.2 7.4 9.5 20.6 10.6
Exports
Mining
Product 17.3 18.9 19.6 7.0 4.9 12.4 0.2 2.2 6.6 5.5
Exports

Sector
Unspecified 7.2 0.7 0.9 0.3 4.0 3.5 1.5 0.8 0.5 0.3
Export

Source: MEDaC

From the commodity and Sectoral structures of the country’s exports we see that coffee, raw

hides and skins, chat and pulses & oil seeds dominate the raw agricultural product export

sector. During the period 1991/92-2000/01,the share of the sector from the total merchandise

exports went as high as 90.9 percent in 1997/98. The lowest share recorded was in 1991/92

which was about 54.7percent.

The industrial export sector is dominated by processed hides and skin, food and textile

products. Manufactured exports are negligible, and have declined from around 21 percent of

the merchandise exports in1991/92 to less than 8 percent in 1997/98(although some

improvement was seen in 1999/00). Manufactured exports, which are used as a driver for a

rapid structural transformation of production is very weak in Ethiopia. The country is the

least industrialized country in the world and ranks lowest in terms of manufacturing value
20
added (MVA) per capita in 1998.Ethiopia’s relative position in terms of manufactured

products and exports is reported below.

Table 2.3c Relative Performance of Ethiopia's manufacturing Sector (1998)

Country MVA per Manuf. Total Share of complex* Share of complex*
Capita ($) Exports Manufac. products in MVA products in MVA
per Capita Exports ($m.) (Percent) (Percent)
($)
Ethiopia 7.9 0.8 49 9.0 0.1
Kenya 36.6 28.3 829 24.0 7.6
Tanzania 15.8 2.9 93 25.0 1.5
Uganda 24.3 0.9 19 15.0 0.8
Egypt 326.1 36.5 2,242 39.0 8.8
Mauritius 738.9 1,433.7 1,602 12.0 1.4
China 287.0 135.4 167,681 51.0 36.6
Source: Developing Exports to Promote Growth World Bank (2001)

*
The share of high and medium technology products in MVA or manufactured exports
measures technological sophistication. The remainder is made up of resource based and low
technology products.

Table 2.3c illustrates the small size of the manufacturing sector and the minimal

technological sophistication of the export sector, suggesting that the economy has to struggle

harder to build the necessary capabilities-operational, technical, managerial-to operate in

export markets[World Bank, 2001].

The performance of the mining sector (dominated by gold and petroleum products) declined

in recent years from that of the early 1990s. The highest share was recorded in 1993/94,

21
which was about 19.6 percent of the total merchandise exports, the lowest being 0.2 percent

in 1997/98.

Generally much is expected from the country to shift from predominantly primary

agricultural exports to semi-processed and processed exports if structural transformation and

economic growth is to be realized.

2.4 Export Promotion Efforts in Ethiopia

2.4.1 Pre 1991/92

The policy adopted in the pre-1991/92 period (both in the Imperial and Military government

of Ethiopia) was characterized by strongly inward-oriented development strategy, which used

a prolonged over valuation of the Birr, high tariff rates, extensive foreign exchange control

and other non tariff barriers as well as heavy taxation on exports. These polices are likely to

have a detrimental impact on export by influencing directly or indirectly the profitability and

competitiveness of exports.

Even though both previous government of Ethiopia were commonly pursuing import

substitution strategy and export sector was secondary for them in their economic

development plans, it doesn’t mean that they didn’t make any effort to promote and diversify

the country’s exports. They made effort to promote exports and diversify the entire export

commodities as shown in the three different five-year development plans of the Imperial

Government of Ethiopia (IGE) and in the Derge’s ten year perspective plan. The first five-

year development plan (1957/58-1962/63) gave priority to import-substitution industrial

22
promotion and infrastructural facilities like road development while it gave minor attention

for export promotion.

Regarding exports the plan stated the need for increasing exports by making full use of the

potential of agriculture and increased imports to be financed as much as possible by

increased exports. The measures under taken to achieve the plan were relaxation of the then

existing foreign exchange regulation, adjustments in export duties and freight rate and

improvement in export goods quality and sanitary conditions. But in practice no adequate

attention was provided to export development. As a result, at the end of the plan period

expanded deficit in merchandise trade balance was observed. To correct this imbalance the

second-five year plan was set in place with adequate attention paid for export development

through diversification.

The second five-year development plan stated the export sector to rely mainly on traditional

export products such as coffee, hides and skins, oil seed and pulses and others. It also

stipulated an important role to be played by new export products of industrial origins and

mining products. In this process, exports were expected to attain a larger degree of

diversification in which semi-manufactured and manufactured industrial goods would play an

increasing role. Such a move would facilitate the expansion of the country’s export through

strengthening their world markets acceptance and competitiveness and greatly contribute to

the improvement of the already affected terms of trade and balance of payment status of the

nation. This plan also set the share of agricultural exports to exhibit a decrease from 93.6

percent in 1962/63 of the total export to 72.3percent in 1967/68 while the share of

manufactured products was planed to increase from 5.2 to 24.2 percent during the same year.
23
To implement this plan, incentives like profit/income tax holidays, export trade licensing

simplification, restructuring and strengthening of chamber of commerce, establishment of

trade attaches in Ethiopian Embassies and missions all over the world, and provision of

market study trainings were offered for investors who engage themselves in the production of

non-traditional export items.

The third-five year development plan (1969/70-1973/74) gave a great deal of attention for

foreign trade in general and for the export sub-sector development through diversifying

variety of export items in particular. In this plan period, agricultural product exports were

expected to decrease to75 percent in1973/74 from that of 86 percent in1967/68. Through the

addition of new agricultural products in the export basket, the share of coffee was envisaged

to fall from 55 to 40 percent at the end of the plan period.

In order to increase the exports of manufactured item, special emphasis was given for the

strengthening of hides and skins processing. The plan also stipulated diversification of

exports mineral products such as potash, gold and others. It was believed that such a move

would help the country realize balance of payment improvement. During the plan period,

export was envisaged to double with reduced cost of production and improved quality. In

addition the plan stipulated a three-fold increase in non-agricultural exports such as textiles,

wood products, building materials, non-metallic products and chemical industry products. To

implement the plan, the then existing system of duty draw back on direct raw materials and

other components of export product was revised. In addition, both fiscal and monetary

incentives were offered for both domestic and foreign investors engaged in export-goods

production.
24
In sum, although attempts were made in all the three development plans of the Imperial

Government of Ethiopia, this didn't bring the anticipated export promotion and

diversification.

The military government who came to power in 1974/75 under took a ten-year perspective

plan of 1985/86-1994/95. The main objective of the plan was to orient the country’s export

structure towards manufactured products from the already existing primary exports of

agricultural product, to expand substantially the country’s foreign exchange earnings through

exporting diversified industrial, mining and agricultural products and to diversify export

markets and reduce over dependence on traditional ones.

In order to achieve its objectives, the military government employed a multitude of strategies.

These include promotion of exports through the provision of favorable tax, tariffs and foreign

exchange rate measures, improving exports in terms of quality, quantity and variety and

providing current information on World market prices and other factors in the international

market to exporters and producers. In order to publicize and expand the market for the

country’s export products, the government took part in international trade fairs and

encouraged the export of manufactured products and strengthened chambers of commerce

and other institutions which are directly engaged in promoting export trade.

The ten-year perspective plan stipulated the share of traditional exports (coffee, hides &

skins, and oil seeds and pulses) to decrease from 73.5 percent in 1985/86 to 53.2 percent in

1994/95,while the share of other export products to rise from 26.5 to 46.8 percent in the plan

25
period. In addition, to counter balance the negative effect of distortionary polices and hence

to secure growth in export, the government introduced an export subsidy in 1983/84.

The fund for the subsidy was raised by imposing a 5 percent ad-valorem tax on import.

Furthermore, the export sector was given a preferential interest rate of 6 percent on bank

loans against 8 percent for importing activities. However, the subsidy introduced was not

sufficient in terms of coverage and amount to neutralize the anti-export bias incentive

structure.

To summarize, despite the measures taken by both the Imperial and the Derge regimes to

diversify the export basket and promote exports, the Ethiopian export products remain

undiversified and are still concentrated on very few primary products like coffee, hides &

skins oil seeds & pulses and chat. This is because both regimes used overvalued exchange

rate, high rate of tariffs and other trade restrictive commercial policies that developed strong

anti-export bias, and strongly in-ward oriented trade policies favoring import substitution

than export promotion. Although export promotion incentives like export subsidy and others

were provided, these have neither resulted in the export diversification nor in the expansion

of the existing export volume. This was so because the incentive provided were not enough

to counter-balance the anti-export-bias caused by currency overvaluation, high duties

(tariffs), taxes and others.

2.4.2 Post 1991/92

In 1991 the transitional government of Ethiopia (TGE) together with the IMF and the World

Bank has undertaken liberalization and structural adjustment program to address the internal
26
and external imbalances of the economy. In particular trade policy reform was undertaken

which aimed at promoting exports through diversifying the country’s commodity exports.

Among the measures undertaken the following were important ones.

a) Devaluation of the Ethiopian currency by more than 140 percent in terms of US dollar to

make exports competitive and promote export trade. In addition a weekly auction of

foreign exchange was introduced and to guarantee that the incentive pass to the peasants,

the government set a floor price for coffee, haricot bean and sesame seed.

b) The tariff regime was continuously revised and was reduced on a stage basis from a

maximum of 230 percent to50 percent. Similarly, to nullify the anti-export bias, the state

lifted a 2 percent transaction tax on non-coffee exports and abandoned the direct financial

subsidy on export.

c) The import and export licensing system were simplified and become more transparent so

as to encourage new entrants in the export market. The range of goods and services

covered by the auction has been progressively extended and finally fully liberalized.

d) A duty draw back scheme was introduced where by exporters are re-funded the tax and

duty they paid on the inputs and raw materials used in export production. This is to

provide exporters a free trade status on their import of intermediate inputs and encourage

non-traditional export products, especially that of manufactured goods. But the

effectiveness of the scheme on export is constrained by lengthy administrative

requirement to get re-funded.

27
e) A foreign exchange retention scheme has been introduced which entitles exporters to

retain 10 percent of their earning to hold in their account and to sell the 40 percent at a

competitive rate, while submitting the remaining 50 percent directly to the National

Bank. But the scheme may not be beneficial in view of the usual control over the use of

the retained 10 percent and for the fact that it ties up the working capital.

f) A preferential interest rate scheme is also introduced for exporters, which is less by 3.5

percent compared to the interest rate paid on non-export activity loans. Such low

preferential interest rate scheme is provided for exporters because it is believed to

strengthen the country’s export diversification efforts.

g) State exporting enterprises were provided a managerial autonomy but deprived of a

monopoly power. This creates a conducive environment for private exporters and puts

them at equal footing with public enterprises.

h) The Ethiopian Export Promotion Agency is established very recently as an autonomous

body by proclamation No.132/1998. The main objective of the agency is to promote the

country’s exports. By doing so it is believed to achieve export diversification in

agricultural, industrial and mining sectors of the nation’s economy.

As a result of these trade policy reforms a remarkable decrease in the anti-export-bias

incentive structure and an increase in export volume and earning was realized.

28
Table 2.4.2 shows the effective exchange rate for exportable (EER*x)4, the effective exchange

rate for importables (EERm) and the anti-export-bias incentive structure (EERm/EERx) for

the pre and post 1991/92 periods.

Table 2.4.2a Effective exchange rate (Birr/Dollar) and anti-export-bias (EERm/EERx).

Item 1985/86-1991/92 1992/93-1998/99 Percentage Change
Average Average (%)

Effective Exchange 1.89 5.82 208% increase

rate for exports (EERx)

Effective Exchange 3.74 8.39 124% increase

rate for imports
(EERm)
Bias Against Exports 1.98 1.44 27% decrease
(EERm/EERx).

Source: MEDaC.

As can be seen from the Table, the reform measures raised the EERx by 208% compared to

the pre-reform period. This indicates that the reform has created an enabling environment for

the export sector. In addition, since the reform, the anti-export bias incentive structure

(EERm/EERx) has shown enormous improvement (a decline by 27% from the pre-reform

period). However, there is still an anti-export bias incentive structure, which implies a need

to take additional measures to fully achieve neutral incentive system that is conducive for

effective export promotion.

_______________________________
4
For definition and calculation of EERx, EERm and EERm/EERx see Annex 1.
29
3. REVIEW OF LITERATURE

3.1 Theoretical Literature

The idea that international trade brings economic growth and increases the welfare of a

nation started during the 17th century by a group of merchants, government officials and

philosophers who advocated an economic philosophy known as mercantilism. According to

mercantilists, for a nation to become rich and powerful, it has to export more than it imports

where the resulting export surplus is used to purchase precious metals like gold and silver.

Thus the government in its power has to control imports and stimulate the nation’s exports.

Adam smith attacked the main mercantilists' views and proposed the classical theory of

international trade based on the concept of absolute advantage model. According to him

stock of human, man-made and natural resources rather than stock of precious metals were

the true measure of the wealth of a nation and argued that the wealth of a nation can be

expanded if the government would abandon mercantilist controls. In addition, he showed that

trade can make a nation better off with making another worse off [see Mannur, 1996 p.21].

Absolute advantage, however, explains only a very small part of the world trade today i.e.

trade between developed and developing countries. Most of the world trade especially trade

among developed countries could not be explained by absolute advantage [Salvatore,1990].

The model of comparative advantage was later articulated by David Ricardo to replace the

principle of absolute advantage. According to this model, a country will specialize in the

30
production and export of the commodity in which it has a comparative advantage i.e. the

commodity that it can produce at the lowest relative cost.

The comparative advantage model is based on a set of assumptions one of which is the labor

theory of value. According to the labor theory of value, (a) either labor is the only factor of

production or is used in the same fixed proportion (b) labor is homogeneous i.e. of only one

type. Since neither of these assumptions is true the labor theory of value must be rejected. In

addition to the above argument, the comparative advantage model states that trade depends

on the terms of trade which in turn is determined by internal cost ratios in two trading

countries i.e. by supply conditions alone. This obviously is flawed since terms of trade are

not only determined by supply factors but also by demand forces. In order to modify the

Ricardian theory, the principle of reciprocal demand was formulated by J.S Mill and later

was developed by Edgeworth and Marshall. According to the reciprocal demand theory, it is

both the demand and supply conditions which determine the terms of trade and hence trade

between countries. However, the theory says nothing about the gains to be obtained through

trade, it merely fills some gaps which existed in the early classical theory.

As an attempt to modify the classical theory of trade, the factor endowment theory of Eli

Hecksher and Berti Ohlin (H-O), of external trade evolved. According to this theory,

different products require productive factors in different relative proportions and Countries

have different endowments of factors of production. Some countries have large amounts of

capital (capital abundant) while others have little capital and much labor (labor abundant).

31
This theory argues that each country has a comparative advantage in that commodity which

uses the country’s abundant factor. Thus capital abundant countries should specialize in the

production and export of capital-intensive goods while labor abundant countries should

specialize in the production and export of labor-intensive commodities.

This theory, which played a predominant role in the early literature of trade theory,

encouraged third world countries to focus on their labor and land intensive primary product

exports. It was argued that by exchanging these primary products for manufactured goods of

the developed countries, third world nations could realize enormous benefits obtained from

trade with the richer nations.

Although the factor endowment theory contributed a lot to the modern theory of international

trade, the validity of the theory is based on a set of assumptions that are unlikely to hold.

Specifically, six basic assumptions of the neo-classical trade model are criticized in

explaining trade between the developed and the developing countries5.

In recent years economists have therefore challenged the static neo-classical model and

developed new models that explain trade between developed and developing countries.

Unlike the traditional model which is assumed to apply to all nations, the so called North-

South trade models focus specifically on trade relation between rich and poor countries [see

Ocampo, 1980].

_________________________
5
For the critics of the basic assumptions in explaining trade between the developed and the developing
countries see Todaro (1994) p. 454
32
Other theories of trade have also been put forward which attempt to either supplement the

neo- classical trade theory or replace it with different approaches. These include the vent for

surplus theory of international trade [Myint, 1958], preference similarity or overlapping

demand theory developed by Linder (1961), the technological gap and the product cycle

theory articulated by Posner (1961) and Vernon (1966), respectively 6.

These theories that are referred as complementary (alternative) theories do not suggest that

the neo-classical trade model should be discarded. They are not comprehensive and try to fill

a portion of the gap in the international trade that the traditional classical theory couldn’t

explain.

The notion of trade as an “engine of growth” is given much emphasis by many economists.

Proponents of the traditional theory of trade still contend that trade can contribute

substantially to the development of primary-exporting countries. It is argued that the growth

of many developed countries like the United States, Canada, Australia, New Zealand

(referred as regions of recent settlement), which were once developing nations is mainly

attributed to international trade. However other economists strongly believe that the accrual

of the gains from international trade is biased in favor of the advanced industrial countries

and that foreign trade has inhibited industrial development in poor nations. Thus these

economists contend that international trade as being completely irrelevant for developing

nations and the development process.

_____________________
6
For a good discussion of these trade theories and others see [Sodersten (1994) and Salvatore (1990)].

33
The controversy on the notion of trade as an engine of growth led developing countries to

pursue different trade strategies for development. In what follows we present the two trade

policies adopted by many developing countries namely, import substitution and export

promotion and the discussion follows Todaro (1994).

3.2 Trade Strategies for Development: Export Promotion Versus

Import Substitution
In a broader sense the trade policies that have been undertaken by developing countries for

the past four decades can be categorized as outward-looking and inward-looking

development policies. Proponents of the view that trade brings development encourage

outward looking development policies. According to Streeten (1973), quoted in Todaro

(1994), the outward-looking development policies “ encourage not only free trade but also

free movement of capital, workers enterprises and students,…, the multinational enterprises,

and open system of communications” (p. 484). In contrast, opponents of the traditional view

advocate an inward-looking development policy. This policy stresses the need for LDCs to

implement their own styles of development and adopt indigenous technologies appropriate to

their resource endowments. Thus by restricting trade, the movement of people and

communication, multinational enterprises and their wrong technology, greater self-reliance

can be realized.

Within these broad trade strategies, we have protectionists who advocate inward-looking

import substitution strategies and free traders that are proponents of outward looking export

promotion strategies. According to advocates of import substitution (IS), substitution of

imported items by domestic production is supposed to take place in two stages. The first

34
stage involves substitution of previously imported simple consumer goods while the second

involves substituting wider range of more sophisticated manufactured items by domestic

production using high tariffs and quotas on these imports.

According to Salvtore (1990), three advantages of the IS strategy can be cited:

“1) The market for the industrial product already exists, as evidenced by import of the

commodity, so that risks are reduced in setting up an industry to replace imports. 2) It

is easier for developing nations to protect their domestic market against foreign

competition than to force the developed nations to lower trade barriers against their

manufactured exports. 3) Foreign firms are induced to establish the so-called “ tariff

factories” to overcome the tariff wall of developing nations” (p. 327).

Furthermore, the IS strategy enables a country to have greater domestic industrial

diversification and provides the ability to export previously protected manufactured goods as

economies of scale, low labor costs, and learning by doing cause domestic prices to become

more competitive with world prices.

During the 1950s and 1960s, the first export pessimism which characterized the thinking of

most influential development economists Raul Prebish (1952) and Ragnar Nurkse (1959), led

to the adoption of the IS trade strategy by many developing countries. According to Prebish,

the terms of trade for primary product exports are deteriorating and hence the main exports of

LDCs are declining regardless of the policies of developing countries. Nurkse’s export

pessimism arose from the view that markets of developed countries could not accommodate

imports on a sufficient scale as developing countries accelerated their development [see

35
Bhagwati, (1988)]. This old brand of export pessimism linked external conditions with

internal expansion. Specifically, Todaro (1994) argues that at least five external (demand

side) factors inhibit the growth of LDC's primary exports.

“ First, the per capita income elasticities of demand for agricultural food stuffs

and raw materials are relatively low compared with those for fuels, certain

minerals and manufactures. …Second, developed-country population growth

rates are at the replacement level, so little expansion can be expected from this

source. … Third, the price elasticity of demand for most non-fuel primary

commodities appears to be relatively low. … The fourth and the fifth factors

working against the long-run expansion of LDC primary-product export

earnings- the development of synthetic substitutes and the growth of

agricultural protection in the developed countries are perhaps the most

important (p. 487-88).

In addition it is argued that markets would not be able to absorb all the exports if developing

countries shifted to an EP strategy at the same time. However, the success of some

developing countries especially the four Far Eastern economies has refuted the validity of the

first export pessimism. According to Riedel (1984), unlike the view of export pessimists, the

export performance of these and other countries is explained by domestic incentives (supply)

more than by external (demand) conditions.

In sum, although the first export pessimism was unjustified, it provided the rational for the

adoption of the IS strategy in many developed countries. This strategy attempts to replace
36
commodities that are being imported, usually manufactured consumer goods by domestically

produced items. Hence by erecting tariff barriers or quota on certain important commodities,

it is argued that domestic industries will be able to reap the benefits of large-scale production

and lower costs (infant industry argument) or that the balance of payment will be improved

as fewer consumer goods are imported. However the growth performance of many

developing countries has proved the IS strategy to be largely unsuccessful. According to

Todaro (1994) the strategy led to five undesirable outcomes.

First, many IS industries remained inefficient and costly to operate as they become

accustomed to protection from foreign competition through tariff walls. Second, the main

benefits of the IS process is accrued to foreign firms that were operating behind tariff walls

and take advantage of free tax and investment incentives. The local industrialists, however,

benefited very little from what is left over by foreign firms after remitting the gains abroad.

Third, the IS strategy led to a setup of capital intensive industries with little labor absorption

and thus aggravated the unemployment problem of LDCs. In addition the increasing need for

imported capital-good inputs and intermediate products further intensified the balance of

payment deficits and the debt problem. Fourth, to lower the domestic currency price of their

imports and hence encourage local manufacturing, many LDCs overvalued their exchange

rate. This penalized the primary product export sector by raising the price of exports in terms

of foreign currencies. As a result many LDCs experienced a decline in their earning from

traditional exports. Fifth, in contrast to what has been conceived the IS strategy inhibited

industrialization. This is because most infant industries didn’t grow behind protective tariffs.

37
The overall result was that those developing nations that tried industrialization through

import substitution grew at much slower rate than the few developing countries that followed

an export-oriented policy. As Kruger (1985, p. 22) stated: “… development strategy that

relies on integration with the world economy, rather than insulation from it, is not only

feasible, but preferable”.

As a result many developing nations began to pay more attention to export-oriented policy.

The benefits or returns of this strategy are thought to be both numerous and widespread

[Bhagwati, 1988 and Kruger, 1985]. It is argued that trade according to the principle of

comparative advantage yields efficiency in terms of resource allocation. However the gains

from efficient resource allocation is realized only when governments remove biases against

exports. Another gain from adopting the EP strategy relates to the economies of scale issue.

Advocates of this strategy argue that domestic markets are too small to allow firms to achieve

optimal scale. It is through production for sale to foreign markets that firms can achieve

increasing returns and, eventually, optimal scale [Grabowski, 1994].

Another benefits of the EP strategy relates to its dynamic effects. According to Bhagwati

(1988), the EP strategy may lead to more competition and less-sheltered markets and hence

more innovations which bring a positive externality to the rest of the economy. According to

Grabowski (1994), the EP strategy removes the distorting impact of government policies in

their attempt to guarantee the availability of the domestic market for domestic producers.

Thus activities by the state that are likely to distort the market are restricted in the EP strategy

of development.

38
Another possible dynamic impact of an EP trade strategy is the ability to import foreign

technology. With the growth of exports new technologies that are important for development

can be imported. Thus, the overall rate of growth is likely to be increased as the rate of

technical innovation increases. For these reasons, advocators of an EP strategy proposes a

move toward outward oriented policies as best strategies for promoting growth. In this

respect, many studies have been made to validate whether an EP trade strategy is a superior

policy or not.

Most of these studies confirmed that countries that adopted export led trade strategy

performed better in terms of growth than those that adopted the IS trade strategy. This led

many developing countries to look towards the EP strategy. However, the desire to shift to

the EP strategy in developing countries is undermined by the second brand of export

pessimism. According to Bhagwati (1988) there are two sets of factors generating this

pessimism:

“a) Objective events such as the slowing down of the world economy since the

1970s and the resurgence of powerful protectionist sentiments in the industrial

countries, and b) New intellectual and academic arguments in support of

inward looking trade policies in the developing countries” (p.41).

This second export pessimism rests on the notion that regardless of the absorptive capacity

for the exports of LDCs, protectionist measures in the industrial countries will inhibit exports

of developing countries and makes the pursuit of the EP strategy inefficient. However, the

protectionist threat was not serious to force developing countries turn away from embracing
39
the EP strategy. This is evidenced by the fact that the exports from many developed countries

continued to grow in spite of protectionist measures. As Kruger (1980) has stated [quoted in

Meier, 1985]:

“ If such protectionist measures are taken, they will lower the rate of return
to outward oriented trade strategies. They will however, for the foreseeable
future, still leave the rate distinctly above the returns from a policy of
persisting with inward-oriented growth” (p. 482).

The IS and EP strategies have often been viewed as opposites or completely separate

theoretical categories. According to Grabowski (1994), however, these strategies are not

opposite, but simply alternative ways for stimulating growth in the size of markets for

manufactured goods.

In both strategies the main concern is the availability of sufficient market to enable

firms realize an optimal scale of production. The IS strategy through tariff and non-

tariff barriers creates a domestic demand for domestic industries while the export

market permits industries to fully utilize the optimal scale of production. Grabowski

(1994) goes on to argue that:

“… import substitution is generally necessary for outward growth to

succeed. Most important, the success of import substitution is linked to the

development of productivity in the agricultural sector. If the later fails to

grow rapidly, import substitution will fail and successful export-based

growth will not occur” (p. 536).

40
This was justified by the success of countries like Taiwan and South Korea who successfully

completed the transition to outward-oriented development via import substitution.

So far we have dealt the two trade strategies IS and EP, the superiority of the EP strategy

over the IS based on the experience of many LDCs and an alternative strategy which views

IS to precede the EP strategy. Since our aim is to test whether export growth leads to

economic growth, in what follows we examine the channels through which exports may

affect economic growth.

3.3 Export and Economic Growth

The relationship between export performance and economic growth is an area that has been

given much attention by development economists. The results of different studies on export

expansion and economic growth has broadly classified economists into those that support the

hypothesis that export growth has a positive impact on economic growth and those that doubt

the existence of such relationship. The central question to be addressed in this section is

“how does export growth influence economic growth?”

Adam Smith’s theory of international trade assumes that a previously isolated country about

to enter into international possesses a surplus productive capacity above the requirements of

domestic consumption. With trade the country is able to reallocate the given resources as to

provide the new effective demand for the output of the surplus resources. Hence, a surplus

productive capacity suitable for the export market appears as a costless means of acquiring

imports and expanding domestic economic activity [Meier, 1995 and Myint, 1958].

41
One of the export-based models formulated to present a dynamic view of how an economy’s

growth can be enhanced by expansion of its exports is the staple theory of growth7.

According to this theory, the discovery of a primary commodity in which a country has a

comparative advantage or an increase in demand for its comparative advantage commodity

leads to an expansion of resource-based export commodity which in turn induces a higher

growth of aggregate and per capita income. The export of the primary product also has

effects on the rest of the economy through reducing unemployment and underemployment,

inducing a higher rate of domestic saving and investment, attracting an inflow of factor

inputs into the expanding export sector, and establishing links with other sectors of the

economy [Meier 1995, p.460].

Feder (1982) views a given economy as if it consists of two distinct sectors: an export and

non-export sector. According to him the marginal factor productivities are significantly

higher in the former than the latter. This arises from inter-sectoral beneficial externalities

(capacity utilization, economics of scale incentives provided for technological improvement

and efficient management due to competitive pressures from abroad) generated by the export

sector. Thus, growth can be generated by reallocation of the existing resources from the less-

efficient non-export sector to the higher productivity export sector.

According to Kavoussi (1984) and Moschos (1987), export expansion raises factor

productivity and leads to various benefits, such as more efficient use of resources and

__________________________________
7
The term staple according to Meier (1995) designates a raw material or resource- intensive commodity
occupying a dominant position in the country’s exports.
42
adoption of technological innovations, resulting from foreign competition, greater capacity

utilization and gains of scale effects associated with large international markets.

Jung and Marshal (1985) argue that growth in real exports tends to cause growth in real GNP

for three reasons. First, export growth may represent an increase in the demand for the

country’s output and thus serve to increase real GNP. Second, an increase in exports may

loosen a binding foreign exchange constraint and allow increases in productive intermediate

imports and hence result in the growth of output. Third, export growth may result in

enhanced efficiency and thus may lead to greater output.

Chow (1987) suggests that in small open economies, export growth can expand their limited

domestic markets, and contribute to the economics of scale necessary for industrial

developments. Furthermore, export growth integrates domestic economy with regional and/or

global economies thereby expanding the dimension of competition to the international

markets. Competition promotes resource allocation in developing countries as they transform

from less productive farming sector to relatively more productive manufacturing sector.

Therefore, factor productivities are improved through export growth.

According to the “balanced growth” doctrine, there is a vicious circle present, which acts as a

stumbling block in attaining self-sustaining growth. Rosenstein-Rodan (1943) and others

quoted in [Krishna, 1998 et.al] argue that:

43
 “Firms did not industrialize because there was no market for their goods and

there was no market for their goods because income was low and income was

low because firms did not industrialize. This kind of low level equilibrium, it

was argued, could be broken by the simultaneous industrialization of large part

of the economy, and any failure to industrialize was essentially viewed as a

coordination problem. Of course, exports, by breaking this circle of causation,

could provide an important avenue for growth” [p. 1].

On the other hand the “unbalanced growth” doctrine led by Albert O. Hirshman (1958),

while agreeing on the existence of a vicious circle, argue that industrialization of certain

“leading” sectors would pull along the rest of the economy. Hence instead of industrialization

of a large number of sectors, what was needed was the industrialization of the “leading”

sectors. Then through backward and forward linkages these sectors would initiate the

industrialization of the rest of the economy. Exports, especially in such leading sectors, could

start the industrialization process.

Esfahani (1991) emphasized that the first and foremost purpose of exports is to relieve the

import shortage that many developing countries confront. According to him although the

externality effect of exports (efficiency of resource allocation, economy of scale and various

labor training effects) may carry some weights of their own, the major purpose of exports to

GDP growth is alleviating the import shortages, which restrict the growth of many LDCs.

Thus exports can fill the “foreign exchange gap” that was perceived as obstruction to growth.

44
According to [Krishna, 1998 et.al] exporting firms, especially multinationals could provide

externalities by serving as conduits for the dissemination of world class technology to less

dynamic domestically oriented firms. Because international markets for technology and

knowledge are imperfect, exporting helps to overcome some imperfections and permits

access to international best-practice technology through other mechanisms.

In general all the above theories explain the different channels through which exports can

induce economic growth. The empirical studies, which substantiate the above arguments are

presented below.

3.4 Empirical Literature

The contribution of export growth to economic growth has been tested by different

economists using different econometric techniques. All the tests that have been carried out

are broadly classified as those that are based on cross-country analysis and those that are

based on country specific time series studies.

3.4.1 Cross-sectional Studies

Using the data for 41 less developed countries for the period 1950-73, Michael (1977) tested

the hypothesis that “export growth accelerates economic growth” using spearman rank

correlation coefficient. He used the rate of growth of exports share of GDP and output

growth to avoid spurious results stemming from the fact that exports are a component of

GDP. The spearman rank correlation coefficient for the sample as a whole was found to be

0.38, which was significant at 1% level.

45
However the positive association of the economy’s growth with the growth of export share is

particularly strong among the more developed countries and does not exist at all among the

less developed countries. Thus, he concluded that growth in output is affected by export

performance once countries achieve some minimum level of development.

Similarly Balassa (1978) tested the hypothesis using the rank correlation methodology. He

used pooled data on eleven developing countries for 1960-73. Again he found a positive

correlation coefficient between growth of exports and output growth.

According to Edwards (1993) these results were criticized on three accounts:

“ First, by looking at correlation coefficient, the possible role of other factors on

growth was ignored. Second, no attempt was made to distinguish between

endogenous and exogenous variables. Third, these analyses were not based on

firm theoretical grounds (p. 1379-80).

In order to deal with the above criticisms, a number of authors attempted to carry out the test

by formulating a conceptual framework based on neo-classical production function.

For instance, Tyler (1981) took a sample of 55 middle-income developing countries for the

period 1960-77 to investigate the impact of exports on economic growth. In his study the low

income countries were eliminated because of the idea that a certain level of development is

required for exports to bring economic growth. The regression result demonstrated a positive

and significant association between export growth and economic growth.

46
Feder (1982) tested the export-growth relationship for a group of semi-industrialized less

developed countries for the period 1964-73. He developed an analytical framework,

incorporating the possibility that the marginal factor productivities are not equal in the export

and non-export sectors of the economy. The regression result indicated that the coefficient of

the export variable is about 0.42, which is significant at 5% level. This led to a strong support

to the hypothesis that marginal factor productivities in the export sector are higher than in the

non-export sector. Using the mean values of the sample variables he decomposed the average

rate of GDP growth for the period under consideration. The contribution of exports is further

decomposed into the gain in growth due to beneficial externalities affecting the non-export

sector and the gain due to other elements underlying higher factor productivity in the export

sector. Accordingly, the inter-sectoral externality effect of exports is found to be higher than

the effect of higher factor productivity in the export sector.

In order to examine the relationship between export expansion and economic growth,

Kavoussi (1984) took a sample of seventy-three low and middle-income developing countries

for the years 1960-78. From the regression result he found that in large and heterogeneous

sample of developing countries he considered, higher rate of economic growth is associated

with higher rates of export growth. He also demonstrated that the positive correlation

between exports and growth is not limited to middle-income countries (as argued by the

previous authors) but holds for low-income countries as well. Furthermore, his result has

shown that primary exports play an important role in the growth process of the low-income

countries. On the other hand his finding seem to indicate that exports of manufactured goods

to be strongly associated with economic growth for the middle income countries.

47
Balassa (1985) examined the export-growth relationship during the period of external shock

(1973-78) for a sample of 43 developing countries. It has been suggested that while export

orientation brought benefits during the period of rapid world economic growth, such would

not be the case once the world economic environment deteriorated because of external

shocks. So, having found a positive impact of exports on economic growth for 11 semi-

industrialized countries, for the pre-1973 period, he attempted to re-examine the existence of

such a relationship for the post-1973 period. Utilizing the same framework and extending the

scope of the investigation to countries at lower level of development, he found that the rate of

growth of exports significantly affected the rate of economic growth and the numerical

magnitude of this effect increased compared to the earlier period. His results further indicate

the possibilities for the low-income countries to accelerate their economic growth by

employing an outward oriented policy as well as the advantages of relying on manufactured

exports.

However, Balassa’s (1985) finding that the contribution of exports to growth has increased in

the post-1973 period as compared to the pre-1973 was criticized. According to Rana (1988),

although his ‘pre-post’ comparison may be theoretically right, the empirical results need a

close scrutiny. This is because of the heterogeneity of the samples considered. While the pre-

1973 sample comprised pooled data from 11 semi-industrialized countries, the post-1973

sample is much broader and comprises cross-sectional data from 43 low and middle-income

countries. Hence it is unrealistic to assume that the production function in the pre-1973 to be

homogeneous across such a wide range of countries and over different time periods, and so

the results of his ‘pre-post’ comparison could be misleading.

48
To overcome this problem, he re-examined the export-growth relationship for the pre-1973

and post-1973 periods by using a balanced sample of 43 countries in each period. He also

employed other estimation techniques to adjust for the problems caused by the pooling of

data by developing an error component model. In contrast to Balassa’s finding, the empirical

results indicated the coefficient of the export variable to decline in all cases. He concluded

that exports would have less effect on growth when the world environment is unfavorable.

Ram (1985) used a fairly large sample of 73 LDCs to investigate the contribution of exports

to economic growth for the periods 1960-70 and 1970-77, separately. Furthermore in order to

examine the widely held belief that exports are not important to economic growth in the low-

income LDCs, he considered low and middle-income countries for both periods. The results

of his study indicated export performance to be important to economic growth. In addition

his results suggested that the importance of exports to have increased during the 1970s.

While the impact of export on growth does seem small in the low-income LDCs over the

period 1960-70, the positive impact of exports on growth is large and almost equal in

magnitude for the two groups for the period 1970-77. Again using two plausible growth

models (the “Feder Type” and the standard Neo-Classical) for 88 LDCs and the period for

1960-82, Ram (1987), found a strong positive impact of exports on economic growth.

In order to test the hypothesis of the existence of a critical level of development, which

causes a break in the relationship between export expansion and economic growth, Moshos

(1989) employed a switching regression framework for a selected sample of LDCs. The

result obtained on the basis of cross-sectional data do not support the view that the positive

effect of export expansion on economic growth is limited to more advanced developing

49
economies and that its effect is non-existent among less advanced countries. On the contrary,

the evidence indicates that output growth is influenced positively by export expansion for

both more advanced and less advanced developing countries.

To test the export-growth relationship, Odedokun (1991) employed crossectional data for 85

countries at different levels of economic development and different regions of Sub-Saharan

Africa, Asia, Western Hemisphere and others. From the regression result, export is found to

be an engine of growth for all LDCs combined and the positive relationship between export

expansion and economic growth is stronger in high-income LDCs than in other LDCs.

According to Sheehey (1992), the results of a substantial number of previous studies that

showed the positive impact of exports on economic growth are biased by a built-in

correlation between exports and GDP. Thus, he tested the export-growth relationship for 53

non-oil developing countries for a period the period 1960-81 using an alternative export

variables not subject to bias. He found that rather than offering the widely available benefits

suggested in previous studies, exports could result in a lower rate of output growth during a

period of weak world demand. The positive effects of shifting more resources into exports

were confined to a limited number of more industrialized countries in the 1960s, a period of

strong growth for world trade.

3.4.2 Time Series Studies

All the above studies used cross sectional data to test the impact of exports on economic

growth. However Rati (1987) argues:

50
 “Estimates obtained from cross sectional data are useful in many ways, especially

when the number of observations for individual countries are small. However,

there is evidence of tremendous parametric variations across countries in regard to

estimates of the growth equations in such contexts. Imposition of a common

structure in the form of cross sectional models can be

a drastic simplification, and important parametric differences could be masked in

cross-sectional estimates even when the sample chosen look fairly homogeneous

with reference to certain prior criteria” (p. 52)

Therefore it is important to make an assessment of the export-growth nexus for individual

countries on the basis of time series data. Accordingly, he made a beginning toward

considering the export-growth nexus by estimating two growth models for 88 LDCs

separately for the period 1960-82. The country specific results indicated that the predominant

export-growth connection is positive. In the entire sample of 88 LDCs, the coefficient of the

export variable is positive for more than 80% of the countries, and nearly one-half of these

positive coefficients are statistically significant at least at 10% level.

Other country specific studies were also conducted to test the export-growth relationship.

Begum and Shamsuddin (1998) have tested the relationship for Bangladesh for 1961-92.

They employed the “Feder type” production function in their analysis. Their main finding

was that the sum of the productivity differential and externality effects of the export sector is

positive implying that reallocation of resources from the non-export to export sector will

enhance the productive capacity of the economy. Therefore through this effect export growth

can induce output growth.

51
Dodaro (1993) employed a simple model developed by Fajana (1990) to test the relationship

between exports and economic growth for 43 LDCs for the period 1967-1986. Unlike most of

the previous studies that relied on cross-sectional analysis, the results obtained from the

regression analysis suggest a weak relationship between exports and economic growth.

Although a positive and significant relationship does emerge for slightly more than 50% of

the countries considered, for majority of them the coefficients are significant at the 10%

level.

All of the above studies both from cross sectional and time series have found a significant

positive impact of exports on economic growth. These studies have interpreted results in

regressions of output variables on export variables as providing support for an export

promotion development strategy. According to Jung and Marshal (1985), however, such

interpretation is questionable since these regressions provide no means of determining the

direction of causality. They criticize the usual approach that most of the studies followed

(regressing real growth on contemporaneous real export growth) and to infer support for the

proposition that export growth causes output growth from the significance of the export

growth coefficient. According to them such an approach contains a serious methodological

weakness. Although the hypothesis of export promotion clearly implies a correlation between

export and real GNP growth, an equally plausible hypothesis is that output growth causes

export growth. To substantiate their argument they perform causality test between export and

economic growth for 37 developing countries. The time series results for the countries

considered provided evidence in favor of export promotion in only four instances. Even

countries like Korea, Taiwan and Brazil whose tremendous growth was largely attributed to
52
export growth provided no statistical support for the export-led growth hypothesis. This

strongly suggests that the evidence in favor of export promotion is weaker than previous

studies have indicated.

Chow (1987) investigated the causal relationship between export growth and industrial

development during the 1960s and 1970s for eight newly industrialized countries (NIC)8. The

result of Sim’s causality test indicated that for most of the NICs there is a strong bi-

directional causality between the growth of exports and industrial development. He

concluded that depending on the size of the domestic market, export growth can cause

industrialization, either unidirectionally, or bidirectionally by influencing the development of

manufacturing industries.

Furthermore, Dodaro (1993) employed individual country time-series analysis to establish

the direction of causality between export growth and real output growth. The causality test

offers a very weak support for the contention that export growth promotes GDP growth.

Support for the alternate contention that GDP growth promotes export growth is also

weak.Hence the evidence is weak with respect to the alternate notion of trade as an “engine

of growth” and suggests the need to reconsider the whole relationship between exports and

economic growth within the context of LDCs.

According to Bahmani and Alse (1994), there are three major shortcomings associated with

all the time-series studies just stated above. First, none of these studies have checked for the

________________________
8 He used growth of manufacturing industries in LDCs as a proxy for industrial development.

53
cointegrating properties of the variables considered. The standard Granger and Sims tests are

valid only if the time series involved are not cointegrated. Second, most economic variables

like GDP and exports are non stationary which result in spurious regression. Finally, because

lack of quarterly data most of the previous studies used annual data. If the time delay

between cause and effect is small compared to the time interval over which data is collected,

however, the lack of causation could be the result of temporal aggregation.

Accordingly they used an alternative test for Granger causality, which is based on error-

correction models that incorporated information from the cointegrated properties of the

variables involved. Using this approach they performed causality test for 9 developing

countries based on quarterly data for the period 1973I-1988IV. The results indicated that in

contrast to the previous studies when the cointegrating properties of the time series are

incorporated into the analysis, bi-directional causality between export growth and output

growth receives strong empirical support in almost all countries.

According to Amoateng and Adu (1996), the export-driven economic growth hypotheses

have provided mixed results in a bivariate causality framework. The main shortcoming with

the bivariate causality analysis is the omission of other relevant variables, which could bias

the results. Hence, they introduced foreign debt service as a third variable within a trivariate

causality analysis of exports and economic growth for 35 African Countries for the period

1971-1990. They found that there is a joint feedback effect between export revenue, external

debt service and economic growth. Their main finding is that in the period 1971-90 both the

54
export-driven output growth and output growth-led export promotion hypothesis have found

a strong empirical support. For the sub-period 1983-90, however, the structural adjustment

programs, which removed economic distortions, promoted exports and encouraged

repayment of the external debt, which resulted in economic growth in the countries

considered.

Other methodological frameworks have been developed by some authors to examine the

relationship between export and economic growth. Sprout and Weaver (1993) used a

simultaneous equation model to test the export-economic growth relationship using cross-

sectional data for 72 LDCs for the period 1970-84. Their results indicated that the

simultaneous equation model has advantages over the commonly used single equation

models. They argue that endogenizing export growth in a structural simultaneous equation

model takes care of the simultaneity problem that may arise because of a two-way

relationship between exports and economic growth. In addition following Thomas (1985),

they argue that OLS estimates will underestimate the export-growth coefficient provided

export growth is truly endogenous, and the export-economic growth relationship is positive.

Hence, two-Stage Least Square (2SLS) estimates provide a more accurate measure of export-

growth relationship than the Ordinary Least Squares (OLS). Their empirical finding seems to

indicate a simultaneous relationship between economic growth and export growth among

some LDCs. As with most of other studies undertaken the results are, however, mixed.

Furthermore, they found that those LDCs that export more processed goods and have

diversified exports enjoyed economic growth than those LDCs that export primary products

and fail to diversify.

55
Esfahani (1987) also employed a simultaneous equation model to deal with simultaneity

problem between GDP and export growth. According to his results the positive relationship

between export and economic growth has been mainly due to the contribution of exports to

the reduction of import shortages which restrict the growth of many LDCs. Hence he tested

the export-growth relationship for a sample of 31 semi-industrialized countries from 1983-

1987 using the simultaneous equation framework. His main finding was that the economic

growth of most of the countries considered is largely attributed to the import supply effects of

exports.

With regard to Ethiopia, few attempts have been made to test the export-economic growth

relationship. Girma (1982) carried out country specific regression analysis for Ethiopia by

incorporating GDP as the dependent variable and exports as the only explanatory variable.

His results indicated that GDP and exports are highly correlated with correlation coefficient

of 0.962 and the coefficient of determination (R2) was 0.81. However, his work didn't

consider the effect of other important variables that could significantly influence economic

growth.

One of the country specific time series studies undertaken is that of Rati (1987) for a group of

88 LDCs including Ethiopia. In his study he provided estimates of two models of the export-

growth linkage for the period 1960-82. From the regression result, although coefficients of

the export variables in the two models are not significant, they have the expected sign. The

weak statistical significance could be mainly due to the small sample size and the problem in

the econometric technique applied.

56
In his study on the impact of exports on economic growth for Eastern and Southern Africa

countries, Kedir (1998) estimated two models (conventional and "Feder type") of the export-

growth relationship for Ethiopia. His result confirmed a positive and significant impact of

exports on economic growth in both models. Furthermore, he run the Granger non-causality

test to see the direction of causality and found out that the positive association runs from

exports to economic growth.

One immediate comment on the methodology and results is that he did not take into account

for the cointegrating properties of the variables considered in the Johansen framework. He

used growth rate of the variables and hence the regression result conveys information only

about the short run dynamics. In addition, the Granger causality test did not consider the

possibility that exports and economic growth are cointegrated and hence the results could be

biased. The sample size considered (1967-94) could be small to give reliable estimates.

In sum, all of the above empirical studies reviewed so far indicated that the export-economic

growth linkage is an unsettled issue that needs further investigation. Although most of the

crossectional studies indicated that the export-economic growth nexus is predominantly

positive and significant the time series studies cast some doubt about the existence of such a

relationship. Based on the methodological framework employed, these studies can be

categorized into those that use single equation model (OLS) estimation (both crossectional

and time series), causality test or simultaneous equation models. This study attempts to add

to the existing literature by considering altogether the methodologies employed so far. The

export-economic growth linkage will be studied using the Johansen framework for a

relatively longer period (1960-2000). In addition, the Granger causality test will be
57
undertaken allowing for the presence of cointegration between exports and economic growth.

Furthermore, to take care of the simultaneity problem that may arise between exports and

economic growth and to see the role that trade structures play in export growth and economic

growth, simultaneous equation model will also be estimated.

58
4. METHODOLOGY, MODEL SPECIFICATION AND

DATA SOUCE

4.1 Econometric Methods

4.1.1 Stationary and Non-Stationary Series

The standard classical methods of estimation which are used in the applied econometric work

are based on a set of assumptions one of which is the stationarity of the variables. A variable

is said to be covariance (weakly) stationary if the mean and the variances of the variable are

constant over time and the covariance between two periods depends only on the gap between

the periods, and not the actual time at which this covariance is considered.

A non-stationary series has a different mean at different points in time and its variance

increases with the sample size. So, the first thing in an econometric work is to check whether

a series is stationary or not. Using the classical estimation methods to estimate relationships

with non-stationary variables results in spurious regression. This is a situation in which

results obtained suggest there are statistically significant relationships between the variables

in the regression model when in fact all that is obtained is evidence of contemporaneous

correlations rather than meaningful causal relations [Harris (1995)].

Hence the non-stationary (trend) in variables needs to be removed first before getting into

any econometric work. If the trend in a variable is trend deterministic, then it is perfectly

predictable and can either be removed by regressing the variable on time (with the residuals

from such a regression forming a new variable which is trend-free and stationary) or can be
59
captured by including a deterministic time trend as one of the repressors in the model. If on

the other hand the trend is not deterministic (stochastic) then it is not perfectly predictable. In

such a case the variable needs to be differenced to nullify the trend and make it stationary. A

variable is said to be integrated of order one denoted I(1) if it must be differenced one time to

make it stationary. A level stationary series is said to be integrated of order zero i.e. I(0). In

general if the series need to be differenced d times before it becomes stationary, it is said to

be integrated of order d denoted I(d).

4.1.2 Tests for Unit Roots

There are several ways of testing for the presence of unit root. The most common and

popular one in econometric work is the DF test either because of its simplicity or its more

general nature [Harris (1995)]. Hence the emphasis here will be on using the Dicky-Fuller

(DF) approach to testing the null hypothesis that a series contains a unit root (i.e. it is non-

stationary) against the alternative of stationarity.

The AD test is based on the assumption that the data generating process of the variable being

tested is a random walk [auto regressive process of order one (i.e. AR (1)]. If however, the

variable follows a higher order auto regressive process, the error term will be auto correlated

which will invalidate the use of the DF distribution. The ADF test solves this problem by

considering a higher order and augmenting the random walk equation with some more lags. It

is suggested to allow both an intercept and time trend in the regression model used to test the

presence of unit root. In both tests the null hypothesis is that the variable is non-stationary

against the alternative stationary. The null hypothesis is rejected only when there is strong

evidence against it at the conventional levels of significant.

60
 4.1.3 Cointegration and The Error Correction Model (ECM)

Many macro economic time series are not stationary at levels and are most adequately

represented by first differences. Even though the individual time series are not stationary, a

linear combination of these variables could be stationary (i.e. they may be cointegrated). If

these variables are cointegrated, then they have a stable relationship and can not move “too

far” away from each other. In contrast lack of cointegration suggests that such variables have

no long run link, in principle they can wonder arbitrarily far away from each other

[Rao(1994)].

There are two common methods for testing co-integration and estimating the relationship

among co-integrated variables. These are the Engle and Granger (1987) two-step procedure

and the Johansen's maximum likelihood methods.

In the Engle Granger methodology, the residuals from the long-run relationship are tested for

stationary to determine whether the variables are cointegrated or not. The DF test could be

performed on the residuals to determine their order of integration. If the residuals do not

appear to be white noise, the ADF test can be used instead.

Testing for co-integration using the Engle-Granger procedure has a number of weaknesses.

First the test for cointegration is likely to have lower power against the alternative tests.

Second, its finite estimates of long-run relationship are potentially biased and third,

inferences cannot be drawn using standard t-statistics about the significance of the

parameters of the long run model [Harris (1995)]. In addition to the above the test procedure
61
assumes that there is only one cointegration vector, when in fact there could be more, that is

any linear combination of these vectors is obtained when estimating a single equation. The

Johansen procedure takes care of the above shortcomings by assuming that there are multiple

cointegrating vectors.

The Johansen procedure is a multivariate generalization of the Dickey-Fuller test [Enders,

1995]. Under this procedure the variables under consideration are by vector auto regressive

(VAR) of lag p given by:

Zt = A1 Zt-1 + A2 Zt-2 + …+ Ap Zt-p + εt

Where:

Zt is the (nx1) vector (Z1t, Z2t, …, Znt) and Ai is an (nxn) matrix of parameters. The error term

εt is an independently and identically distributed n-dimensional vector with zero mean and

variance matrix ∑ε
The above equation can be written in vector error correction model (VECM) as:

p −1
∆Z t = ∑Π
i =1
i ∆Zt − i + Π Zt-p + εt

In the above formulation, the rank of the matrix Π is equal to the number of independent

cointegrating vectors. If rank( Π ) = 0, the matrix is null implying no cointegration. If instead,

Π is of rank n, then the vector process is stationary. For cases in which 0<rank ( Π )<p, there

are multiple cointegrating vectors and in particular if rank ( Π ) = 1, then there is a single

cointegrating vector and the expression Π Zt-p is the error-correction factor.

62
The rank of a matrix is equal to the number of its characteristic roots ( λi ) that differ from

zero. Once Π and λi ’s are estimated, the test for the number of characteristic roots that are

insignificantly different from unity can be conducted using the λtrace (r) and λ max (r) statistics [

Harris, 1995].

In the λtrace (r) test statistic the null hypothesis is that the number of distinct cointegrating

vectors is less than or equal to r against a general alternative while in λ max (r) statistics the

null is that the number of cointegrating vectors is r against the alternative of r+1 cointegrating

vectors.

4.1.4 Granger Causality Test

In multivariate time series analysis, causality test is done to check which variable causes

(precedes) another variable. Given two variables X and Y, X is said to Granger cause Y if

lagged values of X predicts Y well. If lagged values of X predict Y and at the same time

lagged values of Y predict X, then there is a bi-directional causality between X and Y.

According to Granger (1988), the existence of cointegration between X and Y must be

checked before running causality test. If cointegrating relationship is found, then there must

exist causality in at least one direction. To test for causality the, first the following

cointegrating equations need to estimated by using the OLS.

Xt = α o + β o Yt + µ t ------------------------------------------------------------- (4.1.4a)

Yt = α 1 + β 1 Xt + µ ' t ------------------------------------------------------------- (4.1.4b)

63
Assuming that X and Y are I (1), Cointegration implies that the residuals µ t and µ ' t be I(0).

Having found that the variables X and Y are cointegrated, the error correction models are

formulated as follows:

M N
∆X t = ao + bo µ t −1 + ∑ C oi ∆X t −i +
i =1
∑d
i =1
oi ∆Yt −i + ε t --------------------------- (4.1.4c)

M N
∆Yt = a1 + b1 µ ' t −1 + ∑ C1i ∆Yt −i +
i =1
∑d
i =1
1i ∆X t −i + ε ' t --------------------------- (4.1.4d)

The error correction terms µ t −1 and µ ' t −1 are the stationary residuals from the cointegration

equations (4.1.4a) and (4.1.4b) respectively. By including these terms in equations (4.1.4c)

and (4.14d), the error correction models introduce an additional channel through Granger

causality can be detected. In equation (4.1.4c) Y is said to Granger cause X not only if the

do’s are jointly significant, but also if bo is significant. The error correction model allows for

the finding that Y Granger cause X as long as the error-correction term carries a significant

coefficient even if the do’s are not jointly significant.

4.3 Model Specification

In the study of the export-growth linkage, a number of variables that might be important in

the analysis can be considered. However, the limited number of available observations often

necessities the use of simple models that capture the basics of the relationships of interest.

The assessment of the effect of export performance on economic growth is carried out in a

production function framework in which exports enter as an additional 'input' in the

production process. Following the works of different authors [Balassa(1978), Tyler(1981),

Kavoussi(1984) for instance], the model to be used can be derived from a general production

function of the type:

64
Yt = f (Lt, Kt, Xt ) -----------------------------------------------------------------------------(4.3a)

Where, Yt is aggregate real output, Lt and Kt are the conventional labor and capital inputs and

Xt denotes real exports, which is introduced as an additional input. Using equation (4.3.a) and

expressing the variables in logs, the model to be used in this study can be specified as:

LGDPt = β 0 + β 1 LLABt + β 2 LINVt + β 3 LEXPt + Ut ----------------------------------(4.3b)

In equation (4.3b) above, LGDPt, LLABt, LINVt, and LEXPt are the logs of output, labor,

investment9 and export variables respectively. The coefficients β 1 , β 2 and β 3 are elasticities

and Ut is the random disturbance term.

A number of authors [Tyler (1981), Sheehey(1990)] argued that if evidence is found in

support of the export-led growth hypothesis, then this could be biased by the built-in

correlation between GDP and exports which is a component of GDP. According to

Sheehey(1990), alternative measures of the export variable not subject to this bias should be

used to test the desired relationship. Hence following this argument equation (4.3b) above

will be re-estimated using the share of exports to GDP as an alternative export variable.

In equations (4.3b), it is expected that coefficient of the export variable ( β 3 ) be positive and

significantly different from zero. In addition coefficient of the investment variable ( β 2 ) is

expected to be positive. However, the labor force coefficient ( β 1 ) will be positive or negative

depending on whether the country is labor surplus or not.

_______________________________
9
Since data on capital is usually not available, instead gross fixed investment is used.
65
The above models have been conventionally used by many authors to test the export-

economic growth relationship. Estimation of the models is based on the assumption that

export is exogenous and there is no feedback relationship between exports and economic

growth. However, if export is truly endogenous, then the OLS will underestimate the export-

growth coefficient. Hence, a simultaneous equation model will be estimated which takes into

account the feedback relationship between exports and economic growth.

Following Sprout and Weaver(1993), the model to be used consists of three equations. The

first equation like the previous models tries to measure the direct effects of export growth on

economic growth. To measure the indirect impact of trade structures on economic growth,

equations two and three will be estimated. Equation two portrays the determinants of

domestic investment while equation three addresses the issue of feedback from economic

growth to exports. Furthermore, it attempts to measure the extent to which the export sector

is influenced by internal supply factors versus external demand factors.

The model is specified as follows:

GRGDPt = α o + α 1 GDI t + α 2 GLAB t + α 3 GEXP t ------------------------------------ (4.3c)

GDI = β o + β 1 PCIt+ β 2 GPCIt+ β 3 XSHt+ β 4 KIt ---------------------------------- (4.3d)

GEXP= γ o + γ 1 GRGDPt + γ 2 PRIt+ γ 3 TPGRWt + γ 4 TPCONt+ γ 5 TSCOMPt--------- (4.3e)

Where

GRGDPt = Growth of real GDP

GDIt = Gross domestic investment as a percentage of GDP

GLABt = Growth of the labor force

66
 GEXPt = Growth of real exports

PCIt = Real GDP per capita

GPCIt = Growth of real GDP per capita

XSHt = Exports share of GDP (exports as a percentage of GDP)

KIt = Capital inflow

PRIt = Price competitiveness

TPGRWt = Trade partner’s growth.

TPCONt = Trade partner concentration

TSCOMPt = Trade structure composite

In equation (4.3c) it is hypothesized that economic growth (GRGDPt) is a positive function

of the growth of the two primary factors of production, capital and labor, as well as the

growth of the export sector.

In equation (4.3d), it is hypothesized that investment (GDIt) depends on the level of real per

capita income (PCIt), the growth of real per capita income (GLABt), the size of the export

sector (XSHt) and foreign capital inflows (KIt).

It is theorized that investment in LDCs is limited by an inadequate rate of domestic savings

and insufficient foreign capital inflows. The domestic saving rate, in turn, is theorized to be a

positive function of the level of income, growth of income and the size of the export sector.

Higher levels of income and greater growth of income generate greater savings as well as

consumption.

67
Hence, it is expected that the coefficients of PCIt and GLABt be positive. It is argued that a

large export sector result in greater savings and hence investments for a number of reasons.

First, a large export to GDP ratio produces a higher propensity to save than does the rest of

the economy. Second, income generated from export is easier to tax than more diffused wage

or profit income thus increasing public savings. Finally, the foreign exchange generated from

exports enables the purchase of the intermediate goods, which are deemed to be important for

investment. In contrast, it is argued that a higher developed export sector is responsive to the

demands of the world market rather than to internal development needs. Hence, while a large

export sector may initially induce investment by attracting foreign capital, most of the

benefits will accrue outside the country’s borders. Therefore a positive or negative GDI t -

XSHt relationship is expected based on the above arguments. Foreign capital inflows are

viewed as a means to augment domestic investment by increasing domestic savings. The

coefficient of KI, hence, should be positive.

In equation (4.3e), it is hypothesized that the growth of exports (GEXP t) is a function of (a)

the economic growth of the country (GRGDPt); (b) the price competitiveness of the country

relative to its competitors (PRIt); (c) the economic growth of the country’s major trading

partners (TPGRWt); (d) the degree to which the country’s exports are confined to a few trade

partners (TPCONt); and (e) a composite measure of the country’s composition and

concentration of exports (TSCOMPt).

Following the arguments of many authors (see Jung and Marshall (1985) a growing output

may be channeled abroad if internal demand is insufficient. Hence, while a growing export

sector can contribute to greater economic growth, the reverse may also hold true. According
68
to this view, the coefficient of GRGDPt in equation (4.3e) should be positive. In addition, the

growth of exports is contingent to a greater extent on how well the country can compete in

the world markets. A high figure for the price variable (PRIt) represents greater price

competitiveness in the country relative to its competitors. Hence, a positive GEXPt - PRIt

relationship is expected. Following the notion of trade as an engine of growth, export growth

will slow down as the demand for LDCs weaken from a slow down in economic growth in

industrial economies. According to this argument the relationship between export growth

(GEXPt) and economic growth in major trading partners (TPGRWt) would be expected to be

positive. Similarly, the greater the trade partner concentration, the more likely demand

constraints for the country’s exports will arise. Hence the GEXPt - TPCONt relationship is

theorized to be negative. In addition, it is theorized that the greater the percentage of primary

product exports to total exports and the greater the concentration of export goods, the lower

will be both export growth and economic growth. Hence it is expected that the coefficient of

the commodity concentration variable (TSCOMPt) to be negative.

4.2 Data Description and sources

The main sources of data in this study are the Ministry of Economic Development and

Cooperation (MEDaC), The National Bank of Ethiopia (NBE), Ministry of Finance (MOF),

the Ethiopian Export Promotion Agency (EEPA) and various publications of International

Monetary Fund (IMF) and World Bank (WB).

Data on Gross domestic product (GDP), GDP per capita, domestic investment, exports of

goods and services and labor force are basic from which various real and other variables used

69
in the regression analysis are derived. Data on labor force participation is obtained from

World Development Indicator (WDI) of the year 2000.

The real value of exports is calculated by computing the export unit value index and then

deflating the nominal export earnings using this index. The growth rate of real GDP, real

GDP per capita, real exports and labor is computed as the annual growth rates of each of

these variables.

Price competitiveness is calculated by taking the ratio of domestic export unit value index to

the export unit value index of major competitors. Trade partner’s growth is calculated by

using a weighted average of the real GDP growth rate of the five leading recipient countries

of Ethiopia’s exports. The weights reflected the proportion of exports received by each of the

five trading partners. Trade partner concentration is computed as proportion of the total

exports received by the country's five major trading partners while The trade structure

composite variable is calculated by taking the average of the value of primary exports as a

percentage of total exports and the value of the two leading export commodities as a

percentage of total exports.

The length of the time series data used in the analysis ranges from 1960/61 to 2000/01.

70
5. EMPIRICAL ANALYSIS

5.1 Result of Unit Root Tests

Before any meaningful regression is performed with the time series variables, it is essential to

test the existence of unit roots in the variables and hence to establish their order of integration.

The variables used in the analysis need to be stationary and/or should be cointegrated in order to

infer a meaningful relationship from the regression.

Estimation of the cointegration relationship to be undertaken in the next section requires all the

time series variables in the model to be integrated of order one. The test results of the standard

Dickey-Fuller(DF) and ADF statistics for all the time series variables used in the estimation are

presented in table 5.1 below

Table 5-1. Results of Unit Root Tests for Order of Integration of the Variables
VARIABLE DF ADF
Without With With Without drift With drift With drift
drift drift drift and trend with lag and trend
and and with lag
trend trend

1 2 1 2 1 2
LGDPt 4.30 0.05 -1.98 3.34 4.59 0.07 0.36 -2.21 -1.09
LEXPt 1.76 -0.24 -1.26 2.14 1.65 0.30 0.15 -0.76 -0.95
LLABt 12.3 -1.18 -2.59 6.02 3.91 0.22 -0.37 -2.40 -2.05
LINVt -0.96 -1.18 -0.12 -0.01 -0.02 -1.80 -1.80 -1.24 -1.19
∆LGDPt -4.21 -5.80 -5.75 -3.8 -1.46 -6.58 -3.05 -6.53 -3.03
∆LEXPt -7.21 -7.83 -7.90 -3.70 -2.45 -4.13 -2.72 -4.26 -2.89
∆LLABt -2.16 -6.93 -6.85 -0.84 -4.11 -4.44 -4.11 -4.35 -4.05
∆LINVt -3.76 -3.71 -3.94 -3.19 -2.25 -3.15 -5.66 -3.42 -2.56
Critical 1% -2.62 -3.60 -4.20 -2.62 -3.60 -4.20 -3.61 -4.20 -4.21
Values 5% -1.95 -2.93 -3.53 -1.95 -2.94 -3.52 -2.94 -3.52 -3.53
From the above test result we see that the variables are non-stationary at levels but are

stationary at first difference hence the variables are considered as I (1) processes.
71
 5.2 Estimation of the Long Run and Error Correction Models

The fact that the time-series variables under consideration are non-stationary implies that,

taken alone, the variables do not have the tendency to revert to their long run levels. Having

found that these variables are non-stationary, the next step is to check whether any linear

combination of the variables is stationary or not.

Using the Johansen's framework, the variables LGDPt, LLABt, LINVt and LEXPt can be

represented as a vector autoregression in full as:

∆LGDPt  ∆LGDPt −i   LGDPt −1 

 ∆LLAB   ∆LLAB   LLAB 
 t 
= Γi 
t −i 
+ Π
t −1 

 ∆LINVt   ∆LINVt −i   LINVt −1 

     
 ∆LEXPt   ∆LEXPt −i   LEXPt −1 

Under the above formulation, the rank of the matrix Π determines the number of

cointegrating vectors between the variables. The method used for determining the number of

cointegrating vectors is the Johansen's (1988) maximum likelihood approach.

The results of the Johansen's cointegration test for the variables under consideration is

presented in Table 5.210.

_____________________
10
The econometric package PcFIML 9.0 version is used for estimation

72
Table 5.2a Result of the Test for the Number of Cointegrating Vectors

Ho: r = n-r λ̂ i -Tln(1- λˆr +1 ) T-nm λ max (0.95) -T ∑ ln(1 − λˆ )

i
T-nm λtrace (0.95)

r =0 4 0.95 116.7** 104.8** 28.1 146.2** 131.2** 53.1

r ≤1 3 0.29 13.63 12.23 22.0 29.42 26.4 34.9

r≤2 2 0.24 10.83 9.724 15.7 15.79 14.17 20.0

r≤3 1 0.12 4.953 4.44 9.2 4.953 4.44 9.2

As presented in table 5.1a the null of no-cointgration is rejected at 1% critical level while a

case of one cointegrating vector is supported by both λ max and λtrace statistics. In addition, the

adjusted λ max and λtrace for small sample size still reject the hypothesis of no cointegration and

identify the existence of a single cointegrating vector11.

Having found one cointegrating vector among the variables, the unrestricted dynamic model

is then estimated from which the long-run model is solved. According to Inder(1993) quoted

in Harris (1995), the unrestricted dynamic model gives precise estimates of the long-run

parameters and valid t-statistics about the significance of these parameters. Using lag length

of 2 and testing to insure that this lag structure is general enough to pass various diagnostic

tests relating to the properties of the residuals Ut, the following long-run model is estimated

using the dynamic modeling approach (with t-values in parenthesis).

_____________________________
11
According to Reimers (1992), quoted in Harris (1995), using the Johansen's procedure with small sample size
problem results in over rejection of the null when in fact it is true. Thus he suggests using test statistics that
take into account the number of parameters to be estimated in the model and making an adjustment for
degrees of freedom.

73
Table 5.2b Result of the Estimated long-run Model

LGDP = 5.852 + 0.6581 LLAB + 0.1353 LINV + 0.01362 LEXP

( 18.94) ( 2.626) ( 1.382) ( 0.324)

WALD test χ 2 (3) = 382.21 [0.0000] **

AR 1- 2 F(2, 23) = 1.349 [0.2793]

ARCH 1 F( 1, 23) = 0.959 [0.3376]
Normality χ 2 = 2.172 [0.3375]
2
Xi F(22, 2) = 0.199 [0.9839]
RESET F (1, 24) = 0.508 [0.4830]

The Wald test rejects the null that all the long-run coefficients (except the constant term) are

zero. The results of various diagnostic tests [the Breush-Godfrey Lagrange Multiplier (LM)

test for serial autocorrelation, the autoregressive conditional hetroscedasticity test, the

Jarque-Bera test for normality, the White's test for hetroscedasticity and Ramsey's general

test of model misspecification] are reported and all tests did not detect any problem of serial

correlation, hetroscedasticity, non-normality and model misspecification. However, the

degree of freedom is not sufficient to conduct the White's hetroscedasticity /functional

misspecification test.

In the above long-run model, all coefficients have the anticipated signs indicating that labor,

capital and exports positively affect output. As can be seen from the t-ratios, however, it is

only labor that significantly affects economic growth in the long-run.

Having already obtained the long-run model and estimated the coefficients, the next step will

be estimation of coefficients of the short-run dynamics that have important policy

74
implications. Hence, an error correction model will be estimated that incorporates the short-

term interactions and the speed of adjustment towards long run equilibrium. In the error-

correction model, the short-run disequilibrium is approximated by the first lag of the

estimated long-run linear combination.

The procedure adopted for estimation is the Hendry's approach of general to-specific

modeling. In this approach a large model is estimated first which includes as many of the

explanatory variables and their lags as possible. Then all insignificant explanatory variables

are continuously dropped until a parsimonious model with few explanatory variables but

acceptable in terms of significance, economic interpretation and diagnostic validity is

obtained.

The error-correction model has been estimated using the OLS technique and the results are

summarized in Table 5.2c.

Table 5.2c Result of the Error-Correction Model

Variable Coefficient t-ratio[prob.]

∆LRGDPt-2 -0.343 -2.860 [0.0075]
∆LLABt 1.491 4.997 [0.0000]
∆LEXPt 0.057 2.649 [0.0126]
ECMt-1 -0.763 -3.329 [0.0023]
Dum -0.121 -4.523 [0.0001]

R2= 0.719 σ = 0.0322 DW= 2.23

AR 1- 2 F( 2, 29) = 5.159 [0.0121]

ARCH 1 F( 1, 29) = 0.134 [0.7181]
Normality χ 2 = 2.287 [0.3187]
Xi2 F (11, 19) = 1.558 [0.1908]
Xi*Xj F (22, 8) = 1.203 [0.4151]
RESET F(1, 30) = 0.029 [0.8669]

75
In estimating the error-correction model, dummy for war (dum) is introduced to capture the

effect of war and policy change from the regression analysis. The results of the various

diagnostic tests are reported and the tests did not detect any problem of hetroscedasticity,

non-normality and model misspecification. However, serial autocorrelation is detected at 5%

critical level.

In the above model, the coefficient of the error correction term is significant with expected

sign and of fairly large magnitude (-0.763). Its magnitude indicates that deviation from the

long run equilibrium is adjusted fairly quickly where 76.3% of the disequilibrium is removed

each period. Coefficients of the short run dynamics show that, labor is statistically significant

at 1% critical level while export at 5%, indicating that labor and exports growth significantly

affect the growth of the economy in the short run. The negative and significant coefficient of

dummy for war (dum) indicates that the prolonged war that took place during the Derg

regime has negatively affected the growth of the economy. In addition, the above model has

been estimated using the share of exports in GDP as one of the explanatory variables and the

regression result still indicate that labor and exports significantly affect the growth of the

economy (see Annex 2). The regression result using these different specifications of the

export variable confirm the hypothesis that exports positively and significantly affect the

growth of the Ethiopian economy12.

___________________________________________________

12
Previous country specific studies used the average annual growth rate of the variables to test the export-
economic growth relationship. In most of these studies the growth of real exports or the growth of exports
share to GDP. Some other studies [Feder(1983) and Ram(1987 )] used growth of exports multiplied by the
share of exports to GDP. Although the models used do not say anything about the long run relationship of the
variables, they are important in the dynamic sense. For completeness, these models have been estimated and
the coefficient of the export variable is predominately significant (see Annex 3 for the results).

76
The results obtained from these regressions explain some of the most important questions

that usually arise while examining the contribution of exports to economic growth in

developing countries context.

The first is the issue of requirements of a minimum threshold level of economic growth in

order to enjoy the benefits of rapid exports growth. Most of the previously undertaken studies

[Michaely (1977), Kavoussi(1987) and Ram(1985) and others] confirmed the export-

economic growth to hold only for those semi-industrialized and middle-income LDCs.

According to theses studies, the positive and significant export-economic growth relationship

is restricted to those LDCs with an income per capita of approximately U.S.$ 36013. With a

per capita income of $110 and export composition dominated by primary products (Coffee,

pulses & oilseeds and hides and skins accounting for about 80% of the total export earning),

the positive association of exports and growth in the Ethiopian context casts doubt on the

argument that a higher level of income and a higher proportion of manufactured exports are

required to reveal such a positive relationship.

The second is the issue of proper specification of the export variable in the regression

analysis. Most of the previous studies used either the growth rate of real exports or the rate of

growth of exports share to GDP. Some other studies like Feder(1983) and Ram(1987) used

growth of exports multiplied by the share of exports to GDP. The test results using these

specifications of the export variable are somewhat mixed.

_______________________________________
13
Studies undertaken by some authors like Balassa(1985) and Odedokun(1991) confirmed the relationship to
hold for lower income LDCs.

77
According to the results of this study the export-economic growth relationship is confirmed

in the Ethiopian context regardless of the specification of the export variable.

The Third is the issue of the trade policy adopted by most of the developing countries.

According to studies like Balassa(1985), countries that exhibited a positive and significant

impact of exports on economic growth were those that were relatively open and followed an

outward oriented trade strategy. The World Bank(1987) report classifies Ethiopia among the

strongly inward oriented countries for both separate periods of 1963-73 and 1973-85. During

these periods, which coincide with the Imperial and Derge regimes respectively, Ethiopia

was following an import substitution rather than export promotion trade strategy. The results

of this study indicate that that the export-economic linkage could still be evident even in the

face of an inward oriented trade strategy.

The other two are the issues of causality and simultaneity. In order to tackle the simultaneity

problem, previous studies either performed causality test or employed a simultaneous

equation model. Simultaneous equation model is estimated in order to take into account the

idea that there is simultaneity or feedback relationship between exports and economic growth

and to examine the indirect impact of exports on economic growth. The simultaneous

equation model is estimated using the two-stage least square and the result is reported in

Annex 4.

From the estimation result, although most of the variables under consideration are not

statistically significant, coefficients of the export growth variable (GEXP), exports share to

GDP (XSH) and GDP growth (GRGDP) are statistically significant. In the first equation, the
78
positive and statistically significant coefficient (at 1% critical level) of the export growth

variable once again supports the hypothesis that exports growth leads to economic growth. In

the second equation the positive and significant coefficient (at 5% critical level) of the

exports share to GDP variable suggests that a large export sector may raise investment either

by augmenting public savings through the tax generated from export proceeds or by

attracting foreign capital from the revenue generated from exports. Thus, in addition to its

direct effect, indirectly exports can induce economic growth via enhancing savings. In the

last equation the coefficient of the GDP growth variable is found positive and highly

significant (at 1% critical value). This result suggests that while growth in export can

contribute to greater economic growth, conversely, a growing economy may result in greater

exports growth. This happens particularly when the domestic demand for particular goods is

insufficient in which case the growing output is channeled abroad.

The causality between export growth and economic growth is the final issue to be analyzed

in this section. The causality test is conducted by taking into account the cointegration and

error-correction formulation of the variables. It has already been shown that both output and

exports are I(1) variables. What remains is to check whether these two variables are

cointegrated in the Engle-Granger sense. The result of the cointegration test based on the

Engle-Granger two-step procedure is reported below.

Table 5.2d Results of the DF and ADF test for the Residuals of Cointegration
Equations
VARIABLE DF ADF

ERRy -1.97[-1.95] -1.99[-1.95]

ERRx -1.17[-1.62] -1.92[-1.62]

79
 In the above table ERRy and ERRx are the residuals from the cointegration equations

LGDPt = f (LEXPt) and LEXPt = f (LGDPt) respectively. The results of the DF and ADF test

statistics are presented with the 5% critical values given in brackets.

From the test results, the residuals ERRy and ERRx are stationary which implies that the

variables LGDP and LEXP are cointegrated. Hence Granger causality which is based on error-

correction models that incorporate information from the cointegrated properties of the variables

is used.

Tables 5.2e and 5.2f below show the results for error correction model for the variables LY

and LX.

Table 5.2e Results of the Error correction Model for the dependent variable LGDPt
Variable Coefficient t-statistic[prob.]
C 0.0421 4.580[0.0001]
ERRYt-1 -0.0750 -1.818[0.0787]
∆LYt-1 -0.0133 -0.090[0.9282]
∆LYt-2 -0.5220 -3.589[0.0011]
∆LXt-1 0.0289 1.011[0.3194]
∆LXt-2 0.0156 0.550[0.5856]

R2 = 0.362
RSS =0.049
F-statistic = 3.519[0.0123]

80
Table 5.2f Results of the Error correction Model for the dependent variable LEXPt
Variable Coefficient t-statistic[prob.]
C 0.013 0.223[0.825]
ERRXt-1 -0.103 -0.968[0.340]
∆LXt-1 -0.158 -0.830[0.413]
∆LXt-2 0.164 0.877[0.387]
∆LYt-1 0.345 0.377[0.708]
∆LYt-2 1.47 1.620[0.115]

R2 = 0.167
RSS = 1.923
F-statistic = 1.249[0.310]

Given the above test results the Wald test statistic for the hypothesis that the lagged ∆LXt-i 's

do not have a joint significance effect on the variable ∆LYt is 0.549 with probability 0.583

which is rejected at both 5% and 10% critical levels. The F-statistic for the hypothesis that

the lagged ∆LYt-i's do not have a joint significant effect on the variable ∆LXt is 1.38 with a

probability 0.267, which is also rejected at these critical levels. The coefficient of the residual

ERRx is insignificant as indicated by the t-statistic. However, significance of the variable

ERRy is weakly supported at 10% critical level, which suggests that the causation runs from

exports growth to GDP growth. Therefore, the result of Granger causality test from the error

correction model indicates a different channel through which exports growth could cause

output growth.

81
In sum, all the above methodologies that are undertaken in this study summarize the different

approaches that have been followed by different authors to test the export-economic growth

relationship. The results from these analyses proved that regardless of the methodologies and

specification of the export variable, exports are found to significantly affect the growth of the

Ethiopian economy. Therefore, export expansion brings economic growth through alleviating

import shortages faced by the country, inducing public investment from the revenue collected

from the export tax, expanding the limited domestic market, contributing to the economies of

scale necessary for industrial developments and generating positive externality to the rest of

the economy.

82
6. CONCLUSIONS AND POLICY IMPLICATIONS

6.1 Conclusions

Recently, the issue of accelerated economic growth is gaining much attention by many

development economists. The decline in economic growth of most of the Sub-Saharan

African countries and other LDCs coupled with the alarming population growth led to

stagnation and even a continual decline in the income of these countries. This led to closer

scrutiny into the economic structure of these countries to determine factors determining the

growth and hence help these countries achieve a sustained economic growth.

Following the traditional trade theory, international trade is recognized to play a decisive role

in the economic growth of many countries. The classical trade theory accepts the notion that

trade acts as an "engine" if not as a "handmaiden" playing a supportive role in the economic

growth of most LDCs. Proponents of this theory argue that trade can contribute substantially

to the development of primary exporting countries while opponents of the theory strongly

contend that international trade as being completely irrelevant for development process of

LDCs. The controversy on the role of trade led to the emergence of the import substitution

(IS) and export promotion (EP) trade strategies. Failure of the IS strategy and success of

developing countries that pursued the EP strategy, led many LDCs to pay more attention to

the EP trade strategy.

Different arguments have been forwarded as to how growth of exports is associated with the

growth of the economy. These are greater capacity utilization, exploitation of economics of

83
scale, technological improvements, efficiency rising benefits of comparative advantage,

better allocation of resources, higher rates of productivity, and provision of foreign exchange.

To substantiate these arguments, a number of cross-sectional and country specific time series

analyses have been undertaken on many LDCs. Many authors have employed different

analytical methodologies to test the export-economic growth linkage in the context of

developing countries. Although, the results from these studies are somewhat mixed,

predominately a positive and significant relationship between exports and economic growth

is observed.

In Ethiopia, owing to structural problems and policies that were pursued by the different

regimes that came to power, the performance of the export sector has been less satisfactory.

The nation's output and exports are highly concentrated in agricultural commodities. And

primary agricultural products accounted for about 80-90 percent of the merchandise export

earnings of Ethiopia in the past four decades. The commodity concentration index for the five

major export products averaged 0.56 still confirming the heavy dependence of the country on

few export commodities. Furthermore, the sectoral structure of exports reveals the reliance of

the country on raw agricultural products for about 80% of its total exports earning.

Manufactured exports, which are crucial for a rapid structural transformation of production is

very weak in Ethiopia and the country is among the least industrialized country in the world.

Different trade policies have been implemented by the different governments that have ruled

the country for the last four decades. The policy adopted in the pre-1991/92 period (both in

the imperial and military government of Ethiopia) was characterized by strongly inward-

oriented development strategy that had a negative impact on export through influencing
84
directly or indirectly the profitability and competitiveness of export. The current government

that came to power in 1991/92 has undertaken trade policy reforms, which aimed at

promoting exports through diversifying the country’s commodity exports. Despite the policy

reforms, however, there is still a bias against exports that calls for active government

intervention to create a conducive atmosphere for an effective export performance.

This study attempted to examine the contribution of exports to economic growth of Ethiopia

using different econometric techniques. Different econometric techniques that tested the

export-economic growth linkage were employed. The main reason was to be able to see the

export economic growth relationship in wider perspective and also check whether the results

are sensitive to the methods applied.

In the first part an OLS techniques was used to test the export-economic growth relationship.

Data from different sources covering the period 1960/61-2000/01 are used for the analysis.

Exports were used together with labor and capital as an input in the neoclassical production

function. Pre-estimation tests of the statistical behavior of the variables (the DF and the ADF

tests for unit root) showed that all the variables used in the analysis are integrated of order

one (I(1)). The Johansen's technique was applied and the result of the cointegration test

supported the existence of a single cointegrating vector. Then, the static long run equation

was estimated and from the result although all the variables have the expected signs it is only

labor that was found to significantly affect economic growth at 1% critical level.

The error-correction model was then estimated using Hendry's general to specific approach in

order to determine a parsimonious model. The regression result shows that exports and labor
85
growth do significantly affect the growth of the economy. Specifically, a one percent change

in exports brings about 0.06 percent change in economic growth whereas a one percent

change in labor is related to more than one percent change in output growth. Furthermore, the

same model was estimated using a different export variable (exports share to GDP) in order

to check whether the results obtained are sensitive to the type of export variable used. The

regression result still supported a positive and significant relationship between exports and

economic growth.

A simultaneous equation model is then estimated which incorporated three separate

equations. The aim is to take care of the simultaneity problem that may arise between exports

and economic growth and also to shed light on the role that trade structures play in export

growth and economic growth. The estimation result using 2SLS is statistically robust, where

a one percent increase in export is related to about 0.16 percent growth of the economy.

In addition, although there is a direct contribution of exports to economic growth, indirectly

exports can foster economic growth substantially by inducing pubic savings, attracting

foreign capital and hence promoting investment. It is also found that there is a positive and

significant output-exports growth relationship i.e. the hypothesis that a growth in output can

positively influence exports growth is statistically supported. Lastly, the question of causality

between exports growth and economic growth is examined using techniques of cointegration

and error-correction modeling. The results showed that the causality runs from exports

growth to economic growth.

86
In sum, this study attempted to test whether exports contribute to economic growth or not in

the Ethiopian context. An error-correction model and the methodological frameworks that

have been employed so far in the literature were used to test the required relationship. The

results suggest that exports can substantially contribute to economic growth of the country

and the results obtained are not sensitive to the methodology used.

6.2 Policy Implications

The immediate policy recommendation that emerges from this study is that the government

in power should attempt to diversify and promote exports in order to fully exploit the benefits

of the sector and promote economic growth. In this regard, the policies towards export

promotion are crucial. As indicated in the calculation of the effective exchange rate for

imports and exports, there is still an anti-export bias (about 44%) that should be eliminated in

order to fully achieve a neutral incentive system that is conducive for efficient export

promotion. To this regard, the following measures should be taken.

a) The tax burden on exports and imports of inputs should be lowered. Specifically, the

duty draw back scheme should be very effective which allows exporters get a refund

of the tax and duty they pay on the inputs they use on export production. In addition,

the current coffee tax needs to be lowered in order to increase the profitability of the

export and reduce domestic consumption.

b) The gap between the official and parallel exchange rate need to be narrowed by

maintaining a realistic exchange rate. This will reduce the implicit import subsidy and

increases the competitiveness of exports.

87
 c) domestic exporters should be given an equal status with their foreign competitors by

enabling them to work in undistorted market and policy environment. By increasing

the competitiveness of exporters in the world market the bias against exports can be

greatly reduced. This can be achieved for instance by providing exporters an

automatic access to foreign exchange for the purchase of their intermediate goods and

also providing them a preferential interest rate on bank loans which is much lower

from the interest rate paid on non-export loans.

In addition to reducing the biases against exports, supply (sector specific) constraints need to

be addressed. Particularly, improvements in the coffee sector need to be made all along the

production, marketing, transport and processing chain in order to increase exports

measurably. For instance the auction system should be modified and coffee buyers should be

allowed to inspect and test coffee before the auction and facilities to allow testing should be

developed. With regard to the export of hides and skins, efforts should be made to control the

parasitic disease, which has greatly reduced the quality and demand for the hides and skins

and finished leather products in the international market. In addition, incentives should be

provided for livestock producers to care for their animals and to increase the number and

quality of hides and skins sold for processing. Special attention need to be given to the export

of chat whose share is enormously increasing during the past 10 years. Price regulation of

chat export need to be relaxed in order to obtain the greatest revenue possible from exports

and reduce the incentive for smuggling.

Finally, the high cost of transportation, which greatly hindered the competitiveness of the

different export sectors need to be greatly reduced. Particularly, much emphasis should be
88
given to infrastructural facilities development. Adequate supply of efficient transport

networks in road, air and railways can greatly reduce transportation cost and increase the

competitiveness of export sectors.

In sum, the above measures to be undertaken in order to reduce and eliminate the supply

constraints deterring the performance of export sector are somewhat general. A closer look

and detailed investigation into each sectors is very important if export promotion and

diversification schemes are to be successful.

89
 BIBLIOGRAPHY

(Note: * indicates those that are not cited directly in the study)

Amoateng, K. and Amako, B.(1996), " Economic growth, export and external debt causality:

the case of African countries", Applied Economics, 28: 21-27

Bahmani-Oskoee, M. and Alse, J. (1994), "Export Growth and Economic Growth: An

application of Cointegration and Error-Correction Modeling", The Journal of

Developing Areas, 27, 535-42

Balassa, B. (1978), "Exports and Economic Growth: Further Evidence", Journal of

Development Economics, 5:181-89

_______ (1985), "Exports, Policy Choices, and Economic Growth in Developing Countries

After the 1973 oil shock", Journal of Development Economics, 18:23-35

Begum, S. and Samsuddin, A.(1998), "Exports and Economic growth in Bangladesh",

The Journal of Development Studies, 35: 89-114

Bhagwati J. (1988), "Export-Promoting Trade Strategy: issues and evidence", Research

Observer, 1:27-57

Chow P.C. (1987), "Causality between export growth and Industrial Development",

Journal of Development Economics, 26: 55-63

Dodaro, S. (1993), "Exports and Growth a reconsideration of causality", The Journal of

Developing Areas, 27: 227-44

Edwards, S. (1993), "Openness, Trade Libralization and Growth in Developing Countries",

Journal of Economic Literature, XXXI: 1358-93

Enders, W. (1995), Applied Econometric Time Series, John Wiley & Sons, Inc., New York.

90
Esfahani, H.S. (1991), "Exports, imports and economic growth in semi-industrialized

countries", Journal of Development Economics, 35: 93-116

Fajana, O. (1979), " Trade and Growth the Nigerian experience", World Development, 7: 73-

78

Feder, G. (1983), "On exports and Economic growth", Journal of Development Economies,

12: 59-74

Girma, E. (1982), An Analysis of the Relationship Between Ethiopia's Foreign Trade and

GDP. MSC Thesis, Addis Ababa University, Addis Ababa, Ethiopia.

Grabowski, R. (1994), "Import Substitution, Export promotion and the state of economic

development", The Journal of Developing Areas, 28: 535-54.

Granger, C.W. (1988), "Some Recent Developments in a Concept of Causality", Journal of

Econometrics, 39:199-211

Harris, R.Z.D. (1985), Using Cointegration Analysis in Econometric Modelling, London,

University of Portsmouth, Prentice Hall

Hirschman*, A.O. (1958), The Strategy of Economic Development, Yale University Press,

New Haven

IMF (Various Issues) International Financial Statistics Year Book

Jung, W.S. and Marshall, P.J. (1985), "Exports, growth and causality in developing

countries", Journal of Development Economics, 18: 1-12

Kavoussi, R.M. (1984),"Export expansion and economic growth further empirical evidence",

Journal of Development Economies, 14, 241-50

Kedir, E. (1998), Exports and Economic Growth: Theories and Evidence from Eastern and

Southern Africa, MSC Thesis, Addis Ababa University, Addis Ababa, Ethiopia.

91
Krishna, K. and Ozyildirim, A.(1998), "Trade, Investment and Growth: Nexus, Analysis and

Prognosis", Pennsylvania State University Working Paper

Kruger*, A.O. (1980), "Trade Policy as an Input to development", American Economic

Review, Papers and Proceedings, (May): 288-92

_______ (1995), " The role of Trade in Growth and Development: Theory and Lessons from

Experience", in Sustaining Export-Oriented Development Ideas from East Asia,

P:1-30, (eds) Garnaut R., Grill E. and Riedel J., Press Syndicate of the University of

Cambridge, Australia.

Mannur, H.G. (1996), International Economics (Second Edition), Vikas Publishing House

Pvt. Ltd. Delhi.

MEDaC (1997), A study on Export Promotion Schemes, Addis Ababa

MEDaC (2000), Efforts and Achievements on Export Diversification: An Assessement

Meier, G.M. (1985), Leading issues in Economic Development, Sixth Edition, Oxford

University Press.

Michaely, M.(1977), "Exports and economic growth: An empirical Investigation", Journal of

Development Economies, 4: 49-53

Milner, C. (1994), " Trade Strategy and Revealed Trade Bias: An Evaluation for a small

Industrializing Economy", World Development, 22:587-99

Moschos, D. (1989), "Export expansion, growth and the level of economic development",

Journal of Development Economies , 30: 93-102

Myint, H. (1958), "The 'Classical Theory' of International Trade and the Underdeveloped

countries", Economic Journal, 68:317-37

Ocampo, J.A. (1986), "New Developments in Trade Theory and LDCs", Journal of

Development Economies, 22: 129-70

92
Odedokun M.O. (1991), "Differential impacts of export expansion on economic growth in

the LDCs: A comparison of evidences accross regional income groups and

between the decades of 1970s and 1980s", Eastern Africa Economic Review, 7:69-

93

Posner, M.V. (1961), "International Trade and Technical Change", Oxford Economic Papers,

13:323-41

Ram, R. (1985), "Exports and economic growth: Some additional evidence", Economic

Development and Cultural Change, 33: 415-25

_______ (1987), "Exports and economic growth in developing countries: Evidence from time

series and cross-sectional data", Economic Development and Cultural Change 36:

50-71

Rana, P. (1988), "Exports, policy changes, and economic growth in developing countries

after the 1973 oil shock", Journal of Development Economies, 28: 261-64

Rao, B.B. (1994), Cointegration for the Applied Economist, University of New South Wales,

Australia, Macmillan.

Reimers*, H.E. (1992), Comparisons of tests for multivariate cointegration, Statistical

Papers, 33: 335-59

Riedel, J. (1984), "Trade as an engine of growth in developing countries, Revisited",

Economic Journal, (March): 56-73

Rodney, W. (1982), How Europe Under Developed Africa, Howard University Press;

Washington, D.C.

Rosenstein-Rodan*, P.N. (1943),"Problems of Industrialization of Eastern and South-Eastern

Europe", Economic Journal, 53:202-11

93
Salvatore, D. (1990), International Economics (third edition), Macmillan publishing

company, New York.

Sheehey, E.J. (1992),"Exports and Growth: additional evidence", The Journal of

Development Studies, 28:731-35

Sodersten. and Reed. (1994), International Economics third edition, Macmillan, London

Sprout, R.V. and Weaver J.H. (1993), "Exports and economic growth in a simultaneous

equations model", Journal of Developing Areas 27: 289-306

Thomas, R.L. (1995), Introductory Econometrics, Second Edition, Longman, London

Todaro, M.P. (1994), Economic Development, Fifth Edition, Longman Publishing, New

York

Tyler, W. G. (1981), "Growth and Export Expansion in Developing Countries: Some

Empirical Evidence, Journal of Development Economies, 9: 121-30

Vernon, R. (1966), "International Investment and International Trade in Product Cycle",

Quarterly Journal of Economics, 80:190-207

World Bank(1987), World development report, Oxford University press

World Bank (2001), Developing Exports to Promote Growth, Report No: 23294-ET

94
95