You are on page 1of 116

CHAPTER ONE INTRODUCTION Background to the study The debt crisis was perhaps the dominant macroeconomic problem

of the 1980s for many highly indebted developing countries. Analysis of the causes of the crisis, its domestic consequences, and the nature of the desirable policy responses on the part of both the international community and domestic has generated an enormous literature.

governments in the indebted countries

This crisis evolved from a comple! combination of factors in "ub "aharan Africa, some of which are e!ternal while others were the result of the internal policies pursued by the African governments. A ma#or cause of the African debt crisis was the two oil price shoc$s of the early and late 19%0s. To be precise, the oil price hi$es of 19%& 19%' and 19%9 1980 changed the international environment, leading to macroeconomic deterioration in Africa that lasted until 198(. The unprecedented increase in oil prices distorted the trade balance of oil importing countries and created fiscal deficits that undermined domestic investment )*+,TA-, (00'.. The situation became worse during the second oil shoc$ which happened #ointly with a sharp increase in world real interest rates. The global recession of 1981 198( compounded the problem by adversely depressing demand for the $ey

e!ports of developing countries. /oreover, the deteriorating terms of trade caused by the recession created balance of payment problems for oil e!porters and worsened those of oil importers. 0owever, based on the assumption that the global recession would be shortlived and that prices of non fuel commodities would recover quic$ly, most of these countries resorted to heavy e!ternal borrowing to finance the fiscal and e!ternal imbalances. 1n Asia, for e!ample, many countries whose industrial base is strong decided to hold bac$ on their e!ternal borrowing, and e!panded their volume of e!ports through a variety of e!port promotions and industrial policies )*+,TA-, 19882 3alassa, 1981 and 19842 5u6net, 1988.. 7n the contrary, many African countries had no alternative to e!ternal borrowing due to mostly their non diversified economies and the continued decline in their e!port commodity prices during the global recession period of 1981 and 198(. The debt ratios of "ub "aharan Africa rose sharply between 1980 and 198%. The debt to 8-9 ratio rose from &8 percent to %0 percent, while the debt to e!port ratio rose from 140 percent to &(4 percent ):orld 3an$, 1989.. -uring this same period, income per capita declined by 1' percent. /ultilateral and bilateral creditors reviewed their lending policies to low income countries, especially to Africa by stipulating that further loan disbursement will be based only on economic reforms in the conte!t of structural ad#ustment programmes. 3efore the launch of the 019, 1nitiative, the total long term outstanding debt of "ub "aharan Africa increased by more than (00 percent between 1980 and 1994 .That same period multilateral and official debt stoc$s

increased by more than 400 percent and &00 percent respectively ):orld 3an$, (001.. The structural ad#ustment programmes failed woefully to deliver the promised growth and development in many African countries due to heavy debt service obligations. ;i$e most developing countries of the world, "ierra ;eone relies substantially on e!ternal funds for financing its development pro#ects < iron and steel mills, roads, electricity generation plants etc. "uch e!ternal funding usually ta$es the form of e!ternal loans. 1n the early years of political independence )i.e. 19=1 through 1980., the si6e of such loans was small, the rate of interest concessionary, the maturity was long term, and the source was usually bilateral or multilateral in nature. >or instance, "ierra ;eone?s e!ternal debt in 19%0 was about @49 million2 however, beginning in the year 1980, the situation changed. "ierra ;eone, at the lure of the international financial centres, started to borrow huge sums from private sources at floating rates and with shorter term maturities. The 1980 A7A* loanB alone was estimated at *" @'&& million. 3y 1990, the value of "ierra ;eone?s e!ternal indebtedness was *" @1.( billion, which represented over 1&'.' percent of "ierra ;eone?s gross domestic product )8-9. for that year. The situation precipitated a debt crisis that progressively worsened over time. 3y 1991, "ierra ;eone had to adopt a :orld 3an$C1nternational /onetary >und )1/>. sponsored "tructural Ad#ustment 9rogram )"A9., with a view to revamping the economy and ma$ing the country better able to service her debt.

&

The stoc$ of debt has increased sharply since the inception of the "tructural Ad#ustment 9rogram )"A9. in 1991. 1t rose from *"@901.4 million in constant dollar terms in 1989 to *"@1.( billion in 1991 )3an$ of "ierra ;eone 3ulletin, 199'.. 7ne significant factor that had influenced the level and growth of e!ternal debt since the inception of the "tructural Ad#ustment 9rogram )"A9. was the government?s determined effort at macroeconomic ad#ustment. The program has been supported with substantial levels of loans and credits. The sharpest increase in the level of debt was e!perienced in 199= when e!ternal indebtedness rose by 104 percent in constant dollar values. This was to be e!pected as the level of shortages in the economy at the start of the program was such that "ierra ;eone needed massive in#ections of capital inflows if reforms were to ta$e root. /ore significantly, the significant shortfalls in e!port earnings between1990 (00(, made increased borrowing a necessity to furbish the economy. The level of debt actually fell in (00= because of the 9aris ,lub debt rescheduling agreements, which the country signed in Duly 199'. The twenty first century ushered in an era of change in the attitude of richer nations towards their poorer counterparts in the developing world. :ith the adoption of the /illennium -evelopment 8oals )/-8s. in (000, the international community has set itself a target of reducing poverty, income inequality, hunger and increasing human development by (014. 0owever, many observers have e!pressed doubts that, on the current trends, it is unli$ely that this ob#ective can be achieved in the poorer countries, especially in Africa and "ierra ;eone in particular at any time close to that date )3rown, :olfensohn, (00'..

'

1n order to meet the target of poverty reduction in "ierra ;eone by (014, the current 8-9 growth rate in the country would have to be increased and sustained to levels of at least % to 8 percent per annum, implying doubling the current aid inflow and maintaining it at that level for at least a decade )*+,TA-, (008.. 9ursuing such an action, with the right mi! of international support and appropriate domestic policies, would generate enough savings for investment that would put "ierra ;eone on a sustainable growth trac$ and graduate it from debt burden and aid dependency. "ierra ;eone?s e!ternal debt problems and its resource requirements are directly related to its capacity to accumulate capital and to grow. 0owever, since the inception of the debt crisis in the early 1980s, the country has been transferring billions of dollars every year to the creditor nations in debt service at the cost of providing domestic services in the country. -ebt service is accounting for a larger percentage of "ierra ;eone?s 8-9 annually. /eanwhile, the debt crisis facing "ierra ;eone has consistently occupied the international policy ma$ers. "ince its inception in the 1980s the international financial community has been providing help to the country in an attempt to reduce its e!ternal debt burden, foster growth, reduce poverty, and attain e!ternal viability )1/>, (004.. This assistance has ta$en the form of providing concessional financing from the international financial institutions, debt relief from official creditors mainly in the conte!t of the 9aris ,lub rescheduling and restructuring, and in some cases, through bilateral action by creditors. These measures have resulted in considerable success in alleviating the e!ternal debt

burden of the country. -espite these efforts, "ierra ;eone still continue to suffer from unacceptable levels of poverty, civil conflicts, high e!ternal debt burden and low economic growth. Today, the international community is constantly engaged in discussions and analysis of the problems of the country. The :orld 3an$ produced several reports during the 1980s diagnosing in detail the problems of "ierra ;eone and recommended several policy measures for restoring economic growth and long term development. /ember countries in the region also carried out series of studies and adopted numerous comprehensive programmes for achieving specific regional goals and ob#ectives )1qbal and 5abur, 199%.. There was a continuous flow of information about the country?s problems but opinions were, however, different as to the cause and what should be done to restore growth and promote sustainable development. The :orld 3an$, for instance, blamed domestic mismanagement and other factors for the country?s poor economic performance. 0eavy e!ternal indebtedness and rampant corruption in "ierra ;eone have not only had a negative effect on economic growth and development, but also derailed efforts at economic recovery and mar$et enhancing initiatives )Eichards, +wanna, and +wan$wo, (00&.. The country is the world?s most relatively indebted and aid dependent country of the world. /uch of the resources that could have been ploughed bac$ into investments are used to service debt or embe66led by corrupt government officials. 7ut of the '' countries classified by the :orld 3an$ as Aheavily indebted poor countries,B "ierra ;eone is among them. The country is so burdened and overwhelmed by debt service obligation

that its economy and financial systems have practically crippled as they struggle to evolve a way of servicing its foreign debt. 1n this process, its economic and social welfare systems have deteriorated beyond e!pected levels )Eichards, +wanna and +wan$wo, (00&.. Although the international creditors seem to be fle!ible in debt rescheduling and restructuring, nonetheless, no signs of mar$ed improvement have been noticed in "ierra ;eone. The continued deterioration of economic conditions rallied sympathetic calls by a networ$ of activists, civil society and +87s for debt relief and cancellation.

Statement of the ro!"em The growing problem of debt accumulation in "ub "aharan Africa has received considerable attention in the literature and is now recogni6ed as a serious economic issue. -ifferent authors have emphasi6ed different aspects of the debt crisis. >or e!ample ,line )1984. focuses on the global macro economic considerations and "achs )1984. stresses not only the importance of the global shoc$s but also the country specific factors. 8reene )1989. attributes this problem to both domestic policies and e!ternal factors. :hat is very important about these views is that the bac$ground to the e!ternal debt problem varies from region to region and country to country. "ierra ;eone?s severe indebtedness situation is not an e!ception to the seemingly combined effects of factors that are generally associated with sub "aharan African countries? huge debt burden. "ierra ;eone, li$e others, resorted to e!tensive borrowing to finance its development agenda after independence in

19=1. F!ternal loans were usually acquired to support the balance of payments or hasten the pace of economic development by augmenting domestic savings to the e!ternal debt obligations are met without undue pressure on available resources. The gradual increase in the e!ternal debt, from 14.' percent of 8-9 in 19%0 to &9.& percent of 8-9, 1&'.' percent of 8-9 and 1=41 percent of 8-9 in 1980, 1990 and (008 respectively suggest that the country is one of the highly indebted countries in Africa. "ierra ;eone?s debt problems have been attributed to both e!ternal and domestic factors. The e!ternal causes of "ierra ;eone?s huge debt can be traced as far bac$ as the 19%0?s. The most important factors were the liberal lending postures of the international commercial ban$, negative real interest rate during the 19%0s and the sharp decline in foreign e!change earnings as a result of the collapse of commodity mar$ets. The growth in world trade coupled with the low interest rate created a situation where international ban$s persuaded themselves to believe that Acountries never go ban$rupt. ,onsequently, they increase their e!posure regardless of the quality and ris$iness of their loans. The plummeting of the primary agriculture products and the disproportionate increase in oil prices compounded the problem. As mentioned above, the accumulation of huge debt was also caused by important domestic factors. >irstly, there was lac$ of appropriate debt management strategy. Although the government of "ierra ;eone loans Acts of 19=4 and the general loan )Amendment. Acts of 1984 attempted to consolidate the authority of contracting both domestic and international loans under the

/inistry of >inance, line ministries and other government agencies failed to comply. This short coming arose primarily because the role of e!ternal debt as a complementary instrument of economic transformation even though recogni6ed was not seriously persuaded. The acquisition, deployment and retirement of e!ternal loans were not systematic enough to ensure that the pro#ect were self financing. 7ften, medium term loans were used to finance long term pro#ects. Thus repayment fell due even when pro#ect had hardly ta$en off. 1n some cases the pro#ect were $nown to be economically viable but embar$ upon for political reasons such as, the hosting of the 7A* in 1980. This study, therefore, see$s to find out empiricallyG what has been the impact of e!ternal debt on the economic growth of "ierra ;eone since there is too much controversy as to the relationship between e!ternal debt and economic growth2 and whether "ierra ;eone suffers from a debt overhang problem.

O!#ect$%es of the Study The main ob#ective of this study is to investigate the relationship between e!ternal debt and economic growth using time series dataset in "ierra ;eone from 19%0 to (008. "pecifically the study see$s to2 1nvestigate the long run relationship between e!ternal debt and economic growth in "ierra ;eone. >ind out whether short run relationship e!ists between e!ternal debt and economic growth in "ierra ;eone.

F!plore the nature of relationship between e!ternal debt and economic growth in "ierra ;eone

-etermine policy options based on the outcome of the study.

Hy otheses of the study This study see$s to test the following hypothesisG that 07G There is no long run relationship between e!ternal debt and economic growth. 07G There is no short run relationship between e!ternal debt and

economic growth. 07G There is no causality between e!ternal debt and economic growth.

S$gn$f$cance of the Study "ince growth is the engine of economic prosperity of any nation, this study will contribute to the development of policies and strategies that see$ to promote growth in "ierra ;eone. A growth in annual output is what translates to an increase in income per capita. A higher per capita income in turn, increases the mar$et bas$et of goods a citi6en can afford. ;arge e!ternal debt burden has, however, encouraged a continuous outflow of resources from "ierra ;eone in debt service payment, and thus, retarding the country?s economic growth and development. The study is therefore, relevant for the social, political, and economic development of "ierra ;eone economy. /eanwhile, as every nation is today striving hard to increase its national output, reduce income inequality and improves the standard of living of its

10

citi6ens, "ierra ;eone is no e!ception. All things remaining the same, saving and investment are prerequisites for economic growth while heavy e!ternal debt burden depletes domestic savings and retards growth. The study will contribute to the scholarly discussion on the implications of heavy e!ternal indebtedness especially in the conte!t of "ierra ;eone. ;ow levels or debt free "ierra ;eone will benefit the country in the following waysG firstly, it will encourage voluntary capital inflows from abroad in the form of foreign direct investment and eliminate the uncertainties associated with heavy debt burdens. This will create #ob opportunities for the unemployed masses, improve the standard of living of the citi6ens and reduce crime rates among the youth. >oreign and domestic entrepreneurs, for e!ample, may not hesitate to invest for fear of restrictions on their abilities to finance imports or repatriate profits. "econdly, the social conditions and quantity and quality of public services in such crucial areas li$e education, health, transport and infrastructure will be improved through increased government spending. The spread of poverty and decline in the provision of public and social services in "ierra ;eone is largely blamed on the outflow of capital in debt service payments. >inally, with adequate $nowledge and understanding of the causes and effects of heavy e!ternal debt burden on an economy, government leaders of "ierra ;eone will adopt appropriate policies geared towards minimi6ing macroeconomic imbalances and eliminate economic distortions caused by heavy debt obligation. Trade and investment policies and regulatory framewor$s will be streamlined. 9rivate savings and investment will be encouraged through financial

11

liberali6ation. 9ublic investment programs will compliment private sector initiatives and concentrate on infrastructural deficiencies and human resource development. Sco e of the Study This study e!amines the relationship between e!ternal debt and economic growth in "ierra ;eone using time series data set from 19%0 to (008. 1t utili6es the recently developed Autoregressive -istributed ;ag )AE-;. model otherwise $nown as the bounds testing approach to cointegration developed by 9esaran and 9esaran )1999.2 9esaran, "hin and "mith )(001.. The study employed the following variablesG Eeal 8-9 as a pro!y for economic growth, e!ternal debt, labour force, investment as a ratio of 8-9, and e!port. 1nvestment as a ratio of 8-9 is pro!ied by investment2 including a dummy variable to capture the effect of civil unrest in the country.

Organ$sat$on of the Study This study is divided into five chapters. ,hapter one, which is the introductory chapter, focuses on the bac$ground to the study, problem statement, ob#ectives of the study, hypotheses, significance and scope of the study as well as organisation of the study. ,hapter two presents review of relevant literature, both theoretical and empirical that underpins e!ternal debt and economic growth with a view of identifying all its possible influences. This certainly leads to the formulation of an appropriate model for analysis. ,hapter three presents the methodological framewor$ and techniques employed in conducting the study. 1n

1(

addition, it deals with data and techniques adopted in carrying out the study. ,hapter four e!amines and discusses the results and main findings with reference to the literature. The final chapter presents the summary, conclusions and recommendations of the study

1&

CHAPTER T&O RE'IE& O( RE)ATED )ITERATURE Introduct$on The broad aim of this chapter is to present the review of relevant literature on the relationship between e!ternal debt and economic growth. The chapter is organised into three sections. The first section presents an overview of the "ierra ;eonean economy with specific focus on both domestic and e!ternal debt. The rationale is to give a fair view and draw attention to the various creditors that characterised the e!ternal debt stoc$ in the economy. 1n addition, the section discusses briefly the main features of "ierra ;eone?s e!ternal debt spanning the early 19%0s to (008 and the causes of e!ternal debt accumulation. The second section presents and discusses the theoretical foundations of the relationship between e!ternal debt and economic growthG 0arrod -omar growth model, the +eo ,lassical growth model and Fndogenous growth model are discussed2 while the empirical literature is reviewed in the third section. The empirical literature is categorised around econometric approaches. This is as a result of the fact that many of the strides in empirical studies of debt and growth have been methodological. The categorisation is done into cross country regressions, time

1'

series, panel data and country case studies. 0owever, greater attention is devoted to time series studies. 0ence the main purpose of the review is to identify relevant and ma#or variables for a growth model lin$ing e!ternal debt to long term growth.

O%er%$e* of the S$erra )eonean Economy The economy of "ierra ;eone en#oyed moderate growth rate of nearly & percent annually in the first half of the 19%0s. The macro economy was relatively stable with low inflation rates, a healthy foreign e!change position and moderate levels of e!ternal debt. The deterioration in economic performance began in the mid 19%0s when 8-9 growth slowed down to one percent due mainly to poor performance in the mining sector. Thus by 199' domestic public debt stood at ;e 1&.1&8 billion and 8overnment borrowing from the ban$ing system accounted for 41 percent and non ban$ing borrowing at '9 percent. This coupled with e!ternal debt brought the total debt stoc$ to *"@1.( billion. 1t rapidly accelerated with population growth e!ceeding the annual average 8-9 growth of 1.4 percent resulting to a fall in real per capita incomes )3an$ of "ierra ;eone, 1999.. The poor performance of the economy in the 1980s reflects, in part, inappropriate government macroeconomic and fiscal management policies including e!ternal debt policy. The e!pansion of civil service, rising subsidy and interest payments gave rise to mounting fiscal imbalances. The budget deficit, on a commitment basis, averaged more than 1( percent of 8-9 in the 1980s, reaching an unsustainable level of 18 percent of 8-9 in 198=C8%.

14

9oor economic and fiscal mismanagement resulted in e!cessive borrowing to finance fiscal deficit and development pro#ects resulting in high e!ternal debt levels beyond the servicing capacity of the economy. 3y 198%, "ierra ;eone was unable to service its e!ternal debt. The inability of the government to honour e!ternal debt obligations resulted in the accumulation of e!ternal debt arrears, which hampered relations with the international financial community and hence hampered economic growth. 1n the absence of e!ternal budgetary support, the government resorted to domestic borrowing and money creation to finance the huge and persistent fiscal deficits. /onetary policy was essentially accommodative, e!panding domestic credit in response to increasing fiscal deficits. As a consequence, inflation surged and reached 1=% percent by 198=C8%. >ollowing the prolonged years of economic decline and failed remedial measures, the government in 198% began to implement an economic recovery programme that was intensified in 199( with support from the 1/>, :orld 3an$ and the Africa -evelopment 3an$ )Af-3.. A number of other international development agencies, notably the Furopean *nion )F*. and the 3ritish 7verseas -evelopment Administration, also supported 8overnment?s reform efforts. "ome of the reform measures include liberali6ation of trade and e!change rate policies, public enterprise, financial sector, civil service, fiscal and e!ternal debt management reforms. -espite the adverse security situation caused by the civil armed conflict, significant progress was made in stabili6ing the economy. The overall budget

1=

deficit on a commitment basis reduced from 1( percent of 8-9 in 1989C90 to =.& percent of 8-9 in (008. The annual inflation rate fell from 114 percent in 1991 to = percent in (008. 0owever, the deterioration in the security situations in 199% completely reversed the gains made in stabilising the economy and disrupted the implementation of $ey structural reforms. This adverse security situation to a large e!tent hampered economic growth.

Com os$t$on of S$erra )eone+s E,terna" and Domest$c De!t Stocks As standard practice, a nation?s debt profile is classified into two broad categories of F!ternal and -omestic -ebt. *nder each of these profiles, there are further classifications of debt by way of identifying the creditor as a multilateral organi6ation, bilateral arrangements )between one creditor nation and a debtor nation., commercial )mostly non governmental and private. institutions2 or in the case of domestic debt by way of financial instruments used to obtain the loan. 3oth bilateral and multilateral debts have increased steadily since 19%0, even though the bilateral debt has been higher. The immediate consequences of this increase in official debt were that total debt rose rapidly relative to e!port earnings and this led to the accumulation of arrears. >or e!ample, the :orld 3an$?s credit, of the multilateral sources, increased from &1.4 percent on average from 198= 1991 and over '0 percent in 199( (00=. >or the same period, 1/> credit fell from over '0 percent on average to (%.9 percent. 7ther multilateral creditors li$e the African -evelopment 3an$ )Af-3. and the F, accounted for &1 percent for the period 199( (00=. 1-A credit also accounted for a greater

1%

proportion among the :orld 3an$ sources. 1n 199=, for e!ample, 1-A credit accounted for %%.= percent. Tables 1 and ( give the composition of "ierra ;eone?s F!ternal and -omestic -ebt "toc$, covering selected periods )(000 (004.. The bilateral creditors are distinguished as 9aris ,lub )9,. ,reditors and +on 9aris ,lub ,reditors. The 9aris ,lub is an informal group of creditor governments that has regularly met in 9aris, >rance, since 194= to reschedule bilateral debts. The >rench Treasury provides the "ecretariat for the 9, of ,reditors. The ,reditors meet with a debtor nation to reschedule its debt as part of the international support to countries e!periencing severe debt service burdens and implementing an ad#ustment programme supported by the 1/>. The 9, does not have a fi!ed membership and its meetings are open to all official creditors that accept its practices and procedures. The core creditors are mainly 7rgani6ation for Fconomic ,ooperation and -evelopment )7F,-. member countries but other creditors attend as relevant to a debtor nation. +on 9aris ,lub countries are those outside the membership of the 9,. "ince debt is a stoc$, it reduces as principal repayments and interest payments are honoured, given that additional debt is not contracted. 1n this case, it will be noticed in Table ( that domestic debt stoc$ under :ays and /eans Advances )in million ;eones. is 6ero for the = year period. This does not mean that no debt under this category was obtained during a given year, but that it was reduced to 6ero by the end of that year. The emphasis is that the figures in Tables 1 and ( show debt stoc$ as at end of each year.

18

Ta!"e -. Com os$t$on of S$erra )eone/s E,terna" De!t Stock 0$n m$""$on US 12 for the se"ected er$od 0344-534462

,reditors /ultilateral 9aris ,lub +on 9aris ,lub ,ommercial

(000 %49.00 (81.88 ='.1( 11=.&=

(001 %8'.10 &11.10 =0.(0 (%(.(0

(00( 8%8.=' &'&.90 =(.80 (40.%0

(00& (00' 9'4.90 1,0'&.40 &%1.=0 =&.10 ('9.80 &=(.00 4&.%0 (4&.=0

(004 1,040.00 &'8.&0 44.00 ('4.00

Total 1,((1.&= 1'(%.=0 1,4&=.0' 1,=&0.'0 1,%1(.80 1,=98.&0 "ourceG /inistry of >inance H 3an$ of "ierra ;eone 3ulletin )(00%. Ta!"e 3. Com os$t$on of S$erra )eone/s Domest$c De!t Stock 0$n m$""$on )eones2 for the se"ected er$ods 0344-534462

,reditors Treasury 3ills Treasury 3earer 3onds :aysH /eans Advances

(000 9',%8&

(001 1&9,&4'

(00( 18%,='=

(00& (&1,4&(

(00' &0',''%

(004 &%4,'''

'%,&9%

=9,=1'

84,(4&

11',0'(

1&1,'88

1&%,418

0 41(,9=(

Total 1'(,180 (08,9=8 (%(,899 &'4,4%' '&4,9&4 "ourceG /inistry of >inance H 3an$ of "ierra ;eone 3ulletin )(00%.

S$erra )eone+s e,terna" de!t !urden s$nce -784

19

"ierra ;eone is one of the heavily indebted low income non oil producing countries in "ub "aharan Africa. >ew years after independence, it resorted to development finance from abroad to supplement national savings for investment and achieving sustainable growth. ,onsidering its narrow e!port base mainly from primary agricultural products and unprocessed minerals, "ierra ;eone has a continuing need for foreign e!change not only to finance the development process, but also to reduce the widening gap in current account balances, which had adversely affected the economy. The country has a huge debt overhang which serves as disincentive to investment, and economic growth. 1n 1980 )a year after the second oil price hi$e., e!ternal debt amounted to *"@'&& million from *"@49 million in 19%0 and *"@&%9 million in19%9 respectively. This increase was connected with increase in foreign e!change demand to pay for oil imports, a situation caused by the 19%9 oil price increase as well as foreign e!change required for financing the 7rganisation of African *nity )7A*. summit held in 1980. 1n order to finance the summit, short term loans were contracted at non concessional terms, e!acerbating the upward trend in e!ternal debt obligations. 1t is asserted that the total cost that was involved in hosting this summit did not only reduced the country?s e!ternal reserves, but also severely depleted domestic savings by (%0 percent from 19%9 to 1980. )Af-3, :orld 3an$, 1994.. The trend in e!ternal debt accumulation continued unabated during the 1980s, and at the beginning of the 1990s )see figure '., government policy decisions were significantly influenced by huge debt burden. 1n 1990, a year after

(0

an 1/>C:orld 3an$ programme was agreed upon, e!ternal debt accumulated to *"@1.(0= billion in 1990, *"@1.'4( billion in 199&, *"@1.4&( in 199', while it moderately reduced to *"@1.((= billion in 1994 )an amount substantially higher than the 19%0s and 1980s figures.. 1t can therefore be #udged that e!ternal debt significantly increased during the 1990s )a period of "tructural Ad#ustment 9rogramme., a situation mainly attributed to the acquisition of new loansCdisbursements from 1/>, :orld 3an$ and African -evelopment 3an$ to finance critical imports and development activities. At the same time, 8overnment normalised its relations with the e!ternal creditors mainly 1/>, :orld 3an$ and the African -evelopment 3an$. This resulted in the inflow of new resources at concessional terms. These loans were contracted mainly from the multilateral institutions and were used to resuscitate the physical and economic infrastructure and to finance the importation of essential items such as petroleum, rice, raw materials and spare parts for industries. "ierra ;eone?s e!ternal debt is classified into short and long term debt. The short term debt requires a repayment period of twelve months or less while long term debt requires repayment periods beyond one year. 9rior to 1980, short term debt was of less significance as it only accounted for 8 percent of total debt owed. 1n 1980, short term loans amounted to *"@4' million, while it increased to *"@''= million, *"@'98 million and *"@=(0.( million in 1990, 199' and (008, which represents &8 percent, &' percent and &( percent of the total debt stoc$ respectively.

(1

Although the absolute values tend to suggest that long term debt accumulated has been greater than short term debt, the debt service burden of the latter can be more severe than long term debt. This is so, because the terms of commitment for short term debt would have adversely impacted on economic management and growth than those whose repayment period e!ceed one year. /ultilateral institutions are the main significant group of creditors accounting for about 49.8 percent of the debt followed by 9aris ,lub creditors with (= percent and +on 9aris ,lub bilateral creditors )4.9 percent., while the remainder 8.& percent is owed to other creditors. The principal multilateral creditors are the :orld 3an$ )(1 percent., the 1/> )1= percent. and the Arab ban$ for Fconomic -evelopment in Africa )3A-FA., 79F, >und, ,ommonwealth -evelopment ,orporation ),-,. and the 1slamic -evelopment 3an$ )1-3.. 3ilateral debt is largely owed to 1& countries, of which eleven are 9aris ,lub creditors. 1t is evident that the structure of e!ternal debt significantly changed from pre "A9 period of 1990 and during "A9. /ultilateral creditors, which accounted for a mere 18.= percent of the total debt stoc$ in 1990 increased to 49.8 in (008. This change reflected the deliberate effort during "A9 as government borrowings were closely guided by debt policies wherein loans were contracted from multilateral financial institutions on concessional terms with long maturity periods. This is clearly reflected in >igure 1 and ( which show percentage distribution of e!ternal debt in both 1990 and (008. /oreover, the trend in e!ternal debt is depicted in figure & from 19%0 (008.

((

>igure 1G 9ie charts showing percentage distribution of e!ternal debt by creditor type in (008.

>igure (G 9ie charts showing percentage distribution of e!ternal debt by creditor type in 1990. 9ey. /* /ultilateral -ebt 31 < 3ilateral -ebt 9E < 9rivate -ebt "T < "hort Term -ebt Source. 3an$ of "ierra ;eone (008 (&

($gure :. E,terna" De!t trend; -7845344< "ourceG :orld 3an$, :orld -ebt Tables, Iarious 1ssues >igure & depicts a plot of e!ternal debt for the period 19%0 to (008. The figure shows that e!ternal debt was low in the first decade after independence. 0owever, after the oil price shoc$ in 19%9, the economy started e!periencing high e!ternal debt. This continued in the ne!t two decades where, during the civil war, the economy recorded the highest e!ternal debt throughout the review period )i.e. the highest e!ternal debt was recorded in 199(.. This could be attributed to factors including the decline in prices of "ierra ;eone?s main e!ports, particularly iron ore and cocoa )7sei, 1994..

Domest$c de!t s$tuat$on s$nce -784 The persistent and rising fiscal deficits which the economy has been faced with since the 19%0s has , to a large e!tent, determined the level of government?s indebtedness to the domestic economy as well as money creation. This is largely due to e!pansionary fiscal policy stance, the delay and sometimes postponement ('

of donor disbursement and the difficulty of mobili6ing domestic resources owing to wea$ ta! administration capacity. At the same time, the rebel war e!acerbated the situation causing shrin$age in government?s revenue base such as custom revenue, cash crop and e!port earnings in the mining sector. As a result, the level of domestic financing, which includes central ban$ borrowing, has closely followed the trend in budget deficits over the review period. "ierra ;eone, li$e most developing countries, uses a limited number of debt instruments to borrow from the domestic economy. These include, Treasury bills )T3., Treasury 3earer 3onds )T33. and sometimes government stoc$s. Treasury 3ills are short term domestic debt instruments with a 90 days maturity period, while Treasury 3earer 3onds are debt instruments of up to one year. As part of the financial sector reforms under the "A9, interest rates on these instruments were liberalised and mar$et determined in the 1990s. 9rior to this, interest rate was strictly determined by the ,entral 3an$, which often resulted to fi!ed and negative real interest rates. The Treasury 3ills mar$et has also undergone significant development since the deepening of the reform program in 199(. These include the introduction of wee$ly auction at the 3an$ of "ierra ;eone and the unification of the ban$ and non ban$ing mar$et for Treasury 3ills in Danuary and August 199( respectively. ,onsequently, the direct sale of T3 to the public was stopped in the same year. The 3an$ channels all private sector participation in the Treasury 3ills mar$et through commercial ban$s.

(4

1n August 199&, a one year treasury bearer bond was introduced as a means of enhancing open mar$et operations and increasing the variety of debt instruments or savings instruments available to the public. A clearing house mechanism for the buying and selling of government debt instrument was also introduced in 199' in order to provide the basis for a secondary inter ban$ money mar$et. Total outstanding domestic debt including short term advances to government as at end -ecember (008 stood at ;e (&= billion. Total debt held in the form of treasury bills amounted to ;e 100 billion. 7f this ;e 18 billion is held by 3an$ of "ierra ;eone, ;e %'.8 billion by the commercial ban$s and ;e% billion by the non ban$ public. Treasury 3earer 3ond held for the same period amounted to ;e&0.8 billion. 7f this, ;e1 billion is held by 3an$ of "ierra ;eone, ;e4.8 billion by commercial ban$s and ;e(' billion by the non ban$ public. 1n many cases, the total domestic debt held in the form of treasury bills and bearer bonds fall far short of the domestic financing requirement of the budget deficit. 0ence, government often resorts to borrowing from the ,entral 3an$ resulting in a large stoc$ of :ays and /eans advances to the tune of ;e104.14 billion as at end -ecember, (008. Addressing the arrears problem does not mean the end of the debt problem. 1t is interesting to note that with the regulari6ation of the country?s relationships with creditors, ways for new loans to be made available are now open to the country. Although the arrears were almost eliminated between 1989 and 1999 yet the total debt by the end of (008 had risen to about *"@=(0.(

(=

million with loans from the :orld 3an$ of *"@(=0 million, 1/> of *"@1%1 million and other multilateral organisation of *"@(08 million. 1t must be noted that most of these loans were contracted on concessional basis. 0owever, despite the concessionality of the loans, "ierra ;eone still has an obligation to pay on all its loans. Theoret$ca" "$terature re%$e* Theoretical wor$ on economic growth dates bac$ many centuries ago. Although classical economists li$e "mith )1%%=., /althus )1%98. and Eicardo )181%. did not formally study economic growth, however, they outlined the basic ingredients of economic growth. Eecent neoclassical economists regarded as pioneers of economic growth analysed economic growth with rigorous models. They include Joung )19(8., Eamsey )19(8., "chumpeter )19&'., 0arrod )19&9., 5night )19''., and -omar )19'=.. The main contributors to neoclassical growth models are Eamsey )19(8., "olow )194=., and "wan )194=., while Eomer )198=. and ;ucas )1988. initiated economic growth theories that are $nown as endogenous growth theories. Fconomic 8rowth is that part of economic theory that e!plains the rate at which a country?s economy grows over time. 1t is usually measured as the annual percentage rate of growth of the country?s ma#or national income accounting aggregates, such as the 8ross +ational 9roduct )8+9. or the 8ross -omestic 9roduct )8-9. with appropriate statistical ad#ustments to discount the potentially misleading effects of price inflation )Dohnson, (000.. Fvery country?s economy will show si6able year to year and quarter to quarter fluctuations in its economic

(%

growth rate, however, theorists tend to concentrate their efforts on analysing and e!plaining the variations in the longer term trend or average rate of economic growth over longer periods such as a decade or more. "pecialists in 3usiness cycle theory are often left with the e!planation of the shorter term fluctuations around the longer term trend because investigation has shown that the predominant influences on short term growth rates seem to differ in important ways from the determinant of an economy?s long term average growth performance. The Harrod5Domar =ro*th >ode" The 0arrod -omar growth model shows through a mathematical equation, the e!istence of a direct relationship between savings and the rate of economic growth, the indirect relationship between capital and economic growth. The model, which attempts to integrate 5eynesian analysis with elements of economic growth, assumes that economic growth is a direct result of capital accumulation in the form of savings. The 0arrod -omar growth model has been used by development economists to estimate the financing gap of a developing economy. Assuming that there is abundant supply of labour, they claim that scarcity of capital is the only constraint to production )Fffendi, (001.. The model itself predicts that growth will be proportional to the rate of investment. That is, growth will be equal to investment divided by the 1ncremental ,apital 7utput Eatio. 1f a target growth rate is set, the required investment to meet this target can be estimated by multiplying the target with the incremental capital output ratio.

(8

The financing gap is thus, the gap between the available financing for investment and the required investment. 3y filling this gap with imported foreign capital such as foreign debt or aid, developing countries will get the required investment that will yield the target growth rate. Fmpirical evidence does not, however, support this theory because the massive e!ternal debt that has accumulated in developing countries since the 19=0s was not accompanied by an increase in per capita income. 7n theoretical grounds, the 0arrod -omar growth model suffers a shortcoming, which is that the equilibrium is unstable since it requires the equali6ation of warranted and natural growth rates and uses production functions with little suitability among the inputs. The production function used in this model is of ;eontief type with fi!ed proportion of inputs. These wea$nesses have made economists to pay less attention to the 0arrod -omar growth model in favour of growth models that are less rigid and empirically more applicable. -espite these flaws, the 0arrod -omar growth model was very popular after the great depression until the early 1940s and has generated many papers in the field of economic growth.

The Neo5c"ass$ca" =ro*th >ode" The shortcomings of the 0arrod -omar growth model was solved by the growth model of "olow )194=. and "wan )194=., which is referred to as the neoclassical growth model. They used production functions that e!hibit constant returns to scale, diminishing returns to each input, and positive substitutability to

(9

inputs. The production function is assumed to be a function of capital, labour and technology. 3y assuming a constant rate of saving the model predicts that growth in the long run is a function of only technical change and not of saving or investment. "aving will have an effect on the level of income but not on its growth rate. This prediction implies that in the absence of continuous improvement in technology, per capita growth will eventually cease. +eoclassical growth theory suggests that two countries with the same technology will have the same steady state growth rate of per capita income. This implies that a country with lower capital labour ratio will have a higher per capita growth rate than a country with a high capital labour ratio. This phenomenon is called an absolute convergence. ,onditional convergence, on the other hand, is a situation in which a country will have a higher growth rate if that particular country has a lower starting level of real per capita income relative to the long run or steady state position. This conditional convergence prediction that comes from the "olow "wan )neoclassical. model has been e!ploited e!tensively as an empirical hypothesis with mi!ed results. >or e!ample, 3arro and "ala 1 /artin )199(. observe convergence among the '8 states of the *" in terms of the growth rates of their per capita income and per capita gross state product, while Eomer )198=. and ;ucas )1988. cite the lac$ of convergence of the growth rates of different countries. 3oth 3arro and "ala 1 /artin argue, however, that conditional convergence is more li$ely if regions or countries are more homogenous as in the case of the '8 states in the *".

&0

Another important neoclassical growth theory, beside the "olo "wan model, is the Eamsey model. The Eamsey model is a refinement of the "olow "wan model. ,hronologically, Eamsey?s growth theory was developed before the "olow "wan?s. 1n the literature, however, his model is usually put after the "olow "wan?s. 7ne of the $ey features in his model is the assumption that households optimise their utility over time. This assumption essentially ma$es the model dynamic. *sing Eamsey?s model as their starting point, ,ass )19=4. and 5oopmans )19=4. recast the saving rate that is e!ogenous under "olow "wan model as endogenous. Fven though this is considered a refinement of the neoclassical growth model, it does not eliminate the dependence of the long run growth rate on e!ogenous technological progress. The wor$s of ,ass and 5oopmans mar$ the end of the basic neoclassical growth era.

Endogenous =ro*th >ode"s 1n recent years, growth theory has received renewed attention with the rise of a new line of research that e!plains growth rates endogenously. 1t was first started by the wor$ of Eomer )198=. and ;ucas )1988.. *nli$e the "olow "wan model in which the long run growth rate is determined by an e!ogenous technological progress, Eomer and ;ucas suggest a model that endogenises the growth rate of the economies. /odels that have endogenous source of per capita economic growth are called endogenous growth models. Fndogenous growth models are similar to neoclassical models but they differ considerably in their underlying assumptions and suggested conclusions. Three distinctions between

&1

the two are obviousG >irst, the neoclassical assumption of diminishing marginal returns to capital is discarded. "econd, the model also envisages increasing returns to scale in aggregate production, and third, the model recogni6es the role of e!ternalities in determining the rate of return on capital. Fndogenous growth theory can be e!pressed in a simple equation Y = AK . :here A can be interpreted as representing any factor that affects technology, while K represents both human and physical capital. +otice here that there are no diminishing returns to capital, a feature that can be achieved by invo$ing some e!ternality that offsets any propensity to diminishing returns. 1nvestment, whether physical investment by a firm or human capital investment by an individual, leads to an increase in productivity that e!ceeds the private gain. This model leaves open the possibility that an increase in the investment rate, either in physical or human capital, can lead to sustained growth if strong e!ternal economies are generated by investment itself so that in the neoclassical model becomes unity. Fndogenous growth theory can help e!plain anomalous international flows of capital that e!acerbate wealth disparities between developed and developing countries. -eveloping countries potentially have high rate of return on investment due to the law of diminishing returns. 0owever, this is eroded by lower levels of complementary investments in human capital, infrastructure, or research and development. As a result, developing countries benefit less from broader social gains associated with each of these alternative forms of capital e!penditure. Fndogenous growth models also suggest an active role for public policy in

&(

promoting economic development through direct and indirect investment in human capital formation and foreign private investment. :ithin this new growth theory, one approach posits that the government supplies productive services that increase the marginal product of private capital and, thus, positively influence economic growth. 1f the production function is homogenously linear in private and public capital simultaneously, and government e!penditures are endogenous, this approach yields an endogenous rate of growth. Arrow and 5ur6 )19%0. were the first to introduce productive public capital in growth models, but in their model growth is still determined by e!ogenous factors. 3arrow )1990. used this approach to present a model with endogenously determined growth rates and a balanced budget. 0owever, he assumed that government spending as a flow variable enters the macroeconomic production function, whereas Arrow and 5ur6 suppose that the stoc$ of public capital shows productive effects. Two important shortcomings of the endogenous growth models are that the models do not predict convergence either in absolute or conditional and that it has only limited empirical support. >ocusing on the ob#ective of our study, the theoretical literature on the relationship between e!ternal debt and economic growth tends to point to a negative relationship. 8rowth models augmented with public agents issuing debt to finance consumption or capital goods tend to e!hibit a negative relationship between e!ternal debt and economic growth, particularly in a neoclassical setting.

&&

/odigliani )19=1., refining contributions by 3uchanan )1948. and /eade )1948., argued that the national debt is a burden for ne!t generations, which comes in the form of a reduced flow of income from a lower stoc$ of private capital. Apart from a direct crowding out effect, he also pointed out to the impact on long term interest rates, possibly in a non linear form Aif the government operation is of si6able proportions it may significantly drive up )long term. interest rates since the reduction of private capital will tend to increase its marginal productB )p. %&9.. Fven when the national debt is generated as a counter cyclical measure and Ain spite of the easiest possible monetary policy with the whole structure of interest rates reduced to its lowest feasible levelB )p. %4&., the debt increase will generally not be costless for future generations despite being advantageous to the current generation. /odigliani considered that a situation in which the gross burden of national debt may be offset in part or in total is when debt finances government e!penditure that could contribute to the real income of future generations, such as productive public capital formation. -iamond )19=4. adds the effect of ta!es on the capital stoc$ and differentiates between e!ternal and internal debt. 0e concludes that, through the impact of ta!es needed to finance the interest payments, both types of public debt reduce the available lifetime consumption of ta!payers, as well as their saving, and thus the capital stoc$. 1n addition, he contends that internal debt can produce a further reduction in the capital stoc$ arising from the substitution of government debt for physical capital in individual portfolios.

&'

The channels through which e!ternal debt can potentially affect economic growth are diverse. /eade )1948. was drawing attention to the fact that the removal of the Adeadweight debtB wouldG )i. raise the incentive of households to save )the 9igou effect.2 )ii. improve the incentives for wor$ and enterprise2 )iii. possibly allow for a decrease in income ta!ation at a later stage as a result of saving interest payments on the budget )improving even more the incentives for wor$ and enterprise.. Theoretical literature of recent times on open economies also shows that reasonable levels of e!ternal borrowing by a developing country can enhance its economic growth through capital accumulation and productivity growth. This in turn, leads to faster convergence of per capita income between nations )3arro et al. 19942 9attillo, 9oirson and Eicci, (00'.. ,ountries at their early stages of development have small stoc$s of capital and are li$ely to have investment opportunities with higher rates of return. 0owever, due to the low volume of savings in developing countries, which reflects their low incomes, domestic savings should be supplemented by foreign resources. As long as these countries use the borrowed funds for productive investment and do not succumb to macroeconomic instability and policies that distort economic incentives, or si6able adverse economic shoc$s, growth should increase and allow for timely debt repayments )9attillo, 9oirson and Eicci, (00'.. +otwithstanding the above, economists and policy ma$ers have had doubts about the beneficial impacts of foreign resources on the economies of developing countries, especially since the 19%0s. +either the presumed positive

&4

impact of foreign resources on the volume of savings, nor the presumed growth is supported by empirical evidence. 1n "ub "aharan Africa, recurrent economic crises have led to poor growth performance and most of these countries continue to e!perience difficulties in servicing their e!ternal debt ),laessens, 199=.. Although the literature on debt hardly e!plains why these countries have not been able to use the borrowed funds to generate sufficient output, critics of foreign aid in general maintain that it does more harm than good. 0owever, it is often argued that an increase in a country?s e!ternal debt beyond a critical level can lead to unsustainable debt and insolvency of the country. The debtor country may lose its creditworthiness, leading to a sudden reversal of capital flows. This in turn may lead to a currency and financial crisis and a large output loss. /oreover, debt cycles e!ist in international borrowing and lending )+ishimura and 7hyama, 1994.. >urthermore, the e!ternal debt and growth literature suggests that there is a unidirectional causal lin$ between indebtedness and growth ),houdhry and Arvin, 1998.. -e 9inies )1989. focuses on debt sustainability and ad#ustment, and argues that as the international credit mar$et imposes severe credit rationing on the heavily indebted ;ow 1ncome ,ountries )01;1,s., they were in effect imposing e!cessive import restraints, which is one of the ma#or obstacles to a debtor country?s growth prospects. "elows$y and Ta$ )198=. indicate that since heavily indebted low income countries have been servicing their debt by squee6ing imports and investments to generate trade and equivalent saving

&=

surpluses, they have also been paying a high price in terms of forgone consumption and output growth. "ince these countries cannot be e!pected to sustain such policies for very long, "elows$y and Ta$ advocate a growth oriented debt policy wherein creditors provide new financing, at least until the structural ad#ustment programmes ta$e effect. Eoubini )1984., "achs )1984., 5aminarides and Talbert )1989., and 5aminarides and +issan )199&. all e!amined the consequences of debt within a multi country framewor$. Their studies suggest that at the aggregate level, debt adversely affect economic growth and the causality is a uni directional one, that is, growth is adversely affected by a higher debt burden. 1t is shown in various theoretical models that reasonable levels of current debt inflows are e!pected to have a positive effect on growth. 1n traditional neoclassical models, which allows for capital mobility, and the possibility to borrow and lend from foreign sources, debt increases transitional growth. There is an incentive for capital scarce countries to borrow and invest since the marginal product of capital is above the world interest rate )9attillo, 9oirson and Eicci, (00(.. "ome endogenous growth models have similar implications. >or instance, Faton )199&. e!tends the *6awa ;ucas models and shows that an increase in the cost of foreign capital that lowers e!ternal borrowing leads to lower long run growth. 3arro and "ala i /artin, )1994.2 3arro, /an$iw, and "ala i /artin, )1994., assert that models with perfect international capital mobility are based on unrealistic assumptions and have empirical implications that are counterfactual.

&%

They argue that a more realistic assumption is that countries may not be able to borrow freely because of the ris$ of debt repudiation or moral ha6ardsG 1f the borrowing country can hide its actions from the lender it may choose to consume or reinvest abroad some of the borrowed funds )8ertler and Eogoff, 1990.. 7ne important question on debt growth theory that would come to mind is that, why would large levels of accumulated debt lead to lower growthK A well $nown e!planation comes from the Adebt overhang hypothesisB, which state that e!ternal debt burden provides a disincentive to domestic investment in developing countries and thus slows economic growth since any additional foreign e!change earnings would have to be stashed away by foreign creditors. 5rugman )1988. defines Adebt overhang as a situation in which the e!pected repayment on e!ternal debt falls short of the contractual value of debt.B 1f the country?s debt level is e!pected to e!ceed the country?s repayment ability, the e!pected costs of debt service will discourage further domestic and foreign investment, which consequently result to lower economic growth )5rugman, 19882 "achs, 1989.. 1n 5rugman?s specification, the e!ternal debt overhang affects economic growth through private investment, as both domestic and foreign investors are deterred from supplying further capital. 7ther channels may be total factor productivity, as proposed in 9atillo et al. )(00'., or increased uncertainty about future policy decisions, with a negative impact on investment and further on growth, as in Agenor and /ontiel )199=. and in line with the literature of partly irreversible decision ma$ing under uncertainty )-i!it and 9indyc$ 199'..

&8

3ecause of the fear that the domestic economy will be effectively Ata!edB the higher the volume of output that is produced to service the e!ternal debt, potential domestic and foreign investors will be unwilling to increase their today?s investment e!penditures for the sa$e of future increased output, and thus, economic growth is discouraged. The debt overhang theory spins around the adverse effects of e!ternal debt on investment in physical capital. 0owever, the theory has a much broader scope. A high level of e!ternal debt can also reduce the government?s incentive to carry out structural and fiscal reforms, since any strengthening of fiscal position could intensify pressure to repay foreign creditors ),lements, 3hattacharya, and +guyen, (00&.. Adam and 3evan )(004. find interaction effects between deficits and debt stoc$s, with high debt stoc$s e!acerbating the adverse consequences of high deficits. 1n a simple theoretical model integrating the government budget constraint and debt financing, they find that an increase in productive government e!penditure, financed out of a rise in the ta! rate, will be growth enhancing only if the level of )domestic. public debt is sufficiently low. >ischer )199&., while e!plaining the deficit debt growth relationship posited that larger budget surpluses are associated with more rapid growth through greater capital accumulation and greater productivity growth. 0e posited further that, high deficit may be consistent with low inflation for a while, but that a more detailed assessment of debt dynamics may be needed to see if the deficit is sustainable and therefore consistent with macroeconomic stability.

&9

Em $r$ca" "$terature re%$e* The empirical evidence on the relationship between debt and growth is scarce and primarily focused on the role of e!ternal debt in developing countries. Among more recent studies, several find support for a non linear impact of e!ternal debt on growth, with deleterious effects only after a certain debt to 8-9 ratio threshold. "avvides )199(. asserts that if a debtor country is unable to pay its e!ternal debt, debt payments become lin$ed to the country?s economic performance. The country benefits only partially from an increase in output or e!ports because a fraction of the increase is used to service the debt and accrues to the creditors. Thus, from the perspective of the debtor country as a whole, the debt overhang acts li$e a high marginal ta! rate on the country, thus lowering the return to investment and providing a disincentive to domestic capital formation. The disincentive effect of the debt overhang may have repercussions on private saving and investment, even when all e!ternal debt is held by the government. The government has little incentive to institute policies to promote domestic capital formation or to reduce current consumption in e!change for higher future economic growth when the benefits from such policies go to creditors in the form of higher debt payments. The author estimated models by using Two "tage ;imited -ependent Iariable model )(";-I. procedure by cross section time series data from '& less developed countries for the period of 1980 198=. "avvides )199(. also tried to test the significance of several factors affecting investment rates and the li$elihood of less developed countries );-,s.

'0

encountering debt servicing problems. This study concludes that debt overhang and decreasing in foreign capital inflows have a significant negative effect in investment rates. :hen capital inflows are divided into commercial and non commercial flows, the conclusion differs. A decrease in commercial inflow is a significant factor in investment rates. 7n the other hand, non commercial inflow is a not a determinant factor in lessening investment rate. This result is also consistent with the 1/>?s )1989. conclusion. 1t concluded that debt overhang is a significant factor influencing slowdown in investment. 1t is argued that the reasons behind borrowing during the mid 19%0Ls and early 1980Ls varied from country to country. 5ruger )198%. states that after the rise in oil prices, the oil importing developing countries faced large current account deficits. 7n the other hand, oil e!porters had large current account surpluses, which they lent to the commercial ban$s, which in turn financed the deficits of oil importing countries, thus the surpluses of the oil e!porting countries were used by oil importing developing countries. 3auerfreund )1989. used a computable general equilibrium model to measure the cost of e!ternal debt to the Tur$ish economy. 0e e!plains the issue of the debt overhang, using a multi sector, non linear general equilibrium model. The approach ta$en to measure the debt overhang is to compare the growth rate of the Tur$ish economy following hypothetical debt forgiveness. The year 1984 was selected as the base year for measuring the debt overhang because the Tur$ish economy in 1984 was classified as being representative of the post 1980 Tur$ish liberalisation era.

'1

-eshpande )199%. attempted to e!plain the debt overhang hypothesis by an empirical e!amination of the investment e!perience of 1& severely indebted countries. The severely indebted countries are Algeria, Argentina, 1vory ,ost, Fgypt, 0onduras, 5enya, /e!ico, /orocco, 9eru, 9hilippines, "ierra ;eone, Iene6uela and Mambia. The author e!plains that debt overhang, which in contrast to the normal debt obligations is the actual amount of paid debt service is determined by creditors and debtor countries. 0ence, any increase in production and e!ports are used for debt payment to creditors. As a consequence, this gives a disincentive to investors. 1nvestors are not willing to invest a large amount of money. The author argues that the ad#ustment measures, which are applied by severely indebted countries, have an impact on the indebted countries, since the investment crisis has typically implied a growth crisis for the highly indebted countries. This has further worsens severely indebted countries? debt service capacity. The period for evaluation is 19%1 to 1991. The author uses two period, the first period is 19%4 198& and the second period is 198' 1991 with 7;" estimation for panel data. >or a variety of reasons, e!ternal debt is found to e!ercise a negative effect on the investment. >irstly, for the period of 19%1 1991, the investment ratio for the sample countries shows a rising and then a declining tendency at the end of the eighties. The relationship between e!ternal debt and investment is negative. /oreover, the first period has a positive influence on investment, in the second half of this period, time effects turn to negative. ,ohen )199&. estimated an investment equation for a sample of 81 developing countries over three sub periods using 7.;." method. The author

'(

shows that the level of debt does not e!plain the slowdown of investment in highly rescheduling developing countries. :arner )199(. tried to measure the si6e of debt crisis effect on investment with ;east "quare estimation for 1& less developed countries over the period 198( 1989. 0e affirmed that the reasons behind the decline of investment in many heavily indebted countries are declining e!port prices, high world interest rates and sluggish growth in developed countries. Eoc$erbie )199'. critici6ed :arner )199(. of various shortcomings. Eoc$erbie )199'. used 7.;." for each of the 1& countries over a sample period 19=4 1990 and the results affirm that the debt crisis of 198( had significant effects in terms of dramatic slowdown of domestic investment in less developed countries. Af!entiou and "erietis )199=. in furtherance to Af!entiou )199&. e!amined 44 developing countries facing debt service difficulties. The study?s ob#ective was to find out the relationship between foreign borrowing and productivity over the period 10%0 1990. The results show that during the period 19%0 1980, the relationship between indebtedness and national productivity is not negative. They submitted that the developing countries used the foreign loans to absorb the shoc$ from oil price increase as painless as possible. 0owever, for the period 1980 1990 when the debt forgiveness and rescheduling started, the debt crisis and debt overhang affected some indebted countries economic growth. 3arfour 7sei )1994., in his study on 8hana, argued that debt repayment inevitably imposes constraints on a debtor countryLs growth prospective since it

'&

involves the transfer of resources to other countries. Therefore, in order to adequately appreciate the problem of indebtedness, it is essential to relate the debt with its repayments of some income resources generated by the debtor out of which the repayments could be made. 0e noted that movements in the ratio of debt service payment to e!port of goods and services )debt service ratio. and total e!ternal debt to income )8+9. are the two most important inde! used to assess the debt burden2 the higher the ratios, the greater the burden. /#ema )199=. attempts to analy6e the impact of foreign debt in Tan6ania by using a simultaneous equation model, ta$ing in to account the direct as well as the feedbac$ effect of debt. 0e found that the impact of the debt service ratio on real growth in 8-9 is negative. "imilarly 7caya )1999. argued that, the debt variable has negative significant impact on percapita income for both highly indebted and less indebted countries. 1yoha )1999., using a simultaneous equation and simulation method on sub "aharan African countries found that there is a debt overhang and crowding out effect in "ub "aharan African countries because of the e!cessively high stoc$ of e!ternal debt that depress investment and lowers the rate of economic growth. Therefore, the debt servicing capacity is lin$ed to economic growth as we have seen in the case of 8hana and others. :hen the e!ternal debt is great more resources will be committed to the payment of the e!ternal debt leading to a lea$age from the economy hence slowing down economic growth. 1n the case of +igeria we also learn that accumulation has a negative effect on investment and growth.

''

Eegarding the effect of e!ternal factors on private investment it can be viewed from two anglesG >irst, a raise in international interest rates on debt will increase the burden of debt and thus, reduce the import capacity, which may have a direct negative effect on the level of private investment )Alemayehu, 199%.. "econd, the domestic private sector itself holds foreign assets and if the burden of future debt is viewed as a heavy ta! burden, this can lead to capital flight. /any studies relating to the above mentioned issues have been carried out. 1n the case of "ierra ;eone, Dibao )199(. used a simultaneous equation method to test the debt overhang effect on investment. 1n his conclusion, the debt overhang hypothesis does not seem to hold in "ierra ;eonean case. *sing similar methodology, "esay )1998., in his masterLs thesis concluded that the effect of long term foreign borrowing on growth had had a mi!ed result. These mi!ed results can be attributed to a number of factors. >irst, it is not the volume of e!ternal capital that matters but rather how it was used. 0e indicated that between the period of 19=' to 19%% and between the period of 199& to 1994 e!ternal capitals contributed positively to the growth of "ierra ;eonean economy while between the period 199= to (004 the contribution was negative. 0e #ustified therefore during the civil conflict era a large proportion of foreign capital was used in the procurement of arms. Thus, from this he concluded that the debt crisis is more of a policy problem. This is in close lin$ with A#ayLs findings that +igeria had implemented policies that led to the accumulation of debt in e!cess of what was sustainable.

'4

:alelegne and Asafa )(001. using Fngel 8ranger F,/ regression found out that there is no evidence of debt overhang effect and a crowding out effect of debt on investment in Fthiopia. The limitation of this approach is that since there is controversy about the cause and effect relationship between output equation, investment equation and import demand equation they used, using simultaneous equation may be more appropriate had it been used than using single equation. Af!entiou and "erletis )199=. e!amined 44 developing countries facing debt service difficulties. The main ob#ective of this study was to find out the statistical relationship between foreign borrowing and productivity. -ata coverage includes 19%0 1990 period. The period of this study )19%0 1990. has two sub periods. The first sub period is from 19%0 to 1980 and is characterised by the rapid increase of foreign debt2 the second sub period is from 1981 to 1990, and is characterised by the years of debt servicing problems and debt overhang. The effects of debt are e!amined for all four categories of developing countries and for each sub period as well as for the entire period. The 44 developing countries were classified into ' categories, 19 of them are classified as indebted low income countries, 1( severely indebted middle income countries, 10 moderately indebted low income countries and 1' moderately indebted middle income countries. 1n this study, each of the four classifications is treated as a separate specific case and the effect of si! debt indicators on the growth of its per capita income is investigated. The results show that during the period of 19%0 1980, the relationship between indebtedness and national productivity is not negative. -eveloping

'=

countries used the foreign loans to ta$e time and absorb the shoc$ from oil price increases as painlessly as possible. These findings were proved by all four groups of developing countries. 7n the other hand, the debt forgiveness and rescheduling began during the 1980 1990 sub period. >or this period, the debt crisis too$ place and the debt overhang affected some indebted countries? economic growth. The results reveal that there was a negative relationship between indebtedness and national productivity for two groups of the severely indebted developing countries. "everely indebted developing countries used their foreign loans improperly. Therefore, they faced debt service difficulties when they were required to pay their debt obligations. 1n this period, developing countries failed to meet their debt payments, as they had both resource wasting and failure to improve their foreign e!change earnings. 8eiger )1990. e!amined the relationship between 8+9 growth rate and debt burden. The debt burden represents debt service ratio )the sum of interest payments and repayments of principal on e!ternal debt to e!ports of goods and services., the ratio of debt service to 8-9, and the ratio of net transfers to 8-9, and the ratio of net transfers to 8-9 in highly indebted countries in south America where the problem of debt is serious. This study focuses on the specific country to determine the impact of debt burden and capital inflows on the economic growth. The ratio of net transfers to 8-9, debt service to 8-9, and debt service to e!ports were regressed on real 8-9 growth rate over a 1& year period, from 19%' to 198=, for nine highly indebted "outh America countries. The

'%

countries included were Argentina, ,hile, 3ra6il, 9eru, ,olombia, Fcuador, 9araguay, 3olivia and Iene6uela. The results of this study confirm that there is a statistically significant inverse relationship between debt and economic growth. >urthermore, intra country analysis shows that the marginal effects of the debt burden on the economy decrease when the debt burden increases. Fven though there is an important variation in the model from country to country, many different factors affect economic development in each of the countries and there are also different reactions to the debt burden. >or all countries e!amined, the lagged model is the most highly correlated. 7n the other hand, the burden of the principal and interest payments has a greater impact on the economy in the following year rather than in the current year. 1t is also not surprising that the lagged equations model results have more statistical significance than the linear equations. >osu )199=. tested the relationship between economic growth and e!ternal debt with an empirical study for the sample of sub "aharan African countries over the 19%0 198= period by employing the 7;" method. This study e!amined to which degree debt had a negative impact on economic growth of sub "aharan African )""A. countries. This study estimates the direct effect of debt hypothesis and indirect debt hypothesis. The direct effect of debt hypothesis proposed that if debt service payments do not decrease investment and saving levels considerably, the debt negatively affects growth directly by reducing productivity. 1t is also argued that the direct effect of debt hypothesis suggests that both debt service payments and debt outstanding may affect 8-9 growth rate negatively even if

'8

debt outstanding and debt service payments do not affect investment levels. The results show that by using a debt burden measure, direct effect of debt hypothesis reveals that 8-9 growth is negatively influenced via a diminishing marginal productivity of capital. The findings of this study also show that on average a high debt country faces about one percentage reductions in 8-9 growth rate annually. This e!plains one third of all reduction of growth rate in sample countries. 7n the other hand, the results do not support the adverse indirect effect of debt indirect effect of debt hypothesis states that the relationship between debt and economic growth is indirect, via reduced. >osu )1999. has also employed an augmented production function to investigate the impact of e!ternal debt on economic growth in sub "aharan Africa for the 1980 1990 period. The author has used tests to measure the direct effect of debt overhang hypothesis namely that e!ternal debt negatively affect economic growth even if it has little or no effect on the level of investment. The findings show that as debt variable is included, measured over 1980 1990, in the equation, debt e!hibits a negative coefficient. /oreover the author questions the negative association between e!ternal debt and economic growth. 1t is claimed that sub "aharan countries entered into structural ad#ustment programmes in the mid 1980s. The negative relationship between economic growth and debt might be due to a poor performer receiving large e!ternal debt. 0ence, the growth equation is re estimated for only the first half of the decade from 1980 to 1984. The results are reassuring in that the coefficient of debt is still negative and significant.

'9

/eanwhile, additional evidence shows a rather wea$ negative impact of debt on investment levels. ,unningham )199&. e!amined the association between debt burden and economic growth for 1= heavily indebted nations during the period 19%1 to 198%. 1t is predicted that the growth of a nation?s debt burden has a negative effect on economic growth because of the impact on the productivity of labour and capital. As a nation has a significant debt burden, the debt burden needs to be serviced. This will influence how capital and labour will be used in production. This study concludes that the growth of a nation?s debt burden had negative effect on economic growth during the period of 19%1 19%9. 7n the other hand, the results for the 1980 198% period offer little support for the inclusion of the growth of debt burden in the economic growth model. "awada )199'. investigated whether the heavily indebted countries )01,s., concerned with their e!ternal debt repayments, stay solvent. A direct test of the solvency condition derived from the usual in temporal budget constraints shed light on the sustainability of their current policies. This study employed annual time series data for sample period from 1944 1990 and estimated the cointegration regression using the 7;" method. The findings of this study show that heavily indebted countries )01,s. have debt overhang problems. "ince their current e!ternal debts are above the e!pected present value of the future gains. The 1/> )1989. reveals that debt overhang e!ists in the problem debtor countries in 1980s. These countries include Argentina, 3olivia, 3ra6il, ,hile, ,olombia, Fcuador, 1vory ,oast, /e!ico, /orocco, +igeria, 9eru, 9hilippines,

40

*ruguay, Iene6uela and Jugoslavia. There are two pieces of evidence supporting the debt overhang proposition. >irst, the saving ratio decreased when e!ternal finance dried up. "econdly, in a comparison of a group of countries with debt problem with a group of other heavily indebted countries, which did not e!perience a debt servicing problem, saving ratios decreased in the former group. There was an important drop in saving ratios and investment ratios in problem debtor countries during 1980s. According to A#ayi and 5han )(000., sustainable foreign borrowing is measured by several ratios, such as debt to e!port, debt service to e!port, debt to 8-9 )or 8+9., and e!ternal debt to 8ross +ational 1ncome among others. 0owever, the determination of the sustainable level of these ratios is indeterminable and their usefulness is reduced to a warning of potential e!plosive growth in the stoc$ of foreign debt. >or instance, if the acquisition of additional foreign debt increases the debt servicing burden more than it increases the country?s capacity to bear the burden, such an acquisition becomes undesirable and the situation must be reversed through e!port e!pansion. 1f e!port is not e!panded, more borrowing will be necessitated for servicing debt and e!ternal debt will pile up above the country?s capacity to bear. 0ameed, Ashraf and ,haudhary )(008. analy6ed the long run and short run relationships between e!ternal debt and economic growth of 9a$istan. They have e!amined the dynamic effect of 8-9, debt service, capital stoc$ and labor force on the economic growth by fitting the production function using annual data for the entire period of 19%0 (00&. The basic model is derived from the

41

neoclassical production function by incorporating the e!ternal debt service variable as suggested by ,unningham )199&.. The results show that debt servicing has a negative effect on the productivity of labor and capital2 and debt service ratio tends to affect negatively 8-9 and thereby the rate of economic growth in the long run, which in turn, reduces the ability of the country to service its debt. The estimated error correction term shows the e!istence of a significant long run causal relationship among the specified variables. :hile in the short run, unidirectional causality is reported from debt service to 8-9. These suggested that debt as an important factor in overall debt scenario in 9a$istan. 9erasso )199(. has tried to compare the impact of debt servicing obligations and domestic policies on domestic investment in highly indebted less developed countries );-,s. for the 198( 1989 periods. -ata are obtained from (0 middle income severely indebted countries. The results of this empirical study show that appropriate domestic policies have a stronger impact on increasing investment and growth in highly indebted countries than decreasing )debt relief. debt servicing obligations, for e!ample, a 1 percent real devaluation causes up to a 0.&( percent increase in propensity to invest. 7n the other hand a 1 N reduction in interest payments causes a 0.19 percent increase in propensity to invest. Af!entiou )199&. has e!amined the negative impact of foreign indebtedness on the growth of 8+9 for twenty middle income developing countries between 19%1 and 1988. The 8ranger causality test was employed to investigate the relationship between economic growth and foreign indebtedness. "tatistical evidence shows that in seven out of twenty countries, the debt service

4(

ratio )Total debt serviceCF!ports of goods and services. seems to be as a growth suppressing factor. 1n si! out of twenty, the interest service ratio )Total interest paymentsCF!ports of goods and services. is a significant growth inhibiting factor. 1t is concluded that there is a strong debt overhang effect which too$ place in the sampled countries in the sample period of 19%1 1988. The author claims that the large debt accumulation of sampled countries was the result of bad domestic management. ,ausality test results support inferentially source mismanagement caused negative effect on 8+9. 1f foreign resources were not productively used, 8+9 growth rate would be negatively affected by indebtedness. There are several studies in the literature which test whether indebtedness impacts on the economic activity of developing countries. 1t is argued that if foreign loans are converted into capital and other necessary inputs, development will occur. 7n the other hand, if borrowing countries misallocate resources or divert them to consumption, then economic development is negatively affected )Af!entiou and "erletis, 199=a.. This study employed the framewor$ of the 8ranger causality test. 1n doing so, si! measures of indebtedness were used as pro!ies for the multiple mechanisms. The data used for the 19%0 1980 period

and 44 countries were tested in this study. All 44 countries were classified according to the :orld 3an$ debtor country classification system. The :orld 3an$ grouped these countries into ' categories, 19 of them being severely indebted low income countries )"1;1,s., twelve severely indebted middle income countries )"1/1,s., ten moderately indebted low income countries )/1;1,s. and fourteen moderately indebted middle income countries )/1/1,s.. 1n the

4&

framewor$ of 8ranger causality test, it is hypothesised that each measure of indebtedness causes the per capita income. 3roadly spea$ing, the causality test should be used, depending on the statistical behaviour of the time series. To determine which causality test were )in terms of growth and levels. carried out, the integration and cointegration properties of the data should be e!amined. >or this reason, they applied an accepted statistical procedure. The 8ranger causality test shows that there is no causal relationship between debt and income in a sample of 44 developing countries. 1t means that debt overhang is an e!aggeration. The tests findings show that indebtedness is not a specific factor of per capita income growth. 0ence, foreign resources can have a positive effect on the economic development if resources are transferred into inputs since borrowing countries need to have these scarce resources. The causality between e!ternal debt and economic growth is e!amined in many studies. Amoateng and Amoa$o Adu )199=. have investigated the relationship between e!ternal debt servicing, economic growth and e!ports for the total sample of &4 African countries. These countries were grouped into sub samples of &1 souths of the "ahara countries, (' low income African countries and 11 middle income countries. 1n this study 8ranger?s causality test is employed to analyse the interrelationship between e!ports, 8+9 growth and foreign debt servicing during 19%1 1990 for the &4 countries, using data pooled into time series and cross sectional form. The authors e!amined the #oint effect of two variables on the third variables.

4'

The empirical results declared that there is a unidirectional and positive causal relationship between foreign debt service and 8-9 growth after e!cluding e!ports revenue growth for Africa and "outh of "aharan countries during 198& 1990. The same result is obtained from middle income African countries for the 19%1 1990 period. They found that foreign debt service has a positive causal relationship with 8-9 growth. 1t is found that the impact of foreign debt service on 8-9 growth is negative and unidirectional for the period of 198& 1990. 1t means that for these countries )middle income Africa. e!ternal debt causes decrease in economic growth. :hen they applied empirical tests to the low income African countries, they found the same result for the 19%1 198( sub periods. Their findings demonstrate that there is an uni directional causality between 8-9 growth and foreign debt service after e!cluding e!ports revenue growth. This implies that foreign loans had a positive impact on the economic growth before the 198( debt crisis in these countries. The results from 198& 1990 indicate that there is a bidirectional and positive causality between foreign debt service and 8-9 growth after e!cluding e!ports revenue growth. 7n the other hand, foreign debt service is included as a third variable with a trivarite causality analysis of e!ports and economic growth for &4 African countries. The evidence shows that generally, there is a #oint feedbac$ effect between e!ports revenue, e!ternal debt service and economic growth. -ebt economic growth ne!us has also found significance among several other scholars. Fssien and 7nwioduo$it )1998. e!amine the impact of foreign

44

debt on economic growth and they found that the degree of responsiveness of growth to e!ternal finance in +igeria is elastic. 3y implication government should only put in place appropriate debt management strategies to enhance economic growth. The debt burden of a country and the consequent debt service impose a constraint on the economy in terms of insufficient foreign e!change to finance importation of raw materials and capital goods needed for economic growth. Another serious constraint is found in debt overhang theory which states that accumulated debt burden adversely affect the rate of private investment. The debt stoc$ acts as a ta! on future income and production and discourages investment by the private sector. "tudies including "ach?s )198=. have made a theoretical case for debt overhang effect by analy6ing the crowding out effect of debt on service payments. They posit that many highly indebted poor countries frequently divert foreign e!change resources to meet pressing debt service obligation. 7f interest to policy ma$ers presently is the effect of the debt relief granted some African countries and as put by 3urnside and >ani66a )(00'., it differs from previous ma#or debt relief initiatives in that it requires that budgetary resources saved from debt service be used for poverty reduction purposes. This view however need be interpreted with caution as many countries in Africa have specific country

problems which may not allow the impact of the debt reduction be felt by the common man. >or e!ample in +igeria, rather than having a positive feel of the debt relief, the standard of living of an average +igeria has worsened due to escalating prices of essential commodities and growing food shortages.

4=

,howdhury )199'. tried to resolve the controversy about the cause and effect relationship between e!ternal debt and economic slowdown. The author also tried to resolve the 3ullow and Eogof?s )1990. proposition. They argue that the e!ternal debts of developing countries are a symptom rather than a cause of economic slowdown. F!ternal debt leads to bad management in highly indebted countries, such as, e!change rate mismanagement. The e!pectation of currency devaluation leads to speculative capital flight. -evaluation also causes the currency costs of debt service obligations, deteriorates budget deficit, and affects money supply and inflation. The estimation of the growth rate and debt accumulation rate and the regression analyses in the various stages of the causality tests employs the logarithmic transformations of the time series data on 8+9. >or this estimation 3angladesh, 1ndonesia, /alaysia, 9hilippines, "outh 5orea, "ri ;an$a, and Thailand are e!amined during the period 19%0 1988. >irstly, the author estimated the hypothesis that accumulation of e!ternal debt does not affect the 8+9 growth rate. 1t is fascinating that the long term effect of e!ternal debt accumulation rate on the 8+9 growth rate is found to be positive in 3angladesh, 1ndonesia and "outh 5orea. >or e!ample, an increase of 1N in the e!ternal debt caused an increase of the 8+9 by (0N in 3angladesh. "econdly, the effect of 8+9 growth rate on the e!ternal debt accumulation is e!amined. 1t is found that 8+9 growth rate affect the e!ternal debt accumulation rate of 9hilippines only. >or the 9hilippines, a 1N increase in the 8+9 leads to 1.(4 N increase in e!ternal debt in the long run. 1n addition, during the 1980Ls the 9hilippines? balance of payments is worsened. 3alance of payments deficits were

4%

financed by e!ternal borrowing, yet, e!ternal debt increased to a much greater e!tent than indicated by the worsening of the current account deficit, when domestic residents transferred their capital to the overseas countries. The results of the 8ranger causality tests show that the 3ullow Eogoff )1990. propositions that e!ternal debt of developing countries are a symptom rather than a cause of economic slowdown was re#ected. The results also show that a feedbac$ or bidirectional relationship e!ists between e!ternal debt accumulation rate and 8+9 growth rate for /alaysia and 9hilippines. >inally, there is no causal relationship between 8+9 growth rate and the e!ternal debt accumulation rate. 5aragol )(00(. investigated the long run and short run relationship between economic growth and e!ternal debt service for Tur$ey during the 194= 199=. This study used multivariate co integration techniques and employed a standard production function model. The IAE )Iector Auto Eegression. estimates of the system showed that there is a one co integrating relationship in the long run. -ebt service is negatively related to economic growth in the long run. 8ranger causality test results showed a uni direction causality running from debt service to economic growth. 9attillo et al. )(00(. use a large panel dataset of 9& developing countries over 19=9 1998 and find that the impact of e!ternal debt on per capita 8-9 growth is negative for net present value of debt levels above &4 '0N of 8-9. ,lements et al. )(00&. investigate the same relationship for a panel of 44 low income countries over the period 19%0 1999 and find that the turning point in the net present value of e!ternal debt is at around (0 (4N of 8-9. 7ther previous

48

empirical studies that find a non linear effect of e!ternal debt on growth include "myth and 0sing )1994. and ,ohen )199%.. 7n the other hand, "chclare$ )(00'. finds a linear negative impact of e!ternal debt on per capita growth )and no evidence of an inverted * shape relationship. in a panel of 49 developing countries over the period 19%0 (00(. "chclare$ )(00'. also investigates the relationship between gross government debt and percapita 8-9 growth in developed countries. +o robust evidence of a statistically significant relationship is found for a sample of (' industrial countries with data averaged over seven year periods between 19%0 and (00(. 1n contrast, a recent study by Eeinhart and Eogoff )(010., which analyses )through simple correlation statistics. the developments of public )gross central government. debt and the long term real 8-9 growth rate in a sample of (0 developed countries over a period spanning about two centuries )1%90 (009.,

finds thatG )i. the relationship between government debt and long term growth is wea$ for debtC8-9 ratios below a threshold of 90N of 8-92 )ii. above 90N, the median growth rate falls by one percent and the average by considerably more. A similar change in the behaviour of 8-9 growth in relation to the debt ratio is also found by 5umar and :oo )(010.. 0oa and Eobert )(00=. investigated the role of foreign currency denominated debt )>,--. as a natural hedging instrument, using a sample of Australian firms. Their results show that the incidence of foreign debt use among industrial sector firms is associated with a lower level of e!change rate e!posure. The practice of issuing foreign debt within the industrial sector also conforms

49

better to the hypothesis that firms do so to satisfy a demand for hedging. 1n contrast, although the incidence of foreign debt issues is higher in the resourceCmining sector, the underlying motive for such arises from a demand for financing. 5eunsu$ et al. )(010. investigated the determinants of foreign borrowing costs in a stochastically growing economy. They found that these increase with the debt wealth ratio, depending on the volatilities of domestic and foreign origin, and the length of debt contract. 1n addition, the sensitivity of the short term debt supply to the debt wealth ratio e!ceeds that of the long term debt, and the effects of volatility on the borrowing premium, growth of wealth, and its volatility, depending on the relative si6e of a direct effect and a secondary portfolio ad#ustment effect of the initial shoc$, as well as the length of the debt contract. 9anel regressions suggest that the empirical evidence generally support the theoretical predictions. 5arin )(009. derives optimal public spending and the resulting optimal deficit and debt in an optimal control framewor$ when the government see$s to ma!imi6e the utility of constituents in the presence of the e!ternal revenue that cannot be influenced by the government directly, but is contingent on public revenue and debt. 0e found that in this conte!t, a policy of running budget deficits and accumulating debt becomes optimal. 1n simulations, he characteri6es the si6e and the path dependency of the optimal deficit and debt. /ichael and ,hristopher )(00=., in their article titled AThe role of foreign currency debt in financial crisesG 1880 to 191& versus 19%( to 199%B show that

=0

e!posure to foreign currency debt does not necessarily increase the ris$ of having a financial crisis. "ome countries do not suffer from financial fragility despite original sin. 3efore 191&, the 3ritish offshoots and "candinavia, afflicted with it, avoided financial meltdowns. Today, many advanced countries have original sin, but few have had crises. 1n both periods, aggregate balance sheet mismatches are associated with a greater li$elihood of a crisis. The evidence suggests that foreign currency debt is dangerous when miss managed. This is part of the difference between developed countries and emerging mar$ets both of which borrow in foreign currency. 0ing /an )1999. studies foreign debts of newly industriali6ing countries )+1,s.. These countries develop by imitating foreign technology and e!porting goods that are lower than the Oquality ladder?. -ifferentiated by quality differences, such goods are often substitutes to those e!ported by advanced countries in the west. As +1, incomes increase, their consumers see$ to smooth their consumption over time. The novelty of this study?s finding is that +1,s borrow from abroad only when these goods are substitutes, but they would not have done so had the goods been complements. The intuition, obviously yet ignored in the literature, is the opposite terms of trade effects arising from substitutes and complements in consumption. /ost papers in the Ogrowth debt? literature adopted the Eomer and ;ucas steady state framewor$. They concentrated on issues, such as population growth and foreign debt, but neglected the imitation growth debt relation.

=1

1n middle income countries, :arner )199(. claims that the debt crisis does not depress investment, but this has been countered by several researchers including 8reene and Iillanueva )1991., Flbadawi et al. )199%., >osu )1999. and ,howdhury )(001. who have found some credence in the debt overhang hypothesis. 1n a study of 4' developing countries )including 1' 019,s., 0ansen )(001. observed that the inclusion of three additional e!planatory variables )the budget balance, inflation and openness. leads to re#ection of any statistically significant negative effect of e!ternal debt on growth. -#i$stra and 0ermes )(001. reviewed a number of studies on the Adebt overhangB hypothesis and concluded that the empirical evidence is inconclusive. 8enerally, therefore, a few studies give a clear idea of the level of the debt to 8-9 ratio at which debt overhang effects come to play. Flbadawi et al. )199=. confirmed a debt overhang effect on growth using cross section regression for 99 developing countries spanning sub "aharan Africa )""A., ;atin America, Asia and /iddle Fast. They identified three direct channels in which indebtedness in ""A wor$s against growthG current debt inflows as a ratio of 8-9 )which should stimulate growth., past debt accumulation )capturing debt overhang. and debt service ratio. They found that debt accumulation deters growth while debt stoc$ spurs growth. 7ther studies worth mentioning are those that simply use simulation analysis to demonstrate the impact of debt burden indicators on economic growth under different scenarios )see A#ayi, 19912 /bire and Atingi, 199%.

=(

,howdhury )(001. further contributes to the e!isting literature on the debt growth ne!us by analysing the relationship in two separate country groupsP 019, and non 019,P using e!treme bounds analysis for sensitivity tests and the mi!ed, fi!ed, and random coefficient approach that allows for heterogeneity in the causal relationship between debt and growth. The e!treme bounds analysis shows that the relationship between a debt measure and economic growth is robust to changes in the conditioning set of information included in the regression equations. The mi!ed, fi!ed and random coefficient approach, on the other hand, shows a statistically significant negative causal impact running from each of the four debt measures to economic growth in both country groups. /aghyereh and 7met )(00(. e!amined the impact of e!ternal debt on performance of the Dordanian economy and determined the optimal level of debt using a new econometric technique. The findings of the study indicate that the optimal level of e!ternal indebtedness is about 4& percent of 8-9. 1n other words, when the level e!ceeds this level, its impact on the performance of the Dordanian economy becomes negative. 1n a similar vein, "chclare$ )(00'. empirically e!plores the relationship between debt and growth for a number of developing and industrial economies. >or developing countries, the study finds that lower total e!ternal debt levels are associated with higher growth rates and that this negative relationship is driven by the incidence of public e!ternal debt and not by private e!ternal debt. The study does not find any support for a * shape relationship between e!ternal debt and growth. >or industrial countries, the study

=&

does not find any significant relationship between gross government debt and economic growth. *sing econometric analysis and "ri ;an$a data for the period 194( to (00(, :i#eweera et al. )(004. investigate whether "ri ;an$a faces a debt overhang problem. ;ong run estimations rely on cointegration methodology whereas short run analysis employs an error correction method. The results indicate that "ri ;an$a does not have a debt overhang problem, probably because total e!ternal indebtedness is not too high. ,lements et al. )(00&. e!amines the channels through which e!ternal debt affects growth in low income countries. Their results suggest that the substantial reduction in the stoc$ of e!ternal debt pro#ected for highly indebted poor countries )019,s. would directly increase per capita income growth by about 1 percentage point per annum. Eeduction in e!ternal debt service could also provide an indirect boost to growth through their effects on public investment. 1f half of all debt service relief were channelled for such purposes without increasing the budget deficit, the growth could accelerate in some 019,s by an additional 0.4 percentage point per annum. 3entum Fnnim )(009. used the Dohansen multivariate cointegration method and vector error correction )IF,. to e!amine the effect of growth rate of e!ternal debt on the growth rate of 8-9 in 8hana. The author found out that there e!ists a stable long run relationship among the growth rates of 8-9, labour force, investment, e!ports and e!ternal debt and that the long run effect of growth rate

='

of e!ternal debt on 8-9 growth rate has been positive which implies 8hana is not suffering from debt overhang problem.

Summary This chapter reviewed relevant literature on the relationship between e!ternal debt and economic growth in "ierra ;eonean. 3efore the review of the theoretical and empirical literature, an overview of "ierra ;eonean economy was discussed. This included the composition of e!ternal debt, the e!ternal debt burden and domestic debt burden since 1980. This review showed that the country has relied heavily on foreign borrowing to finance its capital e!penditures. Also despite of the difficulties caused by the security situation, "ierra ;eone has made satisfactory progress in implementing its structural ad#ustment programme. The theoretical literature illuminated many channels through which e!ternal debt affected economic growth. 0owever, the results of empirical studies on the nature and direction of causal relationships between e!ternal debt and economic growth are mi!ed and inconclusive as a consequence of using a variety of e!ternal debt pro!ies as well as the choice of econometric method. "ocial indicators such as war, political instability and cultural diversity all had negative correlation to economic growth. :orld recession and draughts had similar effect in that both of them affected growth negatively. As the review reveals, a growing body of empirical studies demonstrates a strong positive or negative lin$ between e!ternal debt and long run economic growth.

=4

CHAPTER THREE >ETHODO)O=? Introduct$on The purpose of this chapter is to present the methodological framewor$ suitable for conducting the study. 1t discusses the methods and tools of analysis employed in this study. "pecifically, the chapter presents a detailed description of the theoretical and empirical specification of the model, variables in the model, source and data type, estimation techniques as well as tools for data analysis.

Theoret$ca" >ode" S ec$f$cat$on The literature about the relationship between e!ternal debt and economic growth or the effect of debt on growth can be summari6ed by three strands of thoughtG The first sees e!ternal debt as a capital inflow with a positive effect on domestic savings and investment and thus on growth. This argument implies that foreign savings complement domestic savings and investments )Faton, 199&.. The second strand considers e!ternal debt as a substitute to domestic savings and investment and therefore tends to crowd them out )5rugman, 19882 Alesina and Tabellini, 19892 and Tornell and Ielasco, 199(.. The argument here runs as follows. 1f future debt is going to be greater than a country?s ability to repay its loans, the e!pected debt service will be an increasing function of its output level.

==

1n other words, the returns from investing in a country are seen as being sub#ected to a high marginal ta! by creditors and this might discourage domestic and foreign investors. These models or theories are $nown as the debt overhang theories. >inally, more recent studies have attempted to reconcile the above conflicting views by developing models with non linear effects of debt on growth. >or e!ample, ,ohen and "achs )198=. and ,ohen )1991, 199(. present endogenous growth models where capital accumulation is the sole force driving growth. ,ountries? access to international financial mar$ets is limited because of the ris$ of debt repudiation. 8rowth is high in the early stages as the country borrows and invests. ;ater on, growth falls to a lower level, and this new level would be higher than it would have been had there been no international borrowing and lending )financial autar$y.. The stage of repaying countries? debts does not crowd out investment because lenders are more patient and value growth more than the debtor countries themselves. This result, however, depends on the ability of lenders to implement optimal rescheduling policies. 1f they are not able to commit themselves to this policy, a debt overhang scenario will occur and investment and growth in the later stages will be even lower than in financial autar$y. 3ased on these theoretical postulates, an endogenous growth model in the form of a ,obb -ouglas production function is formulated to capture the relationship between e!ternal debt and economic growth denoted by equation )1..

Yt = AK t L t )1.
:here Y denotes the aggregate output at time t , K is the aggregate capital stoc$ at time t , L represents labour stoc$ at time t while A denotes total factor =%

productivity growth )T>9.. and are coefficients of elasticity for capital and labour respectively. The T>9 captures growth in output not accounted for by increase in physical input )capital. in the model. >ollowing Faton )199&., 3arrow and "ala i /artin, )1994.2 3arrow, /an$iw, and "ala i /artin, )1994., the study augmented equation )1. with some variables that are assumed to affect the relationship between e!ternal debt and economic growth. "o equation )1. gives equation )(. and )&. as followsG

& 1 Yt = AEXDTt EXPOt ( INVGt LAFt ' 1 ( & ' EXDTt = AYt EXPOt INVGt DS t

)(. )&.

:here Y is economic growth, EXDT is e!ternal debt, EXPO is e!port, INVG is investment to 8-9 ratio )i.e. capital., LAF is labour force and -" is debt service ratio to 8-9.

Em $r$ca" >ode" S ec$f$cat$on 1t is already clear from the above that the relationship between e!ternal debt and economic growth is mi!ed with some studies finding evidence in support of the debt overhang hypothesis )i.e. negative effect. while others show a positive effect. The following long run regression models are estimated by ta$ing natural logarithm on both sides of equation )(. and )&. above in order to estimate a log linear models of the following form that e!plain the relationship between e!ternal debt and economic growth in "ierra ;eone and are denoted by equations )'. and )4. assuming Yt = LnRGDPt =8

LnRGDP t = 0 + 1 LnEXDTt + ( LnEXPOt + & LnINVGt + ' LnLAFt + 4WAR +t )'.

' LnDSt +4WAR + t

LnEXDTt = t + 0 + 1LnRGDP ( LnEXPOt + & LnINVGt +

)4.

:here Ln denotes the natural logarithm, J is Eeal 8-9 )economic growth., FQ-T is e!ternal debt, FQ97 is e!ports, 1+I8 is the ratio of investment to 8-9 )i.e. capital., ;A> is labour force, -" is debt service ratio to 8-9 and :AE is a dummy. 1n carrying out the analysis in this study, the dependent and independent variables chosen were based on their ability to portray the investigation in a meaningful and consistent manner. Iariables were included, e!cluded, or pro!ied based on theoretical andC or empirical #ustification.

A r$or$ e, ected s$gns The coefficients, 1 , ( , & , and

' in equation )'. and 1 , ( , & , ' in 0 and 0 are the drift

equation )4. are the elasticities of the respective variables, components,

is the error term and t denotes time. Also 4 and 4 are the

coefficients of the dummy in equations )'. and )4. respectively. The following are e!pected

1 R0 or 1 S0, ( S0, & S0, ' S0 and 4 R0 if all things remain 1 S0 or 1 R0, ( S0, & R0, ' S0 and 4 R0. All the

constant. "imilarly,

variables are in natural logarithm e!cept the dummy variable, :AE. ;og transformation can reduce the problem of heteroscedasticity because it compresses the scale in which the variables are measured, thereby reducing a =9

tenfold difference between two values to a twofold difference )8u#arati, 1994.. 1t is important to note that the model is a multiplicative one where all parameter coefficients represent constant elasticities. The bul$ of the literature suggested that e!ternal debt is harmful to economic growth and therefore we e!pect the coefficient to be negative. 1nvestment is believed to e!ert a positive impact on economic growth since it is a ma#or channel for sustained economic growth therefore we e!pect the coefficient to e!hibit a positive sign. ;abour force is also e!pected to be positive. F!port is also another channel that feeds economic growth so we e!pect the sign to be positive. An increase in debt service ratio to 8-9 has a positive relationship with e!ternal debt and since e!ternal debt is inversely related to economic growth, we e!pect that increase in debt service increases e!ternal debt therefore we e!pect debt service to e!ert a positive impact on e!ternal debt. The :AE dummy by every indication is e!pected to be negative.

Data descr$ t$on and sources Fconomic 8rowth refers to sustained increases in the economy?s real gross domestic product or real gross national product overtime. Eeal 8-9 is the economy?s total output of goods and services, normally measured over a period of one year in monetary terms. Eeal 8+9 is the economy?s total output of goods and services net of factor income from abroad, normally measured over a period of one year in monetary terms. The literature suggests that we can use real gross domestic product as an efficient measure of economic growth )8elb 19892

%0

Eoubini H "ala i /artin, 199(2 5ing H ;evin, 199&2 -emetriades H 0ussein, 199=2 ;evine et al, (000.. The economic growth is used as the dependent variable in the model. Thus, real gross domestic product is employed as a pro!y of the economic growth. Ayadi and Ayadi )(008. used real gross domestic product as a pro!y of economic growth in e!amining the impact of e!ternal debt on economic growth for +igeria and "outh Africa. F!ternal debt may be broadly defined as the outstanding amount of those actual current and not contingent liabilities that require payment)s. of principal and or interest by the debtor at some point)s. in the future and that are owed to non residents of an economy )1/>, (00&.. F!ternal debt is one of the ma#or sources of financing the development programs. These programs generally focus on infrastructure development that is necessary to lay down the foundations of sustainable economic growth. 1t is hypothesi6ed that there e!ists a positive relationship between e!ternal debt and economic growth. >or this study F!ternal debt has been translated into "ierra ;eonean ;eones at constant prices of (000 and then logarithm of e!ternal debt is ta$en to construct the variable. 1t is hypothesi6ed that there e!ists a negative relationship between e!ternal debt and economic growth. >ollowing 8ounder )(001., the e!port coefficient in our model relates to the output elasticity of e!ports and this variable reflects the degree of AopennessB of the economy and constitutes an AinputB in the production function. Fdwards )1998. observes that e!ports play a positive role in the growth process by increasing total factor productivity after including factor productivity and

%1

institutional factors. 1t is hypothesi6ed that there e!ists a positive relationship between e!port and economic growth. The theory suggests that properly used foreign debt acts as a lubricant in the economy creating more employment opportunities and increasing the national output that can be used for domestic use as well as for e!port. 1n a similar way, labour force contributes to the economy as each and every nation and firm need the wor$ force to run its production activities. ;abour force is directly involved in the production process and large labour force can help to enhance production of goods and services. >or purpose of this study ;abour force of the age between 14 and =' has been ta$en from Fconomic "urvey of "ierra ;eone. 1t is hypothesi6ed that there e!ists a positive relationship between labour force and economic growth. The share of investment is used as a pro!y for the ratio of gross fi!ed capital formation to nominal 8-9. /ba$u )199&. and 1slam )199(. utili6ed the total investment to 8-9 ratio as opposed to the savings to 8-9 ratio employed as a pro!y for investment. The use of total investment to 8-9 is in conformity with earlier studies )such as 8ounder, (001.. 1nvestment as a share of 8-9 had always been one of the most significant factors associated with the output growth. As mentioned by "ala i /artin )(00(. countries that invest more tend to grow faster than those countries that save and invest less. "ome ""A countries li$e "ierra ;eone which had low growth rates accounted for only 10 percent of investment in 8-9. /eanwhile, in the Asia countries li$e the four Asian Tigers )"outh 5orea, "ingapore, 0ong 5ong and Taiwan. investments yield up to 40 percent of 8-9.

%(

1nvestment is one of the main sources for growth in "ierra ;eone. 1t is hypothesi6ed that there e!ists a positive relationship between investment to 8-9 ratio and economic growth. -ebt service ratio to 8-9 is another traditional indicator of indebtedness, which compares an economy?s debt service e!penditure to its level of productivity. 8enerally, the higher the ratio of debt service to a nation?s productivity, the more serious the debt burden on the economy )7motoye, (00=.. :ar as a dummy variable is included to capture the structural brea$s that e!isted during the period under consideration. 1t is a binary variable which ta$es the value of 0 for no war and 1 in periods of war in the country. This variable is e!pected to have a negative effect on economic growth. The e!pected sign is negative. 1n line with the above theoretical framewor$, we need the following variables for testing the hypothesesG real 8-9, e!ternal debt, e!ports, investment as a percentage of 8-9 and labour force. -ata on all the variables are acquired from various issues of the 3an$ of "ierra ;eone?s annual reports as well as various issues of national accounts from the /inistry of >inance and Fconomic -evelopment )/>F-., /inistry of ;abour, Jouth and Fmployment )/;JF., :orld -evelopment 1ndicators ):-1., "tatistics "ierra ;eone )"";. and the 1nternational >inancial "tatistics )1>".. The period covered in the study )19%0 (008. depended mainly on the availability of consistent data for all the variables. >inally, the choice of the data coverage was informed by the fact that %&

developments within the "ierra ;eonean economy have been characteri6ed by political instability and military regimes. Thus any meaningful analysis of e!ternal debt and economic growth should evaluate the impact of these regimes. Est$mat$on rocedure To establish the long run relationship and the direction of causality between e!ternal debt and economic growth the study employed the Autoregressive -istributed ;ag )AE-;. model approach to cointegration and error correction model. The Augmented -ic$ey >uller )A->. and the 9hillips 9erron test statistics were employed to analyse the time series properties of the data set. This was done by conducting unit roots test to determine whether the variables are stationary. >inally, the study performed causality test by employing the 9airwise 8ranger causality tests. Autoregress$%e D$str$!uted )ag 0Bounds Test2 A roach to Co$ntegrat$on

F!istence of the long run equilibrium relationship between e!ternal debt and economic growth can be investigated by using several methods. The most widely used methods include Fngle and 8ranger )198%. test, fully modified 7;" procedure )>/7;". of 9hillips and 0ansen?s )1990., ma!imum li$elihood based Dohansen )1988, 1991. and Dohansen Duselius )1990. tests. All these methods require that the variables in the system are integrated of order one, 1)1.. >urther, these methods are considered as wea$ as these methods do not provide robust results for small samples or structural brea$s. 1n fact, these methods are still been employed by many researchers who strongly argue that they are the most accurate methods to apply on I)1. variables. 0owever, researchers li$e 5remers et al %'

)199(. and 0a$$io H Eush )1991. find some disadvantages with the above techniques. They argued that two common misconceptions e!ist in the standard ,ointegration techniques. >irst, long run relationships e!ist only in the conte!t of cointegration of integrated variables of the same order. "econd, standard methods of estimation and inference )e.g. > test. will render inconsistent and inefficient parameters in cointegrated relationship. -ue to these problems, a newly developed autoregressive distributed lag )AE-;. approach to cointegration has become popular in recent years. The AE-; modelling approach was originally introduced by 9esaran and "hin )1999. and further e!tended by 9esaran )(001.. This approach has numerous econometric advantages in comparison to other ,ointegration methods. 7ne ma#or advantage of AE-; approach is that it can be applied irrespective of degree of integration. "econdly, AE-; approach provides robust results in small sample si6es and estimates of the long run coefficients are well consistent in small sample si6es )9esaran and "hin 1999.. >urthermore, a dynamic error correction model )F,/. can be derived from AE-; that integrates the short run dynamic with the long run equilibrium without losing long run information. 1n view of the above advantages, we use AE-; approach for cointegration analysis and the resulting F,/. The choice of this methodology is based on several considerations. >irstly, as shown by 9esaran et al. )(001., the AE-; models yield consistent estimates of the long run coefficients that are asymptotically normal irrespective of whether the underlying regressors are 1)1. or 1)0.. "econdly, this technique generally provides unbiased estimates of the long run model and valid t statistics even when

%4

some of the regressors are endogenous )0arris, (00&.. 1nder )199&. and 9esaran )199%. have shown that the inclusion of the dynamics may help correct the endogeneity bias. Thirdly, given the si6e of the sample used in this study )&9 observations. and the number of parameters to be estimated )4 in total, e!cluding the constant term. the bound approach appears more appealing than the Dohansen cointegration technique, which would have required the estimation of a system of 4 equations and thus a considerable loss in degree of freedom. /oreover, unli$e the Dohansen cointegration technique, the ARDL approach avoids problems relating to the large number of choices which must be madeG including decisions such as the number of endogenous and e!ogenous variables )if any. to be included, the treatment of deterministic elements, as well as the order of VAR and the optimal number of lags to be used. The estimation procedures are very sensitive to the method used to ma$e these choices and decisions )9esaran H "mith, 1998.. >urthermore, as long as the AE-; model is free of residual correlation, endogeneity is less of a problem. 9esaran H "hin )1999. showed that the appropriate lags in the AE-; model correct for both serial correlation and endogeneity problems. 1ndeed, the AE-; method is capable of distinguishing between endogenous and e!ogenous variables that estimation is possible even when the e!planatory variables are endogenous )9esaran H 9esaran, 19992 9esaran et al., (001.. This is important in the present conte!t as the share of investment to 8-9 and labour force in our model could be endogenous. To implement the bound test procedure, Fquations )%. and )8. are modelled as conditional AE-; error correction modelsG

%=

LnRGDPt = o + 1i LnRGDPt i + ( i LnEXDTt i + &i LnEXPOt i +


i =1 i =1 i =1

i =1

'i

LnINVGt i + 4i LnLAFt i + 1 LnRGDPt 1 + ( LnEXDTt 1 + & LnEXPOt 1 +


i =1

' LnINVGt 1 + 4 LnLAFt 1 + WAR + )=.


LnEXDTt = o + LnRGDPt i + (i LnEXDTt i + &i LnEXPOt i + 1i
i =1 i =1 i =1 p p p

i =1

'i

LnINVGt i + 4i LnDSt i +1 LnRGDPt 1 +( LnEXDTt 1 +


i =1

& LnEXPOt 1 +' LnINVGt 1 +4 LnDSt 1 + WAR +t )%.

:here denotes the first difference operator, 0 and 0 are drift parameters while

t and t are white noise error terms. The parameters ij are the short run

parameters in equation )=. and ij are the long run multipliers while the parameters ij and ij are the short run and long run parameters in equation )%. respectively. The long run relationships are ascertained by restricting the coefficients of the lagged level variables to 6ero. To investigate the e!istence of cointegration, an 7;" regression is estimated on the first differences part of equations )=. and )%. using the A$ai$e 1nformation ,riterion )A1,. to select the optimum number of lags. "ubsequently, an > test or :ald test is conducted for the #oint significance of the parameters of the lagged level variables when added to the first regressions. This involves testing the null hypotheses of the non e!istence of a long run relationship among the variables in equation )=. against the alternative hypotheses of the e!istence of a long run relationship among the variables specified asG H 0 G 1 = ( = & = ' = 4 = 0

%%

H 1 G 1 ( & ' 4 0 "imilarly, the null hypothesis of no cointegrating relationship among the variables in equation )%. is tested against the alternative hypothesis of the presence of cointegrating relationship. This is given asG H 0 G 1 = ( = & = ' = 4 = 0 H1 G 1 ( & ' 4 0 These hypotheses are tested against the alternative by the mean of an > test with an asymptotic non standard distribution. Two asymptotic critical value bounds provide a test for cointegration when the independent variables are I )d. with 0 R d R 1 9esaran et al. )199=.. The lower bound assumes that all the regressors are I)0. , and the upper bound assumes that they are I )1. . 1f the computed > statistics lies above the upper level of the band, the null is re#ected regardless of whether the variables are 1)0. or 1)1., indicating cointegration or a long run relationship among the variables. 1f the computed > statistics lies below the lower level band, the null hypothesis of no cointegration relationship among the variables cannot be re#ected, supporting the absence of cointegration. 1f the statistics fall within the band, inference would be inconclusive and depends on whether the underlying variables are 1)0. or 1)1.. This necessitates the testing for unit root on the variables under investigation )9esaran H 9esaran, 199%.. 0owever, given that 9esaran?s critical values are based on simulated large sample si6e, this study uses the critical values developed by +arayan )(00'. since it is more appropriate for small samples.

%8

After confirmation of the e!istence of a long run relationship between the variables in the model, the AE-; methodology estimates )! + 1. regressions. :here ! is the ma!imum number of lags and
+1

number of

is the number of

variable in the equation )"hrestha H ,howdhury, (0042 9esaran H 9esaran, (009.. The orders of the lags of the AE-; models are selected using either "chawrt6 3ayesian ,riteria )S"#., A$ai$e?s 1nformation ,riterion )AI#., the R ( criterion or the 0annan and Tuinn ) H$. criterion. The "3, uses the smallest possible lag length and is considered as the most parsimonious model whereas the A1, chooses the ma!imum necessary lag length )"hrestha H ,howdhury, (0042 Dalil et al., (008.. ;ong Eun Fstimates of the selected AE-; /odel 7nce cointegration is established, we then proceed to estimate the selected long run AE-; model in order to obtain the long run coefficients and their asymptotic standard errors. 1n this study, concentration is given to the economic growth model because it is the only model that gave us cointegrating relationship.

LnRGDPt = 0 + 1 LnRGDPt i + ( LnEXDTt i + & LnEXPOt i


i =1 i =0 i =0

%1

a1

a(

+ ' LnINVGt i +
i =0

a&

LnLAF
4 i =0

a'

t i

+ WAR Ut

)8.

This is followed by the estimation of the short run elasticities of the variables with the error correction representation of the AE-; model. 3y employing the error correction side of the AE-;, we then determine the speed of ad#ustment to equilibrium. The e!istence of long run relationship among the

%9

variables necessitates the estimation of the unrestricted AE-; error correction representation asG
LnRGDPt = 0 + 1i LnRGDPt i + ( i LnEXDTt i + &i LnEXPOt i
i =1 i =0 i =0 p p p

+ ' i LnINVGt i + 4i LnLAFt i UWAR U F,/ t 1 + t


i =0 i =0

)9.

:here is the speed of ad#ustment parameter and E#& t 1 is the error correction term, the residual obtained from equation )%. which is the cointegration residual lagged one period. The coefficients are the short run dynamics while ) . is the speed of ad#ustment to long run equilibrium following a shoc$ to the system. The coefficient of this lagged error correction term ) . is e!pected to be negative and statistically significant in order to further confirm the presence of cointegrating relationship among the variables in the model. The reliability of the goodness of fit of the model is also determined by conducting the diagnostic and stability tests of the model. The diagnostic test ta$es care of heteroscedasticity2 autocorrelation, normality and the functional form that are lin$ed with the model selected. The ,*"*/ and

,*"*/"T are used to perform parameter stability tests as suggested by )9esaran H 9esaran, 199%.. >inally, the AE-; approach to cointegration does not require the pre testing of the variables, included in the model, for unit root unli$e other techniques such as the Dohansen approach )9esaran et al., (001.. 0owever, as remar$ed by 7uattara )(00'., if the order of integration of any of the variables is greater than one, for e!ample 1 )(. variable, then the critical bounds provided by

80

9esaran et al. )(001. are not valid. They are computed on the basis that the variables are 1 )0. or 1 )1.. 9ut differently, it is necessary to test for unit root to ensure that all the variables satisfy the underlying assumption of the AE-; methodology before proceeding to the estimation stage. :ith this in mind, we start the econometric analysis, in this study, by analysing the order of integration of the variables.

Un$t Root Tests +on stationarity of time series data has often been regarded as a problem in empirical analysis. /ost economists have pointed out the inappropriateness of applying conventional econometric techniques to non stationary data because of Aspurious relationshipB. 8ranger and +ewbold )19%'. re emphasi6ed the warnings of Jule )19(=. concerning the use of non stationary data. They coined the term Ospurious? regression for the result obtained by using two trended variables in a regression when the variables were actually unrelated. O"purious relationship? occurs when a regression model involving two non stationary series may produce high E( because of some common trend even though there is no true economic relationship between them. :or$ing with non stationary variables lead to spurious regression from which further inferences are meaningless. 1n this respect, testing for unit root is the first step in time model building. *nit root test are important in e!amining the stationarity of a time series. 1n view of the Ospurious regression? problem, Fngle and 8ranger )198%. suggested a two step method of integrating the cointegration technique with the

81

error correction mechanism. 0owever, Fngle and 8ranger two step procedure is inappropriate if the variables being tested for the cointegration are more than two and a better alternative to testing for cointegration between more than two variables is offered by Dohansen?s test )Dohansen, 1988.. This test is based on the vector autoregressive )IAE. or vector error correction model )IF,/. representation of the system. Iariables are cointegrated if the linear combination of their data series is stationary even though the individual series may be non stationary. Time series model are useful for forecasting, but most standard method for fitting such model requires the series to be stationary )Fnders, (001..The simplest e!ample of a non stationary series is the random wal$ process e!pressed asG
X t = X t 1 +t )10.
( :here t : iid )0, . and is generated by a white noise process. 1f X 0 V 0, we

have X t = i =1 i
t

The variable of Qt is t ( and this becomes indefinitely large as t increases. >orecasting with such a series becomes quite unreliable since the variances of the forecasting error diverge to infinity as the forecast hori6on recedes. This renders the concept of a mean value for Q t meaningless, that is if at point Qt V / )/ ta$en as mean value for Q t. then the e!pected time until Qt again returns to / is infinity. >or such process then, it would be inappropriate to use classical statistical method to estimate parameters and perform significant tests. "ince the main thrust 8(

of econometrics modelling and statistical theory develops within a stationary framewor$, it is imperative to perform unit root test since many of the time series variable, which are of interest in applied economics, are obviously non stationary. The most popular tests for e!istence of non stationary is the -ic$ey and >uller )->. and the closely associated Augmented -ic$ey fuller )A->. as well as the 9hilip and 9erron )99. tests. The A-> and 99 tests will be employed in this study. The tests will be done for the level of the variables to determine whether they are stationary or 1 )0.. 1f the null hypothesis of nonstationary cannot be re#ected, then the process of differencing the variables is repeated till stationary is achieved. 1t should be noted that A-> tests is better than the original -> test since the augmentation with lagged difference ensures that the errors are white noise. The A-> test involves running one of the following regressionsG
Yt = 0 + Yt 1 + i Yt i + t )11.
i =1 p

Yt = 0 + Yt 1 + t t + i Yt i + t )1(.
i =1

:here J is the variable under consideration, t is the time period2 p is the number of lags and t is the stochastic error term. >or the -> lagged dependent variables are included on the right hand side. 0owever in A-> test the lagged are added on the right hand side. The augmentation improves the statistical fit of the equation and is more efficient now with the added information in the equations as seen in the above equation )11 and

8&

1(.. 0ence with A-> test we can assume that error term is independently and identically distributed )iid.. The number of lags is carefully chosen so as to produce serially uncorrelated error terms. 8ordon )1994. noted that if too few lags are included, the test changes in un$nown manner, conversely, if the lags are too many the power of the test is reduced. The right side of p will be estimated using the highest significant lag order of the series by the use of A$ai$e information ,riterion )A1,. and "chwart6 3ayesian ,riterion )"3,.. 1n testing for the unit root, we shall use equations )1(., which is a more general model. 0owever, this does not imply that it is always preferable since in a situation where we have a series with a 6ero drift and non 6ero means, equation )11. is recommended because it has a greater power. 1f the series is generated by a random wal$ with 6ero drift and has 6ero mean than the test based on equation )11. but where V0 are appropriate. 1f the series has non 6ero mean and non 6ero drift the estimation should include both a constant and a trend term as in equation )1(.. 1mportant to note is that the appropriate equation for testing for unit root in a particular series depends on the series being tested. *se of the wrong equation definitely ma$es the test biased and hence unreliable. :e also apply the 9hillips 9erron test of unit root as the A-> tests are unable to discriminate well between non stationary and stationary series with a high degree of autocorrelation. The 99 test has an advantage over the A-> test as it gives robust estimates when the series has serial correlation and time dependent heteroscedasticity. >or the 99 test, we estimate the equation as belowG
X t = + ( X t 1 + ( t T C ( ) + i X t 1 + ( t )1&.
i =1 !

8'

Thus, the A-> and the 99 test the null hypothesis that a series contains unit root )non stationary. against the alternative hypothesis of no unit root )stationary.. That isG H0 G = 0 H1 G 0 1f the tau value or t statistic is more negative than the critical values, we re#ect the null hypothesis and conclude that the series is stationary. ,onversely, if the tau statistic is less negative than the critical values, we fail to re#ect the null hypothesis and conclude that the series is non stationary. The critical values for this t statistic are given in /ac$innon )1991..

=ranger Causa"$ty Test The detection of causal relationships among a set of variables is one of the ob#ectives of empirical research. 8ranger )19=9. formulated a procedure for detecting a causal relationship among the variables. The concept of causality in the 8ranger sense is mainly based on the assumption that the future cannot cause past, it is the past and the present which cause future and that detection of causality is only possible between two stochastic process, which should be linear and covariance stationary. 8ranger?s approach to the question of whether variable X t causes variable Yt is to see how much of the current Yt can be e!plained by past values of Yt and

84

then to see whether adding lagged values of X t can improve the e!planation. Yt is said to be 8ranger caused by X t if X t helps in the prediction of Yt , or equivalently if the coefficients on the lagged X t s are statistically significant. +ote that two way causation is frequently the case2 X t 8ranger causes Yt and Yt 8ranger causes X t . 1t is important to note that the statement A X t 8ranger causes Yt B does not imply that Jt is the effect or the result of X t . 8ranger causality measures precedence and information content but does not by itself indicate causality in the more common use of the term. 1n 8ranger sense then, a series X t is said to cause Yt , if Yt is better predicted by a model using the past values of X t and Yt than by a model using Yt alone )9indyc$ and Eubinfeld, 1991.. 1nclusion of variable X t enhances the predictive power of the model in a statistical sense. To conclude that X t causes Yt , we must re#ect the hypothesis that AQt does not cause Yt B, and accept the hypothesis that A Yt does not cause X t B. The empirical literature reviewed in chapter two identified that the direction of causation between e!ternal debt and economic growth is inconclusive. 1n this respect, each of these variables will be tested for 8ranger causality based on the following two equationsG
X t = iYt i + i X t i + t )1'.
i =1 i =1

Yt = i X t i + iYt i + t )14.
i =1 i =1

8=

:here i V )i = 1, (,..., . so that Jt fails to cause X t . The error terms are assumed to fulfil the criteria.
( E ) t . = E )t . = E ) t ' . = E ) t ' . = 02 and E ) t t . = , E )tt . = ( .

,ausality in equation )1'. should run from Yt to X t on the proviso that the estimated coefficients on the lagged variable ) Yt . are significantly different from 6ero. 1n other words, the coefficients i are different from 6ero, i.e., i 0 . "eemingly, causality in equation )14. runs from X t to Yt provided the estimated coefficients on X t as a group are significantly different from 6ero, i 0 . 3idirectional causality occurs if X t causes Yt and Yt causes X t . 1n other words, the lagged values of both X t and Yt as a group in equations )1'. and )14. are significantly different from 6ero.

Summary The chapter has succinctly elaborated on the theoretical underpinnings of the relationship between e!ternal debt and economic growth and proceeded to formulating both theoretical and empirical models. The theoretical model adopted in the study is the endogenous growth model which posited that economic growth depended on one $ey variable2 the rate of return on both physical and human capital. The endogenous growth model in the form of a ,obb -ouglas production function was then formulated to capture the relationship between e!ternal debt 8%

and economic growth. The empirical models were formulated by ta$ing natural logarithm on both sides of the theoretical model. 1t further described the data set and sources used in carrying out the study and also described the variables used in the research. The chapter also delved into the estimation techniques in which the AE-; approach to cointegration has been elaborated on and deemed as an appropriate technique in carrying out a study of this nature as a result of the numerous advantages it has over the other estimation techniques some of which are discussed inter alia. Thus, the unrestricted error correction models of the empirical model specifications were formed to obtain the AE-; model. *nit root test was conducted to ensure that the variables are not integrated of an order higher than one to avoid spurious regression and 8ranger causality test was conducted to establish if there e!ists any causal relationship between e!ternal debt and economic growth and in which direction does the causality e!ist. The systematic framewor$ of this chapter now establishes the relationship between e!ternal debt and economic growth which guides us in our estimation. This has a lin$ with the estimation of our models as well as guiding us to interpret our estimation results in the subsequent chapters and ma$e policy

recommendations based on the outcome of findings.

88

CHAPTER (OUR RESU)TS AND DISCUSSION Introduct$on This chapter presents a thorough analysis and discussion of the results of the study. The ob#ective is to investigate empirically the relationship between e!ternal debt and economic growth in "ierra ;eone. The study e!amined the time series properties of the data in order to determine the stationarity status of the variables using the Augmented -ic$ey >uller )A->. and 9hillips 9erron )99. tests and then tested for cointegration and causality using the Autoregressive -istributed ;agged /odel )AE-;. and the 9airwise 8ranger causality test respectively. The analysis of these tests then aided us to ascertain the relationship between e!ternal debt and economic growth and thus guided us in discussing the results e!plicitly.

Presentat$on of resu"ts *nit Eoot Tests Eesults

89

3efore analy6ing the estimated results using AE-; approach to cointegration, unit root test was conducted in order to investigate the stationarity properties of the data. As a result, all the variables were e!amined by first inspecting their trends graphically )Appendices F and >.. The time series plot shown in appendices F and > provide some idea about whether the trend is stationary or not. 1t is clear all the logs of the variables in levels show trend over time and none seems to suggest some amount of stationarity. 0owever, the first differences of the variables show no trend. Thus a quic$ e!amination of the series plot seems to suggest some amount of stationarity in first difference. +otwithstanding this, a more formal tests were carried out to establish whether the variables are integrated of order one, 1)1. or otherwise for the variables in levels. The Augmented -ic$ey >uller )A->. and 9hillips and 9erron )99. tests were applied to all variables in levels and in first difference in order to formally establish their order of integration. These tests e!amine the null hypothesis that the considered variable has a unit root versus the alternative hypothesis that the variable is stationary. 0owever, a necessary but not sufficient condition for cointegration is that each of the variables should be integrated of the same order, and the order must be greater than or equal to one. The "chwart6 3ayesian ,riterion )"3,. and A$ai$e 1nformation ,riterion )A1,. were used to determine the optimal number of lags included in the test. The study presented and used the critical values ),I. for ma$ing the unit root decision. Appendices A, 3, , and - show that the A-> and 99 statistics for the levels of all the variables do not e!ceed the critical values )in absolute terms.

90

implying that the variables are non stationary at levels. 0owever, when first differences are ta$en on each of the variables, the A-> and 99 statistics are higher than their respective critical values )in absolute terms. implying stationary after first differences. The study concludes that )real 8-9, e!ternal debt, e!ports, investment as a ratio of 8-9, labour force and debt services. are each integrated of order one or 1 )1..

Co$ntegrat$on Ana"ys$s "ince the focus of this study is to establish the relationship between e!ternal debt and economic growth, it is important to test for the e!istence of a long run equilibrium relationship between these two variables within the framewor$ of the bounds testing approach to cointegration. 8iven that the study employs annual data, a lag length of ( is used in the bounds test. 9esaran and "hin )1999. suggest a ma!imum lag length of ( for annual data in the bounds testing approach to cointegration. After the lag length was determined, an > test for the #oint significance of the coefficients of lagged levels of the variables was conducted. Thus, each of the variables in the model is ta$en as a dependent variable and a regression is run on the others. >or instance, real 8-9 is ta$en as the dependent variable and it is regressed on the other variables. After that another variable for instance e!ternal debt is ta$en as the dependent variable and it is also regressed on the other variables. This action is repeated for all the variables in the model. :hen this is done the number of estimated regressions would be equal to the number of variables in the model.

91

9esaran et al )199%. indicates that Athis 7;" regression in the first differences are of no direct interestB to the bounds cointegration test. 1t is however, the > statistic values of all the regressions when each of the variables is normali6ed on the others which are of great importance. This > statistic tests the #oint null hypothesis that the coefficients of the lagged levels are 6ero. 1n other words, there is no long run relationship between them. The essence of the > test is to determine the e!istence or otherwise of cointegration among the variables in the long run. The results of the computed > statistic when real 8-9 and e!ternal debt variables are normali6ed )that is, considered as a dependent variable. in the AE-;<7;" regressions are presented in Table &. Ta!"e :. Bounds test to co$ntegrat$on ,ritical Ialue 3ounds of the > 5 90N ;evel 1)0. 1)1. ' (.''( &.'0 0 ,alculated > "tatisticG FLnRGDP ) LnRGDP W LnEXDT , LnEXPO , LnINVG , LnLAF . '.44&' )0.0('.XX "tatisticG intercept and no trend ),ase 11. 94N ;evel 99N ;evel 1)0. 1)1. 1)0. (.90% &.98( &.98&

1)1. 4.''8

FLnEXDT ) LnEXDT W LnRGDP , LnEXPO , LnINVG , LnLDS . 1.8880 )0.1''. +oteG ,ritical values are obtained from +arayan )(00'., Appendi! A1 A&, pp. (= (82 XXdenotes statistical significance at the 4N level and 5 is the number of regressors in equations = and %.

>rom Table & above, the > statistic that the #oint null hypothesis of lagged level variables )i.e. variable addition test. of the coefficients is 6ero is re#ected at the 4N significance level. >urther, since the calculated > statistic for

9(

FLnRGDP ).. = '.44&' e!ceeds the upper bound of the critical value band )&.98(., the null hypothesis of no cointegration )i.e. long run relationship. between economic growth and its determinants is re#ected. This result indicates that there is a unique cointegration relationship among the variables in "ierra ;eone?s economic growth model )equation =. and that all the determinants of economic growth can be treated as the Along run forcingB variables for the e!planation of economic growth in "ierra ;eone. "imilarly, when e!ternal debt is ta$en as dependent variable )equation %. the calculated > statistics is less than the lower bound critical values, thus accepting the null of no cointegration with FLnEXDT ).. = 1.8880 . "ince this study is based on growth theory, ;nE GDPt is used as the dependent variable and consequently the result of the other regression is neglected. Therefore, there is e!istence of cointegration among the variables in the growth equation and hence we therefore proceed with the growth equation.

Stat$c "ong5run resu"ts 0)nR=DP $s de endent %ar$a!"e2 Table & reports results of the bound test for the e!istence of a long run relationship. The > statistics is above the 4 per cent critical bounds computed by +arayan )(00'., thus implying that the null hypothesis of no cointegration can be re#ected. 9ut differently, there e!ists a long run relationship among the variables of our model. Table ' shows results of the long run estimate based on the "chwart6 3ayesian criteria. The selected AE-; )1, 0, 0, 0, 1, 0. passes the

9&

standard diagnostic tests )serial correlation, functional form, normality and heteroscedasticity.. The coefficients indicate the long run elasticities. >rom Table ', the long run relationship results show that e!ternal debt e!erted a negative and statistically significant effect on economic growth while e!ports and investment as a ratio of 8-9 showed a positive and statistically significant effect on economic growth. ;abour force however, e!erted statistically insignificant effect on economic growth in the long run. Also, from Table ', it follows that any disequilibrium in the system as a result of a shoc$ can be corrected in the short run by the error correction model. 0ence the error correction model is generated as follows which estimated the

short run ad#ustments to equilibrium. F,/ V ;nE8-9 U 0.&'(1(;nFQ-T 0.411=1;nFQ97 0.118%9;n;A> U 0.&'8=1:AE 8.&4=',7+"TA+T 0.'1=%&;n1+I8

Short5run dynam$c resu"ts 0)nR=DP $s de endent %ar$a!"e2 The fact that the variables in our model are cointegrated provides support for the use of an error correction model mechanism )F,/. representation in order to investigate the short run dynamics. Fstimation results, still based on the "chwart6 3ayesian criterion )"3,., are presented in Table 4. The R "quared is 0.%8 suggesting that such error correction model fits the data reasonably well. /ore importantly, the error correction coefficient has the e!pected negative sign and is highly significant. This helps reinforce the finding of a long run relationship among the variables in the model.

9'

94

Ta!"e @. )ong5run est$mates !ased on SBC5ARD) 0-; 4; 4; 4; -; 42 De endent %ar$a!"e $s )nR=DP


'ar$a!"e ;nFQ-T ;nFQ97 ;n1+I8 ;n;A> :AE ,7+"TA+T Coeff$c$ent 0.&'(1( 0.411=1 0.'1=%& 0.118%9 0.11'9& 8.&4=' Standard Error 0.08'='% 0.191=& 0.%%=08 0.(9&(9 0.040''= '.0&49 -iagnostic Test Test "tatistics "erial ,orrelation >unctional >orm +ormality 0eteroscedasticity ;/ Iersion ,0"T ) 1. V 0.19==4 )0.=4%. ,0"T )1. V 0.'=('' )0.'9=. ,0"T )(. V 1.84=8 )0.&94. ,0"T ) 1. V (.14'' )0.1'(. > Iersion >) 1, (8. V 0.1'9=1 )0.%0(. >) 1, (8. V 0.&4'&9 )0.44=. +ot applicable >) 1, &4.V (.1='0 )0.140. T5Rat$o 0Pro!A2 '.0'1% )0.000.XXX (.==98 )0.01(.XX 4.&=9% )0.000.XXX 0.'040( )0.=88. (.(%8& )0.0&4.XX 1.0%04 )0.0=4.X

"ourceG ,omputed by Author using /icrofit Iersion '.1 developed by 9esaran and "hin )1999.. +oteG XXX, XX )X. imply significance at the 1, 4 H 10 percent levels respectively.

9=

Ta!"e 6. Short5run dynam$c resu"ts 0)nR=DP $s de endent %ar$a!"e2 'ar$a!"e d;nFQ-T d;nFQ97 d;n1+I8 d;n;A> :AE ,7+" F,/) 1. E "quared Coeff$c$ent Standard Error T5Rat$o 0.1=8'9 0.040944 &.&0=% 0.(419% 0.084&'' (.94(' 0.148(8 0.0%=(81 (.0%40 0.1&19% 0.0'8(&1 (.%&=( 0.(=198 0.101'% (.4818 '.1144 1.409& (.%(=8 0.'9('9 0.1(1&0 '.0=0( 0.%81%= E 3ar "quared 0.1(490 > stat. >) =, &0. BPro!C )0.00(.XXX )0.00=.XXX )0.04'.XX )0.01&.XX )0.018.XX )0.011.XX )0.000.XXX 0.=8080

".F. of Eegression

=.%((9 )0.000.XXX (8.=808

/ean of -ependent Iar. 0.'%%9F & Eesidual "um of "quares 0.'49== A$ai$e 1nfo. ,riterion (0.=808

".-. of -ependent Iariable 0.1%'%( Fquation ;og li$elihood "chwar6 3ayesian ,riterion 1'.(&%1

-: statistic (.08&% "ourceG Fstimation results using /icrofit Iersion '.1. +oteG XXX, XX imply significance at the 1 and 4 percent levels respectively.

,onsistent with the long run results in table ', the coefficient of e!ternal debt in table 4 has the theori6ed negative impact on economic growth while investment as a percentage of 8-9 and labour force have the theori6ed positive impact on economic growth respectively. F!ports on the other hand have a negative impact on economic growth in the short run. The dummy, :AE, is negative as in the long run and e!erts statistically significant coefficients on economic growth. The coefficient of the error correction term is negative as e!pected. 8ranger causality tests results

9%

1n order to e!amine the predictability of e!ternal debt on economic growth, 8ranger causality test was applied to measure the linear causation among the variables. The result of the 8ranger causality test is reported in Table =.

Ta!"e D. Pa$r*$se granger causa"$ty tests resu"tsA


+ull 0ypothesisG > "tatistics 9rob. Eemar$sG

;nFQ-T does not 8ranger ,ause ;nE8-9 ;nE8-9 does not 8ranger ,ause ;nFQ-T ;nFQ97 does not 8ranger ,ause ;nE8-9

(.4(&48 0.09=8'X

+ull re#ected

(.%9%89 0.=%489 +ull not re#ected 1.&9&1' 0.(=(94 +ull not re#ected

;nE8-9 does not 8ranger ,ause ;nFQ97 1.(8&9( 0.(908& +ull not re#ected ;n1+I8 does not 8ranger ,ause ;nE8-9 &.%'99( 0.0&''(XX +ull re#ected

;nE8-9 does not 8ranger ,ause ;n1+I8 0.'1%0' 0.%0'4' +ull not re#ected ;n;A> does not 8ranger ,ause ;nE8-9 ;nE8-9 does not 8ranger ,ause ;n;A> 0.1('(% 0.88&4% +ull not re#ected 0.=9&'4 0.40%(0 +ull not re#ected

"ourceG Fstimation results using F views =. +oteG XX and X denote 4 H 10 percent significant levels respectively. The 8ranger causality test results in Table = clearly suggest that causation relation between economic growth and e!ternal debt is from one side, that is, there is a unidirectional relation between economic growth and e!ternal debt. Therefore, the null hypothesis of e!ternal debt does not 8ranger cause real 8-9 can be re#ected, implying e!ternal debt does 8ranger cause economic growth. 0owever, the null hypothesis that investment does not 8ranger cause economic growth can be re#ected2 implying that investment does 8ranger causes economic

98

growth. 0owever, the null hypotheses that real 8-9 does not 8ranger cause labour force and e!ports2 vice versa cannot be re#ected.

D$scuss$on >rom the results in Table ', it is seen that e!ternal debt, e!ports and investment as a ratio of 8-9 have the e!pected signs and e!ert statistically significant effects on real 8-9 in the long run. This implies that a 1 percent increase in e!ternal debt decreases real 8-9 by 0.&' percent in the long run. The negative relationship between e!ternal debt and economic growth as revealed in this study is in line with 5rugman?s )1988. Adebt overhangB theory and other researchers. The Adebt overhangB theory argues that when foreign debt becomes e!cessive2 actual payments to creditors become lin$ed to the economic performance of the debtor country. Therefore potential increases in debt payments depress the returns to productive investment and discourage capital formation. 1n such circumstances, the debtor country shares only partially in any increase in output and may be e!ports because a fraction of that increase is used to service the foreign debt. This may wea$en incentives to invest from the point of view of the debtor country as a whole, because the effective return to investment is reduced. The negative and statistically significant effect of e!ternal debt on economic growth is an indication that e!ternal debt adversely affected economic growth in "ierra ;eone. This result is also consistent with the results by "afdari and Abouie )(011. for 1ran that e!ternal debt has a negative effect on economic growth. /ali$ et al )(010. found a statistically significant long run negative

99

relationship between e!ternal debt and economic growth for 9a$istan. 5aragol )(00(. also found a statistically significant negative effect of e!ternal debt on economic growth for Tur$ey. The results however contradict the findings by Ayadi and Ayadi )(008. in their comparative study of +igeria and "outh Africa and 3entum Fnnin )(009. for 8hana. The findings by Ayadi and Ayadi )(008. showed a positive and statistically significant long run relationship between e!ternal debt and economic growth for "outh Africa but however found insignificant long run relationship between e!ternal debt and economic growth for +igeria. 3entum Fnnin )(009. on the other hand found a positive and statistically significant long run relationship between e!ternal debt and economic growth for 8hana implying that 8hana is not suffering from debt overhang problem. Eesults by :i#eweera et al. )(004. also showed that "ri ;an$a does not have a debt overhang problem, probably because total e!ternal indebtedness is not too high. The coefficient of e!ports in the long run growth equation is positive and significant at 4 percent significance level. The sign of the e!ports variable supports the theoretical conclusion that trade openness contributes positively to economic growth. >rom the results in Table ', the coefficient of e!ports is 0.41 which means that a 1 percent increase in e!ports leads to appro!imately 0.41 percent increase in real 8-9 growth. This implies that e!ports have a very high significant impact on economic growth in "ierra ;eone in the long run. This result is consistent with the theoretical e!pectation of the classical views on the role of trade in the macro economy. 1t is also consistent with other empirical

100

studies such as Jani$$aya )(00&., :ac6iarg )(001., F$anaya$e )1999., Fdwards )199(. and "achs and :arner )199%.. The results obtained goes to suggest that the trade liberali6ation adopted as part of the economic reform programme in 1991 in "ierra ;eone has helped open up the economy and had raised economic growth. This emphasi6es the fact that trade enhances competition and efficiency as well as transfer of technology and $nowledge and hence growth. 9rior to the period chosen for the study, "ierra ;eone?s e!ports were mainly diamonds, bau!ite and iron ore whose prices were constantly fluctuating with a general downward trend. 0owever, during the trade liberali6ation period )the period covered by this study., e!port diversification became one of the primary ob#ectives. Traditional products such as palm oil and handicrafts became important e!port products. 1t can thus be said that this diversification policy has raised earnings or revenue from e!ports which has impacted positively on economic growth. The importance of investment )pro!ied by investment as a share of 8-9. is emphasi6ed by the strongly positive and statistically significant relationship that it e!hibits with economic growth. This implies that a 1 percentage increase in investment as a share of 8-9 is associated with a 0.'( percentage increase in real 8-9. This result validates the neoclassical growth theory )"olow, 194=2 0unt, (00%. and is consistent with the empirical findings by "chmidt 0ebbel H "olimano )199=.. Fconomic theory holds that higher rates of savings and investments are important determinants of the long run growth rate. The suggestion behind "olow?s )194=. framewor$ is that higher investments and savings rates lead to more accumulated capital per wor$er and hence this results 101

in an increase in economic growth, but at decreasing rate. *nder endogenous growth theories that emphasi6e the broader concepts of capital )Eebelo, 1991., economic growth and investment tend to move together. The positive and significant impact is consistent with the findings by 5argbo and Adamu )(010. for "ierra ;eone. The dummy, :AE, has e!pected sign and statistically significant. The negative impact is an indication that economic activities shran$, with rebels holding all the mineral rich and e!port crop production districts. This contributed to a collapse in the fiscal revenue base and to significant increases, in the budget deficit, ban$ financing, and e!ternal payments arrears. 1nflation surged, and the e!change rate depreciated sharply. As a result, besides the humanitarian assistance that the country received from various sources, it had to go into series of loan contracts and deferred loan repayments which adversely affected economic growth. >rom Table 4, the coefficients of the variables provide interesting results since they maintain their signs as in the long run equation e!cept the coefficient of the e!ports variable whose sign changes from positive to negative. The coefficients indicate the short run elasticities. The coefficient on e!ternal debt suggests that short run changes in e!ternal debt have negative and statistically significant effect on economic growth. This result is also consistent with the findings by F6eabasili et al. )(011. for +igeria. The results further indicate that the coefficient of short run changes in e!ports e!erts a negative and statistically significant impact on economic growth. This negative and significant impact of

10(

e!ports on economic growth in the short run may be due to the unfavourable terms of trade. The e!ports of "ierra ;eone are mainly raw primary products which e!perience fluctuating prices in the period under review. 1n addition, the unfair competition on some of the sectors of the economy such as the te!tiles, poultry, and rice, among others that results from trade liberali6ation could e!plain the negative impact of e!ports on economic growth. Also, in the short run, some of the domestic sectors of the economy are not able to compete more favourably with their more efficient counterparts of the advanced countries. This is usually because most of the domestic industries are infant industries which produce at a relatively higher average cost. This situation reduces the productivity in these sectors which subsequently affects growth negatively. This behaviour of the
e!ports variable is consistent with both classical and protectionist arguments. The classical argument that e!ports resulting from comparative advantage leads to growth is valid in the long run while the protectionist argument that e!ports have a detrimental effect on growth is valid in the short run, as far as the "ierra ;eonean economy is concerned.

Also, short run changes in labour force e!ert a positive and statistically significant impact on economic growth. This implies that a 1 percent increase in the labour force in the short run will increase economic growth by 0.1& percent. This results show that an increase in the labour force positively impacted economic growth in "ierra ;eone in the short run. This also may be due to international connections between "ierra ;eone and a host of countries such as Fngland, 8hana, +igeria and others who imparted new s$ills and $nowhow to the country labour force. 1nvestment as a share of 8-9 also e!erts a positive and 10&

statistically significant impact on economic growth. This means that an increase in investment in the short run e!erts a positive impact on economic growth in "ierra ;eone. This positive and statistically significant impact of investment on economic growth is in line with the findings by 5argbo and Adamu )(010. for "ierra ;eone. The estimated coefficient of the lagged error correction model )F,/. is highly significant at the 1 percent level of significance and also has the appropriate negative sign. This is an indication of #oint significance of the long run coefficients. >rom the results in Table 4 the estimated coefficient of the error correction model is <0.'9('9. This negative and significant coefficient is an indication that cointegrating relationship e!ists among real 8-9, e!ternal debt, e!ports, investment as a ratio of 8-9 and labour force. The si6e of the coefficient on the error correction term )F,/. denotes that about '9 percent of the disequilibrium caused by previous years? shoc$s converges bac$ to the long run equilibrium in the current year. Thus, the study discerns that the variables in the model show evidence of moderate response to equilibrium when shoc$ed in the short run. 1t is theoretically debated that an error correction mechanism e!ists whenever there is a cointegrating relationship among two or more variables. The error correction term is thus obtained from the negative and significant lagged residual of the cointegration regression. 1t determines the speed of ad#ustment to long run equilibrium. The negative coefficient is an indication that any shoc$ that ta$es place in the short run would be corrected in the long run. The rule of thumb

10'

is that, the larger the error correction coefficient )in absolute term., the faster the variables equilibrate in the long run when shoc$ed )Acheampong, (00%.. The 8ranger causality test results in Table % suggests that the null hypothesis of e!ternal debt does not 8ranger cause real 8-9 can be re#ected, implying e!ternal debt does 8ranger cause real 8-9. 0owever, the null hypothesis that real 8-9 does not 8ranger cause e!ternal debt cannot be re#ected, implying real 8-9 does not 8ranger cause e!ternal debt. Thus, a unidirectional causality has been identified from e!ternal debt to economic growth at the 10 percent significance level. This implies e!ternal debt can theoretically impact on economic growth through a variety of avenues. 7n the one hand, if converted into capital and other domestically unavailable inputs, which are productively used, the development benefits will soon manifest themselves. 1f, on the other hand, the borrowed resources are misallocated or wasted on consumption the negative effects on productivity will, in time, haunt the economy )As!entiou and "erletis, 199=a.. -ebt burden has an output retarding effect through investment crowding out, and reduction of available public e!penditure funds in areas such as education, health and infrastructural wor$ all which have an output promoting impact. This result leads us to conclude the occurrence of a strong debt overhang in "ierra ;eone. The unidirectional causality between e!ternal debt and economic growth is in line with the findings of ,haudhary et al )(008. for 9a$istan and 5aragol )(00(. for Tur$ey. F6eabasili et al. )(011. also found a unidirectional causality between e!ternal debt and economic growth running from e!ternal debt to economic growth for +igeria.

104

There is also a unidirectional causality between investment as a percentage of 8-9 and economic growth and runs from investment to economic growth. This is an indication that investment is an important variable in achieving economic growth in "ierra ;eone.

D$agnost$c and arameter sta!$"$ty tests To complement this study it is important to investigate whether the above long run and short run relationships we found are stable for the entire period of study. >or this purpose, one needs to test for parameter stability. The methodology used here is based on the cumulative sum ),*"*/. and the cumulative sum of squares ),*"*/"T. tests proposed by 3rown et al. )19%4.. *nli$e the ,how test, that requires brea$ point)s. to be specified, the ,*"*/ tests can be used even if we do not $now the structural brea$ point. The ,*"*/ test uses the cumulative sum of recursive residuals based on the first n observations and is updated recursively and plotted against brea$ point. The ,*"*/"T ma$es use of the squared recursive residuals and follows the same procedure. 1f the plot of the ,*"*/ and ,*"*/"T stays within the 4 per cent critical bound the null hypothesis that all coefficients are stable cannot be re#ected. 1f however, either of the parallel lines are crossed then the null hypothesis )of parameter stability. is re#ected at the 4 per cent significance level. Appendices 8 and 0 clearly indicate that both the ,*"*/ and ,*"*/"T plots lie within the 4 per cent critical bound thus providing evidence

10=

that the parameters of the model do not suffer from any structural instability over the period of study.

Summary This chapter e!amined the time series properties of the data used for estimation, presented and discussed the results vividly. *nit root test employing both the A-> and the 99 techniques essentially showed that all the series had to be differenced once to achieve stationarity. This implied that all the series are integrated of order one, I)1.. The presence of non stationary variables implied the possibility of the presence of long run and causal relationships, which the study verified using the Autoregressive -istributed ;agged /odel )AE-;. and 8ranger causality test. The results disclosed a long run cointegrating relationship between e!ternal debt and economic growth in "ierra ;eone. The long run estimates, however, reveal a negative and statistically significant impact of e!ternal debt on economic growth #ust as in the short run. The negative impact is an indication that "ierra ;eone suffered from debt overhang problem during the period under consideration and this had harmful effect on economic growth in the country. 0owever, e!ports e!erted a

statistically significant positive effect on economic growth in the long run and a statistically negative effect in the short run. The positive impact suggests that e!ports had played significant role in fostering economic growth in "ierra ;eone. Also the negative impact was due to unfavourable terms of trade the country e!perienced during the short run. 1nvestment e!erted a positive and statistically

10%

significant impact on economic growth in both short run and long run. The positive impact suggests that investment is a $ey channel for economic growth in "ierra ;eone. ;abour force e!erted insignificant effect on economic growth in the long run while in the short run it e!erted positive and statistically significant effect on economic growth in "ierra ;eone. The diagnostic and parameter stability tests reveal that the model passes the tests of serial correlation, functional form misspecification, non normal errors and heteroscedasticity and the graphs of the ,*"*/ and ,*"*/"T indicate the absence of any instability of the coefficients because the plots of these graphs are confined within the 4 percent critical bounds of parameter stability suggesting that all the coefficients of the estimated AE-; model are stable over the study period. >inally, the 8ranger causality test result revealed a unidirectional causality between e!ternal debt and economic growth.

APPENDICES APPENDIE A

108

UNIT ROOT TESTS 0AD( AND PP2 (OR ORDER O( INTE=RATION )E'E)S &ITH INTERCEPT AND TREND

Iar. A-> "tats Y;;Z [,.I.\


;nE8-9 0.%(44=9 Y0Z ;nFQ-T 0.0'(&%4 Y0Z ;nFQ97 &.1&90=( Y0Z ;n1+I8 &.109&0% Y0Z ;n;A> ;n-" 0.=9411= Y1Z (.890=94 Y0Z &.((1%(8 &.((1%(8 &.((1%(8 &.((1%(8 &.((1%(8 &.((1%(8

)9.I.. 99 "tats. Y3:Z [,.I.\


)0.9=&%. )0.99'1. )0.11((. )0.1(&0. )0.9=&%. )0.1%=%. 0.48(((8 Y(Z 0.=1%='0 &.10%&4( &.1=9=41 1.8&49(1 (.8((&'= Y=Z Y1Z Y&Z Y'Z Y'Z &.((1%(8 &.((1%(8 &.((1%(8 &.((1%(8 &.((1%(8 &.((1%(8

)9.I..
)0.9%''. )0.999&. )0.119(. )0.1101. )0.==10. )0.1984.

"ourceG ,omputed by Author using Fviews =. +oteG ;; is lag length2 3: is bandwidth, ,I is critical value and 9I is probability value.

APPENDIE B UNIT ROOT TESTS 0AD( AND PHI))IPS PERRON2 (OR ORDER O( INTE=RATION -ST DI((ERENCES &ITH INTERCEPT AND TREND

109

Iar. A-> "tats Y;;Z [,.I.\


;nE8-9 ;nFQ-T ;nFQ97 ;n1+I8 ;n;A> ;n-"

)9.I..

99 "tats. Y3:Z [,.I.\


Y'Z Y4Z Y&Z '.((=814 '.((=814

)9.I..
)0.001'.XXX )0.0001.XXX

4.1'4&8& Y0Z '.((=814 )0.0014.XX X 4.1=8&% =.10='== Y0Z %.(9=41( Y0Z '.((=814 )0.0001.XXX =.1=899 '.((=814 )0.0000.XXX %.&1%&&

'.((=814 )0.0000.XXX

9.=%(&89 Y0Z '.((=814 =.'41'=4 Y&Z '.((=814 4.1=%&48 Y0Z '.((=814

)0.0000.XX X 8.1('41 Y1(Z '.((=814 )0.0000.XXX )0.00'&.XXX '.=4494 Y&Z '.((=814 )0.00'=.XXX )0.001'.XXX

)0.001'.XXX 4.1%&&8 Y1Z '.((=814

"ourceG ,omputed by Author using Fviews =. +oteG XXX, denotes 1 percent level of significance, ;; is lag length, 3: is bandwidth, ,I is critical value and 9I is probability value.

APPENDIE C UNIT ROOT TESTS 0AD( AND PHI))IPS PERRON2 (OR ORDER O( INTE=RATION )E'E)S &ITH INTERCEPT ON)?

Iar. A-> "tats Y;;Z [,.I.\

)9.I.. 99 "tats. Y3:Z [,.I.\

)9.I..

110

;nE8-9 ;nFQ-T ;nFQ97 ;n1+I8 ;n;A> ;n-"

1.'1&(9& Y0Z 1.%'04%= Y0Z 1.'414'( Y1Z (.0=&=40 Y0Z 0.%=48&= Y'Z 0.9((14= Y0Z

(.=((989 (.=((989 (.=((989 (.=((989 (.=((989 (.=((989

)0.4=(0. )0.'0&'. )0.4'(8. )0.&'08. )0.991(. )0.%==4.

1.4('&=0 Y1Z 1.%'04%= Y0Z 1.09'4=( Y(Z (.=11=%0 Y(Z 0.91%&4= Y'Z 1.11'014 Y&Z

(.=((989 (.=((989 (.=((989 (.=((989 (.=((989 (.=((989

)0.40%&. )0.'0&'. )0.%0'(. )0.(41%. )0.99'&. )0.=9=4.

"ourceG ,omputed by Author using Fviews =. +oteG ;; is lag length2 3: is bandwidth, ,I is critical value and 9I is probability value.

APPENDIE D UNIT ROOT TESTS 0AD( AND PHI))IPS PERRON2 (OR ORDER O( INTE=RATION -ST DI((ERENCE &ITH INTERCEPT ON)?

Iar. A-> "tats Y;;Z [,.I.\

)9.I..

99 "tats. Y3:Z [,.I.\

)9.I..

111

;nE8-9 &.84(&(0 Y0Z &.=(10(& )0.00=8.XX X ;nFQ-T 4.4'9008 Y0Z &.=(10(& ;nFQ97 %.&%900' Y0Z &.=(10(& ;n1+I8 9.'%00=4 Y0Z &.=(10(& ;n;A> ;n-" (.=9%48% Y0Z '.9&0%9' Y0Z )0.0000.XXX )0.0000.XXX

&.8'&(0= Y1Z 4.4'=1=8 Y(Z %.&894(' Y(Z Y(Z

&.=(10(& &.=(10(& &.=(10(& &.=(10(& (.=(41(1 &.=(10(&

)0.00=9.XXX )0.0000.XXX )0.0000.XXX )0.0000.XXX )0.09=4.X )0.000'.XXX

)0.0000.XX X 10.4&&8(

(.=(41(1 )0.0=8=.X &.=(10(& )0.0004.XXX

(.='&9&= Y&Z '.9'884' Y(Z

"ourceG ,omputed by Author using Fviews =. +oteG XXX, X denote 1 percent and 10 percent level of significance respectively2 ;; is lag length, 9I is probability value, ,I is critical value and 3: is bandwidth.

APPENDIE E Data "ot of %ar$a!"es $n "e%e"s

11(

"ourceG :orld -evelopment 1ndicators and 3an$ of "ierra ;eone.

APPENDIE ( Data "ot of %ar$a!"es $n f$rst d$fference

11&

"ourceG :orld -evelopment 1ndicators and 3an$ of "ierra ;eone.

APPENDIE = P)OT O( CU>U)ATI'E SU> O( RECURSI'E RESIDUA)S 0CUSU>2

11'

15 10 5 0 -5 -10 -15 1973 1978 1983 1988 1993 1998 2003 2008

>igure 'G 9lot of ,umulative "um of Eecursive Eesiduals The straight lines represent critical bounds at 4 percent significance level "ourceG Author?s computation using /icrofit Iersion '.1

114

APPENDIE H P)OT O( CU>U)ATI'E SU> O( SFUARES O( RECURSI'E RESIDUA)S 0CUSU>SF2

1.5 1.0 0.5 0.0 -0.5 1973 1978 1983 1988 1993 1998 2003 2008

>igure =G 9lot of ,umulative "um of Eecursive Eesiduals The straight lines represent critical bounds at 4 percent significance level "ourceG Author?s computation using /icrofit Iersion '.1

RE(ERENCES

11=