FACULTY OF COMMERCE

DEPARTMENT OF ECONOMICS

The Relationship Between Public External Debt and Private Investment in

Zimbabwe (1980-2016)

BY

CHIPUNZA JOHANNES A. T

R147114A

SUPERVISOR- MR R. MANDISHEKWA

This dissertation is submitted to the Department of Economics in partial fulfilment of the

requirements for the Bachelor of Commerce in Economics (Honours) Degree at Midlands
State University, Gweru, Zimbabwe

May

©2018
 SUPEVISOR’S APPROVAL FORM
The undersigned certify that they have supervised Chipunza Johannes‟ dissertation entitled:
The Relationship Between Public External Debt and Private Investment in Zimbabwe
(1980-2016), submitted in partial fulfilment of the requirements for the Bachelor of
Commerce Economics Honours Degree at the Midlands State University.

Supervisor’s Signature

Chapter ONE

Chapter TWO

Chapter THREE

Chapter FOUR

Chapter FIVE

Date:

i
 APPROVAL FORM

The undersigned certify that they have supervised, read and recommend to the Midlands state
University for acceptance, a project entitled, “ The relationship between public external debt
and private investment in Zimbabwe (1980-2016)”, submitted by Chipunza Johannes
(R147114A) in partial fulfilment of the requirements of the Bachelor of Commerce
(Honours) Degree in Economics.

………………………………………………… date /…………/…………/..............

(Signature of Student)

..................................................................... date /…………/…………/.................

(Signature of Supervisor)

………………………………………………… date /…………/…………/...............

(Signature of the Chairperson)

……………………………………………… date /…………/................./.................

(Signature of the Examiner(s))

ii
 DECLARATION FORM
I, Chipunza Johannes Registration number A.T R147114A do hereby declare that this project
is an original copy of my own work and has not been published before or submitted to any
other institution/university.

iii
 DISCLAIMER
This dissertation is submitted in partial fulfilment of the Bachelor of Commerce Honours
Degree in Economics at Midlands State University. The ideas in this dissertation represent
solely those of the author. Therefore, the University, Economics Department and the
Supervisor are not liable for errors and mistakes in this dissertation.

iv
 DEDICATION
This dissertation is dedicated to my mother, Lynette Hoyi for her unbound and unconditional
support towards my academic journey. You have turned this dream into a reality. I gratefully
thank you for the financial and emotional support. Utmost dedications also go to my lovely
sisters Johanna, Paidamoyo and Tawonashe. I thank you all for the unconditional love and
care. Special dedications also go to my grandfather Jackson Hoyi who has been a
fundamental source of motivation and guidance.

v
 ACKNOWLEDGEMENTS
The achievement of this academic success borders on a series of very gracious and helping
hands. These hands have been the light upon which this beautiful dream has been achieved. I
am extremely grateful to my supervisor, Mr R Mandishekwa, for his patient guidance and
continued assistance and review of this research project. Regards also belong to the
Economics Department for its assistance assisting in this journey of academic growth.
Special thanks also belong to my beloved friends and colleagues Taonga, Achbald, Tafadzwa
and Tariro for your priceless assistance on this research study and the economics degree as a
whole. Special regards also go to my dear friend Varaidzo for the moral support and belief in
my intellect. I greatly thank you.

vi
 ABSTRACT
Progressive public indebtedness in Zimbabwe particularly in the form of foreign financing
has culminated into debt distress in the nation. The country has accumulated stocks of unpaid
public foreign debts over the years and as a consequence debt burden began to develop in the
nation. Apparently such accumulation of external debt stocks is posited to lead to debt
overhang and determent of private investment. As such this research study examined the
relationship between public external debt and private investment in Zimbabwe over the
period 1980-2016. The study regressed private investment (PINV) on current external debt
(EXD), past external debt (LAGEXD), public investment (PUBI), household final
consumption expenditure (HFCE), debt servicing (DEBTSERV), trade openness (TO) and
real interest rate (R). The study aimed at addressing whether external debt accumulation
(debt overhang) deters investment and also addressing the crowding out role of external
indebtedness on private investment. The study found a negative relationship between public
external indebtedness and private investment. Both current and past external debt flows were
seen to deter private investment although current debt flows were seen to have insignificant
influence on private investment. However, debt service was found to promote private
investment. The study therefore concluded that the existence of debt overhang as a
consequence of the accumulation of past external debt stocks is discouraging private
investment into the economy. The study recommend on the formulation of policies aimed at
increasing national savings at the same time reducing public sector access to foreign lending
such that private sector confidence in the economy can be promoted. Also, policies that
maintain desirable servicing of the external debts can mitigate the negative effects of debt
accumulation on private investment.

vii
 TABLE OF CONTENTS
SUPEVISOR’S APPROVAL FORM .......................................................................................
APPROVAL FORM ................................................................................................................ii
DECLARATION FORM ...................................................................................................... iii
DISCLAIMER......................................................................................................................... iv
DEDICATION.......................................................................................................................... v
ACKNOWLEDGEMENTS ................................................................................................... vi
ABSTRACT ............................................................................................................................vii
TABLE OF CONTENTS .................................................................................................... viii
LIST OF ABBRVIATIONS ................................................................................................... xi
LIST OF APPENDICES .......................................................................................................xii
LIST OF FIGURES ............................................................................................................. xiii
LIST OF TABLES ................................................................................................................ xiv
CHAPTER ONE ...................................................................................................................... 1
INTRODUCTION.................................................................................................................... 1
1.0 Introduction ................................................................................................................. 1
1.1 Background of the Study .................................................................................................. 1
1.2 Problem Statement ........................................................................................................... 4
1.3 Objectives of the Study .................................................................................................... 4
1.4 Research Hypothesis ........................................................................................................ 4
1.5 Significance of the Study ................................................................................................. 4
1.6 Limitations of the Study ................................................................................................... 5
1.7 Study Delimitations .......................................................................................................... 5
1.8 Organization of the Rest of the Study .............................................................................. 5
CHAPTER TWO ..................................................................................................................... 6
LITERATURE REVIEW ....................................................................................................... 6
2.0 Introduction ................................................................................................................. 6
2.1 Theoretical Literature Review .......................................................................................... 6
2.2 Empirical Literature Review .......................................................................................... 11
2.3 Conclusion...................................................................................................................... 15
CHAPTER THREE ............................................................................................................... 16
METHODOLOGY ................................................................................................................ 16
3.0 Introduction .................................................................................................................... 16

viii
 3.1 Model Specification ....................................................................................................... 16
3.2 Measurement and Justification of Variables .................................................................. 17
3.2.1 Private Investment (PINV) ...................................................................................... 17
3.2.2 External Debt (EXD) ............................................................................................... 17
3.2.3 External Debt Lagged (LAGEXD) .......................................................................... 17
3.2.4 Debt Service (DEBTSERV) .................................................................................... 18
3.2.5 Public Investment (PUBI) ....................................................................................... 18
3.2.6 Household Fixed Consumption Expenditure (HFCE) ............................................. 18
3.2.7 Real Lending Interest Rate (R) ................................................................................ 19
3.2.8 Trade Openness (TO) .............................................................................................. 19
3.3 Data Types and Sources ................................................................................................. 19
3.4 Diagnostics Tests............................................................................................................ 19
3.3.1 Unit Root Test For Stationarity ............................................................................... 19
3.3.2 Cointegration Test ................................................................................................... 20
3.3.3 Multicollinearity Test .............................................................................................. 20
3.3.4 Heteroscedasticity Test ............................................................................................ 21
3.3.5 Autocorrelation Test ................................................................................................ 21
3.3.6 Model Specification Test ....................................................................................... 21
3.3.7 Normality Test ......................................................................................................... 22
3.5 Conclusion...................................................................................................................... 22
CHAPTER FOUR .................................................................................................................. 23
PRESENTATION AND INTERPRETATION OF RESULTS ......................................... 23
4.0 Introduction .................................................................................................................... 23
4.1 Summary Statistics ......................................................................................................... 23
4.2 Diagnostic Tests ............................................................................................................. 24
4.2.1 Unit Root Testing Results........................................................................................ 24
4.2.2 Cointegration Test Results....................................................................................... 25
4.2.3 Multicollinearity Test Results ................................................................................. 25
4.2.4 Heteroscedasticity Test Results ............................................................................... 26
4.2.5 Autocorrelation Test Results ................................................................................... 27
4.2.6 Model Specification Test Results ............................................................................ 27
4.2.7 Normality Test Results ............................................................................................ 27
4.3 Presentation of Regression Results ................................................................................ 28

ix
 4.4 Interpretation of Results ................................................................................................. 29
CHAPTER FIVE ................................................................................................................... 32
SUMMARY, CONCLUSIONS AND POLICY RECOMMENDATIONS ....................... 33
5.0 Introduction .................................................................................................................... 33
5.1 Summary of the Study .................................................................................................... 33
5.2 Conclusions .................................................................................................................... 34
5.3 Policy Recommendations ............................................................................................... 34
5.4 Suggestion for future researches .................................................................................... 35
REFERENCE LIST ............................................................................................................... 35
APPENDICES ........................................................................................................................ 41

x
 LIST OF ABBRVIATIONS
ADF Augmented Dickey Fuller
BPG Breusch-Pagan-Godfrey
CLRM Classical Linear Regression Model
DW Durbin-Watson
ESAP Economic Structural Adjustment Programme
EVIEWS Econometric View
GDP Gross Domestic Product
GNP Gross National Product
GFCF Gross Fixed Capital Formation
HIPC Highly Indebted Poor Country
IMF International Monetary Fund
LDC Less Developed Country
OLS Ordinary Least Squares
RBZ Reserve Bank of Zimbabwe
WAMZ West African Monetary Zone
ZIMSTAT Zimbabwe National Statistical Agency

xi
 LIST OF APPENDICES

Appendix 1: Data Set 40

Appendix 2A: Stationary Tests 41
Appendix 2B: Cointegration Results 46
Appendix 2C: Multicollimearity Test Results 46
Appendix 2D: Heteroscedasticity Test Results 46
Appendix 2E: Autocorrelation Test Results 47
Appendix 2F: Model Specification Test Results 48
Appendix 2G: Normality Test Results 49
Appendix 3: Regression Results 49
Appendix 4: Summary Statistics 50

xii
 LIST OF FIGURES
Figure 1.1: External debt indicators and Private Investment Trends2

xiii
 LIST OF TABLES

Table 4.1: Summary Statistics 23

Table 4.2: ADF Unit Root Test Results 24

Table 4.3: Johansen Cointegration Test Results 25

Table 4.4: Correlation Matrix Results 26

Table 4.5: Breusch-Pagan-Godfrey Test Results 26

Table 4.6: Breusch-Godfrey LM Autocorrelation Test Results 27

Table 4.7: Ramsey RESET Test results 27

Table 4.8: Jarque-Bera Normality Test results 28

Table 4.9: OLS Regression Results 28

xiv
 CHAPTER ONE
INTRODUCTION
1.0 Introduction

External debt has progressively become a major source of funding for the majority of
developing nations most of whom do not have sufficient funds for capital formation
(Abdullahi et al, 2016). Developing countries require supplementary foreign capital to bridge
the resource gap and stimulate savings in capital formation. External debt in Zimbabwe has
increased heavily owing to inadequate supply of local and foreign currency, high levels of
consumption and low domestic savings (Bayai and Nyangara, 2013). Private Consumption in
Zimbabwe was 76.42% of GDP in 2016 having had averaged at 73.08% of GDP from 1975-
2016 (Global economy data). Gross domestic savings on the other hand stood at -1.818% of
GDP having improved from -10.276% of GDP in 2015 (World Bank, 2017). This has
emphasised the need for increased foreign financing. External debt as a ratio of GDP in
Zimbabwe stood at 60.04% in 2016 having increased from 59.33% in 2015. This growth in
the access of foreign credits has been aided by the high domestic debt burden in the country
making the International community lender of last resort to Zimbabwe. The IMF (2016) has
posited that the effects of external debt burden on the economy‟s private investment are
emanating from debt distress and liquidity constraints related to servicing of the debt burden.

1.1 Background of the Study

Indebtedness in Africa can be traced as far back as the 1960s and 1970s where a host of
African countries received foreign credit for political and economic stabilization purposes, in
the post-independence era. The loans were meant for investment in economic development
and social equity, Colgan (2001). In the 1970s, loans were received from western banks
which had experienced revenue surpluses during the oil crisis. However, the loans were made
with little regard to the debtor nation‟s ability to repay and inevitably led to the debt crisis of
1980s period due to the accumulation of unpaid debt and growing debt servicing costs,
Nkamleu (2006). The development of the debt crisis in the 1980s was aided by the shocks of
the 1973 oil crisis and rising world interest rates.

According to Colgan (2001), debt crisis in Sub Saharan Africa worsened in the 1980s as the
ratio of continent‟s external debt to its GNP rose from 51% in 1982 to 100% in 1992.
Consequently, the region‟s debt grew to four times its export receipts in the early 1990s. In

1
1998, debt stock in the region was estimated at $236 billion, whilst that of the whole African
continent was over $300 billion. By 2001, external debt burden in Africa was estimated to be
twice that of any other region in the world carrying 11% of the developing world's debt. This
reflects a situation of high debt distress in developing nations as theyfail to effectively
manage and service their external debt stocks.

The first approach to address the issue of growing debt burden in low income countries was
the inauguration of the Highly Indebted Poor Country (HIPC) initiative in 1996. The
initiative was introduced to assist developing countries with debt management and servicing.
The HIPC initiative aimed at reducing debt overhang and restore sustainability of the debt of
eligible nations. Debt relief was offered through debt restructuring, rescheduling and debt
forgiveness.

Zimbabwe has over the years accumulated an external debt burden through the restructuring
of its debt and access to new financing. Accumulation of arrears as a result of debt
rescheduling has also contributed massively to the prevailing debt crisis due to increasing
interest payments. Debt rescheduling was administered to the nation as it failed to meet its
debt obligations due to falling export revenues aided by the withdrawal of foreign investment,
IMF (2017). The IMF (2016) has acknowledged that Zimbabwe has failed to receive debt
relief under the HIPC initiative as the country does not qualify as a highly indebted poor
nation thereby contributing to the ever-growing external debt stocks. This has promoted the
development of debt distress in the nation as shown by the external debt indicators in the
Figure 1.1;

300.00

200.00

100.00

0.00

External Debt % of GDP

External Debt % of Exports of primary goods, services and income
Private investment % of GDP

Figure 1.1: Trend of External debt indicators and Private Investment in Zimbabwe
from 1980-2016.

Source: World Bank (2017)

2
Figure 1.1 shows an unattractive situation where external debt occupies high percentages of
GDP and export income. External debt as a percentage of exports of goods, services and
income has increased from 45.8% in 1980 to 207.5% in 2015. This growth in the ratio depicts
debt servicing constraints where majority of export receipts are used to clear the foreign
credits. During the 1980s decade, an average of 30% of Zimbabwe‟s exports where spent on
debt repayment causing a lot of resources and foreign currency to flood out of the country,
(Jones, 2011).

External debt is that part of total public debt in a nation representing arrears to foreign
creditors, (Akomolafe et al, 2015). defined as debt the as a percentage of GDP has increased
overally from 9.33% in 1980 to 60.04 in 2016 signalling the development of debt distress in
the nation. Adegbite et al (2008) posited that the higher the ratio of a nation‟s debt stock to its
output (GDP), the greater the debt burden. The increase in debt burden has been attributed by
Saungweme et al, (2016) to poor debt management and poor budgetary control.

Private investment on the other hand has not shown much positive growth. The trend in
private investment reflects a rather fluctuating and unstable pattern over time. Having
achieved independence in 1980, private investment in Zimbabwe increased from 12.31% of
GDP to 15.18% of GDP in 1981 (World Bank 2017). This increase was due to the
introduction of intervention policies by the State which aimed at shaping the growth path for
private investment, (Makuyana, et al, 2014). However, by 1989, private investment had
reduced to 10.76% of GDP (World Bank data). The decline of private investment during the
1980-1990 decade in developing nations has been attributed to the emergence of the debt
crisis in the mid-1980s facilitated by the growth of public expenditures (Iyoha, 1999).

The growth of private investment in the 1990s era was attributed to the adoption of Economic
Structural Adjustment Programme (ESAP) which promoted the privatization of state
enterprises (Makuyana, et al, 2014). However, the negative growth of private investment up
to 2008 has been accredited by Makuyana et al, (2014) to the impacts of the fast track land
reform policy, high inflation, high interest rates and unstable political environment. This has
led to capital flight during this period. The low and weak growth of private investment after
2000 has been attributed by the IMF (2017) to massive liquidity constraints emanating from
high debt service.

3
1.2 Problem Statement
Public indebtedness has been on the increase over the years due to the accumulation of
unpaid foreign debts. This has ultimately led to the development of debt distress in
Zimbabwe as shown by the external debt indicators in Figure 1.1. The external debt
indicators portray an upward trend implying growth of debt burden in the nation. In contrast,
private investment has declined on declined over the years. This divergence between external
debt and private investment poses serious consequences for economic growth as private
investment is regarded as the major driver of economic growth.This therefore warrants a
research on the nature of the relationship between external debt and private investment in
Zimbabwe.

1.3 Objectives of the Study

The main objective of the study is to undertake an empirical investigation of the relationship
between public external debt and private investment. Specifically, the study seeks:

 To analyse how external debt and private investment are related by considering the
role of debt overhang and crowding out effects on private sector capital formation.

 To provide policy recommendations.

1.4 Research Hypothesis

The study will seek to test the following hypotheses

1. H₀: External debt has no effects on private investment

H₁: External debt fully affects private investment

2. H₀: debt overhang effects do not reduce private investment

H₁: debt overhang effects reduce private investment

3. H₀: debt service payment does not crowd-out private investment

H₁: debt service does crowd-out private investment

1.5 Significance of the Study

There is enormous empirical literature on debt distress in Africa complimented by IMF and
World Bank publications. However most of these empirical studies have focused on the
external debt-economic growth nexus rather than the linkage between external debt and
private investment, for example; Iyoha (1999), Erhieyovwe, et al (2013), Siddique (2015) and

4
Ijirshar, et al. (2016), to name a few. Very few studies have however explicitly explored the
impacts of external debt on private investment. Some of the studies include the works of
Apere (2014), Akomolafe et al. (2015) and Abdullahi (2016).

Nonetheless, the most alarming factor is that the bulk of these studies have been performed
outside Zimbabwe in countries like Nigeria and Kenya therefore creating an information gap.
This study therefore seeks to address that information gap and offer more literature on debt
distress in Zimbabwe. The study also seeks to address how external debt can be more
efficiently managed in Zimbabwe therefore reducing its pressures on the fiscus. The study
will also provide advice on how to stimulate private investment in the face of increasing
external debt burden.

1.6 Limitations of the Study

Considering that the research study uses secondary data, some problems were encountered in
terms of access and availability of the data. For example some incomplete data sets were
collected from statistical institutions like the World Bank but were complemented by other
agencies like ZIMSTAT and the RBZ. Also since data on actual levels of private investment
is difficult to acquire, the study uses Gross Fixed Capital Formation (GFCF) as a proxy for
private investment

1.7 Study Delimitations

The study uses time series data for Zimbabwe for the period 1980 to 2016.

1.8 Organization of the Rest of the Study

The rest of the research study is integrated in such a way that; Chapter Two gives a review of
theoretical and empirical studies that have addressed the external debt-private investment
relationship. Chapter Three follows by outlining the methodology used by the researcher in
undertaking the econometric study. Chapter Four highlights the econometric procedures that
were used for estimation and analysis. Chapter Five concludes the research by offering a
summary on research findings and offering recommendations for future policy derivation and
implementation.

5
 CHAPTER TWO
LITERATURE REVIEW
2.0 Introduction

In this chapter the study outlines the various theoretical and empirical abstractions on the
relationship between external debt and private investment. The theoretical discuss on the
external debt-private investment problem borders from the Two Gap model, Classical view,
Keynesian theory, Debt overhang hypothesis and the Ricardian equivalence. The chapter
further goes on to summarize some of the empirical works on the relationship between public
external debt and private investment.

2.1 Theoretical Literature Review

Akomolafe et al, (2015) defines external debt as that part of total public debt representing
debt owed by citizens of a nation to creditors outside the country. The traditional role behind
public access to foreign debt accrues from its complementary role on domestic savings in
capital formation. However, there seems to be no unanimous consensus on the relationship
between external debt and private investment from theory.

The basic idea of foreign financing is derived from the Two Gap model which posits that
external finance supplements domestic resources in capital formation. The Two-Gap model
is an extension of the Harrod-Domar model of investment and growth which states that
investment (I) is a direct function of savings (S), such that: . Harrod (1939) and Domar
(1946) advocated that the amount of money invested today is equal to the amount of money
saved today and that the growth rate is dependent upon the savings rate. That is,

(2.1)

Where g is growth rate, s is the savings rate and k is the capital to output ratio.

In equation 2.1 above, high growth rates are achieved through the availability of high
domestic savings. Chenery and Strout (1966) therefore augmented the Harrod-Domar model
to explain the complimentary role of foreign savings (external debt) on domestic savings. The
Two Gap model posits that developing nations face two gaps namely the savings gap and
trade gap (also referred to as foreign exchange gap). The savings gap is the difference
between available savings and desired level of investment whereas the foreign exchange gap

6
is the difference between imports (outflow of foreign currency) and exports (inflow of
foreign currency).

Foreign capital inflows therefore act as a bridge within which insufficient domestic savings
and foreign currency are complemented to allow for higher levels of investment as shown by
equation 2.2 below:

(2.2)

Where E = National Expenditure, Y= National Output or Income, I= Investment, S= Savings,

M= Imports, E= Exports and F= Foreign Capital Inflows

In equation 2.2, foreign capital inflows eliminate the foreign exchange gap by allowing new
investment projects to be implemented, allowing the importation of capital goods and
technical assistance (Forgha et al.2014). Elimination of the savings gap promotes investment
and growth. In this case the growth rate would be determined by both domestic savings and
foreign capital as shown by the equation below;

(2.3)

Where = foreign savings (foreign debt) and and are as explained before.

Thus, from equation 2.3 it can be observed that an increase in foreign debt inflows adds to the
domestic supply of savings which therefore stimulates investment and growth.This is
supported by Akomolafe et al. (2015) who found external debt to promote investment.

Were (2001) also established the existence of a positive relationship between external debt
and private investment, but in the short run through current debt flows. This study therefore
expects to find a positive relationship between current external debt and private investment.

The theory has however been criticized by Forghaet al.(2014) for not taking into
consideration the distribution of the borrowed funds and their subsequent utilization. The
authors argue that distribution of the foreign credit towards unproductive investments like
political campaigns, purchase of luxurious vehicles, houses and serving of wage bill may not
necessarily stimulate investment and growth. These arguments also align with the arguments
put forward by the classical theory.

Classical theorists like Ricardo and Smith seem to be predominantly unfavourable to public
borrowing as they propose a laisez fare market system where economic activity is run by

7
private individuals. They denounce public expenditures as being unproductive and inefficient
as resources are ultimately wasted by the public sector relative to private investors. The
classical theorists blame state indebtedness for diverting resources from the productive
private sector to the non-productive public expenditures like the traditional roles of national
defence, diplomatic ties and public order. The classical school argued that state indebtedness
distorts private sector capital formation thereby negatively affecting the accumulation of
capital and overall economic development.

Smith (1936) for example denied the state‟s right to incur debt arguing that indebtedness
delayed the natural process of wealth creation and prosperity within the nation by diverting
resources away from the private sector towards the financing of public expenditures that do
not bear any hope of future production (Bilan, 2016). Modighiani (1961) also posited that
public debt is an intergenerational burden that leads to progressive depletion of the capital
outlay for the future generation, that is, debt burden is hereditary. This is supported by the
Loanable funds theory which states that a substantial increase in public access to loanable
funds would decrease the amount of funds available to potential private investors. This
therefore implies that increased access to foreign debt would decrease private access to
foreign credit thereby ultimately crowding out private investment.

Ricardo, together with other classical economists like Adam Smith, expressed fears of capital
flight as a result of excessive taxation for debt service, (Churchman, 2001). Ricardo (1960)
made two standing points that is, under no conditions should public expenditures be financed
by public borrowing and that immediate debt redemption was crucial. The classical theory
therefore attribute a negative relationship between public external debt and private
investment. This is supported by Iyoha (1999) who found external debt to depress capital
formation. This has however been argued against by Keynesian theorists who argue that
external indebtedness assists the economy in gaining momentum after experiencing a
recession. They therefore vote for public indebtedness as a tool for stimulating aggregate
consumption and investment in the economy.

The school attaches an optimistic role to public sector deficit financing through government
borrowing as a means for the regulation of economic activity precisely through aggregate
demand. The school gave high prominence to the derivation of demand side policies as a
means of financing economic recovery from a recession (Bilan, 2016). This lies on parallel
ground to the classical connotations on public indebtedness.

8
Some empirical researches by Bayai and Nyangara (2013), and Abdul Rashid have proved
that public investment crowd‟s in private investment. These results are in conjunction with
the Keynesian view of public debt and investment. As such, public indebtedness is viewed as
an indispensable tool for balanced growth in the economy.

Despite advocating for the positive impact of public indebtedness on economic recovery from
a recession or stagnation, some Keynesians worry that persistent public loans into the long
term may ultimately be problematic. According to Filip (2010), Beveridge, in his systematic
deficit theory, admitted that „„getting out of the crisis is based precisely on public loans to
finance an increase in government spending and therefore a budget deficit‟‟, after which the
state budget must be restored to equilibrium. Keynes (1982), in this regard, posited that it is
possible to pursue favourable levels of aggregate demand and employment using budget
deficits for the budget will correct itself and debt will be reduced. Keynesian theory
attributes a positive relationship between public external debt and private investment.

The Keynesian perspective is however viewed by modern day conventionalists like Krugman
(1988) and Elmendorf and Mankiw (1998) to be a short run phenomenon. They argue that in
the long run, the classical connotations are evident.

Modern day economist have developed a new faith in which they hold fast to both the
principles of both the Classical and Keynesian schools of thought. Coventional theorists, as
portrayed by Bilan (2016), separate their propositions in terms of time horizons, that is short
run and long run. This new faith has been conveniently term the “conventional view” because
it combines both the Keynesian support for public indebtedness and the classical disapproval
for public borrowing to stimulate economic activities in the long term (Bilan, 2016).

The conventionalists believe that in the short run, the Keynesian propositions apply but
however continued and unsustained growth of public indebtedness will lead to the eventual
crowding out of private investment, falling growth rates and weakening economic
performance, (Bilan, 2016).

Bilan (2016) attributes this conventional view to the works of Krugman (1988) and
Elmendorf and Mankiw (1998). Although Krugman does offer respect for the Keynesian
view of public expenditures by accepting that modern day economists have to accept that
Keynesian economics remain the best framework for understanding recessions and
depression, he also holds true to the fact that continued public indebtedness will eventually
lead to debt distress and reduction of private capital and weakened economic performance.

9
The threshold relationship between external debt accumulation, and private investment and
growth stems from the debt overhang theory by Krugman (1988). Krugman defined debt
overhang as being a case where the external debt burden is so inherently large that foreign
lenders cannot expect with full confidence to be fully repaid. Krugman (1988) further asserts
that when the expected present value of potential resource transfers, say export earnings is
less than the debt stock, then current debt service will be insufficient to fully cater for the
actual, total debt stock thus creating disincentives for investment.

The theory was developed from the firm level debt overhang theory by Myers in 1977. Myers
was looking at the debt overhang effect on firm level investment. Myers positioned that an
unsustainable debt will force the firm to make promised payments to the creditors which must
be serviced via the transfer of revenue streams. This transfer of cash flows was posited to
ultimately lead to the decline in the firm‟s investment as the firm derails on intended
investment projects. Ochhino (2010) notes that, „„a debt overhang distortion acts like a tax on
the increase in the firm‟s value generated by new investment projects, and this may lead them
to forego investment projects with positive net present values‟‟.

Krugman therefore likened a debtor nation to something like a debtor firm, although he
recognised that the parallel is not exact. Consequently, Krugman stated that the debtor nation
will allocate some portion of its national income towards debt service allowing it to
effectively manage its debt. The debt overhang theory however postulates that the
accumulation of such debt burden will provide disincentives for investment as the debt acts as
a tax on domestic private investors. The returns from local and foreign investors in the
domestic economy are thus transferred (taxed away) to the creditors as debt repayment. This
therefore makes private investment less lucrative thereby reducing the incentives for
investment (Claessens et al., 1996).

The secondary effects of debt overhang on private investment accrue as a result of poor credit
ratings and reduced access to future borrowings. A debt overhang implies growing inability
of the indebted nation to payback its debts which may discourage further lending which will
then ultimately deter investment. Various scholars have established the existence of a non-
linear, threshold relationship between external debt and private investment. These include:
Apere (2014), Adegbite (2008) and Ayadi and Ayadi (2008). Apere (2014) postulated that the
relationship between external and private investment is non-linear, where debt first promotes
investment but however accumulation of external debt will end up discouraging private

10
investment. Thus the expected relationship between external debt and private investment
entails a positive one in the short run (where current debt inflows crowd in investment) and a
negative relationship in the long run where foreign debt accumulation is seen to crowd out
private investment.

The Ricardian equivalence theory however argues against the aforementioned theorists by
positing that the effect of public indebtedness leads to a neutral effect on private investment.
Challenging the Keynesian ideology of public debt, Barro (1974) augmented Ricardo‟s
(1960) views of the negative effects of public indebtedness on capital formation by
promoting neutrality between tax financed and deficit financed government spending.
Specifically, Barro (1974) posited that an increase in government borrowing (or reduction in
taxation) today, will likely result in the accumulation of public debt which will have to be
serviced at some point in the future. This will therefore force the government to increase
future taxes so as to acquire revenue for future debt service.

The equivalence theory therefore attributes that as rational consumers can accurately predict
future tax increases, an increase in government borrowing will lead to a rise in current private
savings equal to the value of future taxes. This is so because instead of consuming or
investing the extra revenue from the tax cut, rational private individuals will simply save for
future tax payments, (Caron, 2007). Thus the positive effect arising from the Keynesian
perspective is cancelled out thereby emphasising the neutrality of public debt.

2.2 Empirical Literature Review

A variety of empirical researches have been carried out since the onset of the debt crisis in
the early 1980s. The main objective of these studies was to empirically assess the effects of
external debt on private investment and/ or economic growth. There seems to be a lot of
consensus on the existence of a rather negative relationship between external debt and private
investment. As such, most empirical studies have either confirmed the existence of debt
overhang or crowding out effects as shown below.

A research study by Jarjuet al.(2016) investigated the relationship between external debt and
economic growth in the West African Monetary Zone (WAMZ) using panel data analysis for
the period 2000-2014 and concluded that there exist significant debt overhang and crowding
out effects of external debt on economic growth and investment in the region. The study used
both a fixed effects model and a random effects model. External debt stock was found to
negatively affect economic growth in both models thereby proving the existence of a debt

11
overhang. Debt servicing also negatively affected growth in both models entailing the
existence of a crowding out effect which implied that resources were diverted from
productive investments and channelled towards debt servicing thereby reducing growth. This
aligns with Ricardo‟s (1960) tax incidence connotations.

An empirical study by Akomolafeet al.(2015) researched on the relationship between public

debt and private investment in Nigeria using time series data for the period 1980-2010. The
study divided public debt into domestic debt and external debt. Private investment was then
regressed on domestic debt, external debt, real interest rate and real GDP. The study then
implemented a Vector Error Correction Model (VECM) to address the long run relationship
between the variables. Results from the normalized cointegrating equation supported the Two
Gap model by establishing a positive relationship between external debt and investment in
the long run where a 10% increase in external debt would lead to a 2% increase in Gross
Domestic Investment. External debt was however found out to crowd out investment in the
short run. Domestic debt on the other hand was found out to crowd out investment in both the
short run and long run.

A research study by Apere (2014) also investigated the effects of total public debt on private
investment in Nigeria using a non-linear model where private investment was regressed on
domestic debt, domestic debt squared, external debt, external debt squared, and private
consumption expenditure as a percentage of GDP. The study used time series data for the
period 1980-2012. The relationship was estimated using the instrumental variables technique
and boot strapping technique for the estimation of standard errors for the turning points.
Domestic debt was found to have a linear but positive relationship with private investment.
External debt was found to have a turning point of 124.69%. Apere concluded that „„unless
external debt as a ratio of GDP reaches some threshold value (124.69%) that is large enough
for meaningful investment, the impact of external debt on private investment in Nigeria will
always be negative‟‟. Apere (2014) therefore highlighted the existence of a debt overhang in
Nigeria.

Applying an endogenous growth model in studying the relationship between public external
debt and growth (Casares, 2015) finds a non-linear but U-shaped relationship between debt
and growth in Mexico. The results are consistent with the findings of Cohen (1983) and
Sachs (1986) which are also in line with the debt overhang theory. Casares states that beyond
a certain level, public external debt becomes unsustainable such that increases in the external

12
public debt will lead to a subjective increase on the country‟s risk premium and interest
payments such that domestic savings decline thereby ultimately leading to a reduction of
capital formation. Therefore, Casares like Apere (2014) also established a debt overhang
relationship between external debt, capital accumulation and growth.

A study by Hunte (2003) found out that debt servicing has contributed to the decline of
domestic savings and investment in Sub-Saharan African countries. This was supported by
Easterly (1999) who postulated that the provision of debt relief without sufficient
improvement in the domestic saving culture may be ineffective in eradicating debt distress in
highly indebted countries.

An empirical investigation by Chaudhry et al.(2009) investigated the impact of foreign debt

on savings and investment in Pakistan using time series data for the period 1973-2006. The
study utilized a Vector Autoregressive (VAR) model and found partial evidence that foreign
debt had contributed favourably to domestic savings and investment in Pakistan. The study
recognized that some portion of external debt has been channelled towards consumption
rather than investment expenditures thus the limited positive effect on investment and growth
in the economy.

A study by Adegbiteet al. (2008) examined the impact of Nigeria‟s external debt on
economic development. The study implemented both linear and non-linear simultaneous
models for economic growth and private investment. The study used generalized least
squares to estimate both the non-linear growth and private investment models. The non-linear
growth model was found to suffer from serial autocorrelation rendering parameter estimates
inefficient. The non-linear private investment model was found to be fit for data analysis as
shown by the absence of autocorrelation. In the model private investment was regressed on
the traditional debt burden indicators, that is, external debt ratio of GDP and debt service
ratio of exports. External debt squared was found to have a non-linear effect on private
investment thereby proving the existence of a debt overhang. Debt servicing was also found
to crowd out private investment.

An empirical research by Chowdhury (1994) used a structural simultaneous equation model

to capture the relationship between external debt, capital accumulation and outputin selected
developing countries of the Asia and Pacific regions. The study used panel data for the period
1970-1988 and found out that external debt had a direct and substantially large effect on
Gross National Product (GNP) whereby a 1% increase in external debt would raise GNP by

13
0.24% in the Asia-Pacific region. On the other hand, debt was found as having an indirect
effect on GNP by lowering private investment. The indirect impact of crowding-out private
investment was observed to be substantially larger than the direct effects.

An empirical study by Iyoha (1999) which modelled the macroeconomic impacts of external
debt on economic growth in Sub-Saharan Africa concluded that there exist significant debt
overhang and crowding-out effects in the region. The study used simultaneous equation
models for output and investment demand. Huge stocks of external debt and high debt
servicing were found to depress capital formation, features of debt overhang burden and
crowding out effects respectively.

Another study by Mjema (1996) also analysed the impacts of foreign debt on economic
growth in Tanzania using simultaneous equation models. The scholar observed that debt
service had a negative impact on growth. Amoating and Amoaku-Adu (1996) observed that if
a greater proportion of export earnings is used to service external debts, then little foreign
currency will be made available for investment. This therefore subscribes to the classical
disregard for public access to foreign debt.

Investigating the impact of external debt on economic growth in Kenya,Were (2001)

implemented a simultaneous equation model for the time period 1970-1995. The study used
simultaneous equations for growth and investment. In the growth equation, current debts
stocks had a negative coefficient implying that an increase in current debt stocks would lead
to a decline in economic growth. Current debt flows had however, a positive impact on
private investment in the investment equation. Were concluded that foreign debt supplements
domestic resources in capital formation, if used productively. External debt accumulation had
a negative impact on private investment thereby implying the existence of a debt overhang.
Current debt servicing was found to lead to the crowding out of private investment.

Some studies have failed to estimate a clear relationship between external debt and domestic
investment levels. With some authors like Forgha et al.(2014) finding unclear results. The
direction of external debt in Cameroon was found to be unclear as most of the funds were
presumably spent on manpower training which was hugely undermined by the existence of
high levels of unemployment and brain drain (Forgha et al., 2014).In their study they
recognise that external debt is channelled towards unproductive expenditures. They also
concur that corruption and embezzlement, and capital flight affective greatly the productivity
of the foreign loans.

14
However, some empirical studies have found evidence of the crowding in of private
investment as a result of increased debt service. Bayai and Nyangara (2013) in their study of
the determinants of private investment in Zimbabwe for the period 2009-2011 found a rather
positive relationship between external debt service and private investment. In their original
model of private investment, the study included independent variables like rate of inflation,
interest rate, savings rate, trade terms, GDP, political risk, public investment and debt service.
After using the forward stepwise model fitting technique, savings and inflation rates were
found to have insignificant influences on private investment. Results from the fitted model
produced a positive relationship between external debt service and private investment
whereby a 1% increase in the level of debt servicing, was found to crowd in private
investment by 4.42%.

A variety of increasing research studies have analysed the effect of foreign debt on private
investment by rather addressing its role on capital flight. A growing proportion of these
studies have confirmed the existence of a „„revolving door‟‟ hypothesis where external debt
comes into the country as some form of capital but simultaneously leave in the form of
private capital flight (Ajilore 2005; Chipalkatti and Rishi 2001)

2.3 Conclusion
Theories of public access to foreign financing have offered diverging views on the
relationship between external debt and private investment. As such empirical literature has
consequently produced diverging conclusions on the nature of the relationship between
external debt and private investment. However, despite lack of such unanimous consensus,
most recent empirical studies have subscribed to the Debt Overhang Theory and confirmed
the existence of both a positive and negative relationship, in the short run and long run
respectively. These include Were 2001; Apere 2014; Casares 2015 andAdegbite et al.,
2008,to name a few. Despite the growing external debt burden in Zimbabwe, little has been
researched on its implications on the economy‟s investment levels. The study therefore seeks
to bridge that information gap provide more evidence on the effects of external debt on
investment in the country. The study therefore adapts an empirical model from the reviewed
literature which will ultimately be specified and clearly justified in the succeeding chapter.

15
 CHAPTER THREE

METHODOLOGY
3.0 Introduction
This chapter highlights the methodology used in the study. The chapter starts off with the
specification of the empirical model adopted, then goes to justify the inclusion of selected
variables and finally highlights the diagnostics checks done to ensure that the model is fit for
regression analysis.

3.1 Model Specification

The main aim of the model is to determine the relationship between external debt and private
investment in Zimbabwe. As such, the study adapted a model from Apere (2014) where
private investment is regressed on domestic debt, external debt and private final consumption.
The model for this study added other variables that influence private investment such as real
interest rates and trade openness to effectively capture the determinants of private investment
(Were 2001). External debt lagged by one period (LAGEXD) is adopted from Were (2001) to
capture the effects of past debt accumulation on private which is used to proxy debt
overhang. Domestic debt variables were removed from the model since the objective of the
study is to focus on the foreign aspect of public debt. The model will be as follows:

₀ ₁ ₂ ₃ ₄ ₅ ₆
₇

Where;

β₀ is the intercept term representing autonomous investment, Uᵢ is the stochastic error term

PINV is private investment as percentage of GDP represented by gross fixed capital

formation in the private sector

EXD is external debt as percentage of GDP

LAGEXD is external debt as percentage of GDP lagged by one period to measure impact of
past debt flows that is, the accumulation of debt stocks (the variable is used as a proxy for
debt overhang)

PUBI is public investment as percentage of GDP

16
DEBTSERV is debt service measured by interest payments of external debt as percentage of
exports of goods and services capturing the crowding out effects

HFCE is household final consumption expenditure as percentage of GDP, formerly private

consumption expenditure

TO is trade openness measured as ratio of sum of imports and exports to GDP

R is the real lending rates

β1, β2, β3, β4, β5, β6, β7, and β8 are the coefficients of the explanatory variables in the
model.

3.2 Measurement and Justification of Variables

3.2.1 Private Investment (PINV)
Private investment is defined by Chibber and Leechor (1993) as investments by private
investors on new buildings, plant and equipment that are used in the production of goods and
services. Akomolafe et al., also describe private investment as the change in the value of
fixed assets plus the change in stocks of private firms. The model above specifies private
investment to be the dependent variable. Generally, literature adopts Gross fixed capital
formation, in the private sector as a proxy for private investment, (Apere, 2014). As such,
private investment covers gross outlays by the private sector on additions to its fixed
domestic assets (IMF). This study therefore used Gross fixed capital formation by the private
sector, as a percentage of GDP to represent private investment.

3.2.2 External Debt (EXD)

External Debt is that part of a nation‟s total debt that is owed to foreign creditors,
(Akomolafe et al., 2015). Foreign debt is argued to provide desirable resources for financing
profitable investment projects especially in developing nations because of their lack of
sufficient savings for capital formation, (Hunt, 2007). The foregoing view is supported by the
dual-gap concept which states that foreign capital allows developing nations to invest more
than they save domestically (McKinnon, 1964). The study therefore used external debt as a
percentage of GDP. The expected sign is positive.

3.2.3 External Debt Lagged (LAGEXD)

The ratio of external debt as a percentage of GDP lagged by one period measures the effect of
past debt flows on private investment that is the effect of external debt accumulation on
private sector capital formation. This variable was adopted from Were (2001) who divided

17
the effect of external debt on private investment by separating its influences using the time
factor. Were (2001) posited that current debt flows (external debt to GDP ratio) stimulate
investment whereas past debt flows (external debt lagged) deter private investment.

The study therefore expects a negative relationship between external debt lagged and private
investment. Such a negative sign would imply the possibility of debt overhang.

3.2.4 Debt Service (DEBTSERV)

Debt service is another traditional indicator of indebtedness, measured by interest payments
of external debt as percentage of exports of goods and services. Some empirical researches by
Ayadi (2008) have used debt service ratio of exports as a proxy for debt service. According to
Ijirshar et al., (2016), debt service is done only with export earnings or further borrowing.
The variable is added into the adopted model to capture the crowding out impacts of external
debt on private investment. Krugman (1988) argues that a high debt servicing ratio implies
growing distress on a country‟s fiscus as the nation depletes for example, its foreign currency
reserves and export receipts to cater for the accumulated external debt. The study expects to
find a negative sign.

3.2.5 Public Investment (PUBI)

There‟s a seemingly hot debate on how public investment influences private investment.
Some argue that investment in public infrastructure by the central government crowds in
private investment (Bayai and Nyangara, 2013). Also, Rashid (2006) investigated the
linkages between public and private investment in Pakistan and found out that public
investment crowd‟s-in private investment and both were complementary. Whereas other
opponents argue that investment in public expenditures crowd out private investments as it
reduces financial support for private enterprises and raises the domestic cost of borrowing,
(Jayaraman, 2008). As such, the expected sign can either be positive or negative.

A negative sign signals crowding out of private investment whereas a positive sign implies
that public investment drives private investment.

3.2.6 Household Fixed Consumption Expenditure (HFCE)

Formerly regarded as private consumption, the variable captures the amount that is consumed
by domestic households on goods and services, rather than saved. The variable is expressed
as a percentage of GDP. In the Harrod-Domar model of investment, the amount of money
invested (I) is equal to the amount of money that is saved (S), ie . Investment is
postulated to be a function of savings. Logic therefore follows that an increase in the amount

18
of money spent on private consumption reduces savings thereby ultimately reducing private
investment. The study therefore expects the sign to be negative.

3.2.7 Real Lending Interest Rate (R)

Real lending interest rate captures the cost of borrowing by the private sector. The variable is
included in the model since most investments require borrowed capital. Under the Keynesian
model of investment, capital formation is an inverse function of real interest rate. They argue
state that an increase in the interest rate makes investment more expensive and less
worthwhile. Thus, when the cost of borrowing rises, investment will fall. The expected sign is
negative reflecting that high interest rates crowd-out private investment.

3.2.8 Trade Openness (TO)

Trade openness measures trade liberalization. The variable is expressed as the sum of imports
and exports, as a percentage of GDP. A higher percentage is argued to be attractive to foreign
investors, especially given a situation of growing exports and thus having a positive effect on
net private domestic investment. The expected sign is positive.

3.3 Data Types and Sources

The study used secondary time series data which was presented in the form of percentages,
for example private investment and external debt were expressed as ratios of GDP. The
yearly time series data was therefore collected from both national and international
institutions like the ZIMSTAT, World Bank, IMF, Trading Economics, Global Economy,
Index Mundi and various RBZ and IMF publications, Secondary time series data normally
suffers from measurement error and data smoothening practices. This was however mitigated
by the adoption of data from reliable, credible, legitimate and authoritative sources.

3.4 Diagnostics Tests

To check whether a model is strong for data fitting and meaningful results can be acquired, a
series of diagnostics checks are carried out. These checks include testing for unit root,
autocorrelation, heteroscedasticity, cointegration and model misspecification.

3.3.1 Unit Root Test For Stationarity

To prevent spurious regression, a test for nonstationarity is required. Non-stationary time
series provide meaningless results thereby regarding further inference meaningless also. To
prevent this problem the study used the Augmented Dickey-Fuller (ADF) test for unit root.

19
The ADF test adopts lagged values of the variables in determining the level and order of
integration, (Gujarati, 2004). If the variables are stationary at level they are regarded as being
I(0), but however if the variables are stationary after first difference are regarded as being
I(1), and so on. The hypothesis for unit root is;

H₀: time series is nonstationary that is, has a unit root

H₁: time series is stationary

If the ADF test statistic exceeds the test critical values, one therefore rejects the null
hypothesis and concludes that the series is probably stationary.

3.3.2 Cointegration Test

Cointegration between variables is crucial for analysing the long run behaviour of the model
when the stochastic variables have a common trend, (Gujarati and Porter, 2009).
Cointegration exists when the linear combination between two non-stationary variables is
stationary. The study utilized the Johansen cointegration test in checking for cointegration.
The Johansen cointegration test is used since it applies effectively to a multivariate model.
The hypothesis of the procedure is represented below;

H₀: there is cointegration

H₁: there is no cointegration

If the t-statistic is greater than the t-critical value, one does not reject the null hypothesis and
conclude that cointegration does exist within the model.

3.3.3 Multicollinearity Test

The study will also test for multicollinearity which is the existence of correlation between
exogenous variables of the model, (Gujarati, 2004). Multicollinearity will be tested using the
correlation matrix. The correlation matrix finds the existence of correlated independent
values. The test uses the r value to measure goodness of fit of the model. A correlation of 0.8
is perceived as strong and undesirable. In such a case the researcher can drop one of the
correlated variables, and then test again whether the R-squared is still significant. The
hypothesis of the test is;

H₀: there is absence of multicollinearity

H₁: there is presence of multicollinearity

20
The null hypothesis of the test is rejected if the pairwise correlation is in excess of 0.8.

3.3.4 Heteroscedasticity Test

Heteroskedasticity refers to the violation of the classical assumption of constant variance of
the error term, that is, homoscedasticity (Gujarati, 2004). Simply put, heteroscedasticity is a
case where the disturbance term varies with some independent variable. The presence of
heteroscedasticity widens the confidence interval such that the t and F tests will produce
inaccurate results. In such a case Weighted Least Squares (GLS) or General Method of
Moments (GMM) will be the solution though simple solutions do exist. This study adopts the
Breusch-Pagan-Godfrey (BPG) test in checking for heteroscedasticity. The hypothesis of the
Breusch-Pagan-Godfrey test is as follows;

H₀: there is absence of heteroscedasticity

H₁: there is presence heteroscedasticity

At the 5% level of significance one does not reject the null hypothesis of no heteroscedastic
error terms if the p value of the BPG test statistic is in excess of 0.05. Otherwise the error
terms would be heteroscedastic.

3.3.5 Autocorrelation Test

Autocorrelation according to Gujarati (2004) refers to correlation between disturbance terms
in the model. One consequence of autocorrelation between disturbance terms is widening of
the confidence interval. As such, to prevent using a model struck with autocorrelation, the
study will use the Breusch-Godfrey (BG) Serial Correlation test for autocorrelation. The BG
test for autocorrelation adopts the following hypothesis

H₀: there is no serial autocorrelation

H₁: there is serial autocorrelation

At the 5% level of significance, one does not reject the null hypothesis of no autocorrelation
if the p value is greater than 0.05. Otherwise one would conclude that the model probably
suffers from the existence of autocorrelation.

3.3.6 Model Specification Test

Incorrect specification of the empirical model may yield incorrect results leading to
inaccurate inferences for policy recommendations. Specification errors arise due to incorrect
specification of the model which may result from over fitting and /or under fitting a model

21
and etc. The study adopted the Ramsey proposed RESET test in checking for model
specification errors. The RESET test involves the following hypothesis;

H₀: the model is probably correctly specified

H₁: the model is probably misspecified

Using the significance level of 5% one does not reject the null hypothesis if the probability
value of the RESET test statistic is in excess of 0.05. Otherwise the model would be
incorrectly specified.

3.3.7 Normality Test

The Jarque-Bera normality test is used to assess whether the disturbance term is normally
distributed with a mean of zero and constant variance. The research tests the following
hypothesis;

H₀: the disturbances are normally distributed

H₁: the disturbances are not normally distributed

We do not reject the null hypothesis of normal distribution if the Probability value of the
Jarque-Bera statistic is greater than 0.05.

3.5 Conclusion
This chapter involved a description of the methodology to be used by the researcher in
identifying the nature of the relationship between public external debt and private investment
in Zimbabwe. This involved first giving an outline of the model to be used in regression
analysis in the following chapters and then giving a justification of the adopted variables. The
chapter also highlighted the diagnostic tests to be carried out during econometric estimation
and finalized by giving a preview of the types of data used and its subsequent sources. This
chapter therefore formed the basis within which estimation, presentation, analysis and
interpretation of data is going to be carried out in the following chapter.

22
 CHAPTER FOUR

PRESENTATION AND INTERPRETATION OF RESULTS

4.0 Introduction
The major thrust of economic research is the application of econometric modelling in
addressing economic phenomenon so as to observe the relationship between two or more
economic variables and offer sound policy recommendations. Theory itself must be validated
by empirical data and findings so that it becomes a relevant picture of reality. The application
of econometric modelling will allow the researcher to practically observe the relationship
between public external indebtedness and private investment. This study therefore employs
Ordinary Least Squares (OLS) technique in estimating the relationship between public
external debt and private investment using EVIEWS 8 software. Preliminary diagnostics tests
were carried out to ensure that the estimated results are accurate and desirable for policy
analysis.

4.1 Summary Statistics

Variable Observations Mean Median Maximum Minimum Std. Dev.
PINV 37 11.5746 12.9000 21.8800 0.21000 6.19541
EXD 37 35.6300 30.9100 70.6500 5.3700 20.5768
LAGEXD 36 34.9519 30.5900 70.6500 5.37000 20.4452
PUBI 37 10.1568 6.3600 69.9500 -2.1000 15.2809
DEBTSERV 37 23.3848 23.1500 40.5600 4.5700 9.7159
HFCE 37 65.7751 64.4500 119.4100 25.3100 20.1423
TO 37 53.1459 50.1100 77.6700 22.5700 16.8399
R 37 82.9984 39.4800 572.9400 4.2600 129.5064
See Appendix 4 for full details

All variables in the model are continuous. The dependent variable, private investment (PINV)
has a mean of 11.5746 and a median value of 12.9. Its minimum value is 0.21 with a
maximum of 21.88, associated with a standard deviation of 6.19541. External debt on the
other hand has a mean of 36.63 and a median value of 30.91, with a minimum value of 5.37
and maximum value of 70.65. The maximum value of 70.65 implies high external debt
burden on the economy‟s GDP. Past debt (LAGEXD) also has a high mean of 34.9519 and
minimum and maximum values similar to those of current external debt flows (EXD). Public
investment has a slightly lower mean of 10.1568 with a maximum of and a minimum of -2.1.
The mean amount of debt service was 23.3848 with a maximum of 40.56 and a minimum of
4.7. Household consumption is high with a maximum final consumption of 119.41 and

23
minimum consumption of 25.31 with a standard deviation of 20.1423. Interest rate has the
highest standard entailed by a maximum lending rate of 572.94 and a minimum of 4.26.

4.2 Diagnostic Tests

This section outlines the diagnostic tests that were carried out to ensure that the estimated
model conforms to the assumptions of classical linear regression analysis. These include
stationarity, homoscedastity, no autocorrelation, and normality in the distribution of residuals.

4.2.1 Unit Root Testing Results

The study employed the Augmented Dickey-Fuller test for unit root in estimating the
stationarity of variables. DEBTSERV and PUBI were stationary at level whilst the rest of the
variables were stationary after first difference. Table 4.1 therefore summarises the unit root
tests for each individual variable in the model.

Table 4.2: ADF Unit Root Test Results

Variable ADF Statistic Critical Values Order of Integration

DEBTSERV -5.171543*** 1% -4.234972
5% -3.540328 I(0)
10% -3.202445
EXD -8.92098*** 1% -2.632688
5% -1.950687 I(1)
10% -1.611059
HFCE -8.257874*** 1% -2.632688
5% -1.950687 I(1)
10% -1.611059
LAGEXD -8.826574*** 1% -2.634731
5% -1.951000 I(1)
10% -1.610907
PINV -5.885158*** 1% -2.632688
5% -1.950687 I(1)
10% -1.611059
PUBI -5.192703** 1% -2.630962
5% -1.950394 I(0)
10% -1.611202
R 4.914022*** 1% -2.623688
5%-1.950687 I(1)
10% -1.611059
TO -7.781079*** 1% -2.632688
5% -1.950687 I(1)
10%-1.611059
*implies significance at 10%, **significance at 5% and ***implies significance at all levels.
Consult Appendix 2A for complete results

24
The stationarity tests in Table 4.1 reveal that all given variables except PUBI and
DEBTSERV are stationary after first difference. All the variables were stationary at all levels
of significance with the exception of PUBI which was stationary at 5% and 10% levels of
significance evidenced by the presence of two asterisks. Since the model consists of non-
stationary variables we must check for cointegration to prevent spurious regression.

4.2.2 Cointegration Test Results

The Johansen Cointegration test was utilized in checking for cointegration since the majority
of the variables were non-stationary at level. The presence of cointegration reveals the
existence of a long run relationship in the model. In such a case OLS would not produce
spurious results. Table 4.2 presents the results from the Johansen cointegration test.

Table 4.3: Johansen Cointegration Test Results

Hypothesized Trace 0.05

No. Of CE(s) Eigenvalue Statistic Critical Value Prob.**
None* 0.826896 182.9607 125.6154 0.0000
At most 1* 0.759074 121.5754 95.75366 0.0003
At most 2* 0.543796 71.76117 69.81889 0.0347
At most 3 0.493093 44.29268 47.85613 0.1010
At most 4 0.319111 20.51272 29.79707 0.3387
At most 5 0.119433 7.060273 15.49471 0.5706
At most 6 0.071823 2.608662 3.841466 0.1063
*denotes number of cointegrating equations. See Appendix 2B for full results

The Johansen cointegration trace test predicts 3 cointegration equations at the 5% level of
significance since the trace statistic exceeds the test critical values at the none, at most 1 and
at most 2 cointegrated equations. The study therefore that a long run relationship does exist
within the model. Ordinary least Squares regression method can therefore be applied.

4.2.3 Multicollinearity Test Results

The existence of high correlation between independent variables makes it difficult to analyse
the extent to which each individual regressors influence the explained variable. The study
therefore adopts the correlation matrix in analysing for the presence of multi-correlation
within the model. A correlation of 0.8 and above shows the existence of high
multicollinearity between the independent variables. Whereas a correlation below 0.8 is
deemed acceptable for regression analysis (Gujarati, 2004). Table 4.2 presents the correlation
matrix.

25
Table 4.4: Correlation Matrix Results

EXD LAGEXD PUBI DEBTSERV HFCE TO R

EXD 1

LAGEXD 0.6328 1

PUBI 0.3499 0.2777 1

DEBTSERV -0.2815 -0.2338 -0.0125 1

HFCE 0.1579 0.2050 0.4635 -0.4856 1

TO 0.4777 0.4922 0.2750 0.0141 0.4039 1

R 0.4138 0.3470 0.6493 0.0475 0.2543 0.3862 1

For full results consult Appendix 2C

The results from the pairwise correlation matrix signify the existence of moderate to weak
multicollinearity between the exogenous variables. The pairwise correlation between all
variables lies below 0.8 and as such the researcher adopts the do nothing approach as
prescribed by Gujarati (2004).

4.2.4 Heteroscedasticity Test Results

The study utilized the Breusch-Pagan-Godfrey test in checking for the presence of
heteroscedastic error terms. The test provided satisfactory results as shown in table4.4

Table 4.5: Breusch-Pagan-Godfrey Test Results

F statistic 0.414792 Prob. F(7,28) 0.8849

Obs*R-squared 3.382364 Prob. Chi-square(7) 0.8475

Scaled explained RSS 1.766958 Prob. Chi-square(7) 0.9716

See Appendix 2D for full results

The Breusch-Pagan-Godfrey test provided satisfactory results as depicted in Table 4.4. Since
the probability that the model is heteroscedastic is greater than 0.05 the study therefore
accepts the null hypothesis and concludes that the model does not suffer from
heteroscedasticity.

26
4.2.5 Autocorrelation Test Results
The Breusch-Godfrey Serial Correlation LM test was adopted in testing for autocorrelation
between disturbance terms. The results of the test are summarized in table 4.5 below

Table 4.6: Breusch-Godfrey LM Autocorrelation Test Results

F-statistic 0.148117 Prob. F(2,26) 0.8631

Obs*R-squared 0.405551 Prob. Chi-Square(2) 0.8165

Consult Appendix 2E for full results

Since the probability value of the test exceeds 0.05, the study does not reject the null
hypothesis and concludes that there‟s no serial autocorrelation within the model.

4.2.6 Model Specification Test Results

The study utilized the Ramsey RESET test in checking for specification errors. A correctly
specified model will generate an adequate picture of the relationship between external debt
and private investment. The results from the test are presented in Table 4.6.

Table 4.7: Ramsey RESET Test Results

F-statistic Probability (p) DW Statistic R² Adjusted R²

3.646378 0.0669 1.893741 0.805117 0.747374

Consult Appendix 2F for complete results

The probability value (0.0669) of the RESET test is higher than 0.05 implying that the study
must not reject the null hypothesis and conclude that the model is probably correctly
specified. Also, the DW statistic of 1.893 is greater than the R² value of 0.805 therefore
implying that the model is free from spurious regression.

4.2.7 Normality Test Results

To check for normality in the distribution of the residuals, the study utilized the Jarque-Bera
normality test. The test uses a histogram to assess normality by showing the disturbance
mean and associated levels of kurtosis and skewness as shown by Table 4.7.

27
Table 4.8: Jarque-Bera Normality Test Results

Mean Skewness Kurtosis Jarque-Bera Probability Standard

Statistic Deviation

-1.48e-16 -0.626158 2.727119 2.464137 0.291689 2.951133

Consult Appendix 2G for full results

The Jarque-Bera statistic of 2.464 has a probability value of 0.29 which is higher than 0.05
thereby implying that the disturbance term is normally distributed. The disturbance term has a
maximum kurtosis of 2.727 which is below 3 and is therefore acceptable. Rose et al (2015)
posits that the maximum acceptable range of peakedness is 3. It is therefore safe to conclude
that the disturbance term is normally distributed.

4.3 Presentation of Regression Results

The empirical model adopted in chapter 3 was estimated using OLS regression technique.
The results of the relationship between external debt and private investment are presented in
Table 4.8 together with the influences of other determinants of private investment.

Table 4.9: OLS Regression Results

Dependent Variable: PINV

Variable Coefficient Std. Error t-statistic Probability

C 2.4044376 4.058449 0.592437 0.5583
EXD -0.081406 0.041346 -1.968914 0.0589
LAGEXD -0.125441 0.037308 -3.362361 0.0023
PUBI -0.022572 0.056807 -0.397352 0.6941
DEBTSERV 0.176938 0.084178 2.101944 0.0447
HFCE 0.133598 0.045714 2.922468 0.0068
TO 0.105415 0.050867 2.072347 0.0476
R -0.022676 0.006019 -3.767336 0.0008
Consult Appendix 3 for full details

R² 0.778798
Adjusted R² 0.723497
F-statistic 14.08298
Prob (F-statistic) 0.000000

DW 1.848305

28
The estimated equation in table 4.8 above can therefore be summarized by equation 4.1 as;

4.1

In equation 4.1 PINV represents private investment, EXD is external debt, LAGEXD is
external debt lagged by one period (representing the accumulation of external debt), PUBI is
public investment, DEBTSERV is debt servicing, HFCE is household final consumption
expenditure, TO is trade openness and R is real interest rate.

4.4 Interpretation of Results

The strength of the adopted regressors in modelling the relationship between public external
debt and private investment is shown by the R² value of 0.7787 which is approximately 78%.
This R² reflects the goodness of fit of the model. The value implies that approximately 78%
of the variations in private investment is explained by the variations in the explanatory
variables while the remaining 22% is captured by the stochastic error term (which captures
influences of other variables not included in the model). Also the adjusted R² is above 0.5
implying that the model still has a good fit even after adjusting for more degrees of freedom.
This implies that the model is a good fit in explaining yearly deviations in levels of private
investment in Zimbabwe. Also the DW statistic of 1.84 is close to 2 implying that the mode
does not suffer from spurious regression and thus valid for further analysis. The regression
results in Table 4.9 however depict that current external debt (EXD) and public investment
(PUBI) have insignificant influences on private investment.

4.4.1 Current External Debt (EXD)

Current debt flows were found to have an insignificant crowding out effect on private
investment as evidenced by a p-value of 0.0589 which is greater than 0.05. This insignificant
impact may imply that it takes time for external debt to significantly affect private
investment.

4.4.2 Past External debt accumulation (LAGEXD)

External debt lagged was found to have the expected negative sign where a 1% increase in
external debt accumulation would decrease private investment in Zimbabwe by 0.125%. This
effect is statistically significant as evidenced by a p-value of 0.0023 which is below 0.05.This
therefore implies that the accumulation of unpaid foreign debts discourages investment by
restricting private access to foreign credit as foreign lenders become reluctant to release

29
further lending to private investors. For example, Zimbabwe cleared off its arrears owed to
the IMF administered Poverty reduction and Growth Trust (PRGT) funds in 2016 after a
decade long of interest accumulation, (IMF, 2017). Zimbabwe was consequently expelled
from the PRGT initiative thus restricted from accessing the poverty and growth funds. Thus,
simply put, accumulation of external debt stocks reduces private sector access to foreign
finance thereby ultimately reducing private investment as shown by the negative coefficient
of -0.125 in table 4.9. This therefore establishes the existence of a debt overhang in
Zimbabwe. This is supported by Were (2001) who also found past external debt to deter
investment in Kenya.

4.4.3 Debt servicing (DEBTSERV)

Prior to the estimation of the model, the study expected a negative relationship between debt
servicing and private investment where debt servicing was expected to crowd out private
investment. Regression analysis however produced a rather positive relationship where debt
servicing was found to have a positive coefficient of 0.1769 which was statistically
significant at 5% with a p-value of 0.0447. This therefore implies that when Zimbabwe
manages to service-off old debts it can be able to access new financing cheaply from both old
and new sources. The IMF (2017) does hold true to the fact that the ability of Zimbabwe to
clear-off old arrears to the international community would promote and fast track its re-
engagement process with international lenders. The IMF (2017) does take note that
Zimbabwe is pressing ahead with its reengagement efforts with international finance
institutions like the World Bank Group, IMF and AfDB thereby explaining the positive
relationship between debt servicing and private investment in the economy. These findings
are supported by Bayai and Nyangara (2013) who also found debt service to crowd-in private
investment in Zimbabwe by 4.42%. They conclude that debt servicing provided positive
incentives to foreign investors and lenders who had previously shunned way from investing
in Zimbabwe.

4.4.4 Public Investment (PUBI)

Public investment was seen to crowd out private investment from the regression results in
Table 4.9 although the effect was insignificance as evidenced by the p-value of 0.6941 which
exceeds the 5% level of significance. Also the t-statistic is less than two thereby further
emphasising the insignificance of the variable.

30
4.4.5 Household Final Consumption Expenditure (HFCE)

Household final consumption expenditure had an unexpected positive sign implying that
private consumption would promote private investment in Zimbabwe. From the regression
model a percentage increase in private consumption would stimulate investment by 0.134%.
This positive but unexpected effect was statistically significantwith probability value of
0.0068. This would imply that an increase in household consumption would encourage
investors to engage in more planned and unplanned investments. Although this is supported
by the demand side Keynesian economics which posit that increase in consumption would
have multiplier effect on economic growth by stimulating investment. Nonetheless, such high
consumption has promoted foreign financing to compensate for the inadequate savings as a
result of excessive consumption.

4.4.6 Trade Openness (TO)

Openness to trade was found to have a positive and expected sign implying that trade
openness encourages private investment into the economy. From the regression model, a
percentage increase in trade openness would drive investment by 0.11. These findings are in
line with the findings of Kamundia et al (2015). Openness to trade promotes the free flow of
imports and exports which encourage private investment especially exports are doing well.
Free flow of raw material imports promote investment.

4.4.7 Real Lending Interest Rate (R)

The findings of real interest rate also referred to as real lending rate does conform to theory
and satisfies the expected negative sign. According to the Keynesian theory of investment,
high interest rates discourage investment. As such the model founds out that a unit increase in
the real interest rate would cause investment to decline by 0.023% approximately. Bayai and
Nyangara (2013) also found increases in the real lending rate to discourage private
investment.

4.5 Conclusion

The estimated model found external debt both in terms of current debt flows and past debt
flows to negatively affect investment in Zimbabwe although current debt inflows had an
insignificant impact evidenced by its p-value of 0.0589. The accumulation of unpaid external
public loans was seen to deter private investment by 0.125% as expected in Chapter 3.
However, debt servicing was found to unexpectedly crowd-in private investment. This was

31
however attributed to the reengagement efforts by the Zimbabwean government where debt
servicing is aimed at restoring the credibility status of Zimbabwe would open old and new
financing alternatives to finance also private investment. The following chapter summarises
the study at the same time offering policy recommendations basing on the findings.

32
 CHAPTER FIVE

SUMMARY, CONCLUSIONS AND POLICY RECOMMENDATIONS

5.0 Introduction
The aim of the empirical study was to analyse the nature of the relationship between external
debt and private investment in Zimbabwe. Zimbabwe is currently perceived to be suffering
from debt distress and as such the study aimed to address how such public indebtedness has
affected capital formation in the private sector. This chapter will therefore offer concluding
remarks on how external debt has affected investment via the accumulation of debt stocks
and its subsequent servicing. With the use of the research findings in the previous chapter,
this chapter will go on to offer policy recommendations and also recommend on areas for
further research.

5.1 Summary of the Study

The main objective of this research was to establish the relationship between public external
debt and private investment in Zimbabwe for the period 1980-2016. This was driven by the
diverging growth of external debt indicators relative to private investment. Theoretical
abstractions from the Two Gap model, Classical and Keynesian theories, Debt Overhang and
Ricardian Equivalence were reviewed to provide a theoretical background of the relationship.
The study therefore established the relationship between external debt and private investment
by taking into consideration the two channels within which external debt affects investment;
via current and past debt stocks (a feature of debt overhang). The inclusion of a lagged
external debt variable allowed the researcher to capture the role of debt overhang on private
investment which accrues from the accumulation of past debt stocks. Ordinary Least Squares
regression technique was therefore used to estimate the adapted model. The results showed
that out of the seven variables, only current external debt and public investment had
insignificant impacts on private investment. The study therefore found debt overhang to
negatively affect private investment since past debt stocks had a negative relationship with
private investment. Debt servicing was however unexpectedly found to crowd-in private
investment.

33
5.2 Conclusions
The concept of public indebtedness has gradually become a major topic of discussion in
heavily indebted and distressed nations with a growing number of researchers trying to
address its effects on domestic investment. This study in trying to estimate the nature of the
relationship between external debt and private investment in Zimbabwe observed that current
debt flows insignificantly affect private investment whilst the accumulation of past debt
stocks leads to debt overhang which ultimately discourages private investment. This
ultimately converges to that public external debt is an evil to private sector capital formation.
The crowding-in role of debt servicing on private leads the researcher to not reject the null
hypothesis that debt servicing does not crowd out private investment.
5.3 Policy Recommendations
The results of this study indicated that external debt does affect private investment
significantly through past debt flows, a feature regarded as debt overhang. Clearly put, the
inability of Zimbabwe to clear off its external debt obligations has consequently led to the
development of a debt overhang which has therefore negatively affected the growth of private
investment in the nation. This has therefore ultimately constrained the achievement of high
growth rates in the country. This converges to the fact that Zimbabwe must develop sound
debt utilization and management frameworks where debt is channelled towards productive
investments where the returns on the foreign debt are enough to pay-off the debt and prevent
further accumulation of unpaid foreign debts.

Zimbabwe has historically borrowed foreign funds for servicing of the high wage bill which
is also argued to be inflated by the presence of ghost workers. This unproductive and
inefficient utilization of these debts has also attributed to the accumulation of external debt
stocks. Thus such debt utilization and management frameworks are essential if private
investment is to be stimulated in the face of high debt distress. This will also reduce further
accumulation of external debt stocks.

Zimbabwe has also suffered poor growth of private investment in the face of the high debt
burden because of the lack of adequate privatization of industries. Privatization of say the
railway industry together with other sectors and government parastatals will definitely
promote private investment at the same time reducing public sector demand for foreign

34
credit. This will reduce the debt overhang effects on private investment and if very successful
the country can even exit the debt overhang.

Considering that debt servicing was actually found to encourage private investment it
therefore implies that if Zimbabwe manages to consistently service its debt, private
investment will be crowded in. This therefore warrants the need for a quarterly debt servicing
management framework where the RBZ undertakes a monthly analysis of debt servicing to
ensure that debt service payments lie in the desirable range of the creditors. The IMF (2017),
for example notes that the clearance of its PRGT loans by Zimbabwe will re-establish the
nation as a benefactor of the scheme. Thus debt servicing will reengage Zimbabwe with
International Financial Institutions. This will open new doors for further access to
international credit. This will also boost both local and foreign investor confidence in the
Zimbabwean economy.

5.4 Suggestion for future researches

Considering that Zimbabwe is currently suffering from debt distress, it is therefore
disheartening that little has been researched on its impacts on private investment in the
nation. The accumulation of external debt is argued to reduce the incentives for investment
into the economy, as per the debt overhang theory. As such the field of external indebtedness
offers an opportunity to clearly define the true role of foreign in stimulating investment and
ultimately its consequences on economic growth in Zimbabwe.

The researcher also feels that the work done has not been too exhaustive in unveiling the
nature and relationship between external debt and private investment mainly due to limited
access to data. Other researches can also analyse the role of external debt on capital flight in
Zimbabwe.

35
REFERENCE LIST

Abdullahi, M.M, Abu Bakar, N, and Hassan, S.B.(2016), „Debt Overhang versus Crowding
Out Effects: Understanding the Impact of External Debts on Capital Formation on Capital
Formation in Theory‟,International Journal of Economics and Financial issues, 6(1), pp.
271-278

Adegbite, S.O, Ayadi, S.F and Ayadi, O.F (2008), „The impact of Nigeria‟s external debt on
economic development‟, International Journal of Emerging Markets, 3(3), pp. 285-301

Ajilore, O.T, (2005) „External Debt and Capital Flight in Nigeria: Is There A Revolving
Door?‟, South Africa Journal of Economics and Management Sciences, 8(2), pp. 211-223

Akomolafe, J.K, Bosede, O, and Achukwu, M.(2015), „Public Debt and Private Investment in
Nigeria‟,American Journal of Economics, 5(5), pp. 501-507

Amoating, K, and Amoaku-Adu, B. (1996), „Economic Growth, Export and External Debt
causality: The Case of African Countries‟Applied Economics, pp. 28, 21-27

Apere, O.T. (2014), „The Impact of Public Debt on Private Investment in Nigeria: Evidence
from a Non-linear Model‟, International Journal of Research in Social Sciences, 4(2), pp.
130-138

Ayadi, S. F. and Ayadi, F. O. (2008),„The Impact of External Debt On Economic Growth: A

Comparative Study of Nigeria and South Africa‟,Journal of Sustainable Development in
Africa, 10(3), pp. 234-264

Bayai, I. and Nyangara, D. (2013), „An analysis of the determinants of private investment in
Zimbabwe for the period 2009 – 2011‟, International Journal of Economics and Management
Sciences, 2(6), pp. 11-42

Barfour. O. (1995), „Ghana: The Burden of Debt Service Payment Under Structural
Adjustment‟,African Economic Research Consortium Research Papers,8, English press
Limited, Kenya.

36
Barro, R.J, (1974) „Are Government Bonds Net Wealth?‟Journal of political Economy, 82(6),
pp. 1095-1117

Bilan, I. (2016) „Overview of the main theories of the Economic Effects of Public
Indebtedness‟, EIRP Proceedings, 11

Bulow, J. and Rogoff, K. (1990), „Cleaning up Third World Debt Without Getting Taken to
the Cleaners‟,The Journal of Economic Perspective, 4(1), pp. 31-42

Caron, M. (2007), „Budget and Budgetary Policy‟, Paris: Editions Breal

Casares, R. E.(2015) „A Relationship between External Public Debt and Economic Growth‟,
Economic Studies, 30(2), pp. 219-243

Chaudhry, S.I (2009),„Impact of Foreign debt on Savings and Investment in Pakistan‟‟,

Jouranl of Quality and Technology Management’, 5(11), pp. 101-115

Chibber, A and Lecchor, C, (1993) „From Adjustment to Growth in Sub-Saharan Africa: the
Lessons of East Asia applied to Ghana‟, Journal of Economics, 4, pp. 54

Chenery, H.B and Strout, A. (1966)„Foreign Assistance and Economic Development‟,

American Economic Review, 55, pp. 638-733

Chowdhury, K (1994), „A Structural Analysis of External Debt and Economic Growth: Some
Evidence from Selected Countries in Asia and Pacific‟,Applied Economics, 26(12), pp. 1121-
1131

Churchman, N (2001)David Ricardo on Public Debt: Palgrave; New York

Claessens, S.E, Detragiache, R.K and Wickham, P (1996), „Analytical aspects of the debt
Problems of Heavily Indebted Poor Countries

Claessens, S. E and Diwan, I. (1990), „Investment incentives: New Money, Debt Relief, and
the Critical Role of Conditionality in Debt Crisis‟,The World Bank Economic Review, 4(1).

Cohen, D (1993) „Low Investment and Large LDC Debt in the 1980s‟, American Economic
Review, 83(3), 437-449

37
Cohen, D. and Sachs, J. (1986), „Growth and External Debt under risk of debt repudiation‟,
European Economic Review, 30(3), pp. 529-560

Colgan, A. (2001), „Africa‟s Debt‟, Africa Action Paper

Domar, E.L, (1946) „Capital Expansion, Rate of Growth and Employment‟, Econometrica,
14, pp. 137-147

Easterly, W (1999) „The Ghost of Financing Gap: Testing the Growth Model used in
International Financial Institutions, 60(2), pp. 423-438

Elemndorf, D.W and Mankiw G.N (1998), „Government Debt', NBER

Erhiyvowe, E. K, and Onomoakpoma, O.D (2013), „External Debt and its Impacts on
Growth: An Assessment of Major Macroeconomic Variables in Nigeria‟, 2(2), pp. 143-153

Filip, G.H (2010), „„Public Finance‟‟. Lasi: Junimea Publishing House

Forgha, G.H, Mbella, E.M and Ngangnchi, F.H (2014), “External Debt, Domestic Investment
and Economic Growth in Cameroon; A system Estimation Approach”,Journal of Economics
Bibliography, 1(1)

Hunte. C.K (2003), „Saving Behavior and External Debt-Service: Evidence from Sub-
Saharan Africa‟, Africa Review of Money Finance and Banking, pp. 63-74

Ijirshar, U.V, Joseph, F and Godoo, M (2016), „The relationship between External Debt and
Economic Growth in Nigeria‟,International Journal of Economics and Management
Sciences, 6(1), pp. 1-5

International Monetary Fund (IMF) (2011) „Joint IMF/World Bank Debt Sustainability
Article IV Report: Zimbabwe‟.

IMF (2016), „Staff Report for the 2016 Article 4 Consultation & Third Review of the Staff-
Monitored Program-Debt Sustainability Analysis; Zimbabwe‟.

International Monetary Fund (2017) „Staff Report for the 2017 Article 4 Consultation – Debt
Sustainability: Zimbabwe‟

38
Iyoha, M. A. (1999), „External Debt and Economic Growth in Sub-Saharan African
countries: An Econometric study‟,African Economic Research Consortium Research Papers
No 90, English press Limited, Kenya

Jayaraman T.K and Evan L. (2008) „Does external debt lead to economic growth in Pacific
Island countries‟, Journal of Policy Modeling, 31, pp. 272–288

Jarju, I., Nyarko, E., Kormay, A., Haffner, O and Olukayode, S. O. (2016), “The relationship
between External Debt and Economic Growth in the West African Monetary Zone: A Panel
Data Analysis”, WAMI Occassional Paper Series no 12

Jones, T. (2011), „Uncovering Zimbabwe‟s debt: The case for a democratic solution to the
unjust debt burden‟,Jubilee debt campaign: London

Keynes, J.M (1982), „Activities 1931-1939, World Crises and Policies in Britain and Arenca
in. The Collected Writings of John Maynard Keynes, 21: London: MacMillan

Krugman, P. (1988) „financing vs. forgiving a debt overhang‟,Journal of Development

Economics, 29(3), pp. 253 – 268

McKinnon, R (1964) „Foreign Exchange constraint in Economic Development and Efficient

Aid Allocation‟, Economic Journal, 74. pp. 388-409

Mjema, G. D. (1996), „The Impact of Foreign Debt Servicing in the Economy of Tanzania: A
Simultaneous equation approach‟,African Journal of Economic Policy, 3(1).

Nkamleyu, B.G. (2006) „On Measuring Indebtedness of African Countries: A Stochastic

Frontier Debt Production Function‟,Economic Research Working Paper Series No 65

Occhino, F. (2010). „Debt Overhang in a Business Cycle‟. Federal Reserve Bank of

Cleveland Working Paper No. 10-03R

Saungweme,T. and Mufandaedza, S. (2013), „An Empirical Analysis of External Debt on

Poverty in Zimbabwe: 1980 – 2011‟. IJER

Rashid, A (2006) „Public-Private Investment Linkages in Pakistan‟, South Asia Economic

Journal, 7(2), pp. 219-230

39
Ricardo, D (2005), On the principles of the Political Economy and Taxation, in The work and
correspondence of David Ricardo, (Ed Straffa, P and Dobb M.H) vol 1, Indianapolis: Liberty
Fund

Rose, S, Spinks, N and Canchoto, A.I (2015)

Saungweme, T and Mufandaedza, S (2013), „An Empirical Investigation of the effects of

External Debt on Poverty in Zimbabwe‟, pp. 20-27

Smith, A. (1936), „An Inquiry into the Nature and Causes of the Wealth of Nations‟, 5th
Edition. London: Methuen & Co. Ltd

Ukpe, U.H., Umeh, J. C.,and Asogwa, B.C. (2017), „Effects of Public External Debt and
Private Investment on Agricultural Growth in Nigeria: 1980 - 2016‟,Agricultural Research
and Technology Open Access Journal, 10(2)

Were, M (2001), „The Impact of External Debt on Economic Growth in Kenya‟, Dicussion
Paper No. 2001/2016

40
APPENDICES
Appendix 1: Data Set

PINV EXD PUBI DEBTSERV HFCE

YEAR (% GDP) (% GDP) (% GDP) (% GDP) (% GDP) TO R (%)
1980 13.49 20.13 1.57 30.24 58.45 25.01 4.26
1981 16.47 15.47 6.42 21.97 61.33 22.57 12.75
1982 16.72 13.50 3.77 12.84 67.12 34.92 18.43
1983 16.53 10.25 8.39 35.33 62.69 37.45 37.53
1984 14.84 17.54 8.11 26.16 48.11 40.33 47.47
1985 11.36 12.66 6.12 29.04 53.46 41.69 41.19
1986 13.01 7.98 12.45 31.14 58.78 31.01 4.60
1987 14.02 5.37 14.68 32.29 58.95 28.49 5.42
1988 13.31 20.13 5.66 28.48 50.43 25.77 4.84
1989 11.37 25.70 -1.20 22.38 64.65 41.91 12.11
1990 15.93 22.68 -2.10 23.15 63.11 40.72 12.75
1991 18.29 21.75 7.17 23.12 65.88 50.10 23.90
1992 19.59 19.47 1.25 32.31 65.65 63.71 39.48
1993 20.84 22.89 7.88 30.11 68.88 63.17 41.70
1994 18.65 15.47 2.73 25.20 56.24 75.33 40.33
1995 21.88 30.91 2.85 31.56 59.31 68.14 30.76
1996 15.19 40.65 3.17 40.25 62.75 72.07 23.17
1997 15.67 35.47 7.55 16.58 70.17 65.11 36.48
1998 19.34 30.11 6.33 40.56 63.08 68.27 84.21
1999 10.73 42.57 4.58 20.15 69.39 65.41 43.87
2000 8.47 60.77 6.15 21.24 64.45 40.15 67.16
2001 2.09 60.39 1.25 18.54 25.72 40.72 38.20
2002 1.78 59.17 7.58 19.87 33.46 50.11 32.87
2003 0.21 60.73 6.36 34.51 46.48 75.95 75.44
2004 0.71 55.00 7.22 27.88 31.37 45.08 252.31
2005 0.84 47.51 4.44 35.12 25.31 61.01 219.28
2006 2.54 31.44 9.55 17.58 95.28 72.99 508.74
2007 3.76 70.65 69.95 30.94 76.06 65.17 572.94
2008 2.96 60.55 66.19 11.72 119.41 75.93 287.00
2009 10.84 67.11 20.10 20.43 100.71 65.11 150.00
2010 15.55 40.04 32.40 25.79 87.85 77.67 81.01
2011 12.90 67.40 3.66 8.11 92.07 63.49 65.58
2012 12.28 40.55 2.54 8.92 90.15 71.25 65.98
2013 8.55 67.84 6.94 5.43 85.77 66.21 55.23
2014 9.54 30.27 12.50 16.44 82.34 45.65 12.50
2015 9.22 8.15 4.23 5.29 76.98 38.69 10.20
2016 8.79 60.04 7.36 4.57 71.84 50.04 11.25
Sources: World Bank, IMF, Global Economy, Data market, ZIMSTAT

41
Appendix 2A: Stationary Tests

DEBTSERV UNIT ROOT TEST RESULTS

Null Hypothesis: DEBTSERV has a unit root

Exogenous: Constant, Linear Trend
Lag Length: 0 (Automatic - based on SIC, maxlag=0)

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -5.171543 0.0009

Test critical values: 1% level -4.234972
5% level -3.540328
10% level -3.202445

*MacKinnon (1996) one-sided p-values.

Augmented Dickey-Fuller Test Equation

Dependent Variable: D(DEBTSERV)
Method: Least Squares
Date: 04/18/18 Time: 23:17
Sample (adjusted): 1981 2016
Included observations: 36 after adjustments

Variable Coefficient Std. Error t-Statistic Prob.

DEBTSERV(-1) -0.919339 0.177769 -5.171543 0.0000

C 29.53026 6.357112 4.645232 0.0001
@TREND("1980") -0.446714 0.157111 -2.843309 0.0076

R-squared 0.449225 Mean dependent var -0.713056

Adjusted R-squared 0.415844 S.D. dependent var 11.24214
S.E. of regression 8.592380 Akaike info criterion 7.219284
Sum squared resid 2436.357 Schwarz criterion 7.351244
Log likelihood -126.9471 Hannan-Quinn criter. 7.265341
F-statistic 13.45776 Durbin-Watson stat 1.986006
Prob(F-statistic) 0.000053

EXD UNIT ROOT TEST RESULTS

Null Hypothesis: D(EXD) has a unit root
Exogenous: None
Lag Length: 0 (Automatic - based on SIC, maxlag=0)

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -8.920298 0.0000

Test critical values: 1% level -2.632688
5% level -1.950687
10% level -1.611059

*MacKinnon (1996) one-sided p-values.

Augmented Dickey-Fuller Test Equation

Dependent Variable: D(EXD,2)
Method: Least Squares
Date: 04/18/18 Time: 23:19

42
Sample (adjusted): 1982 2016
Included observations: 35 after adjustments

Variable Coefficient Std. Error t-Statistic Prob.

D(EXD(-1)) -1.548137 0.173552 -8.920298 0.0000

R-squared 0.699655 Mean dependent var 1.615714

Adjusted R-squared 0.699655 S.D. dependent var 28.72807
S.E. of regression 15.74406 Akaike info criterion 8.378959
Sum squared resid 8427.765 Schwarz criterion 8.423397
Log likelihood -145.6318 Hannan-Quinn criter. 8.394299
Durbin-Watson stat 2.036993

HFCE UNIT ROOT TEST RESULTS

Null Hypothesis: D(HFCE) has a unit root

Exogenous: None
Lag Length: 0 (Automatic - based on SIC, maxlag=0)

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -8.257874 0.0000

Test critical values: 1% level -2.632688
5% level -1.950687
10% level -1.611059

*MacKinnon (1996) one-sided p-values.

Augmented Dickey-Fuller Test Equation

Dependent Variable: D(HFCE,2)
Method: Least Squares
Date: 04/18/18 Time: 23:21
Sample (adjusted): 1982 2016
Included observations: 35 after adjustments

Variable Coefficient Std. Error t-Statistic Prob.

D(HFCE(-1)) -1.335434 0.161717 -8.257874 0.0000

R-squared 0.667273 Mean dependent var -0.229143

Adjusted R-squared 0.667273 S.D. dependent var 29.01676
S.E. of regression 16.73758 Akaike info criterion 8.501346
Sum squared resid 9524.989 Schwarz criterion 8.545784
Log likelihood -147.7736 Hannan-Quinn criter. 8.516686
Durbin-Watson stat 2.026448

LAGEXD UNIT ROOT TEST RESULTS

Null Hypothesis: D(LAGEXD) has a unit root

Exogenous: None
Lag Length: 0 (Automatic - based on SIC, maxlag=0)

t-Statistic Prob.*

43
Augmented Dickey-Fuller test statistic -8.826574 0.0000
Test critical values: 1% level -2.634731
5% level -1.951000
10% level -1.610907

*MacKinnon (1996) one-sided p-values.

Augmented Dickey-Fuller Test Equation

Dependent Variable: D(LAGEXD,2)
Method: Least Squares
Date: 04/18/18 Time: 23:23
Sample (adjusted): 1983 2016
Included observations: 34 after adjustments

Variable Coefficient Std. Error t-Statistic Prob.

D(LAGEXD(-1)) -1.434496 0.162520 -8.826574 0.0000

R-squared 0.702339 Mean dependent var -0.513529

Adjusted R-squared 0.702339 S.D. dependent var 26.20733
S.E. of regression 14.29827 Akaike info criterion 8.187124
Sum squared resid 6746.534 Schwarz criterion 8.232017
Log likelihood -138.1811 Hannan-Quinn criter. 8.202434
Durbin-Watson stat 1.750133

PINV UNIT ROOT TEST RESULTS

Null Hypothesis: D(PINV) has a unit root

Exogenous: None
Lag Length: 0 (Automatic - based on SIC, maxlag=0)

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -5.885158 0.0000

Test critical values: 1% level -2.632688
5% level -1.950687
10% level -1.611059

*MacKinnon (1996) one-sided p-values.

Augmented Dickey-Fuller Test Equation

Dependent Variable: D(PINV,2)
Method: Least Squares
Date: 04/18/18 Time: 23:24
Sample (adjusted): 1982 2016
Included observations: 35 after adjustments

Variable Coefficient Std. Error t-Statistic Prob.

D(PINV(-1)) -0.997353 0.169469 -5.885158 0.0000

R-squared 0.504401 Mean dependent var -0.097429

Adjusted R-squared 0.504401 S.D. dependent var 4.628777
S.E. of regression 3.258604 Akaike info criterion 5.228630
Sum squared resid 361.0290 Schwarz criterion 5.273069
Log likelihood -90.50103 Hannan-Quinn criter. 5.243970
Durbin-Watson stat 2.003746

44
PUBI UNIT ROOT TEST RESULTS

Null Hypothesis: PUBI has a unit root

Exogenous: None
Lag Length: 0 (Automatic - based on SIC, maxlag=0)

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -2.492703 0.0142

Test critical values: 1% level -2.630762
5% level -1.950394
10% level -1.611202

*MacKinnon (1996) one-sided p-values.

Augmented Dickey-Fuller Test Equation

Dependent Variable: D(PUBI)
Method: Least Squares
Date: 04/18/18 Time: 23:25
Sample (adjusted): 1981 2016
Included observations: 36 after adjustments

Variable Coefficient Std. Error t-Statistic Prob.

PUBI(-1) -0.303640 0.121812 -2.492703 0.0176

R-squared 0.150659 Mean dependent var 0.160833

Adjusted R-squared 0.150659 S.D. dependent var 14.58059
S.E. of regression 13.43743 Akaike info criterion 8.061349
Sum squared resid 6319.755 Schwarz criterion 8.105336
Log likelihood -144.1043 Hannan-Quinn criter. 8.076702
Durbin-Watson stat 1.935329

R UNIT ROOT TEST RESULTS

Null Hypothesis: D(R) has a unit root

Exogenous: None
Lag Length: 0 (Automatic - based on SIC, maxlag=0)

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -4.914022 0.0000

Test critical values: 1% level -2.632688
5% level -1.950687
10% level -1.611059

*MacKinnon (1996) one-sided p-values.

Augmented Dickey-Fuller Test Equation

Dependent Variable: D(R,2)
Method: Least Squares
Date: 04/18/18 Time: 23:26
Sample (adjusted): 1982 2016

45
Included observations: 35 after adjustments

Variable Coefficient Std. Error t-Statistic Prob.

D(R(-1)) -0.830413 0.168988 -4.914022 0.0000

R-squared 0.415279 Mean dependent var -0.212571

Adjusted R-squared 0.415279 S.D. dependent var 107.7192
S.E. of regression 82.36968 Akaike info criterion 11.68847
Sum squared resid 230682.0 Schwarz criterion 11.73291
Log likelihood -203.5482 Hannan-Quinn criter. 11.70381
Durbin-Watson stat 1.956129

TO UNIT ROOT TEST RESULTS

Null Hypothesis: D(TO) has a unit root

Exogenous: None
Lag Length: 0 (Automatic - based on SIC, maxlag=0)

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -7.781079 0.0000

Test critical values: 1% level -2.632688
5% level -1.950687
10% level -1.611059

*MacKinnon (1996) one-sided p-values.

Augmented Dickey-Fuller Test Equation

Dependent Variable: D(TO,2)
Method: Least Squares
Date: 04/18/18 Time: 23:27
Sample (adjusted): 1982 2016
Included observations: 35 after adjustments

Variable Coefficient Std. Error t-Statistic Prob.

D(TO(-1)) -1.292763 0.166142 -7.781079 0.0000

R-squared 0.640236 Mean dependent var 0.394000

Adjusted R-squared 0.640236 S.D. dependent var 19.73080
S.E. of regression 11.83460 Akaike info criterion 7.808087
Sum squared resid 4761.962 Schwarz criterion 7.852526
Log likelihood -135.6415 Hannan-Quinn criter. 7.823427
Durbin-Watson stat 1.985106

46
Appendix 2B: Cointegration Test Results

Date: 05/22/18 Time: 01:04

Sample (adjusted): 1982 2016
Included observations: 35 after adjustments
Trend assumption: Linear deterministic trend
Series: PINV EXD PUBI DEBTSERV HFCE TO R
Lags interval (in first differences): 1 to 1

Unrestricted Cointegration Rank Test (Trace)

Hypothesized Trace 0.05

No. of CE(s) Eigenvalue Statistic Critical Value Prob.**

None * 0.826896 182.9607 125.6154 0.0000

At most 1 * 0.759074 121.5754 95.75366 0.0003
At most 2 * 0.543796 71.76117 69.81889 0.0347
At most 3 0.493093 44.29268 47.85613 0.1040
At most 4 0.319111 20.51272 29.79707 0.3887
At most 5 0.119433 7.060273 15.49471 0.5706
At most 6 0.071823 2.608662 3.841466 0.1063

Trace test indicates 3 cointegrating eqn(s) at the 0.05 level

* denotes rejection of the hypothesis at the 0.05 level
**MacKinnon-Haug-Michelis (1999) p-values

Appendix 2C: Multicollinearity Test Results

Appendix 2D: Heteroscedasticity Test Results

Heteroskedasticity Test: Breusch-Pagan-Godfrey

F-statistic 0.414792 Prob. F(7,28) 0.8849

Obs*R-squared 3.382384 Prob. Chi-Square(7) 0.8475
Scaled explained SS 1.766958 Prob. Chi-Square(7) 0.9716

Test Equation:
Dependent Variable: RESID^2
Method: Least Squares
Date: 04/25/18 Time: 18:14
Sample: 1981 2016

47
Included observations: 36

Variable Coefficient Std. Error t-Statistic Prob.

C 0.255534 14.77296 0.017297 0.9863

EXD 0.012397 0.150500 0.082374 0.9349
LAGEXD 0.012760 0.135801 0.093964 0.9258
PUBI 0.091146 0.206782 0.440782 0.6628
DEBTSERV 0.130316 0.306414 0.425295 0.6739
HFCE -0.085834 0.166401 -0.515827 0.6100
TO 0.184565 0.185160 0.996788 0.3274
R -0.011051 0.021910 -0.504361 0.6180

R-squared 0.093955 Mean dependent var 8.467265

Adjusted R-squared -0.132556 S.D. dependent var 11.28552
S.E. of regression 12.01023 Akaike info criterion 8.002525
Sum squared resid 4038.877 Schwarz criterion 8.354418
Log likelihood -136.0454 Hannan-Quinn criter. 8.125345
F-statistic 0.414792 Durbin-Watson stat 2.264686
Prob(F-statistic) 0.884889

Appendix 2E: Autocorrelation Test Results

Breusch-Godfrey Serial Correlation LM Test:

F-statistic 0.148117 Prob. F(2,26) 0.8631

Obs*R-squared 0.405551 Prob. Chi-Square(2) 0.8165

Test Equation:
Dependent Variable: RESID
Method: Least Squares
Date: 04/18/18 Time: 23:37
Sample: 1981 2016
Included observations: 36
Presample missing value lagged residuals set to zero.

Variable Coefficient Std. Error t-Statistic Prob.

C 0.767706 4.420942 0.173652 0.8635

EXD 1.71E-05 0.044434 0.000385 0.9997
LAGEXD 0.002973 0.038942 0.076345 0.9397
PUBI 0.007897 0.062464 0.126432 0.9004
DEBTSERV 0.000315 0.087629 0.003594 0.9972
HFCE -0.005989 0.049847 -0.120147 0.9053
TO -0.012307 0.061839 -0.199022 0.8438
R 0.000811 0.006644 0.122037 0.9038
RESID(-1) 0.080611 0.249375 0.323250 0.7491
RESID(-2) 0.116279 0.238461 0.487624 0.6299

R-squared 0.011265 Mean dependent var -1.48E-16

Adjusted R-squared -0.330989 S.D. dependent var 2.951133
S.E. of regression 3.404678 Akaike info criterion 5.518311
Sum squared resid 301.3876 Schwarz criterion 5.958177
Log likelihood -89.32960 Hannan-Quinn criter. 5.671836
F-statistic 0.032915 Durbin-Watson stat 1.906250
Prob(F-statistic) 0.999995

48
Appendix 2F: Model Specification Test

Ramsey RESET Test

Equation: UNTITLED
Specification: PINV C EXD LAGEXD PUBI DEBTSERV HFCE TO R
Omitted Variables: Squares of fitted values

Value df Probability
t-statistic 1.909549 27 0.0669
F-statistic 3.646378 (1, 27) 0.0669
Likelihood ratio 4.560394 1 0.0327

F-test summary:
Mean
Sum of Sq. df Squares
Test SSR 36.26838 1 36.26838
Restricted SSR 304.8215 28 10.88648
Unrestricted SSR 268.5531 27 9.946413
Unrestricted SSR 268.5531 27 9.946413

LR test summary:
Value df
Restricted LogL -89.53352 28
Unrestricted LogL -87.25333 27

Unrestricted Test Equation:

Dependent Variable: PINV
Method: Least Squares
Date: 04/18/18 Time: 23:39
Sample: 1981 2016
Included observations: 36

Variable Coefficient Std. Error t-Statistic Prob.

C 1.294230 3.922587 0.329943 0.7440

EXD -0.003301 0.056876 -0.058030 0.9542
LAGEXD -0.006875 0.071603 -0.096009 0.9242
PUBI -0.018566 0.054340 -0.341663 0.7352
DEBTSERV -0.010028 0.126731 -0.079127 0.9375
HFCE 0.060769 0.057999 1.047766 0.3040
TO -0.001393 0.074112 -0.018795 0.9851
R -0.005222 0.010801 -0.483448 0.6327
FITTED^2 0.046277 0.024235 1.909549 0.0669

R-squared 0.805117 Mean dependent var 11.52139

Adjusted R-squared 0.747374 S.D. dependent var 6.274713
S.E. of regression 3.153793 Akaike info criterion 5.347407
Sum squared resid 268.5531 Schwarz criterion 5.743287
Log likelihood -87.25333 Hannan-Quinn criter. 5.485580
F-statistic 13.94306 Durbin-Watson stat 1.893741
Prob(F-statistic) 0.000000

49
Appendix 2G: Normality Test Results

Appendix 3: Regression Results

50
Appendix 4: Summary Statistics

51