0

THREE ESSAYS ON THE RELATIONSHIP BETWEEN IN-

TERNAL DEFICIT, EXTERNAL DEFICIT, MACROECO-
NOMIC POLICY AND ECONOMIC GROWTH IN
SUB_SHARN AFRICAN ECONOMIES

By Ayana Zewdie

A THESIS SUBMITTED IN FULFILLMENT OF THE RE-

QUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOS-
OPHY IN ECONOMICS

Principal Supervisor Professor Fung KWAN

Co-supervisors 1. Professor Patrick Wai Hong HO
2.Professor Yibai YANG

DEPARTMENT OF ECONOMICS
Faculty of SOCIAL SCIENCES
University of MACAU
 1

Table of Contents
CHAPTER ONE: INTRODUCTION ............................................................................................................................................. 1
1.1. BACKGROUND OF THE STUDY ............................................................................................................................................. 1
1.2. PROBLEM STATEMENT ....................................................................................................................................................... 2
1.3. RESEARCH QUESTIONS ....................................................................................................................................................... 3
1.4. RESEARCH OBJECTIVES ....................................................................................................................................................... 3
1.5. CONTRIBUTION OF THE STUDY AND BRIEF METHODOLOGY .............................................................................................. 4
1.6. ORGANIZATION OF THE THESIS .......................................................................................................................................... 5
CHAPTER TWO: MACROECONOMIC PERFORMANCE AND EXPERIENCES IN SSA ..................................................................... 6
2.1 INTRODUCTION ................................................................................................................................................................... 6
2.2. TRENDS OF ECONOMIC GROWTH IN SSA IN 1980-2018...................................................................................................... 6
2.3. TRENDS OF MACROECONOMIC IMBALANCES IN 1980-2018 .............................................................................................. 8
2.3.1 TRENDS OF CURRENT ACCOUNT IMBALANCE ........................................................................................................... 10
2.3.2. TRENDS OF FISCAL IMBALANCES .............................................................................................................................. 13
2.3.3. TRENDS OF SAVING-INVESTMENT IMBALANCES ...................................................................................................... 13
2.4. CONCLUDING REMARKS ................................................................................................................................................... 13
CHAPTER THREE: LITERATURE REVIEW ................................................................................................................................ 18
3.1. REVIEW OF THE THEORETICAL AND EMPIRICAL LITERATURE ON THE RELATIONSHIP BETWEEN EXTERNAL DEFICIT AND
INTERNAL DEFICIT ................................................................................................................................................................... 18
3.1.1. INTRODUCTION ........................................................................................................................................................ 18
3.1.2. THEORETICAL LITERATURE REVIEW ......................................................................................................................... 18
3.1.3. EMPIRICAL LITERATURE REVIEW .............................................................................................................................. 20
3.1.4 SUMMARY AND CONCLUSION OF THEORETICAL AND EMPIRICAL LITERATURE REVIEW ......................................... 23
3.2. REVIEW OF THE THEORETICAL AND EMPIRICAL LITERATURE ON THE IMPACT OF TRIPLE DEFICIT ON ECONOMIC GROWTH
................................................................................................................................................................................................ 24
3.2. 1.REVIEW OF THE THEORETICAL LITERATURE ............................................................................................................. 24
3.2.2. REVIEW OF THE EMPIRICAL LITERATURE ................................................................................................................. 31
3.2.3. CONCLUSIONS ON THE THEORETICAL AND EMPIRICAL LITERATURE ....................................................................... 44
CHAPTER FOUR: THE RELATIONSHIP BETWEEN CURRENT ACCOUNT DEFICIT FISCAL DEFICIT AND PRIVATE-INVESTMENT
DEFICIT IN THE SSA ECONOMIES .......................................................................................................................................... 46
4.1. THEORETICAL FRAMEWORK (MODEL) .............................................................................................................................. 46
4.2. METHODOLOGY ............................................................................................................................................................... 48
4.2.1. EMPIRICAL MODEL SPECIFICATION .......................................................................................................................... 48
4.2.2. ESTIMATION METHOD ............................................................................................................................................. 49
4.3. ESTIMATION AND INTERPRETATION OF RESULTS ............................................................................................................. 52
4.3.1. INTRODUCTION ........................................................................................................................................................ 52
4.3.2. PRELIMINARY ANALYSIS ........................................................................................................................................... 52
4.3.3-NON-STATIONARITY TEST ......................................................................................................................................... 58
4.3.4. COINTEGRATION TEST .............................................................................................................................................. 59
4.3.5. CAUSALITY ANALYSIS ................................................................................................................................................ 59
4.3. 6. DYNAMIC COMMON CORRELATED EFFECT ESTIMATION TECHNIQUES .................................................................. 62
4.4. SUMMARY, CONCLUSIONS AND RECOMMENDATIONS .................................................................................................... 66
4.4.1. SUMMARY ................................................................................................................................................................ 66
4.4.2. CONCLUSION ............................................................................................................................................................ 70
4.4.3. POLICY RECOMMENDATIONS ................................................................................................................................... 71
CHAPTER FIVE: THE IMPACT OF THE TRIPLE DEFICITS ON ECONOMIC GROWTH IN SSA COUNTRIES 75
5.1. INTRODUCTION ............................................................................................................................................................ 75
5.2.THE THREE-GAP MODEL ............................................................................................................................................. 75
5.2.1. INVESTMENT - SAVINGS GAP .............................................................................................................................. 76
5.2.2 EXPORT - IMPORT GAP (CURRENT ACCOUNT DEFICIT)........................................................................... 78
5.2.3 BUDGETARY GAP (BUDGET DEFICIT) .......................................................................................................... 80
5.3. THE THEORETICAL LINK BETWEEN THE THREE GAPS AND SUSTAINABLE ECONOMIC GROWTH .............................. 82
5.4 EMPIRICAL MODEL SPECIFICATION AND ESTIMATION METHOD .................................................................... 86
5.4.1. EMPIRICAL MODEL SPECIFICATION ................................................................................................................. 86
5.4.2. ESTIMATION METHOD ......................................................................................................................................... 89
5.5. ESTIMATION AND INTERPRETATION OF RESULTS .............................................................................................. 96
 2

5.5.1. INTRODUCTION ........................................................................................................................................................ 96

5.5.2. PRELIMINARY ANALYSIS ........................................................................................................................................... 96
5.5.3-NON-STATIONARITY TEST ................................................................................................................................... 103
5.5.4. COINTEGRATION TEST ....................................................................................................................................... 104
5.5.5 EMPIRICAL RESULTS OF HOMOGENEOUS PANEL VAR ................................................................................ 105
5.5. 6. ESTIMATES OF THE RELATIVE STRENGTHS OF THE RESOURCE GAPS ................................................................... 113
5.5.7 ROBUSTNESS CHECK USING DYNAMIC COMMON CORRELATED EFFECT ESTIMATION ........................ 114
5.6. SUMMARY, POLICY RECOMMENDATION, AND CONCLUSSION ..................................................................... 118
5.6.1. SUMMARY ............................................................................................................................................................. 118
5.6.2. POLICY IMPLICATIONS AND RECOMMENDATIONS ...................................................................................... 119
5.6.3. CONCLUSION ....................................................................................................................................................... 121
REFERENCES ................................................................................................................................................................ 122
 1

CHAPTER ONE: INTRODUCTION

1.1. BACKGROUND OF THE STUDY
Stabilization policy designing requires the knowledge on the dynamic behavior and relationship among
triple deficits, and economic growth. This help policymakers to craft sound macroeconomic policies to
achieve macroeconomic stability and economic growth. Several studies analyzed the issue of twin deficits
hypothesis, but nowadays, a “triple deficits” hypothesis has surfaced.
The required investment to boost economic growth in developing countries is constrained by low savings
and higher consumption. In this respect, developing countries need foreign savings and foreign capital in-
flow via foreign direct investment and external debts to finance the such investment and consumption(Hub-
bard,2006; Kuijs, 2006; and Gruber and Kamin,2007) . Therefore, the rise of globalization and free capital
mobility creates a good opportunity for those countries to attract foreign direct investment and to have
access for external loan. This causes the saving-investment balance of the country in deficit. Hence, both
the private savings-investment deficit and the budget deficit lead to the rise of a current account deficit. The
deficit of all these three balances and its co-movement or causal linkage among them is called Triple deficit
(Akbas et al. , 2014). Such circumstance prompted the scrutinizing of twin deficit hypothesis in recent years.
Accordingly, triple deficit hypothesis is an all-encompassing rendition of the twin deficit theory and turning
into an issue examined in literature.

In most cases, the actual rate of growth is demand constrained as demand constraints come into effect
before supply constraints are reached (Thirlwall, 2013). In the post-Keynesian literature, the main
constraint to long run growth in an open economy is the balance of payments equilibrium growth rate
(Thirlwall, 2001). This is the growth rate consistent with equilibrium on the current account of the balance
of payments, under the assumption that a current account deficit cannot be financed indefinitely, and debt
“ultimately” needs to be repaid (Thirlwall, 1979; 2001). No studies have been carried out on low income
countries or for the sub-Saharan African region.
In these regards, investigating whether the twin or triple deficit hypothesis holds, or the validity of Ricardian
equivalence or twin divergence and understanding the impact of these deficits and macroeconomic policy
on economic growth is important for policy selection and decision to ensure macroeconomic stability and
sustained growth. In this respect, both monetary and fiscal policies are used as major tools to realize mac-
roeconomic stability and sustained growth. However, in the last five decades, the effectiveness of the two
policies has generated a large volume of both theoretical and empirical literature with a greatly differed
result. This has been a serious debate between the Keynesians and the monetarists.
Neoclassical economics and international trade theory based on static comparative advantage has dominated
economic policies in the sub-Saharan African region. The emphasis has been on trade and financial
liberalization for the efficient allocation of resources, leading to specialization in the production of products
with a “natural or artificial” comparative advantage. These policies were implemented in developing
countries and structural adjustment programs (When measured against key development indicators, economic
 2

performance in sub-Saharan Africa has been poor compared to other world regions. In addition, sub-Saharan
Africa is the only world region that has failed to significantly reduce the level of poverty. Between the 1981
to 2008 period, poverty levels remained persistently high at over 69%, measured at United States Dollars
(USD) 2 a day poverty line. There is empirical evidence that structural adjustment programs lead to increased
inequality (Crisp and Kelly, 1999; Easterly, 2003) and reduce the extent to which the poor share in economic
growth (Easterly, 2003).

1.2. PROBLEM STATEMENT

This thesis addresses three key issues related to macroeconomics in the context of the SSA economies—
a group of 48 African countries that have faced large macroeconomic imbalances since independence and
rising external debt after 2005. The thesis focuses on the 35 economies. The SSA economies consist of
highly small open economies that are significantly dependent upon a few agricultural exports, foreign aid
and remittances. Moreover, these economies are extremely vulnerable to external environmental
characteristics and frequent political instability. An important characteristic of these economies is their
lack of sustained economic growth since independence after 1970. Further, real Gross Domestic Product
(GDP) per capita is less than USD$3,000 for each of the SSA economies (World Bank, 2018). Since
independence, the majority of the SSA economies have experienced volatile and average annual economic
growth rate of no more than 5% (World Bank, 2018). Despite the above features that increase economic
vulnerability, the SSA economies have recorded large current account deficit and thus higher saving-
investment gap that has the potential impact on macroeconomic stability and growth sustainability.
More recently, current account deficit experienced by the SSA economies in the post-1980 was the highest
in the developing world. During the sample period, the average current account deficit-to-GDP ratio was
around -6.32% for SSA countries. Further, a recent World Economic Outlook Report by the International
Monetary Fund (2016) forecasts that the SSA economies will continue to experience large current account
deficit with low economic growth rates. Given the above mentioned economic and structural features of
the SSA economies that increase their economic vulnerability, these figures raise serious questions over
the ability of the SSA economies to ensure macroeconomic stability and sustain economic growth. Despite
these concerns, there is very scanty literature and understanding on the relationship between the current
account deficit, budget deficit and private saving-investment gap as well as their implication on economic
growth in the SSA economies.

These macroeconomic imbalances in the SSA economies also poses significant macroeconomic policy
challenges and is problematic because it (i) is associated with low growth and political instability, which
exacerbates economic vulnerability; (ii) reflects a consumption boom and fiscal imbalance rather than rising
private sector investment; and (iii) has implications for fixed exchange rate regimes and price stability.
Moreover, it is associated with real exchange rate overvaluation that adversely affects export completeness,
increases external debt and further reduces savings levels in the SSA economies. Despite these
 3

macroeconomic policy challenges, only a handful of studies has empirically investigated the relationship
between internal deficits, external deficits, macroeconomic policy and economic growth. In light of these
macroeconomic challenges, a more in-depth regional level empirical investigation of growth effects of these
macroeconomic imbalances in the context of the SSA economies is essential. The three studies in this thesis
are motivated by several macroeconomic policy challenges arising due to high macroeconomic imbalances
faced by the SSA economies. The first is the ability to sustain economic growth in light of vulnerable
features mentioned earlier, in particular low and volatile economic growth rates and rising external debt
since 2005. Milesi-Ferretti and Razin (1996) and Roubini & Watchtel (1999) point out that economies that
experience rapid economic growth are able to sustain persistent current account deficit without increasing
their external debt and are better able to fulfil foreign debt obligations. Higher economic growth leads to
higher expected profitability, increases investment rates and ensures that there is little increase in external
debt (Roubini & Watchtel, 1999). The presence of political instability in the region makes both domestic
and foreign investors vulnerable to sudden macroeconomic policy changes and hence increases likelihood
of capital outflows (Milesi-Ferretti & Razin, 1996).

1.3. RESEARCH QUESTIONS

A review of the theoretical and empirical literature, as well as recent macroeconomic policy developments
in the SSA economies, leads to three important research questions:

a) What is the relationship between current account deficit, fiscal deficit and net private
saving in the SSA?

b) What is the impact of current account deficit, fiscal deficit and net private saving on
economic growth in the SSA economies?

c) What is the relationship between macroeconomic policy and economic growth in

SSA?

1.4. RESEARCH OBJECTIVES

The primary objective of this thesis is to investigate empirically the dynamic relationship between
macroeconomic imbalances (current account balance, fiscal balance and private saving-investment
balance), macroeconomic policy and economic policy in the 35 SSA economies. The specific research
objectives are to:

1. empirically investigate the relationship between current account deficit, fiscal deficit
and net private saving in the SSA economies

2. examine the effect of triple deficits on economic growth in the SSA economies

3. examine the theoretical and empirical relationship between macroeconomic policy

(fiscal policy and monetary policy) and economic growth in SSA
 4

1.5. CONTRIBUTION OF THE STUDY AND BRIEF METHODOLOGY

Motivated by various macroeconomic policy challenges faced by the SSA economies, and because of large
and persistent macroeconomic imbalances(internal deficit and external deficit) and the identified research
objectives, this thesis undertakes three empirical studies focusing on three related themes: the relationship
between external(current account deficit) and the internal deficits(budget deficit and private saving -
investment deficit) , growth effects of internal and external balances, and the growth impacts of
macroeconomic policy The studies are also important in light of macroeconomic policy reforms undertaken
by the SSA economies since independence . More importantly, the empirical studies are important from the
perspective of the extant literature. That is, a review of theoretical and empirical literature on sustainability
current account deficit and growth effects of macroeconomic imbalances and macroeconomic policy (see,
Chapter 3) suggests that overall, there is no consensus on the relationship between external deficit and
internal deficit ; and the impact of internal and external balance on economic growth, and the growth
effectiveness of macroeconomic policy have not been fully explored. In addition, the existing empirical
literature is plagued by econometric limitations which limits the usefulness of results for effective
policymaking. The three studies seek to address some of the limitations in the literature and improve upon
the extant literature on sustainability of current account deficit, growth effects of internal and external
balances, and the growth effects of macroeconomic policy. It is worth clarifying here how the three empirical
studies are closely interlinked. For instance, if a policymaker seeks to develop policies to address the
persistence of current account deficit problem, they would be interested in understanding its relationship
with the internal balances. After understanding their relationship with macroeconomic variables, the
policymaker would be interested in determining the sustainability of current account deficit. Following this,
they will want to know if external and internal deficits have any effects on growth and finally, they would
understand the policy implications of all these macroeconomic foundations. The first study extends the
empirical literature on the relationship between external deficit and internal deficit. Recently developed
econometric approaches were employed to empirically investigate the relationship between these deficits
over the four-decade period (1980–2018). This is an important issue because the majority of the SSA
economies have experienced a significant rise in external debt since 2005 (World Bank, 2018). The 2013
experience of Eastern Africa in defaulting on debt obligations to China provides further motivation for
undertaking an in-depth examination of sustainability of current account deficit for the SSA economies. The
results also indicate if there is a possibility for drastic policy changes in the future and new insights on the
effectiveness of past macroeconomic policy reforms in addressing macroeconomic instability.
In addition, in this chapter, empirical study draws on the recent theoretical literature and advanced econo-
metric techniques to examine the relationship between external deficit and internal deficit that are respon-
sible for the deteriorating macroeconomic imbalances position in the SSA economies over the four-decade
period (1980– 2018). The use of a new reduced-form specification (derived based on recent theoretical
developments) and the latest panel data techniques to address concerns of cross-sectional dependency,
cross-country heterogeneity and endogeneity, and the focus on the SSA economies are an improvement
 5

over previous studies and advances the existing literature. The results provide important insights for poli-
cymakers trying to address macroeconomic instability in the SSA economies and address macroeconomic
policy challenges that arise as a result of it.

The second empirical study investigates the effect of internal and external balance on economic growth in
the SSA economies. This study employs the three-gap theoretical framework to provide new insights into
the implications of macroeconomic imbalances on economic growth in the SSA economies. This in turn
will provide a new explanation for low economic growth experienced by the SSA economies over the last
three decades and will enrich the growth literature.
Thus, this thesis advances the literature by providing new policy perspectives on the relationship between
external deficit and internal deficit, and the effect of macroeconomic imbalances on growth, as well as the
growth impact of macroeconomic policy drawing on recent theoretical and econometric literature for a
group of economies that have faced large and persistent current account deficit but have received no
attention in the mainstream literature. In doing so, the thesis examine the theoretical validities in SSA
countries and addresses the methodological limitations of previous studies. More specifically, the empirical
results from this thesis will provide a sound basis for understanding the factors contributing to worsening
current account positions and assist policymakers formulate better strategies and macroeconomic policies
to manage current account positions and thus sustained economic growth. Finally, the third main outcome
would be providing policy makers with an alternative perspective on how external and internal deficits
constrain economic growth in the SSA economies. All in all, the thesis will provide new perspectives on
three related issues.

1.6. ORGANIZATION OF THE THESIS

This thesis consists of seven chapters. The structure of the thesis is structured as follows. Chapter 2
discusses the macroeconomic performance and trend of the SSA economies and provides a succinct
discussion of recent macroeconomic policy reforms and background information on macroeconomic
performances and trends for the last three decades. Chapter 3 provides review of theoretical and empirical
literature. Chapter 4 presents an empirical study on the relationship between external deficit (current
account deficit) and internal deficits (budget deficit and private saving-investment deficit) in the SSA
economies, and Chapter 5 presents an empirical study on the impact of internal and external balances on
economic growth in the SSA economies. Chapter 6 presents an empirical study on macroeconomic policy
and economic growth in the SSA economies. Finally, Chapter 7 provides the summary, conclusion and
policy implications.
 6

CHAPTER TWO: MACROECONOMIC PERFORMANCE AND

EXPERIENCES IN SSA
2.1 INTRODUCTION
This chapter is developed in order to understand the macroeconomic performance and experience of the
sub- Saharan African region. We present major macroeconomic performance and trend of the sub-Saharan
African countries included in the research; however, we focus on 35 countries. These countries are Benin,
Botswana, Burkina Faso, Burundi, Chad, Cameroon, Cape Verde, Central Africa, Cote de’voire, Congo
Demo. Congo Rep., Comoros, Gabon, Eswatini , Ethiopia, Gambia, Ghana, Kenya, Lesotho, Madagascar,
Mali, Malawi, Mauritius, Niger, Nigeria, Rwanda, Senegal, Seychelles, Sierra Leon, Sudan, Tanzania,
Togo, Uganda, Zambia and South Africa. The chapter covers key macroeconomic statistics.

2.2. TRENDS OF ECONOMIC GROWTH IN SSA IN 1980-2018

The annual growth rate for sub-Saharan Africa in comparison to other world regions and the world average
is shown in Figure 2.1. For the 1980 to 1999 period, the growth performance in the region was on average
below the world growth rate at increasing trend. However, since 2000, growth in sub-Saharan Africa has
been above the world average. Growth in the region has persistently outperformed Latin America and the
Caribbean, Europe and Central Asia, and the world average growth. Although the growth rate of SSA is
above the world average growth rate, since 2004, the trend of the annual GDP growth is declined.

Economic activity in Sub-Saharan Africa (SSA) decelerated from 6.53 percent in 2004 to 2.39 percent in
2018, the weakest performance since 2004, due to a combination of external shocks and domestic constraints
(Table 2. 1). The slowdown was most pronounced among oil exporters. In Nigeria, the region’s largest
economy and oil exporter, growth slowed to 3.3 percent, down from 6.3 percent in 2014. Growth moderated
in several mineral and metal exporters – including Mauritania, South Africa, and Zambia (Table 2.6.2). In
South Africa, the economy expanded by 1.3 percent, compared with 1.5 percent in 2014. With the Ebola
crisis receding, activity rebounded somewhat in Liberia, but remained weak in the other affected countries
(Guinea, Sierra Leone) with GDP falling sharply in Sierra Leone as mining production contracted. Activity
weakened substantially in Burundi and South Sudan amid political instability and civil strife. However, in
other countries, including low-income ones and some fragile states—Côte d’Ivoire, Rwanda, and Tanza-
nia—growth remained robust, reflecting lower exposure to the commodity slowdown, and tailwinds from
large-scale infrastructure investment.

Following their sharp decline in 2004, commodity prices weakened further in 2015 (Figure 2.4). The prices
of oil and metals, such as iron ore, copper, and platinum, declined substantially. Those of some agricultural
commodities, such as coffee, fell moderately, although the prices of cocoa and tea showed small gains. The
region’s pattern of exports makes it particularly vulnerable to commodity price shocks. Fuels, ores, and

6
 7

metals accounted for more than 60 percent of the region’s total exports in 2010-14 compared with 16 percent
for manufactured goods (Figure 2.4).

Lower commodity prices obliged a fiscal tightening in several commodity exporters, which caused a sharp
slowdown. Angola and Nigeria are heavily dependent on oil for fiscal revenues and reserves—oil accounts
for more than 60 percent of their fiscal revenues and more than 80 percent of exports. Governments in the
two countries reduced expenditures sharply, which adversely impacted other areas of their economies. The
decline in metal prices hit Mauritania and Zambia.

Figure 2.1 economic growth comparison

10.00

8.00

6.00

4.00

2.00

0.00
1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018
-2.00

-4.00

-6.00

East Asia & Pacific GDP growth (annual %) Europe & Central Asia GDP growth (annual %)
Latin America & Caribbean GDP growth (annual %) Sub-Saharan Africa GDP growth (annual %)
South Asia GDP growth (annual %) World GDP growth (annual %)

Source World Economic Outlook (IMF,2018)

7
 8

Data on the supply side sectoral composition of output is available from 1981. A summary of the data for
sub- Saharan Africa is shown in Figure 2.2. Over the last four decades, the contribution of agriculture and
industry has decreased by -15.8 and -25.75 percentage points respectively, while services have increased by
15.55 percentage points, On the other hand, the average growth rate of the contribution of agriculture, in-
dustry and service were -0.37%, - 0.67% and 1.1%, respectively.
Figure 2.3. compares the demand side sectoral composition of output for SSA. Sub- Saharan Africa has the
largest share of household final consumption in output, compared to another component of GDP. In addi-
tion, the region has the smallest share of general government consumption in total output. Over the last four
decades, the contribution of household final consumption and general government consumption has de-
creased by 14.24 and 52.5 percentage points respectively, while gross capital formation has increased by
0.51 percentage points, In addition, the average growth rate of the contribution of household final consump-
tion , general government consumption and gross capital formation were -0.31%,0.18% and -1.81%, respec-
tively.

Figure 2.2; Supply side Sectoral Figure 2.3.Demand composition of

composition of output, % of GDP output, % of GDP

150 100.00

50.00
100
0.00
1980
1983
1986
1989
1992
1995
1998
2001
2004
2007
2010
2013
2016
50
General government final consumption
expenditure (% of GDP) (WEO)
0
Gross capital formation (% of GDP)

Services, value added (% of GDP) ..

Manufacturing, value added (% of GDP) ..
Industry (including construction), value added (% of GDP) ..
Agriculture, forestry, and fishing, value added (% of GDP) ..

Source: World Bank (2020)

2.3. TRENDS OF MACROECONOMIC IMBALANCES IN 1980-2018

Since the time of independence, the economy of SSA countries has been facing problems of uncertainty and
economic instability. Towards addressing this, the Governments of SSA countries has been concerned with
structural policies that pertain to economic growth, creation of employment opportunities, inflation man-
agement, and commercial policies directed towards ensuring SSA countries becomes economically a viable
state. SSA countries are prepared to achieve both internal balance (domestic full employment with price
stability); and External balance (equilibrium in the balance of payment), with fiscal and monetary policies
being the principal instruments for achieving economic stability in SSA.
 9

According to Egwaikhide et al., (2002), one of the most important policy that facilitates rapid economic
growth in developing countries is allocation of a significant share of national income to investments. Do-
mestic savings which finance the investments for development are the internal bottleneck for these econo-
mies. Because of insufficient savings occasioned by deficiency of national income, investments cannot be
augmented, level of productivity remains low and low level of national income keeps happening. As a result,
inadequate savings cannot fully finance domestic investments, hence the problem of fiscal deficit arises.
Foreign exchange deficit is an important external bottleneck, as well, as the saving deficit in economic
development. For an economy to experience growth, it first needs to import capital goods. However, because
in developing countries a large proportion of foreign exchange revenues are contingent on export of primar-
ily unprocessed agricultural goods, the level of these revenues remains far below the level required by tar-
geted growth rate. This results in a trade account deficit in the economy (Zengin, 2000). Thus, to sustain the
fiscal deficit and trade deficit, total national savings need to more than proportionately increase (Langdana,
1990). The addition of the savings-investment inequality and its interaction with the twin deficits brings in
the Triplet Deficits Concept.

The reason for emergence of the Triplet Deficits Hypothesis in the literature is that in recent years current
account deficits tend to increase, while the budget deficit also increases. With the liberalization of capital
mobility globally, and opening up of financial markets in most economies, the necesity that domestic in-
vestments be restrained to the level of domestic savings receded. In economies where domestic investments
are higher relative to domestic savings, the financing of the resultant savings-investment (S-I) gap from
outside forces the S-I balance to be a factor. This means that the fiscal, Trade and savings-investment bal-
ance of an economy are all in deficit. In the literature, this scenario is what is referred to as the Triplet
Deficits Hypothesis.

Establishing whether the Triplet Deficits hypothesis is valid for an economy is critical for policy selection.
In economies like SSA countries, where imbalances at the macroeconomic level are at serious levels (high
unemployment rate, huge budget deficit, high inflation rate, trade deficit, increased ownership of domestic
assets by foreigners, and lowest economic growth rate among SSA economies), determining the validity of
the triplet deficits hypothesis and to what extent it occurs is important. Policymakers can make better and
more informed policy decisions to manage the deficits, going for either public or external borrowing or
adjustment of tax rates, or promotion of a saving culture to finance the

deficits. In SSA this study would not have come at a better time, the government is under intense pressure
to cap its borrowing (International Monetary Fund, 2018) after raising the external debt to finance its budg-
etary obligations and key flagship infrastructure projects. This risks a high external interest rate payment in
the future, and a weakening of the currency during repayment. Further, SSA’s gross national savings rate
 10

(%GDP) stands at 11.578% far below the world average of 18.06% (The World Bank, 2016) thus invest-
ments cannot be augmented, level of productivity remains low and low level of national income keeps going.

The parallel negative balances of fiscal, trade and financial account over the study period 1980- 2018 inform
SSA experience of the Triplet Deficits (see Figure 1.0). The evidence of the Triplet Deficits is apparent that
the years have budget balance, current account and Savings-Investment balance in deficit. Stephen (2010)
noted that to attain sustained economic growth and macroeconomic stability, the deficits have to be kept in
control. The World Bank adds that deficits in the three accounts have undesirable effects on the domestic
prices, interest rates, balance of payment, and slow growth of the economy.

Figure 2.4: Annual Fiscal Balances, Current Account Balances and Net Savings (S-I), % GDP (1980-2018)

5.00

0.00
19801982198419861988199019921994199619982000200220042006200820102012201420162018
-5.00

-10.00

-15.00

-20.00

-25.00

Fiscal Balance (% of GDP) Current account balance (As % of GDP)

private saving-investment balance(% of GDP)

Source: Africa Development Bank Database, 2020

The figure shows that for most of the period the three accounts were in deficit with a few years registering
a positive account balance (Surplus).

2.3.1 TRENDS OF CURRENT ACCOUNT IMBALANCE

From the Figure 2.4, it is evident that SSA experienced a protracted current account deficit (trade deficit)
over the period of the study. Figure 2.4 shows that there were only five years which SSA attained a surplus
in current account, 2000(0.54), and from 0.79 in 2004, 1.6 in 2005, 2.62 in 2006 and 0.66 in 2007

when the trade balance shows higher surplus, 4.97%,3.95%,5.42%,5.91% and 5.63% of GDP respectively.
The trade surplus was buoyed by a significant rise in coffee prices attributable to the coffee boom. The
worst performance year in the current account deficit was 6.08 % GDP in 2015 when it recorded a deficit
 11

in net exports of 3.46 percent of GDP. During this period, SSA relied heavily on imported intermediary
inputs and heavy machineries for its domestic production that caused imports to exceed exports.

African countries are heavily dependent on the export of primary goods. Figure 2.5. shows the rapid growth
in the export of primary products since 2000. Over 75% of sub-Saharan African exports are primary prod-
ucts or resource-based manufactures. This contrasts with Eastern, Southern and South-Eastern Asia whose
export of the latter category of goods account for roughly 20% of exports while low, medium and high tech-
nology manufactured goods account for the bulk of exports.

On average, during the 1980 to 2018 period, many sub-Saharan African countries experienced current
account deficits. Countries experiencing on average a surplus were Botswana, Gabon and Nigeria. The
latter three are oil exporting economies which have benefited from relatively high oil prices since the early
2000s. These countries have seen a fall in their current account surplus in recent years following the fall
in oil prices. all sub regions of SSA economies experienced current account deficits.
Since the early 2000s, there has been a deterioration in the current account balance in non-oil exporting
countries. The largest declines, as a percentage of GDP, were in Burundi, Cape Verde, Ghana, Sierra
Leone and Zambia.
 12

Figure 2.5a; Export composition sub-Saharan Figure 2.5b; Export composition for world re-
Africa, Lall classifications, USD gions, Lall classifications

Source: United Nations Conference on Trade and Development (2016)

Sub-Saharan Africa’s import composition mainly consists of medium technology manufactures which
have grown rapidly since the early 2000s. High technology manufactured products have also grown as
shown in Figure 2.6.

Figure 2.6; Import composition, sub-Saharan Africa, USD

Source: United Nations Conference on Trade and Development (2016)

12
 13

2.3.2. TRENDS OF FISCAL IMBALANCES

Figure 2.4 also presents the annual account balances for SSA budget account. From the graph, it is evident
that except for 2006 and 2007, SSA has been running chronic budget deficits. The exception was in 2006
and 2007 when it attained a surplus of 3.08, and 0.475 percent of GDP respectively (see figure 2.4). Ac-
cording to international financial statistics data (IMF, various issues). SSA has been running chronic fiscal
deficits in the last three decades. The budget surplus of 2006 and 2007 is attributable to fiscal austerity
measures carried out by the SSA governments.

Fiscal deficits have various adverse effects on the economy. These include increase in national debt, which
might lead to increased ownership of domestic assets by foreigners, higher debt interest repayments,
crowding out of the private sector, higher interest rates, future tax rises as well as inflation.

The Fiscal deficits in SSA are attributable to the macroeconomic policies adopted after 1980. Fiscal defi-
cits can also be attributed to inefficient tax collection and administration (Koori, 1992), macroeconomic
shocks such as the 1979/80 oil price shocks, the world recession in the 1980s and 2008, the droughts, and
the Post-Election violence. They can also be attributed to deteriorating terms of trade for the country’s
export that meant export earnings were either low or insatiable in many cases (Lesiit, 1990) , and poor
budgetary processes coupled with limited resources (Wawire, 2006). To address the budget deficit, the
government needs to practice sound budgetary practices such as fiscal discipline where the budget is bal-
anced between revenue streams and expenditures.

2.3.3. TRENDS OF SAVING-INVESTMENT IMBALANCES

The levels of investment and savings are important determining factors of the attainable rates of employ-
ment and economic growth. Figure 2.4 shows that SSA has been running a savings- investment gap that
has had to be offset by inflow of foreign capital through the financial account.

As shown in Figure 2.4, SSA’s savings investment deficit has steadily risen from about 13. 6% of GDP in
1980 to 20.07% in the 2018 making the economy become increasingly reliant on external funds to finance
its capital formation. This attribute this widening gap to the slight fiscal deficits experienced by the public
sector which has increased from 4.23% of GDP in 1980 to 4.6% in 2018 and higher foreign direct invest-
ment inflow from 0.84% of GDP in 1980 to 9.56% of GDP in 2018. High reliance on external funding
would give rise to large outflows of investible resources in the form of debt repayments.

2.4. CONCLUDING REMARKS

Deterioration in economic performance can be traced to various adverse exogenous developments, inap-
propriate fiscal and monetary policies, especially in the early and late 2000s, and domestic structural
13
 14

factors, especially the failure to expand and to diversify exports. Exogenous developments include the oil
crises of 1979-80 and the consequent world recession; increased protectionism in developed countries;
high external interest rates and a decline in concessionary capital inflows; the frequent droughts which
adversely affected agricultural production and led to massive food imports; which adversely affected in-
vestment and caused some capital flight.
One consequence of this decline has been severe import compression, particularly in the first half of the
1980s. The situation improved somewhat in the second half of the 1980s, mainly due to an increase in net
external capital inflows in return for the successful adoption of adjustment policies financed mainly by
the international financial institutions (IFIs) and due to a temporary decline in oil prices and an improve-
ment in coffee prices in the mid- 1980s. These deteriorated later, however, with oil prices increasing by
77% and coffee prices declining by 37% between 1986 and 1990.
Interest rates were fully freed from administrative controls in July 1991. Interest rate liberalization has
been accompanied by a restructuring of the institutional framework within which commercial banks and
non-bank financial institutions operate. The investment climate is being made more attractive, particularly
for foreign investors, who provide about a third of the investment finance. any countries tapped the inter-
national bond market to finance their investment programs, taking advantage of the global low- interest-
rate environment, and investors’ search for yield. The external headwinds of low commodity prices, of a
slowdown in major emerging markets, and rising borrowing costs were compounded by domestic prob-
lems. These included severe infrastructure constraints, especially power supply, in several countries.

The growth slowdown was associated with mounting fiscal vulnerabilities in a number of countries. Fiscal
deficits widened in oil exporters (Republic of Congo, Gabon, Nigeria) due to falling revenues. In other
countries, the widening deficits reflected increased government spending, including on arrears (Zambia),
infrastructure projects (Kenya), and subsidies (Malawi). Some countries (Angola, Ghana) implemented
expenditure measures – including removing fuel subsidies and freezing public sector hiring – that reduced
the deficits. In many countries, fiscal deficits are larger relative to GDP than they were at the onset of the
global financial crisis (Figure 2.4). As a result, government debt ratios have continued to rise. While debt-
to- GDP ratios remain manageable in most low- income countries, they rose rapidly in several frontier
markets, led by non-concessional borrowing. By contrast, Nigeria’s sovereign debt has remained low, at
less than 15 percent of GDP. External imbalances widened across the region. Current account balances
turned sharply negative in Angola and Nigeria due to lower oil prices. Deficits remained large among oil
importers because of low commodity prices and rising non- oil imports. In Kenya, the current account
deficit remained high as security concerns weighed on tourism earnings. In South Africa, in contrast, the
current account deficit narrowed on the back of export growth. In addition, the depreciation of the rand
partly offset the decline in commodity prices. In Ghana, Kenya, and Namibia, the twin fiscal and current
account deficits have remained large. High fiscal and current account deficits, combined with strong
 15

demand for the U.S. dollar, kept currencies under pressure. Currencies of commodity exporters and fron-
tier-market economies saw sharp depreciations against the U.S. dollar. However, because of inflation, the
average movement in real effective exchange rates across the entire region was relatively small.

The SSA economy has been facing the problem of macroeconomic imbalances after the 1980s. Most of
the SSA economies are agrarian economy. Thus, subsistence economy was prevalent. Under a subsistence
economy there is usually little for surplus (domestic savings or export) as the households consume almost
whatever is produced. These disrupted economic activities because of the massive transfer of resources,
especially, labour from one place to another. This suggests the existence of resource shortages.

During 2007-2018, the urge for economic growth was high but the unavailability of required resources, in
desired quantity and quality, was a major constraint. For example, capital accumulation was weak. It
should be noted that the capital required was not produced in most SSA countries. The capital had to be
imported. In a situation where they were produced locally, two problems were encountered. The first was
their higher cost of production making prices exorbitant. The second was their relatively poor quality.
Also, financing the required investment was made weaker. By implication, to finance the required invest-
ment, therefore, necessitated borrowing, which have to be paid back including the interests. Furthermore,
technology became much more expensive. This gave birth to further technological dependence and the
elusive transfer of technology syndrome. The above was coupled with the problem of technical skills.
there was also the problem of an initial low productivity arising from little experience. The constraints
have resulted in significant imbalance (resource shortages) and poor economic performance.

To clearly bring into focus the imbalances and the consequent poor economic performance during 2000-
2018. Certain policy measures by the SSA governments, notably the austerity measures, introduced
dist01iions into the allocation of the economy's resources and hampered the growth of the economy. The
aim of the austerity measures was to reduce the high-powered money base. As a consequence, the pur-
chasing power of the people reduced, thereby generating imbalances in aggregate demand and aggregate
supply. Olaloye (1988) reported that capital equipment was allowed into the country with little duties paid
on them and some establishments were granted tax exemptions during the early years of their operations,
while more attention was directed towards the manufacturing sector at the neglect of the agricultural sec-
tor. He argued further that the incomes that should have been used for investment in agriculture and in
strengthening the comparative advantage in the export-producing industries were diverted to uneconomic
and highly capital intensive, often inefficient manufacturing sector. The illusion then was that agricultural
sector could naturally grow unaided while the sector that needed support was the manufacturing sector.
The above suggests that the manufacturing sector was highly subsidized. Nevertheless, Fabayo (1981)
pointed to the fact that substantial unused capacity exists in the manufacturing sector. All these measures
induced imbalance situations and a distortion in the allocation of the. Resource of economy.
 16

The above imbalances manifested in the form of unimpressive economic performance. For example the
growth rate of the economy which stood at 6.53 percent in 2004 fell to 2.39 per cent in 2018. The unim-
pressive economic performance was accompanied by many implications. Among these was the effect on
the Savings- Investment deficit (divergence between investment and saving) as percentage share of GDP,
which stood at 13.6 per cent in 1980 only to rise to a negative 20.07 per cent in 2018 Also, overall fiscal
deficit (divergence between total public expenditure and total public revenue) as percentage of GDP which
stood at a negative 4.3 per cent in 1980 showed no significant improvement as it stood at a negative 4.6
per cent in 2018. Furthermore, the balance of current account deficit as percentage of GDP, which stood
at 0.23 per cent in 1980 rise to a negative 6.08 per cent in 2017, (figure 2.4 refers).

TABLE 2.1 Macroeconomic imbalances (resource gaps) and economic performance in SSA, 1980-
2018

Indi-
cator Fiscal Balance Current account Private saving-in- GDP
Name balance vestment balance Trade balance growth
% of Billion % of Billion % of Billion % of Billion (an-
Year GDP US $ GDP US $ GDP US $ GDP US $ nual%)
1980 -2.98 -12.81 -0.23 -0.98 -13.60 -58.51 4.73 1458.76 4.04
1981 -4.23 -17.57 -4.66 -19.38 -15.28 -63.51 -2.06 -665.12 -0.46
1982 -4.89 -19.72 -4.45 -17.97 -13.02 -52.53 -1.90 -618.38 -1.14
1983 -3.62 -14.80 -2.50 -10.23 -11.51 -47.07 0.47 140.62 -2.50
1984 -2.84 -11.16 -1.46 -5.74 -10.82 -42.56 2.40 570.49 2.60
1985 -1.95 -7.43 -0.32 -1.20 -9.79 -37.35 5.08 1096.75 1.76
1986 -2.49 -7.90 -2.11 -6.68 -12.69 -40.21 2.03 501.64 1.61
1987 -4.79 -14.76 -1.26 -3.89 -13.96 -43.01 3.39 913.41 2.73
1988 -4.28 -14.44 -2.54 -8.56 -14.50 -48.87 2.05 588.11 4.49
1989 -3.23 -11.14 -1.81 -6.22 -14.12 -48.64 2.69 810.48 2.53
1990 -2.88 -11.42 -1.24 -4.92 -12.95 -51.33 2.93 1143.02 2.38
1991 -3.72 -15.00 -1.83 -7.37 -12.92 -52.09 2.14 846.93 0.39
1992 -5.22 -21.15 -1.91 -7.73 -12.53 -50.78 1.59 622.10 -0.38
1993 -4.96 -19.30 -1.78 -6.94 -12.60 -49.03 1.77 699.14 -0.90
1994 -4.34 -15.37 -2.03 -7.18 -15.08 -53.44 1.53 586.87 1.21
1995 -3.13 -12.12 -2.91 -11.29 -15.12 -58.64 0.78 368.88 3.37
1996 -2.44 -9.64 -1.45 -5.71 -15.22 -60.02 1.95 1030.03 5.11
1997 -3.04 -12.44 -1.98 -8.11 -15.39 -62.96 1.30 718.93 3.60
 17

1998 -3.39 -13.42 -4.30 -17.02 -16.14 -63.81 -0.37 -206.06 2.37
1999 -3.09 -12.18 -2.78 -10.99 -16.29 -64.28 1.03 411.96 2.16
2000 -0.81 -3.19 0.54 2.13 -19.58 -77.28 4.97 2030.34 3.50
2001 -1.92 -7.32 -1.40 -5.33 -20.13 -76.89 3.27 1287.99 4.26
2002 -2.11 -8.71 -3.10 -12.80 -20.31 -83.86 2.75 1170.78 6.34
2003 -1.70 -8.79 -2.53 -13.07 -20.61 -106.40 2.53 1350.20 4.25
2004 0.30 1.90 0.79 5.07 -21.23 -136.26 3.95 2598.99 6.53
2005 -0.09 -0.72 1.60 12.32 -21.22 -163.34 5.42 4228.37 6.21
2006 3.09 28.03 2.62 23.74 -22.91 -207.94 5.91 5366.93 6.14
2007 0.47 4.98 0.66 6.98 -21.87 -230.56 5.63 5943.95 6.62
2008 -0.23 -2.79 -0.42 -5.01 -22.89 -276.27 6.05 7347.57 5.36
2009 -4.44 -50.54 -2.60 -29.65 -23.15 -263.63 2.32 2685.37 3.04
2010 -2.34 -31.82 -0.84 -11.42 -21.25 -288.79 4.47 6143.14 5.58
2011 -1.55 -24.47 -0.85 -13.40 -20.74 -327.95 5.49 8657.79 4.45
2012 -2.20 -36.18 -2.11 -34.74 -21.69 -357.40 3.03 4965.14 4.04
2013 -3.21 -55.67 -2.61 -45.25 -21.42 -371.75 2.48 4287.88 5.00
2014 -3.34 -60.79 -3.78 -68.77 -22.02 -401.05 0.38 694.96 4.66
2015 -4.09 -67.09 -6.08 -99.85 -22.14 -363.34 -3.46 -5640.54 2.83
2016 -4.38 -66.30 -3.93 -59.44 -21.12 -319.61 -2.31 -3480.77 1.23
2017 -4.60 -73.92 -2.56 -41.12 -20.77 -333.72 -0.10 -165.34 2.53
2018 -3.55 -60.83 -2.70 -46.41 -20.07 -344.32 0.61 1035.72 2.39

All the problems identified in the previous periods were also persistent during the period 2004-2018, in
some cases with greater dimensions. For example, high level deficit financing had resulted into huge pub-
lic debt. The public debt had to be serviced. This resulted in large transfer expenditures with significant
proportion going into debt servicing. Another worrisome aspect of the SSA expenditure profiles was that
most of them were externally oriented. These were either to service foreign debt or to import resources
for the economy. The implication is that significant leakage occurred in the economy resulting in balance
of payments dis-equilibrium.
 18

CHAPTER THREE: LITERATURE REVIEW

3.1. REVIEW OF THE THEORETICAL AND EMPIRICAL LITERATURE ON
THE RELATIONSHIP BETWEEN EXTERNAL DEFICIT AND INTERNAL
DEFICIT
3.1.1. INTRODUCTION
This section mainly focuses on the underlying theoretical and empirical literature review. It consists of
several sections which are theoretical literature review and empirical literature. finally, conclusion with a
critique of the previous studies presented.
3.1.2. THEORETICAL LITERATURE REVIEW

A) Neoclassical View
The standard neoclassical model has three main assumptions which are: consumers are rational,
farsighted, and have access to perfect capital markets. This would then mean that permanent deficits
significantly depress capital accumulation, and temporary deficits have either a negligible or perverse
effect on most economic variables (including consumption, saving, and interest rates). If many
consumers are either liquidity constrained or myopic, the impact of permanent deficits remains
qualitatively unchanged. However, temporary deficits should depress saving and raise interest rates in
the short run.

B) Keynesian model
The Twin Deficit Hypothesis is grounded within the traditional Mundell- Fleming paradigm. Keynesian
proponents argue that fiscal expansion has an effect of raising absorption. This will push up the appetite
for foreign goods and ultimately diminish the surplus in the current account balance. Fiscal expansion
has also a crowding – out effect on the domestic market which raises the interest rate and the resultant
capital flows will lead to appreciation of currency. Domestic goods will appear to be expensive in the
eyes of foreigners, thus exports will diminish, and the current account worsens. The Keynesian view
made a number of assumptions which are: economic agents are either myopic or liquidity constrained,
individuals have a high marginal propensity to consume out of current disposable income and that the
economy has some resources which are under-employed. Eisner (1989) also argued from the Keynesian
point of view suggesting that increased aggregate demand enhances profitability of private investment
thereby leading to a higher level of investment at any given rate of interest. Budget deficits are therefore
viewed as a tool to stimulate aggregate saving and investment, despite the fact that they raise interest
rates. Eisner assumes underemployment in the economy, thus increased consumption would be supplied
by unutilized resources.

C)The Fiscal Approach to Balance of Payments

The fiscal approach to the determination of balance of payments is based upon the national income
identity which states that the current account is equal to government balance and the private sector
 19

balance between investment and savings (Bartoli: 1989). According to this approach, when domestic
savings and investment are equal then the resulting variations in the current account balance will have
been a consequence of variation in the fiscal budget deficit. Policymakers will thus have to use the fiscal
policy or adjustment to domestic national savings and investment to adjust the national accounts.

The fiscal approach is one-sided in that it only takes into account the causality running from fiscal deficit
and the savings and investment relationship. According to Chu (1989), in highly open economies trade
balance can be transmitted directly to the fiscal sector. However, fluctuations in the expenditures rather
than in revenues were the immediate cause of unstable fiscal deficits only if government expenditures are
based on anticipated future revenue which is also a function of future trade.

D)Ricardian Equivalence Hypothesis

The Ricardian Equivalence Hypothesis (REH) was introduced by Barro (1974) and its arguments arise
from the Neoclassical school of thought. The proposition states that the cuts in taxes are matched by an
increase in savings since people look forward to the government increasing the taxes in future. This
foresight gives rise to Say's Law for deficits that the demand for bonds always rises to match government
borrowing. The proposition is expected to hold under the following conditions: generational linkages,
non-distortionary taxes, rational expectations (perfect foresight concerning the path of taxes and fiscal
policies), identical planning horizons for both private and public sector agents, the availability of deficit
financing as a fiscal instrument does not alter the political process and perfect capital markets with no
borrowing constraint. However, much criticism has been raised concerning the realism of these
assumptions. The Ricardian Equivalence in an open economy will produce the same results as in a closed
economy. In an open economy real interest is determined in the world capital markets and within the
economy individuals are free to borrow and lend. Given that both public and private sector agents face the
world interest rates Ricardian Equivalence is satisfied just like in the closed economy case. An increase
in government debt is fully internalized by the private sector which accounts for the taxes to be paid back
to lenders. In an open economy the private sector’s savings rise by enough to avoid having to borrow from
abroad (Barro, 1989).

Leachman (1996) and David Ricardo (1966) argued that there is no first order difference between tax and
debt financed expenditure. The payment for public debt would be financed by future taxes, money creation
and reduces government expenditure or additional deficits. Barro (1974) considered the effect of bond
values and tax capitalization, finite lives, imperfect capital markets, government monopoly in the
production of liquidity services and uncertainty about future tax obligations. The findings of the paper
revealed that as long as there are intergenerational linkages there would be no net wealth effect and
aggregate demand will not be affected.
 20

Buchanan (1976) was the first person to point out the close relationship between Barro’s proposition and
the work of David Ricardo. Furthermore, Barro (1979) concluded again that the choice between debt and
taxes does not really matter, however the study also sought to identify factors that influence the choice
between debt and taxes. In a later paper, Barro (1989) cited major conjectural objections: that people do
not live forever, and do not care about future taxes, private capital markets are not perfect, future taxes
and income are not certain, taxes are not lump sum and the assumption of full employment. However, a
number of observed findings tend to support Ricardian Equivalence. The study also notes that empirical
analysis involves considerable problems with data and identification thereby rendering empirical literature
to be inconclusive. This was also supported by Elmendorf and Mankiw (1999).

Leiderman and Blejer (1988) and Seater (1993) illustrated the implications of Ricardian equivalence. The
former relaxed the main assumptions of the Ricardian model and the study concluded that debt financing
policies can have an impact on private consumption and aggregate demand. Seater (1993) found out that
Ricardian Equivalence is logically reliable but the restrictions necessary for it to hold are too many and
not likely to be met.
3.1.3. EMPIRICAL LITERATURE REVIEW
The existence of twin deficits has always remained highlighted for the role it plays to lessens the
sustainability of the economy. The vast literature is available that discussed the relationship between
fiscal deficit and current account deficit.
Anoruo and Ramchander (1998) empirically investigated the existence of twin deficits in five Southeast
Asian countries. They used annual data from 1957 to 1993 and employed VAR model for estimation.
Results showed that only trade deficit caused fiscal deficit in all countries except for Malaysia where
bidirectional relationship was found. They concluded that fiscal and trade deficits are affected by various
macroeconomic determinants. Vamvoukas (1999) examined the twin deficits in Greece. He used annual
data from 1948 to 1994 by employing cointegration technique, ECM and Granger tri-variate causality
tests. The results revealed budget deficit has significant impact on trade deficit for long run as well as
short run. Thus, it can be concluded that trade deficit can be reduced by decreasing the budget deficit.
Akbostanci and Tunc (2002) analyzed influence of fiscal deficit on current account deficit for long run
as well as short run in Turkish economy. They used quarterly data from 1987 to 2001 and used ECM and
Cointegration analysis for estimation. Results found the existence of twin deficits in long run and short
run. Hence, Ricardian equivalence hypothesis is rejected for Turkey and twin deficit hypothesis is
considered to be valid for Turkish economy. Basu and Datta (2005) empirically investigated the impact
of the fiscal deficit on India’s external deficit. Authors used quarterly data from 1985 to 2003 and
employed cointegration tests for estimation. Results found fiscal deficit and trade deficit are not
cointegrated. They concluded that both these deficits and net savings casually preserve the national
income identity and the ratio of high fiscal deficit has remained persistent because of the autonomous
increase in the saving ratio. It is also found that Indian consumers are not Ricardian. Saleh et al. (2005)
 21

investigated the existence of internal and external deficit in Sri Lanka. They used annual data from 1970
to 2003 and adopted ARDL bound test for cointegration. The results supported the Keynesian
proposition of twin deficits.
There existed causal link moving from fiscal deficit towards current account deficit. So, policies to reduce
fiscal deficit could also be effective in improving the current account deficit. Saleh (2006) empirically
investigated whether there exist any association between budget and trade deficits for Lebanon. He used
annual data from 1975 to 2003 and conducted cointegration and Granger causality tests for empirical
analysis. Results found evidence in favor of Keynesian proposition and the reversed causal relationship
from trade to budget deficit was found. Thus, budget deficit could be eliminated by diminishing the trade
deficit. Onafowora and Owoye (2006) analyzed short run and long run causality for budget deficit and
trade deficit in Nigeria. They used annual data from 1970 to 2001 by utilizing cointegration estimation
technique, Error correction model and causality analysis. Results supported Keynesian view by refuting
the Ricardian equivalence hypothesis. The unidirectional causal relationship was found from trade to
budget deficit. Hence, with the help of indirect monetary channels both deficits can be reduced in Nigeria.
Samadi (2006) aimed at examining the Keynesian proposition of twin deficit with testing the Ricardian
Equivalence hypothesis in MENA countries. Annual data from 1971 to 2000 has been employed in the
study and cointegration techniques, Error correction model, and the Granger causality test have been
performed. Results found mixed evidence of Ricardian Equivalence and Keynesian view. Therefore,
finding of this paper can be helpful in designing the appropriate fiscal policies in Middle East and North
Africa (MENA )countries, especially in Iranian economy.

Lau et al. (2006) estimated twin deficits hypothesis for nine SEACEN economies. They used annual data
from 1980 to 2001 and adopted panel cointegration, Granger causality test and dynamic OLS panel VAR
for estimation. Results found bidirectional causal relationship between both deficits. However, policy
reforms to lessen the fiscal deficit offers opportunity to reduce current account deficit. Chowdhury and
Saleh (2007) analyzed the degree of association among internal and external deficit in presence of saving
investment gap and free trade in Sri Lanka in long run and short run. The study used data over 1970 to
2005 time period and employed ARDL approach for estimation. The results found evidence in favor of
Keynesian view. While, Trade openness also found to be positively but insignificantly affecting current
account deficit. It is concluded that policies to reduce budget deficit would be beneficial for reducing
current account deficit in Sri Lanka. Marinheiro (2008) empirically investigated twin deficit hypothesis,
Ricardian equivalence hypothesis and Feldstein Horioka Puzzle in Egypt. He used annual data from 1974
to 2003 and performed cointegration, Granger causality test and fully modified OLS (FM-OLS)
estimation techniques. Results rejected twin deficit hypothesis and found reverse causal link moving
from external deficit to internal deficit. Ricardian equivalence and Feldstein Horioka Puzzle are also
rejected indicating the high international capital mobility and also specifying the need for future research.
 22

Acaravci and Ozturk (2008) investigated the impact of internal deficit on external imbalances. They used
quarterly data from 1987 to 2005 by utilizing the ARDL bound test for cointegration. The results supported
the Keynesian view by founding positive relationship between internal and external imbalances. The
unidirectional causality was found running from budget deficit to current account deficit. Javid et al.
(2010) analyzed whether current account balance is effected by shocks of fiscal deficit in Pakistan. The
study used annual data from 1960 to 2003 and employed structural Vector Autoregressive model VAR
analysis. The results found that external deficit is improved by shocks of fiscal deficits while exchange
rate is deteriorated. The Ricardian view is also supported as the output shocks explained twin divergence.
Iram et al. (2011) investigated the presence of Keynesian view of twin deficits in Pakistan. Annual data
has been used from 1972 to 2008 by employing ARDL framework, multivariate causality tests and
seemingly unrelated regressions (SUR) techniques. Results supported the Keynesian view. Further, it is
suggested that reduction in fiscal deficit is essential to eliminate current account deficit for enhancing
economic growth.

Magazzino (2012) empirically investigated the twin deficits hypothesis and Ricardian equivalence
hypothesis for European countries. He used annual data from 1970 to 2010 and employed Generalized
Least Squares-Fixed Effects for static estimation and Generalized Method of Moments (GMM) for
dynamic estimation along with Granger causality test. Mixed results have been obtained. The FE
estimator confirms the Twin deficit hypothesis. Yet, the dynamic estimates reached conflicting results.
In fact, GMM-Difference estimates supported twin deficit hypothesis while the GMM-System method
supported Ricardian equivalence hypothesis. Granger causality test also showed mixed results. Ratha
(2012) investigated the twin deficits hypothesis for India. The study used Monthly data from 1998:1 to
2009:9 and quarterly data from 1998Q1 to 2009Q1 and adopted bound testing approach for cointegration
and Error correction model for estimation. Results found evidence for Keynesian proposition of twin
deficits in short run while supported Ricardian Equivalence Hypothesis in long run. Thus, trade deficit
can be reduced by decreasing budget deficit. While, fiscal policies are not effective in the long run. Roy,A.
et al. (2013) empirically examined the twin deficit hypothesis for Bangladesh. They used annual data
from 1972 to 2012 and utilized cointegration test and VAR and Granger causality tests for estimation.
Results found absence of long run cointegration. While, in the short run causal link existed moving from
budget deficit to current account deficit. However, government should take appropriate reforms to lessen
budget deficit.

Kayhan et al. (2013) examined the association for public spending and trade deficit in Turkey. They used
data from 1987Q1 to 2011Q3 and employed causality test for estimation. Findings showed presence of
causality between public spending and trade deficit. An important conclusion can be drawn from results
that decrease in trade deficit can be achieved by decreasing government spending. To permanently get
rid of trade deficit, additional policy applications are needed. Catık et al. (2015) empirically evaluated
the association between fiscal deficit and current account deficit for Turkey. They used data from 1994:1
 23

to 2012:3 and employed causality test and VAR (TVAR) model for estimation. The results found the
macroeconomic activity to be responsible for any connection between current account deficit and fiscal
deficit. Therefore, external balance of an economy can be improved by appropriately using fiscal and
monetary policies. Ravinthirakumaran et al. (2016) investigated whether there exist any causality
between current account balance and fiscal deficit in SAARC countries. They used annual data from 1980
to 2012 and conducted cointegration technique, ECM and Granger causality test. Findings suggested
existence of causal link between budget deficits and current account balance in SAARC economies. It is
concluded SAARC countries must take appropriate policy measures in external and internal sector to get
rid of both deficits.

Walker (2011) studied the extent to which Japanese households conform to Ricardian equivalence. The
study employed VAR techniques on national accounts data and the results suggested that the Ricardian
Equivalence hold. Moreover, there was some form of private savings off-setting to change in fiscal policy.
Yi (2003) considered South Korea data, the study found no cointegration relationship between the
variables (real exchange rate, current account, and consumption). This implies Ricardian equivalence
holds.

Bernheim (1987), Giorgioni and Holden (2003) used a sample of ten developing countries (Burundi, El
Salvador, Ethiopia, Honduras, India, Morocco, Nigeria, Pakistan, Sri Lanka and Zimbabwe) to test the
Ricardian equivalence. The study applied Bernheim’s framework of private consumption across the panel
of countries and the conclusion was that there was some presence of Ricardian equivalence. However,
they were cautious and unconvinced given the diversity of countries and data limitations within the group.

Berben and Brosens (2007) were interested in finding out whether the observed consumer reactions to
fiscal policy could be explained by the level of government debt. A panel of 17 OECD countries was used
and the ARDL approach to cointegration was applied. The results from the study pointed out that in the
long run consumption is positively related to disposable household income, equity wealthy and housing
wealth. Government debt has a statistically significant negative impact that is to say fiscal expansion is
partly crowded out by a fall in private consumption.
3.1.4 SUMMARY AND CONCLUSION OF THEORETICAL AND EMPIRICAL
LITERATURE REVIEW
The Ricardian Equivalence is grounded in the Neoclassical school of thought, however, both Neoclassicals
and Keynesians agree that budget deficits have real effects. Neoclassicals are mainly concerned about the
long run effects of deficits on capital accumulation while Keynesians are more interested in the short run
effects of deficits and their ability to stimulate consumption and national income. It can be concluded that
one can find support for every conceivable normative position and no single choice of paradigm
corresponds exactly to reality.
 24

The results from both Ricardian Equivalence and the twin deficit are conflicting and are not consistent
across countries and over time. This is likely to be stemming from the different empirical techniques,
data measures and samples.

Econometric methods have however been evolving over the years. Early studies used univariate
techniques while multivariate techniques are a recent development. There is also lack of a common
methodology in the compilation of government spending and government deficits across countries. This
is a data problem which mainly affects studies where a panel of countries is used.

Policy makers have also remained interested in analyzing the possible link between fiscal deficit and
current account deficit. An inclusive literature has discussed the twin deficits hypothesis in both
developed and developing economies. In sum, the literature provides mixed results for the existence of
twin deficits in different countries. But there is limited work in SSA which has examined the relevance
of twin deficits jointly with Ricardian equivalence hypothesis and triple deficits. Moreover, the literature
does not incorporate the private savings investment balance while estimating the twin deficits model. So,
it is essential to incorporate this variable in the model to discover that either budget deficit or private
savings investment balance cause current account deficit. Further, it is also important to determine either
the twin deficits or Ricardian equivalence hypothesis or triple deficit exists in SSA countries.

3.2. REVIEW OF THE THEORETICAL AND EMPIRICAL LITERATURE ON

THE IMPACT OF TRIPLE DEFICIT ON ECONOMIC GROWTH
This section provides a succinct survey of the theoretical and empirical literature on the impact of internal
and external deficits on economic growth.
3.2. 1.REVIEW OF THE THEORETICAL LITERATURE
This section presents a critical review of the different demand side growth models that have been proposed
with a focus on heterodox models, in particular the Keynesian/post-Keynesian demand-led growth
models.

3.2.1.1. THE EXTENDED HARROD-DOMAR MODEL TO AN OPEN ECONOMY

Thirlwall (2001) extended the Harrod (1939) model to an open economy by introducing a fourth growth
𝑥
rate, the balance of payments equilibrium growth rate, GB, 𝐺𝐵 = where, x is the growth of exports
𝜋

and π, the income elasticity of demand for imports. Assuming that deficits cannot be financed
indefinitely, a constant real exchange rate and constant relative prices, GB, is the growth rate consistent
with equilibrium on the current account balance (Thirlwall, 2001). The balance of payments equilibrium
growth rate originates from the Harrod (1933) foreign trade multiplier which shows that under certain
assumptions, income adjusts to restore equilibrium on the current account of the balance of payments.
Taking the simplest case, income, Y, is derived from the production of consumption goods, C, and
exports, X, 𝑌 = 𝐶 + 𝑋. All income is spent on consumption goods, C, and imports, M. There is therefore
 25

no savings or investment. The real terms of trade are assumed to be constant, so when trade is balanced,
X=M. Exports are taken as given, based on the domestic cost of production and world prices and demand.
A constant fraction of income, i, is devoted to imports thus, 𝑀 = 𝑖𝑌 When trade is balanced,
𝑋 𝛥𝑌 1
𝑋=𝑖𝑌.Hence,𝑌 = Therefore, = ,The multiplier, 1/i, returns the balance of payments to
ⅈ 𝛥𝑋 ⅈ

equilibrium through changes in income, Y brought on by a change in exports, X or imports, shows that
the balance of payments matter for income determination and therefore economic growth. If the natural
rate of growth is above the balance of payments constraint growth (G N>GB), there will be a deficit on
the current account leading to capital inflows. This in turn will increase the warranted rate of growth.
However, as a country cannot indefinitely run a current account deficit, long run growth is thus
constrained by the balance of payments equilibrium growth rate. For most countries demand constraints
operate long before supply constraints take effect (Thirlwall, 2013).
If however the balance of payments constrained growth exceeds the natural rate of growth (GN<GB), the
economy will run a balance of payments surplus, leading to capital outflows and a reduction in the
warranted rate of growth. In this case, the actual rate of growth can exceed the natural rate of growth
without facing balance of payments problems. As the natural rate of growth is endogenous, it will increase
following the increase in output.

3.2.1.2 THE BALANCE OF PAYMENTS CONSTRAINED GROWTH MODEL (BOPCGM)

Thirlwall (1979) developed a post-Keynesian the Balance of Payment (BOP) constrained long run growth
model, also known as Thirlwall’s Law and dynamic foreign trade multiplier., which gives a central role
to demand using Harrod’s (1933) foreign trade multiplier. The model shows that BOP can act as a
constraint to economic growth by limiting growth in the level of demand to which supply can adjust; that
is, an increase in imports due to rising domestic income will result in BOP deficit. This may require a
fall in demand or depreciation of real exchange rate to ensure sustainability of external deficit. To grow
faster, countries should increase income elasticity of exports or reduce income elasticity of imports to
relax BOP constraints (Bairam, 1988; Thirlwall, 1997). The model posits that foreign exchange is a
constraint to economic growth in developing economies and suggests that policies targeting improving
productivity capacity with raising demand are likely to worsen unemployment conditions as BOP is
constrained. If, however, these economies raise the BOP constrained growth rate by increasing exports or
reducing income elasticity for imports, demand can be increased with little difficulty. Demand will create
its own supply by raising investment, absorbing underemployment and increasing productivity levels.
The model also has implications for supply side policies because differences in income elasticities for
exports and imports point towards the non-price characteristics of the goods (or structure of production).
The model suggests that supply side policies are needed in economies to change the structure of
production, and to reallocate resources between primary and secondary production, tradeable and non-
 26

tradeable goods, and characteristics of the goods (Thirlwall, 1979). Higher income elasticity for exports
and lower income elasticity for imports implies more favourable non-price characteristics. Hence, to
improve non-price characteristics of exports, supply side policies should focus on R&D (McCombie,
1993). The model also has implications for relative price as a mechanism for efficient BOP adjustment
(Thirlwall, 2011). If actual and BOP constrained growth rate are closely related, then this relative price is
not an efficient adjustment mechanism. In this regard, McCombie’s (1993) analysis suggests that
exchange rate adjustments and imports controls through tariffs and quotas are likely to be self-defeating
and less effective.
Thirlwall (2011) notes that one of the major weaknesses of the foreign trade multiplier model is that it
does not consider savings, investment and government spending, or taxation. As such, the model is
unrealistic. However, the conclusions of the model do not change if these weaknesses are addressed.
García-Molina and Ruíz-Tavera (2009) noted that both the BOPG and dual gap models consider the
external sector as a constraint to growth. Similarly, Thirlwall (2011) noted that both Prebisch and
Chenery viewed foreign exchange shortages associated with BOP deficit as a dominant constraint to
growth. The current survey of the theoretical literature suggests that the BOP constrained growth
model has been one of the major models linking current account (BOP) deficit and economic growth.
It is useful to note that the ‘45-degree rule’ proposed by Krugman—drawing upon new trade theory,
monopolistic competition and the importance of increasing returns—suggests that faster economic
growth leads to specialization and production of new goods. Hence, economic growth rate determines
trade elasticities (see, Krugman, 1989; McCombie, 2012). In contrast, the BOPG model argues that trade
elasticities determine growth. Further,Thirlwall (1991) claims that Krugman’s 45-degree rule is not a
new discovery, but is simply a restatement of the BOPG model.
The BOPG model faced several criticisms in the early 1980s, mostly relating to its assumptions and
empirical tests. For instance, McCombie (1981) claims that the model is based on circular reasoning and
that the law only captures an identity. McGregor and Swales (1985) argue that the model relies on
elasticities based on exports and imports equations that are too aggregative and ignore non-price
competitiveness as determinants; and that the model assumes that relative prices remain constant.
McGregor and Swales (1985) criticized the model for ignoring persistent capital inflows. This
shortcoming is problematic for developing economies that suffer from current account deficit for
prolonged periods where these deficits are financed by foreign direct investment and unilateral transfers.
The BOPG model was criticized as having internal inconsistency as its abstracts from the supply side of
the economy (see , Palley 2003). The long-run growth of an economy may not be constrained by current
account only but also by the requirement that output growth rate equals potential output growth rate.
Futhermore, the Thirlwall model does not allow for internal imbalances as a possible source of constraint.
In light of the European debt crisis, Soukiazis, et.al. (2012) extended the Thirlwall model to one in which
growth is constrained by external and internal imbalances. This model therefore enables an examination
 27

of the effect of budget deficit and external deficit on economic growth within a single theoretical
framework

The BOPG model also assumes that foreign and domestic goods are imperfect substitutes (Razmi, 2011).
Further, it does not distinguish between exportable and exports, importable and imports, and tradeable
and non-tradeable.
A series of studies by McCombie and Thirlwall (1997), Moreno-Brid (1998) , Barbosa-Filho (2001) and
Soukiazis, et.al (2012) have extended the model to consider this issue. Hence, the original model
described above was extended to include capital flows and the terms of trade (Thirlwall and Hussain,
1982). The extension is particularly relevant for developing countries, where capital flows, changes in
the terms of trade and the real exchange rate have been very important. The extended model with capital
flows and the terms of trade was first empirically tested by Thirlwall and Hussain (1982) for 20
developing countries covering the 1951 to 1969 period. Just three sub-Saharan African countries were
included in the study: Kenya, Sudan and Democratic Republic of Congo.

One of the implications of Thirlwall’s (1979) model is that the structure of production and exports
determines the income elasticity of demand for exports which therefore determines the rate of growth of
one country relative to another.

3.2.1.3 Structuralist Models

The Post-Keynesian tradition of growth models was inspired by Harrod, Domar and Kaldor, based on the
importance of the investment and foreign trade multipliers as major indicators of long-run economic
growth. Keynesian models along Harrodian and Kaldorian lines, particularly Thirlwall's balance of pay-
ments constrained growth model, build a link between economic growth and trade via demand-pull aspects
of exports. In Thirlwall’s law, trade plays a crucial role in constraining economic growth due to balance
of payments problems.
The Structuralist tradition also bases its analysis on demand-oriented explanations in investigating the
relationship between economic growth and external balance. Structuralists focus on the importance of
current account deficits and financial aspects in the capital account. This tradition looks at constraints that
can arise on growth from a savings-investment gap, from a foreign exchange gap, and then later from the
fiscal constraint, whereas Thirlwall's balance of payments constrained growth model looks at demand as a
source of growth and operates through multipliers to get from demand to growth. Prebisch (1959) first
questioned the doctrine of mutual profitability of free trade between developed and developing countries.
Foreign exchange rate is a dominant constraint on output, if underutilization of resources offsets real
resources gains from specialization. Less developed economies that specialize in diminishing return
activities with a low-income elasticity of demand, such as land-based primary products, are likely to be on
the losing end, whereas developed economies specializing in increasing return activities with high- income
 28

elasticity of demand, such as processed manufactured goods. According to the center–periphery model, if
foreign exchange is a dominant constraint on growth, then it is likely that real resources gain from
specialization will be offset by underutilization of factor inputs (Thirlwall, 2011).

The dual gap model developed by Chenery and Bruno (1962) is based on the basic idea that growth can be
constrained by either domestic savings or foreign exchange. For most developing economies, foreign
exchange is likely to be the dominant constraint during intermediate stages of economic development.
Therefore, economic growth is likely to be constrained by BOP.
The development of the two-gap model was thus an important contribution to the literature of
economic development. The two-gap model has been widely used by planners in developing
countries and donor agencies. It deals with the interaction between the savings constraint and
the foreign-exchange constraint in the determination of economic growth in an economy. The
savings constraint refers to the situation when the growth of an economy is limited by the
availability of domestic savings for investment. A savings gap appears when the domestic
saving rate is below the level necessary to permit the investment required to achieve the target
rate of growth (while imports are adequate). In this situation, aid covers the savings gap and
permits the economy to achieve some exogenous target growth rate. Alternatively, the foreign-
exchange constraint refers to the growth of an economy being limited by the availability of
foreign-exchange for importing (capital) goods. In this case, aid breaks the import bottleneck
and permits the economy to reach the target growth rate. The key assumption here is that the
country is unable to transform its domestic savings into imports from abroad. The central idea of
the two-gap analysis is that foreign aid can serve as a means of breaking the bottlenecks, thereby
permitting fuller utilization of all resources and a continuation of development in an economy.
Thus, the distinctive feature of the two-gap model is that foreign capital inflows play a
dual role in adding to both investment and foreign exchange resources. In addition, the
multiple requirements for external capital to provide additional savings and also to finance
required intermediate and investment imports were highlighted in the two-gap models proposed
by various economists.

The two-gap models that were prominent during the 1960s and the 1970s disappeared from the
empirical literature for some time. More recently, however, the two-gap models have been
extended into three-gap models -- adding a fiscal constraint to the traditional foreign-ex-
change constraint and savings constraint as a third gap limiting the growth prospects of
highly indebted developing economies. A third gap takes into account the fiscal limitations on
policy choice that have become crucial in many developing economies. In such cases, the fiscal
constraint is intended to reflect potential limitations on the availability of resources to
 29

finance the public investment that may be required to support a given level of output. Hence,
some authors consider the government budget constraint to be the most pressing medium-
term growth limitation, particularly when a developing country suffers from an external financial
shock. Tinbergen (1956), Theil (1958), and Chenery and Bruno (1962) represent a tradition that
policy models should contain variables reflecting the economic goals of a society (e.g. maxi-
mum income, output growth, and full employment) and the main instruments of government
policy (e.g. tax policy, foreign trade policy, investment allocation, exchange control, and foreign
borrowing). They also emphasized that a model should be seen as a specification of the important
relations connecting these goals and instruments. A three-gap model meets these criteria be-
cause it explicitly incorporates output growth and connects the goals and instruments in a mean-
ingful quantitative framework

This section attempts to explain how the three-gaps (the savings gap, the fiscal gap, and the
foreign exchange gap) of an economy can constrained/influence the economic growth of the
country. In general, the saving gap refers to the balance between gross investment and national
savings, and the foreign exchange gap refers to the current account balance of the balance of
payments of an economy. The fiscal gap is intended to reflect the balance between public reve-
nues and public expenditures. This section uses the social accounting framework to derive a
precise picture of the three-gaps and their balances through financial transactions in an econ-
omy. The financial transfers between various sectors can be traced through the flow-of- funds
account in a social accounting matrix (SAM). The flow-of-funds account shows how the savings
of a sector are allocated to investment in that sector and the surplus is transferred to the other
sectors which are in deficit. The main advantage of the flow-of-funds account is that it
shows the complete picture of capital transactions among various sectors. A number of
studies have used the flow-of-funds framework to analyze the financial problems of devel-
oping countries [for example, Jansen (1989) for Thailand, and Sarmad and Mahmood (1994)
for Pakistan].
The classification and disaggregation of accounts in a social accounting matrix (SAM) can take
various forms, depending on how the constituent accounts are defined and depending on one's
analytical interests and specific policy concerns. This section uses the SAM as a reference
table to derive the three-gaps, and the flow-of-funds account of an economy. These gaps
and the flow-of-funds account can be derived by integrating the national accounts, the bal-
ance of payments statistics, and the public finance statistics of an economy in a SAM frame-
work. The analytical framework developed in this section will later serve three main purposes.
First, it will be used to derive consistent time-series data on the flow-of-funds and the three-
gaps in SSA's economy. Second, it will help to describe the three-gaps and their balances
 30

through financial transactions among various sectors. Finally, the consistent time-series data
based on the SAM framework will assist in operationalizing the three-gap model. Since we will
be developing and estimating a three-gap model for SSA's economy, we will describe here
the salient features of three-gap models developed by various authors [namely Bacha (1990),
Taylor (1990b, 1990c, 1993), and Solimano (1990)].

(i) Bacha's three-gap model

Bacha (1990) developed a simplified theoretical framework of a three-gap model for an open
economy. This model paid particular attention to the impact of foreign transfers on potential
output growth and on the rate of inflation of a debtor country. A theoretical discussion is also
provided on the possible role of external conditionalities designed to maximize the
stabilization and other desirable impacts of debt reduction measures.

Bacha assumes that private investment (Ip) depends positively on public in-
vestment (Ig); the private sector's capital surplus (SSp) is assumed to be a
function of the rate of inflation (p) and the propensity to hoard (h). It is
expected that the private capital surplus first increases with inflation but even-
tually decreases with it. The propensity to hoard is assumed to be nega-
tively related to private surplus savings. Bacha also assumes that no mar-
ket exists for government bonds, which leaves money expansion as the
only alternative for domestic financing of the public sector deficit.

Thus, the three-gap model developed by Bacha provides a conceptual framework of inter-
actions among the three-gaps in an open economy, which emphasizes the scope for max-
imization of investment (as a proxy for the output growth rate) in a fixed-price one-period
growth model subject to a number of equality and inequality constraints. Within the model's
context, Bacha claims that the fiscal constraint tends to be the relevant medium-term growth
limitation, when the developing economy suffers an external financial shock.
(ii) Taylor's three-gap model
Taylor (1990b, 1990c, 1993) developed a three-gap model to analyze the effectiveness of
various economic policies on output growth in seventeen developing economies.' Adopting
a model similar to that of Bacha, Taylor developed a three-gap model that can also be
presented in a social accounting framework. As compared to Bacha's model, Taylor’s
formulation provides more details with the flow-of-funds. this shows that aggregate for-
eign capital inflows are divided into two components: foreign capital inflows to the private sector
(FP) and foreign capital inflows to the public sector (F). Taylor also incorporates private
 31

capital flight (KF) in his model, while Bacha assumes that there is no private capital flight to
the rest of world. Mathematical expressions for the row and column totals.
(iii) Solimano's three-gap model

Solimano (1990) developed a simple three-gap model framed in a disequilibrium setting. This
model is calibrated with parameters for the Chilean economy. Solimano examined the effects of
various macroeconomic policies (e.g. an increase in public spending, a reduction of interest
payments on external debt, and a reduction in the mark up through a cut in indirect taxes) on
the rate of GDP growth, the rate of capacity utilization, the real exchange rate, real wages, and
the rate of inflation. Like the other three-gap models (mentioned above), this model can also
be presented through a social accounting framework.

Solimano estimated the growth rate of potential output under three constraints and calibrated
the above model for the Chilean economy. The model is used to examine the effects of
various macroeconomic policies. The main results of the model are as follows. (i) An
increase in government spending would slow down the rate of growth in GDP, appreciate
the real exchange rate, and raise real wages in a capacity constrained-growth regime.
(ii) A reduction in interest payments abroad in a capacity constrained situation would
accelerate the rate of GDP growth, reduce the real exchange rate, and increase real wages. (iii)
Finally, a cut in the mark-up rate increases external competitiveness and real wages sim-
ultaneously, allowing an increase in the rate of capacity utilization and an acceleration in the
growth of potential GDP.

3.2.2. REVIEW OF THE EMPIRICAL LITERATURE

There has been little theoretical work linking current account deficit and growth in the mainstream growth
literature, particularly with respect to the neoclassical growth and endogenous growth theories. However,
following the theoretical work on the balance of payment(BOP) constrained growth model by Thirlwall
(1979), an increasing number of studies have used the model to study the relationship between current
account (BOP) deficit on economic growth in both developed and developing economies. This section
discusses the findings of major and recent empirical studies to identify existing gaps in literature. The
empirical evidence on the balance of payment constrained growth(BOPG) model is inconclusive. For
instance, empirical studies that have focused on developed economies such as the US and Canada have
found that BOP is a constraint to economic growth (Atesoglu, 1993a, 1993b; Hieke, 1997). Likewise,
several studies on emerging economies, in particular Brazil, also report that BOP is a constraint to
economic growth (Bertola, et.al, 2002; Britto & McCombie, 2009, 2013). However, empirical support
for the BOPG model in the case of Spain is debatable. For instance, although early studies by Alonso
(1999) and Leon- Ledesma (1999) provide evidence in support of the BOPG model, recent studies
focusing on a larger sample period by de la Escosura (2010) and Bajo-Rubio (2012) suggest that BOP is
 32

not a constraint to economic growth. These two studies highlight that results based on small sample sizes
can be misleading. A study by Bertola et al. (2002) used a large sample of observations from 1890 to1973
but did not examine stability of the estimated relationship. However, it is unlikely that this parameter
will remain constant over such a long period. Perraton (2003) focused on a large sample of developing
economies and found weak support for the BOPG model. Ghani (2006) also investigated whether BOP
is a constraint to economic growth in 90 developed and developing economies, and on the basis of
individual country analysis found that BOP is a constraint in only 45% of the economies. A recent study
focusing on MENA economies by Khasawneh, Magableh, Khrisat, & Massadeh (2012) also found mixed
support for the idea that BOP constrains economic growth.

Soukiazis et al. (2012) extended the BOPG model such that growth is constrained by external and internal
imbalance and non-neutral relative prices. The extended model was supported for Portugal over the period
1986–2010 using 3SLS (see, for e.g., Soukiazis, Cerqueira & Antunes, 2013). In another study, Soukiazis,
Cerqueira and Antunes (2014) validated the extended model for Italy using similar methodology over the
period 1980– 2010. Lanzafame (2014) examined the validity of BOPG model in 22 OECD economies
using the panel estimation method. The results significantly validated the BOPG model. Alencar and
Strachman (2014) found that economic growth in Brazil was constrained by BOP over the period 1951–
2008. However, a recent study by Podkaminer (2017) investigating the validity of the BOPG model for
59 countries over the period 1960–2012 found that the model does not hold in the majority of the
countries.

Vidal (2016) examined differential growth rates in Cuba (1980–2013) and Vietnam (1990–2012) using
the extended BOPG model of Thirlwall and Hussain (1982), confirming the validity of BOP constrained
growth for both economies. In a very recent study, Bagnai (2016) adopted a multi-country version of the
BOPG model to investigate the decline in Italy’s long-run growth, which was found to be due to
tightening of the BOP constraint. Bagnai, et.al. (2016) employed the BOPG model to examine the rise in
SSA growth rates. They examined the contribution of south–south trade by resorting to the panel
cointegration method using data from 1990 to 2008 for 20 low- and middle-income SSA countries. The
analysis revealed that the BOP constraint has been relaxed due to real growth, market share and terms-of-
trade effects. Romero and McCombie (2016) estimated trade equations for 14 European countries over
the period 1984–2007 using five different technological classifications. The results from their panel data
analysis indicated that the multi-sectoral version of Thirlwall’s Law is valid.
The extant literature is plagued with methodological flaws that possibly explain the mixed evidence on the
BOPG model. First, several studies (see, for e.g., Bairam, 1988; Leon- Ledesma, 1999) applied a two-
stage least square technique for estimation purposes. However, they do not report diagnostic test results
nor justify their choice of instruments. In contrast, a recent study on Portugal by Soukiazis and Antunes
 33

(2011) did report diagnostic test results from the use of two-stage least squares. Thus, it is not clear to what
extent results are robust to choose of instrument. Hence, the extant literature on the BOPG model is weak
from a methodological perspective as it has paid little attention to diagnostic analysis, used incorrect
critical values and barely utilized panel estimation methods.
Economists began about 40 years ago to map the linkages between macroeconomic
constraints and the rate of growth for developing countries. Gradually, their analysis has
become more sophisticated. The two-gap models that were prominent during the 1960s and the
1970s disappeared from the empirical literature for some time.
A formal debate on the two-gap model started with Chenery and Bruno (1962), who extended the
Harrod (1939) model, which recognized two binding constraints on growth (namely the supply
of labor and the supply of capital) by introducing a third, foreign exchange, constraint on eco-
nomic growth. Harrod was not unaware of this potential third constraint, but in his writing
on the subject he does not take the matter seriously. The policy model developed by Chenery
and Bruno (using data for the planning period 1960-65 in the case of Israel) yielded the reduced
form restrictions for each constraint of Full-employment equilibrium, Savings-investment equilib-
rium and Balance of payments equilibrium. Based on the reduced-form model, they found that
the balance of payments -- foreign exchange -- proved to be the binding constraint on
economic growth of Israel within the framework of a planning formulation that sought to
maximize growth. In addition, they argued that the developing countries for which foreign
exchange is the binding limit to growth normally show the highest productivity of external
aid.

Chenery and Strout (1966) applied a similar programming model to that developed for Is-
rael, using data for the period 1957-62 in the case of fifty developing countries (LDCs). They
found their solution of the model suggested two phases. In the first phase, corre-
sponding to a lower development level, these countries faced a savings constraint, which sub-
sequently became a foreign exchange constraint as development progressed. They also cal-
culated aid requirements for fifty LDCs for a given target rate of growth over the period
1962-75. Note that the same study also found that the productivity of foreign aid was much
higher when foreign exchange was the limiting factor on Pakistan's economic growth.
Adelman and Chenery (1966) undertook an econometric investigation of the effects of for-
eign aid on Greece's economic growth using time-series data for the period 1950-61. Based
on their estimated functions, Adelman and Chenery found that, for the period up to 1957, savings
were the binding constraint on growth, while the import-export gap became increasingly
dominant thereafter. The import-limited growth function shows that when savings and invest-
ment are not a limitation on growth.
 34

Mckinnon (1964) provided a general conceptual framework showing how both trade possibilities
and foreign capital transfers affect the growth process in developing economies. His study clarified the
basic economic principles underlying the foreign exchange and savings constraints on target growth
rates. Following Chenery and Bruno (1962), Mackinnon argued that when the foreign ex-
change constraint holds, foreign aid transfers will always have a proportionately
greater effect on the feasible growth rate than if the savings constraint holds.
Landau (1971) adopted a different method for identifying the binding constraint for eight-
een Latin American countries over the period 1950-66. He abandoned the programming
framework of previous authors and relied instead on ex-post saving functions and import
function. Hence it would be possible to infer from empirical estimates for individual
countries in different time periods which of the two modes was operative. On this basis,
Landau identified eight countries (Bolivia, Chile, Colombia, Dominican Republic, Guate-
mala, Nicaragua, Panama, and Uruguay) that were facing a binding foreign-exchange con-
straint, while a savings constraint appeared in the case of four countries (Brazil, Para-
guay, Peru, and Venezuela) over the period 1950-66. The remaining six countries (Costa Rica,
Honduras, Argentina, Ecuador, El Salvador, and Mexico) appeared to have alternated be-
tween the two situations for the same period.

Weisskopf (1972a) examined binding constraints on growth for thirty seven developing
countries using time-series data over the period 1953-68. Like Landau's model, Weisskopf's
two-gap model can also be summarized in a SAM framework. Based on ordinary least squares
estimates of equations, Weisskopf found that twenty-three developing countries had been
subject to a savings constraint on growth. Eight countries were dominated by a foreign
exchange constraint and the remaining six countries were characterized by a hybrid savings
and foreign-exchange constraint. Similarly, Levy (1984) investigated the dominant binding
constraint in the case of the Egyptian economy, using the same test suggested by Weisskopf
(1972a). He found that saving was the binding constraint for Egyptian economic growth dur-
ing the period 1960-79. Blomqvist (1976) criticized Weisskopf's econometric test and argued
that it provides no clear-cut criterion for finding binding constraints in developing countries.
He suggested an alternative method of computing F-ratios from the estimated functions as
specified by Weisskopf (1972a). In order to make the results comparable with those
obtained by Weisskopf, Blomqvist used the same data set covering the same thirty-three
developing countries. He found that twenty-four countries could be classified as savings
constrained economies, while two countries were facing a dominant foreign-exchange con-
straint and seven countries remained unclassified. Thus, the results obtained by Blomqvist
generally supported Weisskopf's main conclusion that most developing countries were facing
a savings constraint during the period under consideration.
 35

Bacha (1984) developed a more policy-oriented two-gap model for pedagogic purposes.
Bacha then argued that for given values of all right-hand variables in both the equations,
growth should be considered to be savings constrained if go < gf, and foreign exchange con-
strained if gf < go. Finally, note that van Wijnbergen (1986) developed a simple open econ-
omy macro model and compared its implications with those of the two-gap model. He sug-
gested that a binding-trade gap should be seen as an excess supply of non-traded goods and
an excess demand for traded goods. Similarly, a binding-savings gap should be interpreted as an
excess demand for non-traded goods and an excess supply of traded goods.
Chenery and Bruno (1962) analyzed the main development alternatives in Israel and showed the theoret-
ical interrelationships among the main instruments of development policy. They were aroused by some
alleged deficiencies in growth models as a basis for development policy. Notable among these alleged
deficiencies included the focusing on the investment- savings relationship and the possibilities of substi-
tution between capital and labour, as well as the exclusion of the changing structure of demand, the role
of foreign trade and the allocation of resources. Therefore, he argued that, formal growth models had failed
to c1arify the theoretical relationships among the several instruments of development policy, which ordi-
narily should be one of its major functions.
For instance, other studies, notably, El-Shibly and Thirwall (1981) , and Mwega et al (1994), have reported
that Investment-Savings and Foreign Exchange constraints could be distinct macroeconomic imbalances
that could limit economic growth. By implication, if domestic savings and foreign resources are easily
substitutable, and the incremental-capital output ratio and incremental output-import ratio are not fixed,
resource gap analysis will no doubt loose quite a lot of its force, (El-Shibly and Tliirwall, 1981).

Mckinnon (1964) constructed a growth model of the Harrod-Domar type, which incorporated in a crude
way the effects of international trade on the growth of developing countries. He was aroused by earlier
studies by Chenery and Bruno (1962), who had argued a dual role for the state of foreign exchange balance
as it affected economic growth. The first role, which in the views of the classical economists was that
foreign aid or investment, only has the effect of supplementing domestic savings in the receiving country.
The second role clustered the modern view argued that many goods have strategic importance in efficient
industrial growth but cannot be produced domestically in the early stages of industrial development, or
after a war. Thus, foreign aid or foreign private investment could have a large favorable impact on the
growth rate of the economy where such a constraint was binding even though the foreign transfers were a
small fraction of available domestic saving.
To show the impact of foreign trade on aggregate productive capacity of the economy, Mckinnon specified
that foreign goods entered as inputs into domestic production function. In the specification he postulated
that total investment at time "t" would equal the sum of domestic savings plus net foreign transfers. He
reported that if net foreign transfers were zero, and domestic savings could be freely exported to obtain
 36

foreign capital goods, domestic savings would be the only constraint to economic growth. Also, if exports
were not sufficiently great to finance needed imports of foreign capital goods, the warranted growth rate
would not be achieved, thus, suggesting the existence of a foreign exchange constraint.
He also argued that foreign capital transfers could have effects on economic growth rate, but in a specified
manner. For example, it could be carried out though increasing import capabilities where foreign exchange
constraint existed, that is, increasing purchases of foreign capital goods. Similarly, foreign capital transfers
could have the effect of increasing domestic savings rate where saving constraints existed, that is, by
increasing the effective level of total investment in both domestic and foreign capital goods. However, in
a situation where foreign exchange constraint existed, foreign aid transfers would always have a propor-
tionately greater effect on economic growth rate than if domestic saving constraints existed. In other
words, it would be possible to get more punch from a foreign aid dollar when there existed a pool of
potential domestic savings, or the marginal propensity to save and the marginal propensity to export were
higher.
He further argued that in order for any economic growth rate to be sustainable, it was a necessary condition
that exports covered immediate import needs on current account. If the contrary were the case, there would
be insufficient exports generated at full capacity output to cover current foreign materials requirement and
the purchase of foreign capital goods would be rather impossible. He contended that in a case where cur-
rent export capability at full capacity output was below current account materials needs, self-sustainable
economic growth would be out rightly impossible. To ameliorate such situation, such that full capacity
growth would be sustainable, he suggested sufficiently large foreign transfers. However, that would only
be a necessary condition, as foreign exchange availability would only permit operation at f\ill capacity
and to permit capital accumulation. The implication from the above is that a foreign aid transfer needs do
not have the effect of reducing a foreign exchange constraint on the growth rate. This holds even if the
savings constraint was not binding, provided there were deficiencies in current account balance needs and
the net foreign transfer was not very large.
McKinnon holding a similar view with Chenery (1961) supported the inefficacy of the traditional static
theory of comparative cost advantage in dealing with the problems of economic growth and development.
In McKinnon’s view, static models which used comparisons of long run equilibrium positions of prices
and commodity flows to compare positions of free trade and autarky were inherently ill suited to the
analysis of continuous change. He did not oppose the use of the static Ricardian model to illustrate the
static notion of comparative cost advantage for trade in final consumer goods. However, he argued that
the model was bedeviled with two defects. The two defects included, (l) the capacity of a growing econ-
omy to move from producing simple to more complex industrial goods had been ignored; and (2) two
commodity worlds spawn excessive preoccupation with the terms of trade. In other words, export expan-
sion and economic growth were always associated with sharp declines in the terms of trade. However,
in the modem, economic environment, growth in output and exports was largely associated with a country
 37

moving up the hierarchy of complexity for industrial goods, thereby actually increasing the demand for
its own product.
He introduced substitution possibilities with a view to making the model less rigid and somewhat more
responsive to price considerations and different resource endowments. In other words, the assumptions of
exogenous determination of international prices, fixed needs for imports of intermediate materials and
capital goods, limited capacity to export depended on the development level of the economy. This was
carried out in two ways. First, the use of import substitution as an alternative to export expansion, and
second, the relaxation of the assumption of fixed input needs in the production function and/or the intro-
duction of more factors of production. The concept import substitution refers to the domestic production
of current materials or capital goods previously imported.
On the first way above, McKinnon, argued that the prospective demand for primary products by the heav-
ily industrialized countries of the world was less favorable now than it was in the last century for the
products of the Less Developed Countries. He justified the argument by the development of synthetics,
the fall in raw materials required for a unit of manufacturing output and the position of resource-rich North
America as the most heavily industrialized area. By implication, Less Developed Countries would have
to focus more on the development of light manufactures for direct export substitution in order to mobilize
additional foreign exchange for the process of capital accumulation. However, the export and import sub-
stitution possibilities would depend on the level of development and production capabilities of an econ-
omy.
Finally, McKinnon argued that irrespective of the level of the propensities to save and export, the existence
of potential domestic savings (capacity to Undertake investment) was rather a necessary condition in ob-
taining a high pay-off on a given foreign-aid transfers. However, the institutional organization required to
ensure that domestic savings- were adequate could take several forms. For example, the presence of an
entrepreneurial class, which appropriates a significant proportion of the final product, and with a high
marginal propensity to save was one solution. Another was the effective capital stock in an economy
consisting of human resources as much as physical plant and equipment.
The argument from McKinnon was that inadequate domestic savings, and foreign exchange, with the latter
exerting greater impact could effectively limit economic growth. Mobilizing sufficient savings for invest-
ment has not been an easy task in LDCs, especially in SSA. It should be noted that an economic agent's
decision to save depends on two principal factors, notably the level of income available to the economic
agent at any particular time and the attractiveness of savings option relative to consumption. The latter is
also referred to as the willingness to save or the marginal propensity to save. The relative attractiveness
of savings to consumption would depend on the rate of interest, the resource distribution and the institu-
tional structures available for resource mobilization. In SSA, the level of per capita income is ridiculously
low. There has been widespread income in- equality as the gap between the few rich and the many poor
have not only been widening, more people have been joining the poor class. The instability in the
 38

institutional structures available for resource mobilization as manifested in the high frequency of dis-
tressed banks has eroded the confidence of even the few rich who should be better disposed to savings.
Most of the few rich people now prefer to save in foreign countries. The sum total of the above is that
domestic savings are very low.
Chenery and Eclcstein (1970) were moved by two factors. The first was the earlier study by Prebisch
(1950), who had argued, that continued dependence on primary exports would place limits on further
economic growth in Latin America. However, while the limits, notably, structural imbalance appeared to
have been real, especially for exporters of tropical agricultural products, the alternative then advocated by
Presbisch (1950) and which was pursued throughout Latin America - especially import substitution
through industrialization - had run into increasing difficulties.
The above difficulties manifested in the form of unimpressive industrial growth, despite protection and
other preferential policies. A major problem with the policy of concentrating too heavily on one factor
limiting economic growth had been the mere replacement of one set of constraints with another, Thus, he
argued underscored the need to establish a more comprehensive framework for diagnosing development
problems and prescribing a set of remedies.
It was against the above backdrop that Chenery and Eckstein developed an aggregate growth model that
brought out the relationship between the internal and external factors in economic development. They
used the model to address four issues. One, to appraise the development performance in Latin America.
Two, to indicate in quantitative terms some of the requirements of economic growth in Latin America.
Three, to provide a framework for the discussion of alternatives facing the United States of America and
Latin American countries. Four, some modifications were attempted by introducing some’ additional var-
iables into the equations of the Chenery - Strout model (Chenery and Strout, 1966), in order to bring out
the interdependence of the two identified factors limiting economic growth: foreign exchange and capital.
On the development performance in Latin America, Chenery and Eclstein estimated capital requirements,
saving, investment, and trade performance of sixteen Latin American countries for the year 1950 -65.
On capital requirements, they argued that if higher growth rates were based on greater investment, the part
of gross investment used to replace old equipment and to construct social overhead facilities usually rep-
resents a smaller share of the total. The regression results obtained conformed to the above prediction as
all the sixteen countries except one (El Salvador) showed lower incremental ratios during periods of higher
economic growth.
On savings requirement, they supported Landau (1966) and Vanek (1967), who had earlier argued that
whenever the trade gap was dominant, foreign exchange was the scarce factor of production. Since invest-
ment was highly dependent upon imports, domestic investment opportunities would vary directly with the
availability of foreign exchange, and domestic savings could be expected to vary in response to those
opportunities. Thus, suggesting that export earnings would have positive effect on saving. They tested the
 39

above a priori expectation and reported that, impact of additional foreign capital on saving was found to
be negative. Also, in all cases but two, the impacts of the export share were positive.
On exports growth, they found a rather unimpressive performance as all the countries under study experi-
enced declines in exports. Also, both exports and external reserves were found to be declining share of
Gross National Product. Nevertheless, virtually all the countries showed higher propensities to import
from increase in investment expenditure, On the whole, they found that investment and imports were the
two limiting factors affecting economic growth in Latin American countries. However, we know that in
Less Developed Countries (LDCs), investment is highly dependent on level of imports, as most of the
investment goods are not produced locally.
To import the investment goods requires foreign exchange, which comes from export receipts. By impli-
cation a declining export receipts will reduce imports (widens the trade gap) and consequently investment,
while declining investment will have negative effect on aggregate output.
On the requirement for sustainable economic growth Chenery and Eckstein projected a growth rate of 8.0
per cent per annum. To bring about the necessary adjustment between the Exports and Imports (trade gap),
they suggested currency devaluation along with more pragmatic decisions on investment allocation. They
argued that if a country was faced with a trade gap in excess of its savings gap it should direct a part of
current’ investment towards reducing following year's trade gap. The above argument would hold, assum-
ing that investment in such a situation would be enough to reduce the following year's trade gap by a
specified share of the current excess trade gap. They postulated further that such investment could exceed
30.0 per cent of total gross domestic investment and that it entailed a capital - output ratio that was 20.0
per cent higher than that applied to ordinary investment.
To improve the savings and investment performance they called for increases in marginal saving rates in
relation to the highest observed performance of other developing economies. They defined high saving as
a movement to a saving rate halfway between the historical rate of 35.0 per cent. However, the ability to
divert a substantially larger share of increased income to savings was considered possible only if annual
increases in income were larger. At larger level of income, the marginal propensity to consume reduces,
thus, making allowance for higher marginal propensity to save.
On the third issue of alternative requirements for sustainable economic growth, they argued that the in-
crease in investment needed to sustain economic growth would have to be financed through a net inflow
of external capital. They however, projected that such inflow needs only be temporary. By implication, if
an increased inflow of capital stimulates improved domestic developmental performance, then long-run
external resource costs of sustainable growth could be less comparatively than those of continued histori-
cal growth rates with past performance. In conclusion, they recommended increased savings, exports and
aid to achieve higher growth rates for the Latin American countries.
In SSA, per capita income remains low with high incidence of poverty and high propensity to consume
almost equal to one. Thus, the propensity to save is consequently low, or almost equal to zero and in some
 40

cases negative. By implication, the above might have given rise to problems of resource gaps in SSA
requiring concerted policy measures to address them. The formulation and implementation of such policy
cannot be based on intuitive judgements but on valid quantitative estimate of the effects of the resource
gaps.
Adelman and Cheneiy (1966) investigated the effects of foreign resources on the recipient's economic
development. In the analysis of the results, Adelman and Chenery distinguished between two constraints
to economic growth. These included (i) savings-limited growth, and (2) import- limited growth. They
argued that the effect of variation in the capital inflow would depend on which of the two constraints set
the limit to growth in given circumstances. To estimate savings-limited growth, they assumed that the
investment determined by the model was essential to sustain the level of Gross National Product, while to
estimate import-limited growth, they assumed that the amount of imports determined by the system were
required by the structure of demand at any given level of GNP.
On import-limited growth, that is, when import requirements were the limiting factor, they reported that
when savings and investment were not a limitation on growth, the increase in productivity of an additional
unit of external assistance was more than proportionate relative to the amount of foreign assistance given.
However, imports were in turn limited to the sum of export earnings and capital inflow. On savings-limited
growth, that is, when available savings were not sufficient for the planned investment, they argued that it
was necessary to reduce foreign capital inflow in order to increase GNP. This was justified by the negative
multiplier effect that dependence on foreign capital inflow had on the Greek economy.
For the second alternative, they assumed that the gain from the additional capital inflow that was forth-
coming could be measured by the difference in GNP that. resulted from raising the growth rate from 2.0
to 6.0 per cent. They found that import gap became increasingly important as the growth rate was raised
and, on the aggregate, it was the larger of the two for growth rates above 6.0 per cent. This result was
consistent with that of Chenery and Strout (1966), who had earlier argued that the productivity of aid
would increase with the time period considered when the savings limitation was dominant, but it would
decline when the import limit was more binding.

Rattso (1994) analyzed import compression as a policy response to balance of payment crisis arising from
foreign exchange situation since the 1980s, for most Sub- Saharan Africa countries, What stimulated
Rattso was the policy conflict between efforts to attain balance of payments equilibrium and efforts to
promote sustainable economic growth. Sub-Saharan Africa countries due to their underdevelopment na-
ture, had being found to be facing a highly inelastic imports (see also Moran, 1989). To finance the im-
ports, require balance of payment surpluses, which unfortunately has been absent.
The consequence of unavailability of balance of payment surpluses had manifested in the form of eco-
nomic stagnation. This had led to calls for structural adjustments in the SSA region. A much more dis-
turbing situation had been the controversial sources of the economic stagnation. Rattso concentrated his
 41

analysis on three linkages between the traded and protected sectors in addition to the conventional spillo-
vers through income generation and consumption demand. The sectors were connected through the foreign
exchange constraints, the wage formation, and the investment determination. The traded sector comprises
agricultural exports, raw materials and industrial exports. The protected sector comprises capital intensive
and import intensive industries, delivering to investment projects and therefore of importance for eco-
nomic growth process. Also, capacity utilization of the two sectors was dependent on access to imported
intermediate inputs. He postulated that import compression could have implication on the link between
the traded sector exports and capacity utilization of the protected sector. Also, import capacity could be
determined by export revenues and foreign savings, and used for imports of intermediate and investment
goods.
However, he found that whenever government as a deliberate policy decided to compress imports, imports
of investment goods would be given priority, such that the intermediate imports to the protected sector
would be rationed according to the current account situation. He also argued that the emerging low capac-
ity utilization would motivate a shift and investment levels would fall, possibly because of rationing of
imports of investment goods. This argument was however consistent with the descriptions of Tanzania
given by Ndulu (1986), of Zimbabwe by, and Ndulu (1991) for the whole Sub-Saharan African region.
El-Shibly and Thirwall (198l) employed the dual-gap analysis with Sudan as a case study. They were
aroused by some other previous studies to investigate the dominant resources constraint limiting economic
growth in developing countries. The study also estimated future resource requirements to achieve a par-
ticular target rate of economic growth. Thus, El-Shibly and Thirwall (1981) aimed at carrying out a quan-
titative estimates of investment requirements in relation to forecasted domestic saving, and import require-
ments in relation to expected export earnings, under two different economic growth rate aspirations. The
first of the two alternative economic growth rate aspirations were 5.5 per cent per annum, which was the
historical average growth rate between 1960 and 1975. The second was 7.5 per cent per annum, which
was the target economic growth rale of the Sudanese six-year plan 1577/78 - 1982/83. They also attempted
to compare the estimates with projections made in the Sudanese six-year plan which in the authors' own
opinion under-estimated the short fall on both savings below investment requirements and export below
imports requirements diming the plan period. Thus, the study aimed at walking an independent assessment
of the consistency of the Sudanese plan.
In determining the dominant constraint assuming a growth rate of 5.5 per cent, two different estimates of
Export-Import gap were employed. The first was based on the use of estimated import coefficients of
investment and consumption. The second was based on the use of incremental- capital output ratio. The
use of import coefficients gave higher estimates of import requirements than the use of incremental output
import ratio. Thus, implying a lower implicit value of incremental output import ratio in the disaggregated
approach. The results suggested that domestic resource gap and foreign exchange gap were of roughly
similar magnitude. However domestic inflation would widen the Investment - Savings gap in absolute
 42

terms and a rise in the price of traded goods would widen the Export-Import gap, barring a sufficient
improvement in the terms of trade. However, deterioration in the terms of trade would widen the Export -
Import gap even more.
On the second assumption of 7.5 per cent economic growth rate, both Investment - Saving gap and Export-
Import gap widened absolutely at the higher growth rate. The two approaches used to forecast import
requirements gave roughly similar results, with the disaggregated approach tending to give a slightly lower
estimate. They argued further that both gaps would be widened in absolute terms by inflation and deteri-
oration in the terms of trade would widen the export import gap.
El-Shibly and Thirwall postulated that the domestic resource gap and trade gap could both be nearly 50
per cent higher than the levels estimated by the plan. However, in estimating the Investment- Saving gap
and the Export-Import gap, the authors treated the two gaps as being mutually exclusive. This was a major
methodological flaw of the study. Taylor (1994) has argued that, for gap models to be effective requires
incorporating the identified gaps in a single model. In fact, the taste of solving the problem has been a
major pre-occupation of most of the recent studies including this present study.
Oyejide and Raheem (1990) in an attempt to emphasize the importance of investment in promoting eco-
nomic growth carried out some projections based on a three - gap growth exercise for the period 1990-
1995. They reported that foreign exchange constraints would limit economic growth most in Nigeria,
especially between 1990-1993.
The three-gap framework used by Mwega et al (1994) extended the traditional two -gap model mainly
associated with Chenery’s model. It distinguished fiscal gap as another potentially important constraint to
economic growth apart from savings and foreign exchange constraints. The specification drew heavily
from the methodological framework by Taylor (1990). A major component of this methodology especially
limped the growth problem with shortages of intermediate imports with the adverse impact of this on
capacity utilization, investment and economic growth. They examined whether it was savings, budget or
foreign exchange gaps, which was the binding constraint on economic growth in Kenya and how these
gaps had evolved in the 1970s and 1980s. The rationale for this was that all the three gaps had increased
over time; thereby it was not obvious which one had been the most binding. Thus, they attempted to
analyze the extent to which the availability of foreign exchange (and the associated import compression)
had evolved to become the binding constraints to economic growth controlled for capacity utilization as
hypothesized by the Kenyan government’s 1986 session paper.
They reported that all the three gaps were binding, though at different levels. However, foreign exchange
was found to be the most binding constraint on the growth of output in Kenya. Thus, it was apparent that
impact of extremal shocks and adjustment policies pursued since the early 1970s was to increase the im-
portance of savings gap vis-à- vis fiscal gap and the importance of foreign exchange gap vis-à-vis saving
gap as constraints on potential growth in Kenya.
 43

The session paper further argued that foreign exchange gap had widened as the demand for imports to
promote growth, outstripped growth of export revenues and supply of foreign capital, thus limiting the
rate at which the economy could grow. The paper therefore advocated for a more aggressive export pro-
motion drives as well as efficient import-substitution in agriculture and industry.
They argued further that Kenyan economy could be in macro-balance provided the three gap equations
intersected at one point. This could be brought about in three ways. One, through a downward shift of the
fiscal gap equation. Two, through an upward shift of the savings gap equation. Three, through a leftward
shift of the foreign exchange gap equation. Thus, the size of the gap triangles (the area enclosed by the
three equations when drawn on the same diagram) could therefore be used to measure the degree of macro-
imbalances in a given period in the economy.
However, they found that policies and forecast outcomes of the 1989-93 Development Plan that intended
to close the gaps were not only inadequate but also inconsistent. This was because the plan envisaged
substantial reductions, in the country's dependency on foreign savings and external grants and increased
dependence on imported capital without a commensurate rise in exports and/or a reduction in debt burden
to take up the slack. The target for the reduction in budget deficit from 4.5 per cent to 3.3 per cent of Gross
Domestic Product was also ambitious so that the increase in macroeconomic imbalances if the plan's tar-
gets were achieved was under-estimated. In fact, they expressed the fear that the policies and forecasted
outcomes could exacerbate the macro-imbalance.
Kenya like Many SSA countries is a primary product exporting country, such that domestic. output is
mainly influenced by the domestic weather conditions. The main exports depend on the state of demand
in the world market. Thus, both demand for and supply of imports and exports are inelastic. It should be
noted that economic theories postulate that both demand for and supply of imports and exports should be
elastic. By this we mean that, if import prices are increasing a country should be able to respond to such
increase by doing without some imports, thereby conserving foreign exchange, Similarly, a country should
be able to respond to more demand for its export by supplying more, this is to earn more foreign exchange.
However, it is very doubtful whether Less Developed Countries, and indeed SSA, could mobilize more
foreign exchange through the above means to alleviate the problems of inadequate intermediate imports.
The Mwega et a1(1994) has two important flaws, among others. One in the estimation of the potential
output, they regressed GDP on time. This is inadequate. We know that lags of GDP for at least three years
are theoretically important as explanatory variables to potential output.
Taylor (1994) examined savings, foreign exchange, investment, and inflations restrictions on potential
output growth and capacity utilization. He assumed that actual output could fall below or equal to potential
output, such that the ratio of actual/potential output would equal to or less than unity. The incremental
capital output ratio which could range between 0.2 and 0.4 could be limited by available savings, foreign
exchange restriction and an economy's own investment demand function and thus could constraint poten-
tial output growth.
 44

On adverse external shock he argued that with reduced external inflows the rate of inflation could have to
rise to generate forced saving and a higher inflation tax to offset lower saving from abroad. In other words,
with a binding foreign exchange limit, demand driven equilibrium would be rather difficult to attain. Thus,
such real output would have to tend towards slower potential output growth, reduced output, and faster
demand inflation, which in turn could stimulate further price pressure from the side of costs. Holding a
similar view with Bacha (1990), he argued that an adequate policy response to an external shock would
be rather difficult to orchestrate. To ameliorate the adverse effects of external shock, Taylor recommended
fiscal restraint, increased public investment, higher exports, import quotas and controls expansionary pol-
icy as well as policy co-ordination.
However, he further argued that each of the above recommendation had its own merits and demerits. For
example, fiscal restraint could permit faster capacity growth and reduced inflation. Increasing public in-
vestment could speed up capacity growth and enhance investment crowding-in, but at the cost of higher
inflation and reduced current output. Higher exports could help release the foreign exchange limit, but to
raise sales abroad in the short run could be rather difficult especially for a raw material exporter. He
nevertheless conceded to the fact that an export push could be easier for semi-industrialized economies in
which domestic recession could create spare manufacturing capacity which could be diverted to foreign
markets, as adopted successfully in Turkey, Brazil, etc.
He also argued that import quotas and contracts could permit capacity utilizations or growth to rise, but
this should be in conjunction with other policies. Expansionary policy could stimulate growth and capacity
utilization but at the cost of spiraling inflation. Finally, he argued that policy co-ordination could be diffi-
cult, especially in the Less Developed Countries. His reason was that few countries could be agile enough
to deploy simultaneous fiscal restraint in current transactions, increased state capital formation, intelligent
manipulation of quota, and export incentives to offset all the ill effects of an external shock.
However, in SSA, the replacement of the hitherto fixed exchange rate with the floating exchange rate and
the ongoing guided deregulation has resulted in significant depreciation of the domestic currency. Simi-
larly, the fact that substantial inputs into SSA's production come from abroad heightens the suspicion that
inflation might have been imported. During inflation, it is possible to have increased savings arising from
the super normal profits of the private sector investor, provided that the consequent higher cost could be
passed on to the consumers in form of higher prices. In conclusion, resource gaps, as an economic problem
has been topical in recent times. The peculiarities of each country will no doubt call for more studies. In
fact, it is the need for more studies that justifies this study.

3.2.3. CONCLUSIONS ON THE THEORETICAL AND EMPIRICAL LITERATURE

In the post-Keynesian school, equilibrium growth is therefore achieved through an adjustment in the
warranted and natural rate of growth. This has consequences for the model, for it means that the full
employment ceiling, within limits, shifts in response to changes in the actual rate of growth. It therefore
 45

becomes illogical to think of growth in terms of the exogenously determined production frontier as done
in the neoclassical model. The production frontier instead shifts with each movement in the actual rate of
growth (Thirlwall, 2013).
Thirlwall (2001) extended the Harrod (1939) model to an open economy model and theoretically showed
that the balance of payments equilibrium growth rate is the long run constraint to growth. Model
performance greatly improved when a balance of payments constraint was added for in the long run, a
country cannot advance faster than the latter. The balance of payments constrained growth model has been
tested extensively for developed countries with the results generally giving support to it(McCombie and
Thirlwall (2004) ). It has also been tested for several developing countries particularly in Latin America
and Asia. There is very little empirical research regarding the relevance of the balance of payments
constraint for sub-Saharan Africa. The few studies that do exist on the region verify the theoretical
expectations of the model (Hussain, 1999; Nell, 2003; Perraton, 2003).
The survey of theoretical and empirical literature leads to the following conclusions. First, in contrast to
mainstream growth models, the three-gap model provides a very promising framework to investigate the
growth consequences of the triple deficits. Second, the empirical evidence on the three-gap model is far
from conclusive; hence further research attention is required. Fourth, the survey of the literature indicates
that there has been no investigation of the growth implications of the triple deficits in the SSA economies.
Hence, policymakers are ill-informed regarding the growth implications of the resource gaps (net
government saving, net private saving and net foreign saving). The current study addresses some of these
limitations, contributes to the empirical literature and provides new growth policy perspectives in the
SSA economies over the post-independence period.
 46

CHAPTER FOUR: THE RELATIONSHIP BETWEEN

CURRENT ACCOUNT DEFICIT FISCAL DEFICIT AND
PRIVATE-INVESTMENT DEFICIT IN THE SSA
ECONOMIES
4.1. THEORETICAL FRAMEWORK (MODEL)
The original idea of twin deficit hypothesis emerged around 1980-90, in explanation of current account
deficit experienced in the United States of America ((Abell, 1990; Adams, Sakyi, & Opoku, 2016; Darrat,
1988; Epaphra, 2017)). The macroeconomic theory of national income accounting identity can better
describe the twin deficits phenomenon. Following Lam (2012), the theoretical model of the study is based
on national income accounting identity. For testing the Ricardian equivalence hypothesis, the
consumption function specification of Bernheim (1987)has been used.
Following Lam (2012), this study uses the national income accounting identity which provides the basic
foundation for examining the relationship among current account deficit, budget deficit and private sav-
ings investment balance The twin deficit hypothesis is derived from the National Income Identity (NII),
which for an open economy is stated below in equation 4.1:

Y = C + I + G + (X − M) (4.1)

Where: Y is gross domestic product, C is private / household consumption expenditure, I is real

investment expenditure in the economy, G is government expenditure, X is export goods and services
and M is import goods and services.

The current account balance (CAB) is defined as:

CAB = NX + NI (4.2)
Where NI is net income, which is a composite of both net primary income (net factor income –
compensations of employees, investment income, etc.) and net secondary income (current transfers from
nonresidents minus transfers remitted abroad from residents). NX is net export, which is the difference
between exports of goods and services and imports of goods and services, which is (X – M).

The current account deficit is an indicator of external imbalance, which depicts the magnitude and
direction of international borrowing and lending. If a country’s current account is in surplus, it means
exports outweigh imports, implying it is a net lender to foreigners. On the contrary, if the current
account is in deficit, then imports outweigh exports, reflecting that it is a net borrower to foreigners.
Such borrowing may be done by the government or the private sector. A country with a current account
deficit has its residents giving out liabilities that are bought by non-residents.

In line with the national income identity, national saving (S) in the open economy equals:
 47

S = Y – C – G + CAB (4.3)
Where Y – C – G = I and I is investment.

Therefore, in an open economy the national saving identity is expressed as:

S = I + CAB (4.4)
However, savings in the economy are done by both private individuals and the government, therefore
national saving is decomposed as:

S = S p + Sg (4.5)

Where Sp is the portion of an individual’s income that is saved after deducting taxes and consumption
expenditure, expressed as:

Sp = Y – T – C (4.6)

Where, T is taxes paid to the government.

Government saving on the other hand is the difference between government tax revenue (T) and
expenditures on goods and services (G), and government transfers (Tr). This is expressed mathematically
as follows:

Sg = T – G – Tr (4.7)

In line with the definition of national saving, equation 6 & 7 are substituted into equation 4, which gives:

S = Sp + Sg = (Y – T – C) + (T – G – Tr) = I + CAB (4.8)

Equation (8) is expressed in the form below to capture the effects of government saving decisions.

Sp = I + CAB – Sg = I + CAB – (T – G - Tr) (4.9)

Rearranging (9) and expressing it in terms of the current account yields:

CAB = Sp – I – (G + Tr - T) (4.10)

Equation (10) shows that the current account balance is a function of the savings deficit (S p – I) and the
fiscal deficit (T –G – Tr ).

Theoretically, three possible scenarios can be deduced from equation (4.10); firstly, with the assumption

that private savings (Sp) and investment (I) are constant over time, variations in the fiscal position T –G
– Tr) have the tendency of triggering fluctuations in the current account, which means the existence of a
twin deficit hypothesis (see Kim and Roubini, 2008 and Gavallo, 2005) – as the fiscal deficit and current
account deficit are correlated.
 48

Secondly, with the assumption that private savings (Sp) and investment (I) are constant over time does
not hold, variations in the fiscal position (– G – Tr + T) and private net saving have the tendency of
triggering fluctuations in the current account, which means the existence of a triple deficit hypothesis
(see Sarlvatore, 2006) – as the fiscal deficit , private saving-investment deficit and current account deficit
are correlated.

The third theoretical inference is rooted in the Ricardian Equivalence Hypothesis, which assumes that
changes in fiscal operations will be perfectly balanced by changes in the difference between savings
and investment. When this happens, the twin deficit hypothesis will not hold, implying that fluctuations
in the fiscal balance are uncorrelated with the current account deficit (Suresh and Tiwari, 2014; Barro,
1989)

To understand the degree of association between budget deficit, private saving-investment gap and
current account deficit the above equation provides a basic context. If It is argued that there is no stable
gap between the private savings and domestic investments, hence budget deficit, private saving-
investment gap and current account deficit to move in same direction. While, if it is assumed that there
is stable gap between private saving and domestic investment, only budget deficit and current account
deficit move in the same direction.

Therefore, the study attempts to explore the twin/triple deficit hypothesis/twin divergent/current account
targeting/Ricardian equivalence hypothesis by applying cointegration analysis, causality test and error
correction estimation technique on annual fiscal, private saving-investment balance, and external
balances for 35 SSA countries. Through dynamic panel vector autoregressive (PVAR), the study attempts
to reveal if there exists a consistent causal relationship between these deficits. In addition, the study try
to check the robustness of the results of PVAR using the dynamic common correlated effect (DCCE)
modelling, which has proven successful for forecasting systems of interrelated panel data analysis
((Ditzen, 2018)).

4.2. METHODOLOGY
4.2.1. EMPIRICAL MODEL SPECIFICATION
The model is constructed to estimate the relationship among deficits. For this purpose, current account
deficit is regressed on budget deficit and private savings investment balance. The study aimed to
investigate the relationship between current account imbalance, private saving-investment imbalance and
budget imbalance. So, following (Basu & Datta, 2005); Chowdhury and Saleh (2007)the study estimated
the econometric model for the three deficits. The model is as follows:

𝑪𝑨𝒊 = 𝜶𝟎 + 𝜶𝟏𝑩𝑫𝒊 + 𝜶𝟐𝑺𝑰𝒊𝒕 + 𝝁𝒊 (4.11)

Where, 𝐶𝐴𝑖 is current account deficit, 𝐵𝐷𝑖 is budget deficit, 𝑆𝐼𝑖 is private savings and investment balance
(Sp – I), and 𝜇𝑖 is an error term. The coefficient of private saving-investment is expected to be positive
 49

and significant. The same is true of the coefficient of budget deficit: if the Keynesian channel is dominant
and thus the triple deficit hypothesis holds, it will be positive; otherwise it would be negative and/or
insignificant. Thus, the signs of the coefficients are best determined empirically. Given the coexistence
of high current account deficit, private saving-investment deficit and budget deficit and the consequent
adjustment lags, however, it appears that the Keynesian model better characterizes the SSA economies.

4.2.2. ESTIMATION METHOD

The estimation of long run relationships is important in the empirical application of
macroeconomics models. Long run relationships describe the steady state solution and how a
change in the steady state affect the long run relationship between variables in an econometric
model, estimating a long run relationship implies to estimate coefficients which capture this
relationship. A standard model in panel econometrics is dynamic common correlated effect estimation
models, in which one term of the equation captures the short and the other the long run
movements. In this section we employed panel data estimation method to examine the relationship
between current account, budget balance and private saving-investment balance in the 35 SSA countries
from 1980 to 2018.
This analysis is carried out within a panel data estimation framework. The preference of this estimation
method is not only because it enables a cross-sectional time series analysis which usually makes provision
for broader set of data points, but also because of its ability to control for heterogeneity and endogeneity
issues. Hence panel data estimation allows for the control of individual-specific effects usually
unobservable which may be correlated with other explanatory variables included in the specification of the
relationship between dependent and explanatory variables (Hausman and Taylor, 1981).
4.2.2.2 CAUSALITY ANALYSIS
we employ the granger causality test using Dumitrescu and Hurlin (2012)), Granger (1988) argued that if
variables are integrated of order one and there is an evidence of strong cointegration between them then
there will be a causal relationship in at least one direction. However, for short run analysis of variables the
Granger causality test has been applied using the following set of equations:

∆𝑪𝑨𝑩𝒕 =𝜶𝟎 +∑𝜶𝟏𝒊∆𝑪𝑨𝑩𝒕−𝒊 +∑𝜶𝟐𝒊∆𝑩𝑫𝒕−𝒊+ 𝜶𝟑𝒊∆𝑺𝑰𝒕−𝒊+𝜺𝒕 (4.12)

∆𝑩𝑫𝒕 = 𝜶𝟎 + ∑𝜶𝟏𝒊∆𝑩𝑫𝒕−𝒊 + ∑𝜶𝟐𝒊∆𝑪𝑨𝑩𝒕−𝒊 + ∑𝜶𝟑𝒊∆𝑺𝑰𝒕−𝒊 + 𝜺𝒕 (4.13)

∆𝑺𝑰𝒕 = 𝜶𝟎 + ∑𝜶𝟏𝒊∆𝑺𝑰𝒕−𝒊 + ∑𝜶𝟐𝒊∆𝑪𝑨𝑩𝒕−𝒊 + ∑𝜶𝟑𝒊∆𝑩𝑫𝒕−𝒊 + 𝜺𝒕 (4.14)

4.2.2.3. DYNAMIC COMMON CORRELATED EFFECTS (DCCE) ESTIMATORS

Given the nature of our dataset, we resort to the Dynamic Common Correlated Effects (DCCE) estimator
of Chudik and Pesaran (2015). While the analysis of macro panel data is still dominated by estimators
developed for micro datasets (primarily the estimators by Arellano and Bond (1991) and Blundell and Bond
 50

(1998), devised for panels where T is small relative to N ), the DCCE estimator is particularly suitable when
both the cross-section and the time series dimensions are sufficiently large. Indeed, our sampling period
spans 39 years, which allows us to exploit temporal variation in addition to cross-country heterogeneity.
Unlike standard estimators, a further advantage of the DCCE estimator is that it is robust to unknown types
of error cross section dependence, which is likely to feature due to the presence of common shocks and
unobserved components. This is highly relevant in our case, as the last few decades have witnessed
increased economic and financial integration that generates strong interdependencies amongst the cross-
sectional units in our sample. Indeed, this period captures several macroeconomic and financial cycles across
all countries in our sample, as well as common shocks. unaccounted for cross-sectional dependence can
lead to severe biases and this problem becomes more acute in dynamic panel settings, as discussed in
(Phillips & Sul, 2007)).
Moreover, the DCCE estimator addresses another potential source of inconsistencies that may arise if the
slope parameters are falsely assumed to be identical across countries (see Pesaran and Smith, 1995). Thus,
we control for heterogeneity by first estimating country-specific effects, which are subsequently combined
through a mean-group (MG) estimator to obtain estimates of the average effects.
Our choice of a dynamic framework is motivated by the literature on current account dynamics, which
suggests that there is considerable persistence in current account balance. In this vein, in order to estimate
the relationship between current account deficit, budget deficit and private saving-investment deficit, while
controlling for variables that are known to affect current account deficit, we adopt as our baseline
specification the following heterogeneous dynamic panel model with a multifactor error structure:

After re-estimating and comparing the baseline model replicating the specifications in previous
studies, we also consider a dynamic version of Eq.(7) below, which includes one lag of the
dependent variable (CAi,t−1).

𝐶𝐴𝐵ⅈ𝑡 = 𝛽0 + 𝛽1 𝐶𝐴𝐵ⅈ𝑡−1 + 𝛽1 𝐵𝐵ⅈ𝑡 + 𝛽2 𝑆𝐼𝐵ⅈ𝑡 + 𝜂ⅈ𝑡 + 𝜀ⅈ𝑡 (4.15)

𝜂ⅈ𝑡 = 𝛼ⅈ + 𝜆′ⅈ 𝑓𝑡 + 𝜀ⅈ𝑡 (4.16)

where CABi,t is the current account balance for country i in year t, BB i,t is a government budget Balance
for country i in year t and SIB is private saving-investment Balance for country i in year t. otherwise, xi,t
is a k-dimension vector of control variables as described in the previous subsection and assumed to be
weakly exogenous, αi accounts for time-invariant unobserved country specific effects, ft is an m×1 vector
of unobserved common factors (capturing common business cycles or exposure to global economic,
political or financial shocks, for example) with corresponding country-specific factor loadings λJi and ei,t
represents the idiosyncratic errors, possibly correlated across countries. Further below, we will also
consider a richer ‘hybrid’ version of the New Keynesian Phillips Curve, in which current account deficit
 51

depends on forcing variables that capture current account deficit pressures, as well as on a combination
of expected future current account deficit and lagged current account deficit

This is an extremely flexible specification that, with suitable restrictions on the parameters, encompasses
several approaches used in empirical practice, e.g. static and/or (partially) pooled panels, some of which
will be considered below. However, these frameworks can lead to biased estimates, particularly in the
presence of common unobserved factors, which is likely to be the case in our application.
Consistent estimation is carried out with the Dynamic Common Correlated Effects estimator of Chudik and
Pesaran (2015), which approximates the unobserved common factors by augmenting the estimation equation
with additional terms containing cross-section averages. Mean Group (MG) estimates can then be obtained
by averaging estimated coefficients across countries, with the corresponding standard errors computed non-
parametrically following Pesaran and Smith (1995). Although MG-type estimators are likely to produce
somewhat larger standard errors than pooled estimators, as a much larger number of parameters is
estimated, they are consistent both if slope parameters are homogeneous or if there is slope heterogeneity
across countries.
In addition, we will also consider an IV extension of the DCCE estimator that accommodates the possibility
of endogenous regressors (following Everaert and Pozzi, 2014), as well as the "half-panel jacknife" bias
correction method of Dhaene and Jochmans (2015)), in which the bias-corrected estimates are obtained
.

The error-correction mechanism for the long run equations is specified as:
∆𝐶𝐴𝐵ⅈ𝑡 =𝜆ⅈ + 𝜆ⅈ 𝐶𝐴𝐵ⅈ𝑡−1 + 𝜆ⅈ 𝑆𝐼𝐵ⅈ𝑡 + 𝜆ⅈ 𝐵𝐵ⅈ𝑡 + 𝜀ⅈ𝑡 (4.17)
We can proceed to run regressions in first difference provided the series of interest are I(1). Although we
may well lose the long-run relationship inherent in the data. There is a need to use variables in their levels
as well in the regressions. The Error Correction Model is designed to fit in variables both in their levels
and first differences and thus captures both the short run disequilibrium and long run equilibrium adjust-
ments between variables. Following Mukhtar et al. (2007), the Error Correction Model showing the rela-
tionship between the CADt, SID and BDt is specified as follows:
∆𝐶𝐴𝐵ⅈ𝑡 =𝜆0 + 𝜆1 𝐶𝐴𝐵ⅈ𝑡−1 − 𝜆2 𝑆𝐼𝐵ⅈ𝑡−1 − 𝜆3 𝐵𝐵ⅈ𝑡−1 − 𝜆4 𝐹ⅈ𝑡 + 𝜆∗ 5 ∆𝑆𝐼𝐵ⅈ𝑡−𝑗 + 𝜆∗ 6 ∆𝐵𝐵ⅈ𝑡−𝑗 + 𝜀ⅈ𝑡
(4.18)
For the estimation of the short run dynamics (𝜆∗ 5 ,, 𝜆∗ 6 the transformation of DCCE model into Error
Correction Representation is required. Error correction term (−𝜆4 ) is the rate of adjustment which
indicates that how quickly variables adjust towards equilibrium and its negative sign represents the
convergence in the short run. This term should be negative and statistically significant to establish the
long run relationship among variables.
 52

4.3. ESTIMATION AND INTERPRETATION OF RESULTS

4.3.1. INTRODUCTION
This chapter presents the results of the empirical estimation and gives an economic interpretation of the
results. We start with data description, test for cross sectional dependency, non-stationarity and go on
to estimate DCCE and then Granger causality test follows.

4.3.2. PRELIMINARY ANALYSIS

4.3.2.1. DATA DESCRIPTION

We employ panel data on budget balance, private savings-investment balance, and trade balance to
construct our three variables used in panel data analysis. Our data set is restricted by the availability of
comparable data, especially at the onset of transition; we limit the scope of our data to the period 19980
to 2018 and 35 of the SSA economies.
All the data related to the variables have been directly taken from the relevant sources in proportion to
GDP. The data on budget balance (surplus/deficit) are taken from the World Bank World Development
Indicators Database, IMF country reports. As can be seen in column 1 of the Table 4.1 below, most of our
sample countries ran sizable budget deficits during the observation period.
The data on trade balance, which refers to the difference between exports and imports of goods and
services, are also obtained from the World Development Indicators Database of the World Bank. To obtain
trade balance as a percentage of GDP, we deduct the imports-of-goods-and-services-to-GDP from exports-
of-goods-and-services-to-GDP. Again, as seen in column 3 of Table 1, most countries under consideration
ran trade deficits during the observation period.
To construct data series on private savings-investment balance, we draw on the African development Bank
Database to obtain data on both domestic savings and gross capital formation (a proxy for gross domestic
investment). We proxy government savings by government general budget balance on a cash surplus/def-
icit basis and deducted government savings from the figures for total domestic savings obtained from
above data sources to arrive at our private savings figure. The balance of private savings over gross do-
mestic investment describes private savings-investment balance.
The study used the panel data at an annual frequency for 33 SSA countries from 1980 to 2018. Due to
limitation of data for all SSA countries the study focused on only 35 countries. The main sources of data
are each countries Central Banks DATA statistics, “International Financial Statistics Yearbook”
published by International Monetary Fund, “World Development Indicators” published by the World
Bank and “Key Indicators of Africa” published by African Development Bank, Using these data, It is
essential to describe the panel time series characteristics of the data employed in the estimation of the
panel regression models (4.18) . Therefore, Table 4.1 indicates that the pooled average annual current
account balance(%GDP), budget balance(%GDP), and private saving-investment balance(%GDP) in 35
 53

SSA countries between 1980 and 2018 stood at -6.33%, -3.92, and 4.72%, respectively. This reflects the
high current account balance and low budget balance and private saving-investment balance in 35 SSA
countries as compared to other low-income countries.

Table 4.1: Descriptive Statistics Pooled Original Data

Std.
Variable Obs Mean Dev. Min Max
CAB_GDP 1,365 -6.32 10.32 -146.61 40.98
BB_GDP 1,365 -3.91 6.31 -51.88 39.37
SIB_GDP 1365 4.72 20.63 -128.93 240.44

The descriptive analysis of SSA’s data highlights two econometric issues. First, it identifies a concern
that the individual country series are characterized by cross-sectional correlation or dependence. Figure
4.1. indicates a high degree of correlation between country- level explanatory variables. This is to be
expected as these variables are linked by a country’s trade and economic relations, and financial
institution status. One would expect that countries that share these common factors would likewise have
systematic error correlations.

Figure 4.1: Scatter Correlation Matrix Plot of Variables

Second, as seen in Figure 4.2, while many countries align along a linear relationship between CAB_GDP
and the other variable such as SIB_GDP, BB_GDP, there are a disproportionate number of countries that lie
outside the 95% confidence interval for this linear relationship. This suggests that there may be
heterogeneity in the CAB_GDP slope parameter across countries.
 54

Figure 4.2A: Scatter Plot of Current Account Balance vs. Budget balance Variables

Figure 4.2B: Scatter Plot of Current Account Balance vs. private net saving balance Variables
 55

These two econometric issues have potential implications for model specification and efficiency of
standard panel data estimators. In subsequent sections, the procedures adopted in addressing the
aforementioned issues are discussed in detail.

4.3.2.2. TEST OF CROSS-SECTION INDEPENDENCE

The empirical work in this study is based on the annual data of 1980 up to 2018 and all of the
variables, that is ,current account balance, budget balance and private saving-investment balance
are taken as the ratio of GDP. To begin the analysis with, first,we start by examining cross-sectional depend-
ence among the series in our panels. Most of the recent theoretical and applied panel data econometric stud-
ies have emphasized the need to address the methodological issue related to cross-section or “between
groups” dependence in error terms when dealing with panel data models.
A) Panel Time Series Cross-Sectional Dependence Test Results
Cross-sectional correlation often emanates from unobserved common “shocks” and unobserved, time-in-
variant heterogeneous error components (Eberhardt & Teal, 2011, 2014; Pesaran & Tosetti, 2011;Sarafidis
& Wansbeck,2012 ). This error component is a sub-component of the error term, incorporating spatial
dependence and idiosyncratic pairwise dependence in the disturbance (De Hoyos & Sarafidis, 2006). Alt-
hough the notion of cross-sectional dependence has been in existence since the 1930s, it is often ignored
by researchers in panel model estimation.
The existence of cross-sectional correlation between error terms can have severe implications for the es-
timation of both coefficients and standard errors using standard panel data estimators. This can lead to
poor policy decisions based on biased parameter estimates. For this purpose, the three most often used
cross-sectional dependence test procedures- Pesaran (2004), Friedman (1937), and Frees (2004) cross-
 56

sectional dependence (CD) tests- were employed to examine the between-group correlation in error terms
(as a post-estimation diagnostic test) and panel time series variables (as a pre-estimation diagnostic test).
It should be noted that Frees and Friedman’s tests were originally designed for static models, unlike Pe-
saran’s CD test for static and dynamic models. Pesaran’s cross-sectional dependence test is more appli-
cable for pre- and post-estimation testing, unlike other tests that are more appropriate as post-estimation
tests (De Hoyos & Sarafidis, 2006).
Accordingly, I test for cross-sectional independence in the data used to estimate models as reported in
Tables 4.3 below. Table 4.3 presents the test results for cross-sectional correlation. It shows the average,
country-specific correlation coefficients for the panel series full matrix and off-diagonal matrix elements,
as well as Pesaran’s cross-sectional dependence test statistics. The results indicate high positive, pairwise
cross-sectional correlation of panel time series for current Account balance, Budget balance, and Private
saving-investment balance (all are as % of GDP). The results further reveal the presence of cross-sectional
dependence based on Pesaran’s CD test statistics for each variable. The null hypothesis of cross-sectional
independence is rejected at the 1% significance level.
Table 4.3: Panel Time Series Cross-Sectional Dépendance Test Results
Variable CD-test p-value corr abs(corr)
CAB_GDP 9.640 0.000 0.063 0.248
BB_GDP 20.770 0.000 0.136 0.229
SIB_GDP 6.470 0.000 0.042 0.308

Notes: Under the null hypothesis of cross-section independence CD ~ N(0,1)

As a result of the tests statistics above, the null hypothesis of cross-sectional independence is rejected for
all variables under consideration. This indicates that the individual country, panel data series employed in
this study are cross sectionally dependent and correlated, likely due to similar patterns of common mac-
roeconomic shocks. The standard (or parametric) average absolute correlation indicates positive pairwise
correlation coefficients of all the estimated residuals from replicated models. Also, the pairwise average
Spearman rank correlation estimates from the models are found to be positive and low below 0.5. This
indicates that the upper-diagonal has low positive and negative elements of country-specific pairwise cor-
relations coefficients, which cancel each other out during averaging. This problem invalidates Friedman’s
cross-sectional dependence (CD) test. As a result, I do not place much weight on the finding that Fried-
man’s CD test does reject the null of cross-sectional independence. In contrast, Frees’ CD test, based on
the average sum of squares of the rank of pairwise correlations, rejects the null hypothesis of cross-sec-
tional independence at the 1% significance level. Similar results are obtained using Pesaran’s CD test. As
a result, I conclude that the models’ error terms are characterized by significant cross-sectional depend-
ence.
 57

Table 4.4 Estimated residual Cross sectional dependence test results

CD tests P-value Average absolute value of the off-diagonal
elements
Pesaran's test 6.9 0.0000 0.221
Friedman's test 94.24 0.0000 0.221
Frees' test 2.27 0.0000 0.221
NB.Under the null hypothesis of cross-section independence, and a normal distribution had been used
to approximate Frees' Q distribution.

Taking into account cross-sectional dependence and country-specific heterogeneity in empirical analyses
is essential as our sample countries are highly integrated and highly globalized in their economic relations.
If cross-sectional dependency does exist, the use of Dynamic common correlated effect estimation
approach should be more efficient than an ordinary least-squares (OLS) approach in estimating panel data
causality. Moreover, the causality results obtained from the estimator developed by Zellner (1962) should
be more reliable than those obtained from county specific OLS estimations.
A further issue to decide is whether to treat slope coefficients as homogenous to impose the causality
restriction on the estimated parameters. The causality from one variable to another variable by imposing
the joint restriction for the panel is the strong null hypothesis and the homogeneity assumption for the
parameters is unable to capture heterogeneity due to country-specific characteristics.

B) Heterogeneous Slope Estimators

The preceding exploratory data analysis has determined that the relationship between current account
balance, budget balance, and private saving-investment balance in SSA countries is likely heterogeneous
due to differential macroeconomic policies, and prices across countries. To account for heterogeneous
effects, I next consider some recent panel data estimators that are designed to address these econometric
issues A test of slope homogeneity was performed as a robustness check using the test statistic suggested
by Swamy (1970). A test of slope homogeneity in panels with a large number of observations of the
cross-sectional (N) and time (T) dimension, which is based on Pesaran, Yamagata (2008) and Blomquist,
Westerlund (2013) is performed. Thus. the null hypothesis of the test is of homogenous slopes, implying
that all slope coefficients are identical across cross-sectional units. The test allows for non-normally
distributed errors, such as serial correlated errors. The test results are reported as follows.
Table 4.5 Test for slope homogeneity (H0: slope coefficients are homogenous)
Delta P value
-3.38 0.00
adj. -3.48 0.00
 58

The result of homogeneity test indicates that the null hypothesis of slope homogeneity is rejected, and
thus there is an evidence on heterogeneous slope.
4.3.3-NON-STATIONARITY TEST
The CADF test is a panel unit-root test that takes cross-sectional dependencies into account. The Fisher’s
combined p-values test proposed by Maddala and Wu (1999) and the Pesaran (2007) cross-sectionally
augmented Im, Pesaran and Shin (2003) (henceforth, CIPS) MultiPurt employed in this study. The test
is based on the augmented individual cross-section ADF (CADF) regressions of with cross-section
averages of lagged levels and first difference.
The Multipurt panel unit root diagnostic results in Table 4.6 indicates that indicate that all variables are
stationary in levels, and in first difference. This also confirms the non-stationarity of those series at levels
and first difference, i.e. for all the panel units, therefore, the variables are integrated of order one.
Table 4.6 Panel Unit root test
Variables Lag Maddala and Wu (1999) Panel Unit Root test (MW) Pesaran (2007) Panel Unit Root test (
Levels Difference Levels difference
Without With trend Without With trend Without With Without W
trend trend trend trend trend tre
- -
CAB_GDP 0 289.87*** 281.01*** 1708.02*** 1455.35*** -8.11*** -8.10*** 27.21*** 26
- -
CAB_GDP 1 210.35*** 196.68*** 990.41*** 819.51*** -4.45*** -4.10*** 21.86*** 19
- - - -
BB_GDP 0 377.70*** 304.11*** 1871.34*** 1605.69*** 12.18*** 10.40*** 27.12*** 26
- -
BB_GDP 1 238.37*** 181.52*** 902.48*** 735.70*** -8.03*** -5.95*** 21.35*** 19
- - - -
360.43***
SIB_GDP 0 371*** 1968.4*** 1689.35*** 13.28*** 11.86*** 26.89*** 26
- -
199.8***
SIB_GDP 1 192.35*** 993.65*** 808.58*** -8.6*** -5.85*** 20.44*** 18
NB> Null for MW and CIPS tests: series is I(1). MW test assumes cross-section independence.
CIPS test assumes cross-section dependence is in form of a single unobserved common factor -mul-
tipart- uses Scott Merryman's -xtfisher- and Piotr Lewandowski's -pescadf-. ***-p<1%, **-p<5% and
*-p<10%

To summaries, this chapter contributes to the body of knowledge by identifying and addressing some of
the estimation and theoretical issues in empirical analysis. In particular, five econometric issues were
addressed: (1) cross-sectional correlation of observable and unobservable series, (2) non-stationarity of
unobservable common factors, (3) homogeneity of slope coefficients across countries, (4) non-correlation
 59

between explanatory variables and the unobservable error term, and (5) correct specification of a test of
the current account model.
In conclusion, based on these robustness checks presented in this chapter accounted for the
aforementioned issues we selected and employ the robust and less restrictive panel data estimators that
account for slope heterogeneity, cross-sectional dependence, non-stationarity of unobservable factors,
and endogeneity emanating from common, unobservable shocks. It should also be noted that the
cointegration teat, and dynamic Common correlated estimator are the robust and less restrictive estimator
to examine the current account sustainability based on the error correction model.

4.3.4. COINTEGRATION TEST

After testing the stationarity of the variables, the possible existence of a long-term relationship between
the variables was analyzed in the following section using panel cointegration test.

Once the time series properties of the data are evaluated then Error based cointegration test is conducted
to test whether there exists any long run relationship among the variables. Since our sample data size is
small, we construct and test the restricted Westerlund error correction test with short‐run dynamics for
all series with a single lag and lead. Results of the restricted ECM model are reported in table 4.7.

Table 4.7 Westerlund Panel Cointegration Test

restricted case with single lag and lead
constant constant and trend
Statis- z- P- Robust P- Z- P- Robust P-
tic Value value value value Value value value value
Gt -2.60 3.63 0.00 0.00 -3.19 4.67 0.00 0.00
Ga -12.44 3.12 0.00 0.00 -16.36 2.20 0.01 0.00
Pt -14.83 4.55 0.00 0.00 -17.78 4.71 0.00 0.00
Pa -10.97 5.40 0.00 0.00 -15.02 3.96 0.00 0.00

Tables 4.7 present the results from the corresponding cointegration tests for the periods 1980-2018.
According to these results of restricted case without and with trend, the null hypothesis of no cointegration
of the group‐mean tests (Ga and Gt) and of the panel tests (Pt and Pa) is rejected at 1% significance level
by simple and robust p value. Thus, we conclude that all variables are cointegrated in group mean and
panel tests.

The fact that the variables are cointegrated indicates that we find the first evidence of the validity of triple
deficit in the panel.
4.3.5. CAUSALITY ANALYSIS
After establishing co-integrating relationships between budget deficit, current account deficit, Private sav-
ing-investment gap, we next tested the direction of the causal relationships between these variables. There
 60

are three commonly used approaches for testing the direction of Granger causality in panel data. The first
approach is based on estimating a panel vector error correction model by means of a generalized method
of moments (GMM) estimator that estimates a panel model by eliminating the fixed effect. This approach
does not account for heterogeneity or cross-sectional dependence. The second approach, proposed by Hur-
lin (2008), is a panel data causality test that allows for slope heterogeneity. This approach does not take
into account cross-sectional dependence, which, if it exists, creates substantial biases and size distortions.
The third approach, proposed by Kónya (2006), allows both heterogeneity and cross-sectional dependence
to be taken into account.
The existence of cross-sectional dependence and heterogeneity across countries supports the suitability of
the bootstrap panel causality approach. Here, we take into account the possible existence of direct
relationship between budget deficits and current account deficits, and/or among budget deficits, private
savings-investment deficits, and current account deficits. For this purpose, we employ the bootstrap
Granger causality approach developed by Kónya (2006), based on bi-variate [budget balance (BB) and
private savings-investment balance (SIB)] and tri-variate [(BB), (SIB), and current account balance
(CAB)] finite-order vector autoregressive models. In our opinion, the bootstrap panel causality approach
is superior to the first two techniques mentioned above in terms of accounting for cross-sectional
dependency and country-specific heterogeneity. In detecting Granger causal relationships, the bootstrap
panel causality approach is based on seemingly unrelated regressions estimation of the set of equations
and Wald statistics with country-specific bootstrap critical values. Notably, Kónya (2006) indicated that
this approach does not require any pre-testing for the panel unit root and cointegration
Table 4.19 reports the results of Dumirescu & Hurlin (2012) Panel Granger Causality (PGC) test. The
tests are performed with level variables. The results are reported for lag augmentations from P=1 to P=3
(inclusive).
Table 4. 8
Lag 1 Lag 2 Lag 3
W Zb Zt W Zb Zt W Zb Zt
CAB_GDP 1.19 0.78 0.47 3.20 3.56 2.78*** 3.66 1.59 0.89
BB_GDP ***

BB_GDP 2.16 4.84*** 4.13*** 2.64 1.88* 1.31 3.19 0.47 -0.06
CAB_GDP

BB_GDP 2.70 7.13*** 6.19*** 3.73 5.11*** 4.13*** 5.13 5.13*** 3.91***
SIB_GDP
 61

SIB_GDP 1.43 1.80* 1.39 2.63 1.86* 1.29 3.41 0.99 0.38
BB_GDP

SIB_GDP 2.32 5.53*** 4.75*** 3.99 5.91*** 4.84*** 5.03 4.89*** 3.70***
CAB_GDP

CAB_GDP 1.55 2.30** 1.84* 3.24 3.68*** 2.88*** 4.68 4.05*** 2.99***
SIB_GDP

The results of granger causality contained in Table 4.8 indicate that the null of BB not causing CAB and
CAB not causing BB both can be rejected at 1% hence there is bidirectional causation between two bal-
ances under first lag order. The causation also runs from saving gap to CAB lending support to the triple
deficit hypothesis in the sense that BB and saving gap jointly determine CAB. The saving gap also causes
BB but the causation is bit weaker in terms of statistical significance. These results are in conformity with
those of Akbas and Lebe (2014) who also find saving gap to play a role in determination of both budget
and CAB for G7 countries. The causation may imply that inadequate savings have a negative effect on
investments, which leads to a decrease in both the export revenues and the tax revenues to be taken from
these investments. Therefore, BB and the CAB are negatively affected (Akbas and Lebe, 2014).

The results also showing that there is a bidirectional relationship between CAB_GDP and SIB_GDP at
1% significant level. The outcome of the PGC tests reveals that CAB_GDP on average, can be used to
predict SIB_GDP and vice versa, and supports the notion that there, on average, are short-term(inter)
dependencies The null hypothesis of that CAB_GDP does not homogeneously cause BB_GDP is rejected
regardless of the number of lags included. whereas the SIB_GDP granger cause CAB_GDP regardless of
the number of lags. This revealed that both budget and private saving-investment balance on, average, can
be used to predict CAB_GDP and thus the three variables have a causal relationship. Therefore, the PGC
test confirms the validity of triple deficit hypothesis in SSA countries during the study period.

To examine the validity of twin deficit, or triple deficit or twin divergence, or Ricardian equivalence in
each country of the panel in SSA, we performed the bootstrap panel Granger causality analysis and the
results from the test are presented in Tables 4. 9 (see appendix 2).The results reported in Table 4.9 report
that there exists a significant, and positive relationship between budget deficit and current account,
Granger causality running from budget deficit to current account deficits at 10% level of significance only
for Benin, Cameroon, central Africa, Congo Democratic, Gambia, Kenya, Rwanda, Sudan for the first lag
order, and for Benin, Botswana, Cape Verde, Congo democratic, Kenya, Rwanda, Senegal for second
 62

order lag, but We do not find any significant relationship for the remaining countries in the sample in the
first lag. In other words, the null hypothesis of non-causality between budget deficit and current account
deficit is rejected for Benin, Botswana, Congo Democratic, Cape Verde, Kenya, Rwanda, Senegal, SSA
countries under consideration. We find empirical support for the validity of twin deficits hypothesis for
these countries.
On the other hand, The results reported in Table 4.9 suggest that there exists a significant, and positive,
Granger causality running from current account deficit to budget deficits at 10% level of significance only
for Congo republic, Eswatini , Senegal, South Africa, Uganda, but We do not find any significant rela-
tionship for the remaining countries in the sample in the first lag. On the other hand, Table 3 indicates that
there is a significant and positive Granger causality running from trade deficit to budget deficit at 10%
level of significance for three countries Comoros, Ghana, Mali, Senegal for lag 3. therefore, we find an
evidence for the validity of twin deficit divergence in these countries. The possible explanation of these
findings might be that widening current account deficits may have decreased aggregate demand in these
countries, resulting in a reduction in output and an increase in unemployment. To overcome this issue,
their governments may have attempted to boost their economies through expansionary fiscal and monetary
policies such as allowing budget deficits, increasing reliance on foreign borrowing, or injecting money
into the economy to eliminate the loss of exports. Thus, these current account deficits may reflect budget
deficits financed by foreign borrowing.
The Granger causality test results for the null hypothesis show that BB and SIB do Granger cause CAB
for Cameroon, sudan5,31 under first order lag, and Benin, Botswana1,2 under second order lag as indi-
cated in the Wald test column of Table 4. In other words, the null hypothesis of non-causality is rejected
for Benin, Botswana, Cameroon, Sudan SSA countries under consideration. We find empirical support for
the validity of triple deficits hypothesis for these countries.
Overall, Table 4.9 summarizes our results of the direction of panel Granger causality among the three
variables for all the countries examined. As can be seen from the table, the empirical results do not support
the validity of the twin or triple deficits hypotheses for Burkina Fast, Burundi, Cote devoire, Ethiopia,
Lesotho, Malawi, Mauritius, Nigeria, Seychelles, Sierra Leon, Tanzania, Togo Zambia of our sample
countries. Specifically, their budget deficits do not Granger-cause trade deficits and the existence of dual
domestic deficits (budget plus savings-investment deficits) does not lead to external deficits. This reveals
that there is no any Ricardian relationship among budget deficit, current account deficit and private saving-
investment deficit in those countries, meaning Ricardian equivalence hypothesis holds in these countries.

4.3. 6. DYNAMIC COMMON CORRELATED EFFECT ESTIMATION TECHNIQUES

After confirming cointegration and granger causality tests, the next step is to estimate the significance
level of the coefficients of short run and long-run relationship using pooled mean group and dynamic
common correlated effect estimators. The estimates of long run coefficients reported in Table 4.10. As
 63

revealed by the table all three estimators indicate a positive and statistically significant impact of BB and
SIB on CAB.
According to the results of the pooled mean group and Dynamic common corrected effect estimator, both
the budget deficit and the net savings gap are positive and statistically significant in all models. The error
correction term (ECT) estimated by inserting the long run coefficients in the short run dynamic specifica-
tion of the model turns out to be negative and statistically significant under all estimation techniques. The
negative ECT shows that the system is driven to its long run cointegration path. The coefficient of ECT
reflecting the speed of adjustment is estimated to be around 43-62% per year. These statistics (ECT) sug-
gest that all the series (BB; SIB and CA) are co-integrated in the long run and hence, there is strong
evidence in support of the triple deficit hypothesis in SSA. In addition, the triple deficit hypothesis per-
taining to fiscal deficit(FD), current account deficit(CA), and net private saving deficit(SIB) has also been
validated in the long run through the Granger causality using F-statistic and t-statistic of the lagged ECM,
This suggests that the three deficits have a long run and positive relationship.
The PMG1 estimator suggests that, on average, a strengthening (deterioration) in BB-to-GDP ratio of 1%
point is associated with an improvement (deterioration) in the current account-to-GDP ratio of about
0.62% in the long run. The impact of a 1%-point strengthening (deterioration) of BB on the CAB is of
order 0.50%,0.70% and 0.79% in the long run under the dynamic common correlated estimators of MG1,
MG2 and PMG2 estimators respectively.
The coefficient of saving gap is consistently positive and significant under all estimators. The private
saving gap exerts a positive effect on CAB though the impact is weaker compared to BB. A strengthening
(deterioration) in saving gap ratio of 1% point is associated with an improvement (deterioration) in the
current account-to-GDP ratio of 0.58%.,0.39%,0.52% and 0.51% in the long run using PMG1, MG1,
MG2, and MG2 estimation models respectively.
Based on these results, it can be said that the triple deficit hypothesis and traditional approach are valid in
SSA countries in the long run. Similarly, an increase of 1% in the budget deficit increases the current
deficit on average by a rate of 0.17%.,0.20%, and 0.25% in the short run using MG1,MG2, and MG2
estimation models respectively. A 1% increase in net savings increases the current deficit on average by a
rate of 0.16%.,0.18%, and 0.24% in the short run using MG1, MG2, and MG2 estimation models respec-
tively.

Table 4.10. Estimated long-run relationship and short run adjustment

(1) (2) (3) (4)
D.CAB_GDP PMG 1 MG1 MG2 PMG2
BB_GDP 0.620*** 0.504*** 0.695*** 0.789***
(0.053) (0.167) (0.126) (0.123)
SIB_GDP 0.582*** 0.389*** 0.517*** 0.513***
 64

(0.029) (0.118) (0.078) (0.154)

D.BB_GDP 0.166*** 0.199*** 0.245***
(0.040) (0.044) (0.045)
D.SIB_GDP 0.159*** 0.178*** 0.240***
(0.035) (0.028) (0.034)
L.CAB_GDP -0.616*** -0.589*** -0.445***
(0.034) (0.040) (0.127)
trend -0.054 -0.043 -0.017
(0.034) (0.031) (0.027)
Obs. 1330 1330 1225 1225
R-squared .z 0.772 0.767 0.720
RMSE 4.12 3.10 4.15
N 1330 1330 1225 1225
cd 2.328 0.233 -0.649
cdp 0.02 0.82 0.52

Standard errors are in parenthesis and *** p<0.01, ** p<0.05, * p<0.1

NB-MG1-dynamic common correlated mean group with no cross-section, MG2- dynamic common
correlated mean group with cross-section, PMG1-pooled mean group PMG2- dynamic common
correlated pooled mean group

As stated earlier, in order to conclude that there is long run relationship between current account deficit,
budget deficit and private saving-investment deficit, meaning triple deficit hypothesis hold both the
error-correction term should be negative and less than one as well as statistically significant and the long-
run coefficient should also be statistically significant and positive. As reported in the above Table 4.10,
both the error-correction terms in all model are negative and statistically significant as well as the long
run and short run coefficients of net government saving, and net private saving are positive and
statistically significant at 1%. This result indicates that there is a long run linear combination
among current account deficit, budget deficit and private saving-investment deficit and thus
triple deficits exist for the whole sample in SSA countries. In addition, the null hypothesis of cross-
sectional dependence is rejected for estimation model of dynamic mean group without cross sections,
however, for the other models, it is not rejected.
As stated earlier, in order to validate the existence of twin deficits or triple deficit or Ricardian equivalence
hypothesis in SSA both the error-correction and the long-run coefficient should be statistically significant
and negative. In the whole panel (Table 4.11), the error correction coefficient is negative but statistically
 65

insignificant for 31 countries out of 35 SSA countries, but it is statistically significant, negative in sign
and less than one in value for four countries (Ghana, Kenya, Mauritius, and Uganda). Hence, for Ghana,
Kenya, Mauritius, and Uganda, our results indicate that long run relationship between current account, net
government saving, net private saving and exists. However, for the remaining 31 countries, these do not
hold. Thus, we suggest that we do not find empirical evidence on the validity of triple deficit hypothesis
for 31 countries in SSA for the period, however, we find that triple deficit holds in four countries in SSA
Ghana, Kenya, Mauritius, and Uganda.

Table 4.11. Dynamic Common Correlated Effect Mean Group Model (DCCEMG
short run longrun
ECT
D.BB_GDP D.SIB_GDP BB_GDP SIB_GDP

Coef. Std.Er Coef. Std.Err. Coef. Std.E Coef. Std.Err. Coef. Std.Err.
country r. rr.
Benin 0.32** 0.62 0.38 0.36 -0.42 0.55 0.61** 1.10 -0.14*** 0.91
Botswana -0.21 0.22 0.01*** 0.23 -0.85 0.16 1.05 0.19 0.76 0.13
Burkina Faso 0.40 0.41 0.16 0.22 -0.29 0.22 0.63*** 1.74 1.03 1.27
Burundi 0.35 0.27 -0.09 0.15 -0.63 0.22 0.93 0.46 0.65 0.27
Cameroon 0.52 0.47 0.23 0.15 -0.46 0.20 0.18*** 0.80 0.14*** 0.42
Cape Verde 0.06*** 0.43 0.19** 0.43 -0.93 0.45 0.37 0.43 0.30 0.36
Central Afri-
-0.13*** 0.50 -0.05*** 0.47 -0.93 0.50 0.84 0.60 0.72 0.31
can Republic
Chad 0.31 0.33 0.38 0.11 -0.76 0.12 0.94 0.44 0.74 0.05
Comoros 0.14*** 0.37 0.18 0.20 -0.86 0.29 0.61 0.39 0.21 0.30
Congo, Dem.
-0.01*** 0.34 0.03*** 0.18 -0.52 0.38 -0.06*** 0.76 0.24 0.23
Rep.
Congo, Rep. 0.38 0.16 0.54 0.11 -0.54 0.09 0.92 0.11 0.61 0.12
Cote d'Ivoire 0.48 0.58 0.27** 0.55 -0.57 0.37 1.39 0.40 1.38 0.70
Eswatini 0.20 0.29 -0.05*** 0.17 -0.62 0.22 0.56 0.57 0.34 0.18
Ethiopia -0.15*** 0.46 -0.05*** 0.17 -1.04 0.35 0.30 0.47 0.08** 0.17
Gabon 0.30 0.30 0.18 0.25 -0.69 0.24 1.38 0.25 0.70 0.25
Gambia -0.21 0.34 0.06*** 0.22 -0.88 0.27 0.36 0.49 0.35 0.21
Ghana 0.54 0.37 0.17 0.23 -0.16*** 0.60 -0.40*** 3.38 0.50*** 1.27
Kenya 0.42 0.57 0.55 0.60 -0.19*** 0.80 1.13*** 3.23 1.44** 3.25
Lesotho -0.02*** 0.18 0.03*** 0.14 -0.89 0.17 0.73 0.16 0.67 0.06
Madagascar -0.01*** 0.54 0.23** 0.50 -0.50 0.45 0.96 1.25 0.85 0.45
Malawi 0.32 0.34 0.17 0.15 -0.65 0.22 0.02*** 0.62 0.08*** 0.24
Mali 0.29 0.32 0.19 0.28 -0.46 0.25 0.42 0.63 0.61 0.50
Mauritius 0.11 0.63 0.46 0.40 -0.16** 0.38 3.12** 6.09 1.78* 2.90
 66

Niger 0.98 0.57 0.14*** 0.49 -0.51 0.45 -0.34*** 1.37 -0.05*** 0.95
Nigeria 0.10*** 0.52 0.25 0.27 -0.58 0.39 1.26 1.05 0.47 0.22
Rwanda 0.07*** 0.44 0.31 0.33 -0.52 0.51 0.90 0.63 0.54 0.34
Senegal -0.09*** 0.81 0.30 0.37 -0.68 0.52 0.88 1.31 0.12*** 0.46
Seychelles 0.61 0.21 0.14 0.10 -0.47 0.18 0.24 0.31 0.55 0.29
Sierra Leone 0.16*** 0.41 -0.06*** 0.22 -0.94 0.22 0.92 0.29 0.79 0.11
South Africa -0.05*** 0.87 0.11*** 0.82 -0.50 0.63 0.54** 1.24 0.53** 1.15
Sudan -0.03*** 0.33 0.07 0.05 -0.69 0.24 0.51 0.44 -0.13 0.11
Tanzania -0.06*** 0.67 0.24 0.14 -0.29 0.35 2.55 3.09 0.01*** 0.41
Togo 0.36 0.30 0.34 0.29 -0.61 0.22 -0.08*** 0.62 -0.39** 0.87
Uganda 0.31 0.40 0.23 0.25 -0.23** 0.55 -0.93*** 3.94 0.72*** 1.42
Zambia 0.23 0.37 -0.03 0.20 -0.59 0.21 0.88 0.41 0.90 0.18

Notes: ***, **, * denote significance at 1%, 5% and 10% respectively

On the other hand, table 4.11 presents that there is a short run relationship between budget deficit and
current account deficit, meaning we find an evidence that twin deficit hold in 5 countries Benin, Comoros,
Mauritius, Nigeria, Rwanda, Sierra Leon in SSA in the short run at 1% significance level, whereas budget
deficit and current account deficit have an inverse relationship(twin deficit divergence) in 5 countries
Central Africa, Congo Democratic, Ethiopia, Lesotho, Madagascar, Senegal, South Africa, Sudan, Tan-
zania in SSA. In addition, as reported in table above, we find an evidence that budget deficit and current
account deficit have a positive long run relationship at 5% significance level only in Kenya , however,
these deficits have an inverse relation (twin divergence) in the long run at 5% significance level in Ghana
countries. We also find the three deficits have a long run positive relationship at 10% significance level in
Ghana, Kenya, Mauritius, and Uganda during 1980-2018, meaning that triple deficits hold only in Ghana,
Kenya, Mauritius, and Uganda.

4.4. SUMMARY, CONCLUSIONS AND RECOMMENDATIONS

The section dwells on summarizing results, making appropriate policy recommendations from the study
and suggests areas of further research which have not been explored by the study.
4.4.1. SUMMARY
The objective for the study was to test for the validity of the Triplet Deficits Phenomena by investigating
the Fiscal, Current and private saving-investment deficit nexus in SSA. The specific objectives were to:
establish if there is any relationship between the budget, Current, and the private saving-investment
Balance; and to establish the direction of causality between the three Balances. These objectives were
complimented by two major research questions; how the budget, Trade and private saving-investment
Balance relate and how they cause one another. As these countries have typically low domestic savings
ratios and limited capital bases, they are open to substantial vulnerabilities in relation both to the continuity
 67

of capital inflows and to changes in capital gains and losses emanating from international capital
movements and exchange rate changes.
Due to the serious implications associated with the persistent BD and CADs, the alleged causation from
BD to CAD has been studied intensively in the empirical literature but the role of private saving gap in
the emergence of CAD has often been ignored. In view of this gap the present study investigates the “triple
deficit hypothesis” - an extension of “TDH” with inclusion of private saving gap for the panel of 35 SSA
countries for the period 1980-2018.

Preliminary investigation of the data was done through summary/descriptive statistics, graphical review
and cross-sectional dependency test. This was followed by unit root tests, which were conducted through
CADEF test was applied as confirmatory test of the Maddala & Wu(1999) and Peasarn(2007) results.
Then, it proceeded to test for cointegration. The lag selection was first established and then a error co
Westerlund error correction cointegration Test was run, followed by a pooled mean group and dynamic
common correlated effect estimation analysis. To be specific, the short-run dynamics and causality
between the variables were conducted using the granger causality tests within the Dumitrescu & Hurlin
(2012) Granger non-causality test results framework.
Motivated by the results of the unit root tests, we tested for cointegration between the series to ascertain
whether there is any long-run equilibrium between the series. The results suggest that there is long-run
relationship between the government budget balance, the current account balance and private saving-in-
vestment balance for the sample. These findings are in support of the Keynesian/Conventional approach
that demonstrates how the three deficits in the budget account, trade account and private saving-invest-
ment balance draw a parallel. The approach posits that a current account deficit would indirectly result in
the other two deficits. The findings of the study indicate that a deficit in the budget account results in an
increase in the current account deficit and widens the private saving-investment balance.
Following the results for the cointegration analysis, we employed the procedure espoused by Dumitrescu
and Hurlin (2012)to assess the short-run relationship between the series. According to the results of this
test, budget balance and private saving-investment balance cause the current account balance. These evi-
dence the validity of the Triplet Deficits Hypothesis for the SSA case. The above results show the exist-
ence of cointegration among current account deficit, budget deficit and private savings investment balance
so triple deficits hypothesis cannot be rejected for the sample. Asrafuzzaman et al. (2013) for Bangladesh,
Basu and Datta (2005) for India, Ratha (2012) for India, Javid et al. (2010) for Pakistan also found evi-
dence for triple deficits hypothesis. The developing countries including SSA are marked for facing huge
fluctuations, budgetary and current account imbalances. The results reveal the fact that in selected SSA
countries variations in the budget deficits and private saving investment balance cause any systematic
changes in current account deficit. Therefore, the Keynesian view of twin deficits is not rejected. The
reasons for the validity of triple deficits in SSA region can be attributed to the fact that the difference of
 68

private savings and investment has not been stable during the period of the study separating both the
deficits. Although the triple deficits hypothesis does not prove to be true in these countries but still each
deficit causes several issues and their unlimited growth could cause severe harms to the economy.

However, in terms of causality test results for each country under consideration, mixed results were ob-
tained showing unidirectional and no relationship for some countries. A unidirectional relationship was
found for Benin, Cameroon, central Africa, Congo Democratic, Gambia, Kenya, Rwanda, Sudan 1by
Granger causality test. Thus, the twin link (budget deficit causes current account deficit) was confirmed
by the Granger causality test for Benin, Cameroon, central Africa, Congo Democratic, Gambia, Kenya,
Rwanda, Sudan the SSA region. Private saving-investment balance causes Budget Balance while Budget
Balance and Current Account Balance Cause private saving-investment balance.
overall, the findings of Granger causality test do not show existence of causal relationship among current
account deficit, budget deficit and private savings investment in the short run for all countries except for
Benin, Botswana, Senegal where unidirectional causality exists between BD and CAB in the short run.
Similarly, a unidirectional relationship exists running from SI to CAB in Benin, Botswana, Senegal in the
short run.

The results of the pooled mean group and Dynamic common correlated effect estimators also support the
triple deficit hypothesis in SSA. The long run coefficients obtained from Pooled mean group and dynamic
common correlated effect estimators indicate a positive impact of BD on CAB. The results based on all
the four estimators also indicate a positive impact of saving gap on CAB. On average, a strengthening
(deterioration) in fiscal balance-to-GDP ratio of 1% point is associated with an improvement
(deterioration) in the current account-to-GDP ratio of about 0.50-0.76 and a strengthening (deterioration)
in private saving gap-to-GDP ratio of 1% point is associated with an improvement (deterioration) in the
CAB-to-GDP ratio of about 0.38-0.54 in the long run. Through causality analysis we find a feedback
relationship between current account and BB. The saving gap is also found to be causing current account
and BB but the causation is bit weaker in terms of statistical significance in the latter case. Our results
hence lend support to the existence of triple deficit phenomenon for the selected SSA economies with
CAB being jointly explained by BB and private saving gap.

Overall, it appears that private saving-investment deficit, budget deficits and current account deficits are
causally dependent variables in our sample. It is worth mentioning that our findings are broadly parallel
to the empirical findings of several earlier studies, including Bolat et al. (2014), for Poland, Portugal,
Spain and Sweden, Kuijs (2006) triple surplus for China, Hatsopoulos et al. (1988) and Eisner (1994) for
US, Roubini (1988) for 18 OECD countries. Similar views were also supported by Hakkio (1995), Higgins
and Klitgaard (1998), Cooper (2001), Mann (2002), Labonte (2005), Hubbard (2006) and Elwell (2008)
 69

and the authors explained that by stimulating budget deficits the saving deficits of US created the negative
effects on the foreign trade balance.
There may be several explanations for these findings. First, the existence of an out-put gap may be a factor.
This suggests the existence of an output gap. If so, increases in aggregate demand following expansionary
fiscal policies may have been masked by increases in domestically produced goods and services, rather
than through imports. The second plausible explanation may be a substantial exogenous increase in
investment. These investment booms might have been generated through foreign technical assistance,
technological innovation, successful market-oriented reforms, or a combination of all three. Third, there
was the external assistance these countries received at the earlier stages of transition from international
financial organizations such as the IMF, World Bank, as well as bilateral donors. More-over, the countries
received substantial financial and technical supports from china. Finally, SSA countries are commodity-
exporting countries and so their export earnings and demand depend mostly on external factors. Over the
observation period, there were several currency devaluations that effectively restrained imports to SSA
countries.

In this study, we investigated the relationship among current account deficit, budget deficit and private
saving-investment gap over the period 1980-2018 in 35 countries of SSA countries, which have
experienced continued negative balances in their external account and significant challenges of adjustment.
Our analysis provides fresh evidence for a set of countries which are relatively under-studied but are
collectively of systemic importance; and it does so in an extensive way, utilizing alternative but
complimentary approaches to testing for their relationship. Moreover, the analysis incorporates measures
and techniques which are designed to address recently identified problems with the traditional approaches
to the issue:

This study broadly relates to possible explanations for the divergent results among the many empirical
papers. The different findings may largely arise from the differences in methodology and data. In some
previous studies, the possibility of cross-sectional dependence and slope heterogeneity was ignored in the
series. Further, analysis of data sets that focus on a short period of time may not yield reliable evidence.
Lack of longer-term data for countries, limits the possibility for clear-cut, differentiated results. Not to put
a fine point on it, but differences in econometric techniques, data measures, samples employed, etc. yield
different results. To overcome such differences, future studies should concentrate on comparison of
various estimation techniques on a common data set.
All in all, based on our empirical findings, from a policy standpoint, such an evidence implies that the
causes of large and persistent external deficits must be sought somewhere else than the budget side of the
economy. Behind this, there might be several reasons, such as the structure of foreign trade, the exchange
rate regime pursued, and the international competitiveness of the particular country in question, the degree
 70

of capital mobility. Nevertheless, it is obvious that the case for the twin or triple deficits hypotheses is
more likely to be seen in countries with economies that are highly integrated with international markets,
open to capital movements, and experience intensive international competitiveness.

But the previous literature does not incorporate the important variable; private savings investment balance
while estimating the twin deficits model. So, the study differentiates from the other studies by including
this variable in the model and reveals the fact that both budget deficit and private savings investment
balance effect current account deficit in SSA countries.

4.4.2. CONCLUSION
The study was interested in investigating the triplet deficits hypothesis for SSA by investigating the Fiscal,
Current and Private saving-investment deficit nexus. The overall finding is that Fiscal deficit increases
both Current Account deficit and private saving-investment Balance. Current Account deficit on the other
hand increases private saving-investment Balance. Private saving-investment Balance increases the fiscal
Budget Balance and the Current Account Balance. In terms of causality, Private saving-investment Bal-
ance causes Fiscal Balance, while fiscal Balance and Current Balance cause Private saving-investment
Account Balance. These findings evidence validity of the Triplet Deficit Hypothesis for the SSA case, and
an indication of verification of the Keynesian/Conventional approach.
This study tested the validity of the twin and triple deficits hypotheses by using an annual time series panel
data from 35 SSA countries, spanning from 1980to 2018. We found no evidence in favor of the twin or
triple deficits hypotheses for the sample countries. In other words, there is Granger causality running from
budget deficits to trade deficits and Granger causality running from internal deficits (budget deficits plus
private savings-investment deficits) to trade deficits when the sample countries were considered. Our find-
ings suggest that the twin and triple deficits hypotheses hold for the countries under consideration over
the observation period. These findings are broadly parallel those of similar studies conducted for non-SSA
countries. Moreover, our findings are in concordance with all but a limited number of studies regarding
this subject on SSA countries, specifically (Gabrisch, 2015; Gurgul, Lach, & Mestel, 2012).
The findings above may be justified based on several arguments. The first argument involves the presence
of an output gap. Indeed, with some minor exceptions, actual output remained well below its potential
level in most sample countries. Given that, increases in aggregate demand following expansionary fiscal
policies may have been masked by increases in domestically produced goods and services, rather than
through imports. The second plausible explanation may be a substantial exogenous increase in private
investment. These investment booms might have been generated through foreign technical assistance,
technological innovation, successful market-oriented reforms, or a combination of all three. Successfully
implemented free-market reforms, in particular, would have conferred the economic benefits of growth,
enhanced trade competitiveness, and inflows of much-needed foreign capital. Third, there was the external
assistance these countries received from international financial organizations such as the IMF, the World
 71

Bank, as well as bilateral donors. Besides, the countries received substantial financial and technical sup-
port from China. Finally, most SSA countries are the major commodity-exporting countries within the
sample countries. These are the most resource-rich countries, particularly in natural resources (minerals,
oil and gases) and agricultural products, and they play leading roles in these products in related interna-
tional markets. This implies that external factors play an important role in the export earnings of these
countries. Due to this fact, the nexus between internal and external deficits may have exist. Another pos-
sible justification may be the co-existence of rises in private savings together with the crowding-in effect,
leading to the triple deficit effect. One more argument could be that increases in the imports of intermedi-
ate goods in the countries in the sample resulted in widening trade deficits. Therefore, one could argue
that imports of intermediate goods in the domestic market boosted domestically produced final products
as well as increased current account deficit. Furthermore, an increase in government spending on infra-
structure and other projects leads to crowding in effect and results in higher budget deficits.

Overall, based on our empirical findings, it may be argued that if the triple deficit holds true, fiscal policy
and monetary policy are higher in its ability to influence trade deficits and private saving and investment.
From a policy standpoint, such a piece of evidence implies that the causes of large and persistent external
deficits should be sought on the budget side and the structural of the economy. Several factors may explain
this, ranging from the structure of foreign trade, the exchange rate regime adopted, the international com-
petitiveness of the particular country in question, and the degree of capital mobility. Cases in which the
twin or triple deficits hypotheses would apply are more likely to occur in countries with economies that
are highly integrated with international markets, remain open to capital movements, and experience in-
tense international competition.

4.4.3. POLICY RECOMMENDATIONS

The outcome of the empirical studies suggests that, monetary policy guided by rules improves capital mo-
bility thereby guaranteeing a regular inflow of foreign savings to finance domestic investment. Our empir-
ical results suggest that a move to a rules-based monetary policy alleviates the burden of the twin/triple-
deficit effect for an effective fiscal policy.
Hence, the study recommends that the state should formulate fiscal and monetary policies that manage the
government’s revenue and expenditure and create conducive environment to encourage export and import
substitution as well as private sectors’ saving and investment. Addressing this deficit in the Fiscal account
will improve the economy’s current account balance; while at the same time reduce the country’s overreli-
ance on external funding as reflected in the country’s Savings-Investment imbalance. Fiscal Deficit can be
addressed through various ways. For example, having proper and efficient tax collection and administration
systems; putting mechanisms to safeguard the economy from macroeconomic shocks and improving the
terms of trade for SSA’s exports ((Lesiit, 1990)), and having fiscal discipline in form of good budgetary
 72

processes ((Wawire, 2006). Also, the findings indicate that Private saving-investment Balance increases
Budget Balance but is negative on the Current Account Balance, points towards insufficient savings in the
economy failing to fully finance investments, and thus resulting in a rise in the budget deficit. This can be
addressed by promoting economic growth, which will allow for increased savings and the promotion of a
saving culture.
The improvement in capital mobility is a prerequisite for effective fiscal policy and how the policy affects
external balance. This implies that inflation targeting as a monetary policy regime enhances the effective-
ness of fiscal policy and how it can be used to affect changes in the external balance of an economy. How-
ever, the effectiveness of inflation targeting in doing this depends on how well it deals with inflation expec-
tations. Introducing inflation targeting during times of high and volatile inflation could result in a persistent
fall in inflation expectations, short-circuiting the link from fiscal policy to external equilibrium or even
creating reverse causality from changes in the external balance to output, and thus the government budget
balance and public debt. if policy is aimed at attracting foreign capital, inflation targeting helps to boost
investor confidence by its ability to control inflation and its volatility. Controlling inflation variability puts
inflation expectations in check which is a powerful incentive for investment.
Moreover, the effectiveness of fiscal policy is enhanced within the economy. If government intends to use
fiscal policy to effect changes in the external balance and debt, monetary policy should be geared towards
controlling inflation expectations and improving capital mobility. Though monetary policy guided by rules
is believed to help achieve other objectives, such as ensuring higher output and employment, the emphasis
should be laid on its ability to control inflation if it is to be coordinated with fiscal policy. For inflation
targeting to accomplish these objectives, the right tools and targets must be included in the policy rule. The
appropriate measurement of inflation should be used as well as the appropriate short-term interest rate,
capable of effecting changes in response to inflationary or deflationary pressures. The policy rule should be
transparent and known to the public to avert any form of speculation regarding the form of the policy rule
being used. These efforts will assist the monetary policy authority to build a credible monetary rule which
will affect the expectations of actors of the economy.
Improvement in current account balance in the SSA countries requires fiscal policy but it by itself cannot
rectify the current account deficit because budget deficits are not fully exogenous, and policy controlled.
Fiscal adjustment itself requires adjustment in current account deficits. Since saving gap has a significant
effect on both current account and budget balance, policy makers need to give importance to policies such
as increase in per capita income growth and improved access to financial system etc. that may help increase
the saving rate in these countries.

In addition, a negative and bi-directional causal relation between current account and private saving-invest-
ment deficits indicates that the two accounts offset each other, in such a way that they promote a balance of
payment equilibrium. More importantly, authorities of this economy must put in place a fully disciplined
 73

fiscal policy that should ensure the drastic curtailment of fiscal deficits and, at the same time, create a
conducive environment to attract foreign remittances and foreign investment, which would help to generate
healthy external balances. In addition, exchange rate stability can promote the exports sector, minimizing
external imbalances through creating critical surpluses in current accounts and including related compre-
hensive discipline policies that may be pursued, which would enable the external, private saving-invest-
ment, fiscal, and monetary sectors to perform without creating adverse imbalances in this economy.
In the light of above conclusions and explanation, following Policy recommendations are summarized and
suggested by the study. These are.
First, the government should encourage perfect capital market integration by improving their relations with
other economies. It will help to attract Foreign Direct Investment and also increases the scope of investment
opportunities abroad. An increase in capital inflows will also boost up employment opportunities and wage
levels specially in developing economies. Fiscal policy is most important stabilizing mechanism available
to government for sustainable economic growth. So, government should design the effective fiscal policies
to create macroeconomic stability and augment economic growth.

Second, the findings of this study has shown a unidirectional relationship between the budget deficit and
the current account deficit, whereas budget balance and current account balance have a bidirectional
relationship with private saving-investment balance. This suggests that policymakers can use fiscal and
monetary adjustments which also address private saving, investment and external imbalances. SSA which
have shown evidence of the twin/triple deficit imply that policymakers must consider fiscal consolidation
(reducing deficit and debt accumulation). Fiscal consolidation includes measures such as efficient
spending monitoring; proficient revenue collection apparatus and restructuring the civil service. Fiscal
consolidation has proved to be helpful in many countries where it has been fully implemented. However,
lax fiscal adjustments are destined to fail. Fiscal strain can be controlled by reducing non-priority
expenditure, strengthening the revenue base and where feasible allowing flexible exchange rate.

Third, there is need for the government to put in place policy measures that encourage domestic savings.
This will have the effect of improving the budget deficit as well as help augment domestic investments.
Increasing total savings will play a significant role in ensuring a sustainable budget and Trade deficit.
Policies that can help encourage domestic savings include reducing the amount of credits by use of
interest rates, which is an important instrument for raising the savings rate, and monetary policy
instruments such as rediscount ratios and reserve ratios. In addition, the state should prioritize practices
such as financial sector restructuring especially with reference to sectorial credits and credit
arrangements. Towards this, the monetary policy authority has an important role to play. Fourth, the
central banks need to carry counter measures to manage either interest rate or exchange rates in response
to increase in government expenditure. For instance, when budget deficit increases, central bank can
increase money supply to reduce interest rate as counter measure. Interest rates are also important in
attracting foreign investment which finances the trade account deficits. High interest rate can negatively
 74

impact economic growth by way of reducing investment and consumer spending, and in parallel with
increasing savings. In this context, there can occur the saving paradox. Thus, any austerity measures or
cutback in expenditures should be undertaken via consumer spending to reduce the trade deficit. This
shrink in consumer spending in the short run makes significant contribution in reducing trade deficit by
boosting investments in the long run.

Fifth, the governments can be encouraged to borrow money on concession rates offshore to reduce the
need for domestic borrowing. Increased domestic borrowing crowds out domestic investment resulting
in a drop in GDP growth.

Lastly, there are also a number of factors which need to be improved such as addressing inadequate
infrastructure, high transport costs, product quality issues, regulatory and other constraints limiting
supply responses, and which improve the business environment. The other two options (drawing down
international reserves and external borrowing) are not feasible since the country is saddled with a large
international debt. In the long run there is need for government to develop new exports, primary products
beneficiation (value addition), use of nanotechnology and nurturing them.

Overall, evidence on the relationship between current account balance, private saving-investment
balance and budget balance is not exact hence complex and unclear for the majority of countries. This
relationship evolves over time depending on the dynamics of the economy. Bartlett (1999) also supports
the notion that the relationship between these balances is not consistent overtime. Again, given the
complexities that are intrinsic in mixed economies, it may not be probable to verify a firm and unwavering
relationship between these balances. However, there is neither a one-size fits all explanation for selected
countries nor ‘a silver bullet’ stratagem for any country. The solution might be a mixture of policies that
tackle the binding constraints faced by countries.
 75

CHAPTER FIVE: THE IMPACT OF THE TRIPLE DEFICITS ON ECONOMIC GROWTH IN

SSA COUNTRIES
5.1. INTRODUCTION
The development of the two-gap model [Chenery and Bruno (1962); Chenery and Strout (1966); McKin-
non (1964); and Weisskopf (1972)] was an important contribution to the literature of economic develop-
ment. The two-gap model deals the interactions between the savings constraint and the foreign exchange
constraint the determination of economic growth in an economy. The savings constraint refers the situation
when the growth of an economy is limited by the availability of domestic savings for investment, and the
foreign exchange constraint refers to the growth economy being limited by the availability of foreign
exchange for importing capital goods. More recently, there has been increasing interest in the three-gap
model, introducing fiscal constraint as a third gap limiting the growth prospects of highly indebted devel-
oping economies [Bacha (1990); Solimano (1990) and Taylor 1994)]. The fiscal constraint is intended to
reflect the impact of the availability resources to finance the public investment required to support a given
level of potential output. These constraints are selected for analysis because of their direct impact eco-
nomic growth of SSA countries.

We extract the salient features of three-gap models developed by Bacha (1990), Taylor (1990,
1993,1994), and Solimano (1990) and extend the theoretical framework developed by Mwega
et.al (1994) to examine the impact of current account imbalance, budget imbalance and private saving -
investment imbalance on economic growth in SSA. Such a formulation more realistically captures the
financing practices of the SSA economies under study. A major shortcoming of the Mwega et.al (1994)
and others approach of measuring potential output is the assumptions that time alone is the independent
variable, which determines the growth of potential output. Potential output is a future or desired phenom-
enon, implying · that some elements of dynamic specification is essential in such modelling. Thus, actual
output in the previous years will have influence on the growth of potential output.

The paper is structured as follows: Section 5.2 presents three-gap model; in Section 5.3, Data require-
ment.5.4 estimation methodology is presented; in Section 5.5, the empirical results are discussed; conclu-
sions are summarized in Section 5. 6.

5.2.THE THREE-GAP MODEL

In this study, the three-gap model developed by Mwega et al (1994) is used to examine the impact of
internal and external imbalances on economic growth in SSA.. The model is a blend of a three-gap model
(besides the traditional savings and foreign gap a fiscal gap is now introduced as a -potentially-additional
constraint to growth, see Bacha (1990), Taylor (1990 , 1993,1994), and Solimano (1990) with a
disequilibrium model extended to deal explicitly with capital formation and growth. A departure from
Bacha -Taylor- Mwega et al is that some key relative prices like the real wage and the real exchange rate
are not fixed but are allowed to change. The analysis distinguishes between different 'growth-regimes,"
 76

where the regimes are classified as demand constrained, and capacity constrained. Demand constrained,
in turn, describes the macroeconomic environment in which the growth process takes place. Analytically
this may prove to be a useful way to link the short and the medium run, and it is also intended to show
how different macroeconomic imbalances can affect the rate of growth in distinct ways, depending upon
the dominant constrained in the economy. The model first identifies which are the main macroeconomic
constraints to growth following the traditional distinction between the external constraint (or foreign gap),
the savings constraint (or internal gap) and the fiscal constraint. Bacha (1990), Taylor (1990,
1993,1994), and Solimano (1990) developed a simple three-gap model framed in a disequilibrium
setting. This model is calibrated with parameters for the SSA economy. They examined the effects
of various macroeconomic imbalances (e.g. an increase in current account imbalance, fiscal
imbalance and private saving-investment imbalance) on the rate of GDP growth. Like the
other three-gap models (Bacha,1990; and Taylor,1994), this model can also be presented
through a social accounting framework. However, A major shortcoming of Mwega et al (1994) and
others approach of measuring potential output is the assumptions that time alone is the independent vari-
able, which determines the growth of potential output. Potential output is a future or desired phenomenon,
implying · that some elements of dynamic specification is essential in such modelling. Thus, to solve such
shortcoming, we use the actual output growth in this study.

In this section we present the analytical framework for the study. First, some concepts are considered
crucial for a thorough understanding of the arguments we are trying to put across; we therefore provide
some clarifications of such concepts. Finally, we provide the theoretical link between resource gaps and
sustainable economic growth. The Clarification of the concept of three gaps are presented as follows.
5.2.1. INVESTMENT - SAVINGS GAP
The role of investment to influence the level of output is less controversial. It has been argued that under
a free enterprise economy, investment if well monitored will generate higher business profits, full em-
ployment and also make workers more productive. The concept of investment in economic theory refers
to domestic savings plus inflow of foreign resources or the change in capital stock, strictly defined, in-
vestment is expenditure on real capital goods. It is also taken to mean purchase of any asset, or indeed the
undertaking of any commitment, which involves an initial sacrifice followed by subsequent benefits,
(Thirlwall, 1983)
We may categories investment into two broad groups, notably physical capital investment and huinan
capital investment. By physical capital Investment it refers to the form of investment, which provides the
worker with more efficient material, notably equipment such as better machines, electrical power,
transport, among others. Human capital investment, on the other hand, refers to the form of investment,
which makes the worker himself more efficient, notably by making him more healthy, skillful and spe-
cialized ill knowledge.
 77

To finance any form of investment requires savings. Savings refers to the setting aside or postponement
of present consumption to future date. Thus, savings equal disposable income minus consumption. How-
ever, planned investment may not always automatically equal available savings. Any disequilibrium be-
tween the two would give rise to Investment-Savings gap, with its attendant consequences on unhealthy
social, political and economic conditions. By implication, every economy would thrive to achieve some
balance between savings and investment, as this is one of the factors that can move the economy towards
full employment and thereby close the gap between actual and potential output.
Assuming savings (S) is some function of national income (Y), such that we have the following:
S =sY.
Recall that investment (I) is defined as change in capital stock: I = AK. Capital stock bears a direct rela-
tionship to income Y, as expressed by the capital-output ratio, k, such that K/Y — k or ∆K = l ∆Y.
At equilibrium total saving will be equal to total investment, such that S = I
Substituting we have:
S = sY = or ∆K = k ∆Y = I or sY = k ∆Y
Dividing through by Y and K we obtained the following expression:
𝛥𝑌 𝑠
= 𝑘,
𝑌

The above expression states that rate of growth of GDP is determined jointly by the national savings ratio,
s and the capital-output ratio, k. However, the rate at which the economy grows for any level of saving
and investment depends on how productive that investment is. The productivity of this investment can be
measured by the inverse of the capital-output ratio, k, since this inverse 1/k is the output-capital ratio. It
follows that multiplying the rate of new investment, s = I/Y, by its productivity, 1/k, will give the rate by
which national income will increase. The productivity of capital is represented by p in our expression
below.
A country will achieve its target growth rate, such that the gap between actual output and potential output
closes or reduces, if there is a minimum level of domestic investment. To purchase the minimum level of
𝑔
domestic investment requires minimum savings, the ratio, which is calculated thus: 𝑠 = Where s -
𝑝

the required minimum saving ratio, g- target growth rate to close the output gap, p - productivity of capital.
If it is established that domestic saving is less than the level of investment required to achieve the target
growth rate, we would have as Investment - Savings (I-S) gap at time "t" defined by El Shibly and Tliirl-
wall (1981) and Thirlwall (1983) as:
𝑔
𝐼𝑡 − 𝑆𝑡 = 𝑠𝑌𝑡 = 𝑝 𝑌𝑡 − 𝑠𝑌𝑡 (5.1)

where
𝐼𝑡 - Investment requirement at time ’t' 𝑆𝑡 - Savings requirement at time 't'
𝑌𝑡 - Output at time 't' Other notations are as previously defined.
 78

The disequilibrium between Investment and Savings (I-S gap) arises essentially from the fact that both are
not necessarily carried out by the same economic agents. Also, investment and savings activities are in-
fluenced by different factors. Resource owners will be willing to forgo their present consumption if they
anticipate higher benefits from such resources when invested in productive activities. Thus, a resource
owner’s decision to save would then depend on the overall level of resources made available to him, as
well as the attractiveness of savings option relative to consumption. The overall level of resources refers
to the after-tax income (disposable income) of the resource owner. The relative attractiveness of savings
to consumption refers to the willingness to save or the marginal propensity to save. The willingness to
same would depend on some factors, notably, the rate of interest especially on deposit, the resource dis-
tribution as well as the effectiveness of institutions mobilizing savings. among others.
To specify savings model, a popular thing to do is to assume the existence of diminishing marginal utility
of expenditure. Thus, suggesting that additional welfare gains would become progressively less valuable
as consumption increases. Two implications are deducible from the above theoretical proposition. The
first is that at low level of income, savings would be low. The rationale being that a proportionately greater
amount would be consumed. The second implication is a corollary of the first. It is to the effect that, at
high level of income, income earners would be better disposed to savings, as the opportunity cost of ben-
efits forgone would be lower if basic consumption needs could be easily satisfied. Nevertheless, irrespec-
tive of the level of income, whether a resource owner would decide to postpone present consumption or
not, would largely depend on the degree to which the consumption/saving trade off can be made to look
attractive from the point of view of the resource owner.
5.2.2 EXPORT - IMPORT GAP (CURRENT ACCOUNT DEFICIT)
Production will take place if required resources are purchased. Some of these resources can be obtained
from within the domestic economy, while others can only be obtained from outside the domestic economy.
Those that are available within the domestic economy can be purchased with the domestic savings. Those
available outside the domestic economy would require foreign currencies to purchase them. The relative
importance of domestic inputs and foreign inputs would be better appreciated if we realize the non-sub-
stitutability of the two types of resources.
lnasmuch as the purchase of foreign inputs (imports) would require foreign exchange, how the foreign
currencies are obtained becomes imperative. They can come from three principal ways. These are by in-
creasing exports, compressing imports, as well as foreign assistance/aid. Each of the above options has its
own merits and demerits.
To increase exports would demand two things. The first is that, domestic output should be increased suf-
ficiently, such that more of the output can be released for export after satisfying domestic consumption
needs. The second is that export promotion strategies be put in place and be pursued with a view to matting
the country’s export more attractive. Domestic output would increase if further stressing of domestic re-
sources takes' place. This may not be advisable except the economy is not operating under full-
 79

employment. The underlining theoretical assumption for a successful export promotion is that both de-
mand for and supply of exports should be elastic. The implication is that demand for and supply of exports
should respond more than proportionately to the export promotion strategies. However, countries that are
noncultural and especially those exporting essentially primary products may find it difficult to meet the
above underlining theoretical assumptions. This is because monocultural economies are much more sus-
ceptible to cyclical downswings common with primary commodities in the international market. Also,
primary products command lower prices relative to their secondary and tertiary counterparts.
Compressing imports suggest the pursuance of Import Substitution Industrialization strategies; such that
what were hitherto imported should be sourced for within the domestic economy. However, there is a limit
to how much emphasis can be put on this, as countries in their infancy stage of industrialization may not
have a better option other than relying on imported intermediate inputs and capital goods. Most developing
countries, especially Nigeria, are still in their infancy stage of manufacturing, (Bamidele, 1995). Whether
a country would get foreign assistance or aid is outside the control of the country. The donor country
decides how much to give as grant, for what purpose as well as to whom. Most of the above decisions are
arrived at though political processes as less economic factors are given consideration, (Chenery, 1967),
The implications from the above analysis are two. To compress imports may not be a very attractive option
for developing countries. Also, foreign assistance may not also be reliable. Thus, a country would have to
rely on its exports to purchase required imports for investment purposes. Two, if an economy would max-
imally explore the inherent benefits from exports and imports, such an economy should be opened up to
international trade.’ Gould and Ruffin (1993)argued that a country open to international trade might ex-
perience faster technological progress and increased economic growth. They anchored their argument on
the fact that the cost of developing new technology falls, as more hi-tech goods are available. The concept
technology refers to the sum of knowledge of the means and methods of producing goods and services.
Technology is not merely applied science because it often runs ahead of science. Technological change
and the diffusion of technology are important in economics because new methods, including those em-
bodied in investment, play an important part in theories of economic growth. There is, however, some
controversy about the extent to which technological development is an autonomous factor in economic
growth. The reason for the above is simply due to difficulties in measurement.
If, however, the import requirements were higher than the level of export earnings available for investment
purposes, we would have Export - Import gap or foreign exchange gap. Reducing importation of interme-
diate inputs and machines could lead to low production and hence low output for the economy thereby
creating an output gap. El Shibly and Thirlwall (1981) and Thirlwall (1983) has given the Export-Import
gap at time ‘t’ as:
𝑟
𝐸𝑡 − 𝑀𝑡 = 𝑖 ∗ 𝑌𝑡 − 𝑖𝑌𝑡 = 𝑚 𝑌𝑡 − 𝑖𝑌𝑡 (5.2)

where
𝐸𝑡 - required export, 𝑀𝑡 - required import, i - the ratio of import to output
 80

i" - the required ratio of import to output. m - productivity of imports

r- target growth rate to close the gap. Other notations are as previously defined.
5.2.3 BUDGETARY GAP (BUDGET DEFICIT)
The mechanism of government intervention in the economy operates through public budget. Budget is an
estimate of income and expenditure for a future period as opposed to an account, which records financial
transactions. Budgets are an essential element in the planning and control of the financial affairs of a nation
or business and made necessary essentially because income and expenditure do not occur simultaneously.
Thus, public budget contains revenue estimates, expenditure decisions of government as well as fiscal
actions of government within a specified period usually twelve calendar months.
We may consider the government as a collective agent, such that it is the organ though which the collective
choice of individuals can be made especially with regard to provision of social goods, augmentation of
merit wants, and other public duties. To execute the above programmes, government obtains funds (reve-
nue) from inhabitants of the country and the private sector in the form of taxes, charges, borrowings,
grants, and profits, among others.
Tax is a compulsory levy imposed by government on individuals, firms, goods and services as well as
legal entities within a country. Taxpayers have to pay the sums irrespective of any corresponding return
of services or goods from government. Thus, a taxpayer does not receive a definite and direct quid pro
quo from government. By implication it is not a price. However, it is possible that taxpayers get benefits
from government, but this is not a right. Tax is imposed on income, wealth or services, etc. However,
intention of government will largely determine whether a compulsory levy is a tax or not. This suggests
that all taxes are compulsory levies, but not all compulsory levies are taxes.
Charges are usually payments that are made for the use of government services. Borrowing, as the name
indicates is the use by the government of fund belonging to another economic agent with the understanding
that the fund will be repaid. Borrowing may be internal or external and both principal and interest accru-
able will be paid back at a later date. Grants are funds made available by other countries or international
agencies and are not supposed to be paid back. A profit as the name indicates refers to the excess of
revenue over the costs on government direct investment.
Charges, borrowing, grants and profits are voluntary, but taxes are compulsory. Taxes account for the
largest proportion of public revenue. Taxes may be imposed on production; they may be imposed on
households or firms and they may be imposed on the sources or uses of the taxpayers' account.
Public expenditure refers to the expenses, which the government incurs for its own maintenance and also
for the society as well as the economy as a whole, Expenses incurred in rendering assistance to other
countries are also part of public expenditure. Public expenditure, in Nigeria, is grouped into four functional
groups, notably Administration, Economic Services, Social and Community Services as well as Transfers.
Administration expenditure, we mean expenditure incurred on day-to-day running of government busi-
nesses. Examples are expenditures on general administration, defense, internal security, etc. Economic
 81

Services on the other hand, refer to expenditure by government’s direct participation in the economy,
principally to propel the economy forward, they include government's participation in transportation, ag-
riculture and natural resources, communication, mining, quarrying and special projects. Social and Com-
munity Services mean expenditure incurred on provision of public goods or semi-public goods. Examples
are provision of education, health, houses, streetlights, roads, bridges, dam, etc. Transfers refer to pay-
ments not made in return for some productive service, e.g. payments made by the state to needy individuals
which, in effect, transfer income from wealthier sectors to the poorer. Examples are old-age pensions,
unemployment benefits and widow’s pensions. They are not a payment in return for productive services,
but rather represent income redistribution. likewise, taxation is a transfer payment to the government.
Dis-equilibrium may occur between public revenue and expenditure. This is what we refer to as budgetary
gap. This may either be positive or negative. It is positive where revenue exceeds expenditure, such that
the economy operates a surplus budget. This will increase the reserves of the country and does not pose
much economic problem. However, where revenue lags behind expenditure we have deficit budget. This
may pose a serious problem to the economy, because accumulation of deficit leads to public debt with its
' attendant consequences. The dis-equilibrium may either be voluntary or involuntary. It is voluntary if the
government out of its own volition decides to bring about the dis- equilibrium, while it is. involuntary
where exogenous circumstances impose the dis- equilibrium on the government, ().
The involuntary dis-equilibrium would be better appreciated if we recognize that government, unlike in-
dividuals, has a rather unique way of going about its revenue and expenditure activities. For example,
while individuals would prefer to be in possession of the means to effectively demand for commodities,
government usually adopts a rather opposite approach. That is, government would estimate what to spend
before sourcing for the revenue to finance such expenditure programmes. By implication, government
would most likely wish to ensure that its expenditure programmes are not faltered and hence look for
funds elsewhere.
The way out of the above, most times, has been deficit financing. This may be defined as the net increase
in the amount of money in circulation, where such an increase results from a conscious governmental
policy designed to encourage economic activities (for economic growth or some other objectives), which
would otherwise not have taken place. Defined this way, deficit-financing amounts to domestic credit
creation, which is not offset by increased taxation, more restrictive bank credit policy and similar defla-
tionary measures.
Measurement of the Budgetary Gap (BG) is given as follows: G-T where G= Total Public Expenditure,
T Federal Government Retained Revenue
The concept Federal Government Retained Revenue comprises of some items. The items are share of
statutory allocation, independent revenue, deduction on loans to state and local governments and others.
The above specification can either be deficit, balance or surplus.
 82

5.3. THE THEORETICAL LINK BETWEEN THE THREE GAPS AND SUSTAINABLE
ECONOMIC GROWTH
Most Less Developed Countries fail to achieve sustainable economic growth because of the absence of
the pre-requisites for same. Sustainable economic growth involves two basic things. One is achieving
economic growth. Two, is ensuring that the growth continuously optimizes the use of natural capital with
an intergenerational dimension, i.e. meeting the needs of the current generations without having to com-
promise those of the future generations. Sustainable economic growth, if achieved, reduces to the barest
minimum the resource gaps which hitherto act as constraints. The essence is to make the potential and the
actual output as close as possible. Three groups of factors have been identified in the management of
resource gaps to ensure sustainable economic growth, (Khan, 1995).
The merits of articulating the link between resource gaps and sustainability are many. The merits are:
(i) it ensures development of appropriate methodological tools to appraise projects and assist invest-
ment and planning decisions by weighing equally the requirements of sustainability; and
(ii) it also puts in place an appropriate sustainable economic growth idea to measure growth in the
context of an integrative framework.
Economic growth will be limited and not sustained if the steady state conditions are not fulfilled. The
steady state growth occurs when the proportional rates of growth of all variables in an economy are con-
stant over time, Thus, suppose we have an economy consisting simply of firms and households, where
firms receive consumers’ expenditure in return for goods and pay this back to households in return for
productive services. If households spend part of what firms pay them on consumption expenditure and
save the rest; and firms spend on new investment exactly the same amount as households save. If, now,
national income, consumption expenditures, saving and investment expenditures were all growing at say,
6.0 per cent per annum, every year without change, then we would say that the economy was in steady-
state growth. The steady-state situation is generally the starting point in the analysis of most models of
economic growth.
The variables of concern for economic growth to take place usually include capital accumulation, labour
force expansion and technical progress. Thus, we may specify our production function as
f(K, L,, T) (5.3.)
where Ys = net output, K = capital inputs L = Labour inputs, T = Level of technology.
A look at the above formulation will make it rather clear that they are the four independent variables that
are crucial for growth in output. Nevertheless, we how that capital accumulation (K) is an essential and
endogenous element in all growth models. The other factors labour (L), research (R) and technology (T)
may be considered as exogenous influences, (Allen, 1973). In other words, while the growth in labour
supply, research and technical progress may provide the momentum for economic growth; capital accu-
mulation is the vehicle of economic growth.
Thus, taking the two leading variables in our model: Ys and K, the growth rates are:
L∆Y=∆1ogY. (5.4.)
 83

similarly,
L∆Y =α+ log K .......................................................(5.5)
Then net investment is simply the increase in the capital stock, this we have:
I=∆𝐾
where
I = net investment. such that
𝐼 1
= 𝑘 ∆𝐾 = ∆1ogY ,. (5.6.)
𝑘

represents both investment as a proportion of capital stock and the rate of growth of capital stock. We look
for steady - state growth when Ys and K grow at a constant proportional rate, the same rate for each
variable. If the common rate is g, called the warranted rate of growth or growth rate of potential output of
the steady - state situation, then, we will have:
∆1og Y3 = ∆log K = g. ............................................ ........ (5.7)
Giving on integration
logY = constant + gt........... . .. ................, .... . . .............(5.8.)
that is 𝑌3 = 𝐴ⅇ 𝑔𝑡
Where A = a constant of integration, t = time
similarly
𝐾 = 𝐴ⅇ 𝑔𝑡 ..... .. . .. (5.9)
Thus, given the initial values Y3o and Ko at time t = 0, we will have:
𝑌 = 𝑌0 ⅇ 𝑔𝑡 𝐾 = 𝐾0 ⅇ 𝑔𝑡 (5.10)
𝐼
Furthermore, since 𝑘 is also equal to g, we will then have:

𝐼 = 𝐼0 ⅇ 𝑔𝑡
where
𝐼0 = 𝑔𝐾0 (5. 11)
The theoretical argument we are putting across from the algebraic manipulations above is that for eco-
nomic growth to occur investment is crucial. Also, investment will take place if required capital is pur-
chased. To purchase the capital, requires absence or significant reduction in the resource gaps. The ra-
tionale for the above lies in the following.’ To purchase capital within the domestic economy requires that
the gap between domestic saving and investment be closed. Also, to purchase capital that is foreign based
requires foreign exchange, which necessitates the reduction of the gap between export and import. Fur-
thermore, in a situation whereby public sector is a major factor promoting economic growth, the reduction
of the budgetary gap is crucial. Therefore, it requires savings from the private, government and foreign
sources.
In a typical developing economy, four flows are identified which include private savers and investors, the
financial system, the government, and the foreign sector. It is assumed that private saving is channeled to
 84

higher bank deposits, increases in the stock of narrowly defined money, or asset holdings abroad through
capital flight. The financial side of the economy is treated as a pure credit banking system. This can be
clearly seen in the Chenery and Strout Two-Gap Model.
Following assumption of Rostow (1956) let Ap be bank assets (credit, loans) advanced to the private sector,
Agbe advances to the government, and eRbe the values of foreign reserves. Bank liabilities are deposits
(D) and narrow money (M). Therefore, the banking system balance sheet is;
𝐴𝑃 + 𝐴𝑔 + ⅇ𝑅 = 𝐷 + 𝑀 (5.12)
Analyzing equation (1) in flow terms,
𝐴𝑃 + 𝐴𝑔 + ⅇ𝑅 = 𝐷 + 𝑀 (5.13)
Where a “dot” over a variable denotes its time derivatives, Savings (s) is some constant proportion, s, of
national income (y) such that
𝑆 = 𝑠𝑌 (5.14)
But Investment (I) defined as a change into the capital stock, K, is in accordance with the acceleration
principle measured and assumed to be a constant of the rate of growth of output so that,
𝐼 = ∆𝐾 = 𝑘(𝑌𝑡 − 𝑌𝑡−1 ) (5.15)
𝐼 = ∆𝐾 = 𝑘∆𝑌 (5.15a)

Where k stands for the marginal capital coefficient or simply the incremental capital- output ratio
(ICOR) and ∆ indicates a change in the variable.
Dividing equation (3) by Y yields.
𝐼 𝛥𝑘 𝛥𝑌
= =𝑘 (5.16)
𝑌 𝑌 𝑌

Rewriting (equ. 4) by substituting ∆𝐾 for I,

𝐼 𝛥𝑌
=𝑘 (5.16a)
𝑌 𝑌
𝛥𝑌
Rewriting 𝑌 as g, the rate of growth of output can be expressed as
1
𝑔=𝐾 (5.17)

On the assumptions that k is fixed, based on the assumption that the production function is of fixed
proportions, the only constraint to growth is investment capital, hence, is seen as a bottleneck to growth
(Chenery and Strout, 1966). Real Investment I in turn is the sum of 𝐼𝑃 + 𝐼𝑔 (private and public capital
formation). Let 𝐼𝑃 = 𝑘𝐾 be the government’s investment decision. Its own capital formation is set as a
share of k of the total capital stock. The value of private investment is (g -k)PK and we assume that the
banks issue a new loan 𝐴𝑃 finance the increase in private capital
𝐴𝑃
= (𝑔 − 𝑘) (5.18)
𝑃𝐾

The overall investment function determining g is presented as a reduced form (depending on credit
availability among other factors). Let real government current spending G+γ K
 85

Where γ as a proxy for the non-investment fiscal deficit or government dis-savings. The government also
borrows abroad. Its outstanding stock of loans is F upon which it pays an interest rate r. If the debt
growth rate F exceeds the interest rate r, the government turns to the banks to finance the part of its
spending it cannot cover with foreign loans expressed below.
𝐴𝑔
=𝑘+𝛾−𝑇 (5.18a)
𝑃𝐾

Domestic borrowing can be a substantial share of GDP when T is less than zero. Debtor country
governments owe large foreign obligations but do not own the resources to generate foreign exchange to
pay. The outcome is a severe fiscal crunch. Finally, on the bank asset side, the increase in reserves is
𝑒𝑅
= 𝑇 + 𝜀 − 𝑎𝑢 − (1 − 𝜃)𝑔 − 𝑄 (5.19)
𝑃𝐾
𝐸
Where𝜀 = 𝐾 is the ratio of exports (net of competitive imports) to capital stock, Imports go only for

intermediates goods ( au ) and capital goods (1 − 𝜃)𝑔 , and Q stands for net acquisition of foreign assets
(relative to the capital stock) by the private sector.
To describe the saving gap, we assume that saving desired by nationals in the absence of price inflation is
directed either to foreign asset accumulation or else to increased bank deposits, such that;
𝐷
𝑄 + 𝑃𝐾 = ⌈𝛱𝑠𝛱 + (1 + 𝛱)(1 − 𝜃)𝑠𝑤 ⌉𝑢 = 𝑠(𝛱, 𝜃)𝑢 (5.20)
𝑒𝑎
where 𝑠𝛱 and 𝑠𝑤 are the saving rates from profit and wage income flows respectively.𝜃 = (𝑒𝑎+𝑊𝑏) is the

share of intermediate imports respectively in variable cost, so that(1 + 𝛱)(1 − 𝜃)u is the ration of the
wage bill to the capital stock. If the share λ of saving flows abroad, the equation above becomes.
𝐷
= (1 − 𝜆)𝑠𝑢 (5.21)
𝑃𝐾

Putting equation (2) and (10) together shows that the increase in the supply of narrow money is
𝑀
= 𝑔 + 𝛾 − (1 − 𝜆)𝑠𝑢+⌈𝜀 − 𝑎𝑢 − (1 − 𝜃)𝑔 − 𝑄⌉
𝑃𝐾
𝑒𝑅
𝑔 + 𝛾 − (1 − 𝜆)𝑠𝑢+⌈𝑃𝐾 − 𝑇⌉ ( 5.22)

where the terms in the bracket represent the balance of payments (net of capital flight) on current and
capital account. Relative to capital stock, we assume that private investment demand is given by the
function.
𝐼𝑃 𝐼𝑔 𝐷 𝐴𝑔
= 𝑔0 + 𝑎𝑢 + 𝛽 ( 𝐾 ) + 𝜑⌈(𝑃𝐾 ) − (𝑃𝐾 )⌉ (5.23)
𝐾
𝐼𝑔
The term au is an instantaneous accelerator, 𝐾 shows that public investment crowds in private capital for-
𝐴𝑔
mation because of complementarities and other external effects. the last term in equation (15) intro-
𝑃𝐾

duces financial crowding-out as a potential counterpoise to direct crowding-in. It is assumed along con-
ventional lines that investment is cut back when government puts pressure on financial markets.
 86

𝐴𝑔
Specifically, IP falls when a new government borrowing𝑃𝐾 grows with respect to the deposit increase
𝐷
. The rationale could be that banks raise interest rate and tighten credit limits more of their deposit
𝑃𝐾

base is absorbed by the government. This simple flow specification is dimensionally equivalent to the
quantity theory of money demand and saves the use of asset stock.
Uniting equation (7), (10) and (12) in its simplified form.
−𝑔 + 𝑔0 + ⌈𝛼 + 𝜑(1 − 𝜆)𝑠⌉𝑢 + (1 + 𝛽 − 𝜑)𝐾 − 𝜑𝛾 + 𝜑𝑇 = 0 (5.24)
Thus, relationship shows that the capital stock growth rate g increases endogenously in response to greater
capacity utilization u due to the accelerator and also because higher private saving creates deposits which
banks use to finance investments. Government investment 𝑘 has an overall crowding-in effect if 1 + 𝛽 >
𝜑 This condition will be satisfied if 𝛽 ≥ 1 And if 𝜑 ≥ 1 Also, an increase in foreign transfers T or a
reduction in government dis-savings 𝛾 cuts back on public borrowing from the banks, again permitting
private investment to rise. A third gap, which is the fiscal gap, was also explored for its independent
constraining influence on the growth of an economy. To an extent, a fiscal gap can directly constrain
economic growth, influence the trade gap and therefore make an indirect impact on the economy. However,
the fiscal constraint in a three-gap model is often predicted on the assumptions that the public and private
investments (Blejer and Khan 1984; Barro 1989) are complementary to each other and the savings are
forced through inflation. Within the three-gap simulation analysis of foreign resources flows, Taylor
(1993) convincingly concludes that a ‘substantial realignment in international payments would be required
to accelerate the growth for developing nations overall’.
Employing these assumptions in a two-gap framework, Bacha (1990) argues that foreign transfers are
more important than foreign savings. This explains why the IMF and the World Bank continues to
emphasize on how to contain the inflationary pressures through light fiscal and monetary policies. In less
developed countries (LDCs), there exists a market for government bonds however thin it may be while
the governments continue to resort to a partial monetization of budget deficits, they also access market
borrowing thorough open market operations.

5.4 EMPIRICAL MODEL SPECIFICATION AND ESTIMATION METHOD

5.4.1. EMPIRICAL MODEL SPECIFICATION
For the econometric analysis, we specified a three-gap model. The specification was anchored on the
endogenous growth model. In this regard an aggregate production function in the endogenous growth
model form for a representative economy as postulated by Romer (1986 and 1994) was adopted. Thus we
have:
Y = (R, K, L, F) ......, ......... ................ (5.25)
where
Y= total output produced by the economy (GDP).
 87

R = research and development carried out by economic agents in the economy. K = the accumulated
capital stock.
L = the accumulated stock of human capital. F = other factor inputs.
The gross domestic product (GDP) is used as proxy for level of total output. The’ level of total output in
any economy could be influenced by the capital stock and productivity of the labour forc as well as the
technology.
Total desired investment is divided into two: private sector investment represented by𝐼𝑃 and public sector
investment represented by 𝐼𝑔 . Total capital expenditure of the Federal Government is used as proxy for
public sector investment, Private sector investment is simply calculated by subtracting public sector in-
vestment from gross investment. This is given as: 𝑙´ − 𝐼𝑔
Investment may either be intended or unintended. Investment is intended if it is the amount that the inves-
tors planned for. On the other hand, it is unintended where the investment recorded is the change in busi-
ness inventories due to discrepancy between aggregate supply and aggregate demand. However, unin-
tended investment could generate unstable equilibrium. Thus, the total gross investment as given above
referred to planned investment that is simply given as:
: 𝑙´ = 𝐼𝑔 + 𝐼𝑃
Public sector investment refers to such investment that takes place in a whole economic set up, where the
government performs, in the sense of procurement and usage or application of funds and other economic
resources for the satisfaction and promotion of improved welfare of the citizenry. The concept private
sector investment on the other hand, refers to such investment that takes place in that compartment in a
whole economic set-up, where the citizens acting as individuals or organizations pursue their economic
activities for the purpose of profit maximization.
Investment expenditure is an important component of the national income. In fact, it is the second largest
component of the national income next to consumption expenditures. Thus, from the national income
accounting identity for an open economy, we have:
𝑌 =𝐶+𝐼+𝑋−𝑀 (5.26)
where
Y = total output produced by the economy/ GDP.
C = aggregate consumption expenditures being the sum of private and government consumption (C0 +
Go)
I = investment expenditures.
T = total exports of goods and services. M = total imports of goods and services.
(T-M) = net exports or external trade balance.
Solving for I from equation (5.25) gives:
𝐼 = 𝑌 − 𝐶 − (𝑋 − 𝑀) (5.27)
 88

To finance desired investment I, requires savings. A saving is defined as the postponement of present
consumption to future date. Thus, a saving is equivalent to disposable income less consumption expendi-
tures. It is assumed that savings could come from two sources, notably domestic savings (DOMSAV) and
foreign transfers/foreign savings (NFT). The domestic savings could also be divided into two: private
savings and public savings.
S = SD + NFT..................................., (5.28)
SD=Sp+Sg
A domestic saving (SD) is simply defined as the difference between aggregate income and aggregate
consumption. It includes the following savings items (i) savings and time deposit with commercial banks;
(ii) national provident fund; (iii) federal savings bank; (iv) time deposit with merchant banks; (v) premium
bonds, savings certificates, and savings stamps; (iv) life insurance funds; (vii) other deposits notably with
peoples bank and community banks. The sum total of (i-vii) gives total savings, which is the proxy used
for domestic savings, (SD). This is given as
SD = Y - C.......................................... „ ...........................(5.29)
The level of investment in any economy depends on factors such as the growth of actual output, , the level
of domestic savings, the difference between potential output and actual output which is simply referred to
as output gap (YG), the prevailing rate of interest especially the lending rate of interest, and the level of
capital imports. Theoretically, a positive relationship is expected between. investment on one hand and
growth rate of actual output, savings and capital imports, on the other hand. It has been argued that invest-
ment determines the rate of growth of output out of which additional savings takes place, Nalo (I 993).
Lewis (1954) argued that a central problem in the theory of economic development is to understand the
process by which a community, which was saving 4 to 5 per cent of its national income, converts itself
into an economy where voluntary savings are raised to about 12 to 15 per cent. By implication the role of
savings in output growth can be seen through its impact on investment. Furthermore, as the level of output
/income increases the marginal propensity to consume reduces thereby releasing more income for savings.
It is less controversial that all developing countries require capital imports. In fact, without the imported
capital goods, production may be very low or in some cases rather impossible.
However, a positive relationship is expected between investment and output growth, whereas, a negative
relationship is expected between investment and rate of interest. A higher output gap will induce increased
level of investment. Thus, as the level of investment increases, the level of output increases and the output
gap consequently reduces. Also, high rate of interest rate especially lending rate usually makes it unattrac-
tive to borrow fund and invest. We decided to use lending rate, which reflects market situation better. The
lending rate of interest refers to the price charged on loans granted to customers, by commercial banks
and other financial institutions.
Existence of investment activities implicitly suggests that savings take place. Existence of government
sector suggests that revenue is raised, and expenditure is undertaken, while openness implies existence of
 89

imports and exports. Thus, the three assumptions implicitly suggest that the three resource gaps co-exist.
The excess of exports over imports is our export-import gap (EMGAP) and is equal to trade balance. Total
import is divided into two different categories. These include capital imports, and non-capital imports.
The rationale behind this division is because we realized the fact that not all aspects of total imports could
have positive impact on economic growth.
It has been argued that in developing countries trade balance, which is determined by world demand (Nalo,
1953). Apart from the capital imports discussed above, the export-import gap will also be influenced by
some other factors, notably the output gap and growth rate of output. A major objective of every economy
is to achieve balance of payment equilibrium. This will be feasible if the domestic economy produces
sufficient output to release significant proportion to export and also reduces import. The implication from
the above is that the output gap in the economy should be small and the growth of the economy should be
high enough to sustain the balance of payment equilibrium. Thus, it is expected that output gap (YG)
should have negative relationship with Export-Import gap while the growth rate of output (GGDP) will
have positive relationship with the export-import gap (EMGAP).
We assume that public budget is an ex-post phenomenon. Thus, the Budgetary gap is simply defined as
the discrepancy between what the public sector’ budgeted to spend and what it actually spent. This de-
pended on the quantity of fund from which the public sector could borrow to address the imbalance. It
should be noted that the broad objective of the study is to examine the implications of macroeconomic
imbalance (Investment-Savings imbalance, current account imbalance and Budget imbalance) on eco-
nomic growth in SSA. To achieve the objective, we incorporate the three macroeconomic imbalances in
an equation. From the theoretical background, the three macroeconomic imbalances combined to cause
the output growth (economic growth-GDPG). Therefore, we expect positive relationships between each
of the resource gaps and GDP growth. Thus, we have:
GDPG= 𝜆0 + 𝜆1 𝐵𝐵ⅈ𝑡 + 𝜆2 𝐶𝐴𝐵ⅈ𝑡 + 𝜆3 𝑆𝐼𝐵ⅈ𝑡 ... .. . .................. ...........(5.30)

5.4.2. ESTIMATION METHOD

5.4.2.1. INTRODUCTION
The estimation of long run relationships is important in the empirical application of
macroeconomics models. Long run relationships describe the steady state solution and how a
change in the steady state affect the long run relationship between variables. In an econometric
model, estimating a long run relationship implies to estimate coefficients which capture this
relationship. A standard model in panel econometrics is the error correction model, in which
one term of the equation captures the short and the other the long run movements.
In this section we employed panel data estimation method to examine the impact of savings, trade and
fiscal gap on economic growth of 35 SSA countries from 1980 to 2018.
 90

This analysis is carried out within a panel data estimation framework. The preference of this estimation
method is not only because it enables a cross-sectional time series analysis which usually makes provision
for broader set of data points, but also because of its ability to control for heterogeneity and endogeneity
issues. Hence panel data estimation allows for the control of individual-specific effects usually
unobservable which may be correlated with other explanatory variables included in the specification of the
relationship between dependent and explanatory variables (Hausman and Taylor, 1981).

5.4.2.2. DYNAMIC PANEL VAR

This section starts out by attempting to produce a basic general way to estimate short run and long run
effects in a regression model. The methodology explained below is aimed to introduce the main method-
ology that are used when estimating the growth impact of macroeconomic imbalances. After establishing
this general way, the paper presents some basic econometric foundations and necessary assumption needed
to understand why certain model choices are made
Recent research regarding the growth impact of macroeconomic imbalances has been diverse in the choice
of methodological approach to the estimation of relationship between variables of interest. Evident from
the theoretical framework, there are numerous ways in how to use econometric methods to compute short
and long run effects of macroeconomic imbalances on economic growth.

We use panel VAR techniques to examine the relationship between economic growth and macroeconomic
imbalances in SSA countries. There are mainly two advantages in using panel VAR model: a) allows
addressing the endogeneity problem and b) overcome the data limitation problem. A Panel VAR approach
also allows for individual heterogeneity and improves asymptotic results. The results provide advanta-
geous insights which go beyond the estimated coefficients, reporting the adjustment and flexibility of
unexpected production shocks as well as the relevance of other different shocks.
In recent studies VAR models applied to panel section data are increasingly popular among research-
ers(Abrigo and Love ,2015). This study benefits from the PVAR methodology evolved by Abrigo and
Love (2015) by modelling the endogenous behavior between current account, fiscal balance and private
saving -investment balance. The PVAR approach combines the traditional VAR approach, treating all the
variables in the system as endogenous, and the panel-data approach, allowing for unobserved individual
heterogeneity by introducing fixed effects, resulting in an improved consistency of the estimation (Love
& Zicchino, 2006).
Abrigo and Love (2015) presents a short example in the practical uses of the pVAR approach on a system
of linear equations based on a panel VAR of lag order p.

(5.31)
Which is in a way, just another form of a dynamic model, presented as below

(5.32)
 91

The lag length on the instruments used in the model specification are broadly chosen first as a general case
where the estimations are robust to changes in the instrument lag length. Longer lags may give more
efficient estimates but as the length of the lag grows longer, the observations left to estimate the model
becomes fewer. With a very long-time dimension there could be a much longer lags on the instruments
but in my model with only 39 time periods I cannot choose too long lags on the instruments. I specifically
limit my sample so that a change in the lag lengths does not affect the size of the sample.
The statistical test that I should anchor my decision against is the Hansen’s J statistics which tests the null
hypothesis that the model specification is overidentified. If the specification does not overidentify or does
not identify at all then the chosen lag structure is acceptable. The Hansen’s J statistics tests the joint null
hypothesis that the included instruments are valid instruments in the sense that the excluded instruments
are correctly excluded and that the included instruments are uncorrelated with the error term (Hansen,
1982).

A) MODEL SELECTION CRITERIA

The first step of the empirical analysis was to choose optimal lag order in PVAR and in the moment
condition (Abrigo & Love, 2015.The choice of the order length on the pVAR is anchored to the tests of
moment selection criterions based on the work of Andrews and Lu (2001). These criterions consist of a
vector construction that aims to minimize the modified Akaike information criterion (MAIC), the modified
Bayesian information criterion (MBIC) and the modified Hannan-Quinn information criterion (MQIC).
The difference between the conventional information criterions and the modified criterions are basically
that the modified versions are based instead of a normal likelihood function, on quasi-likelihood functions,
see McCullagh and Nelder (1989, p. 325).
Michael and Inessa (2015) also propose an overall coefficient of determination (CD) which is supposed
to capture the proportion of the total variance which the pVAR model reproduces. The lag length of the
panel VAR is chosen simply through minimizing the CD, MAIC, MBIC and MQIC. If the criterion tests
are inconclusive, that is if there is no actual lag length that has the lowest information criterion on all the
different criterion models, then emphasize is put on the MAIC suggested by Serena and Perron (2001).

In the Panel VAR framework, it is important to impose some restrictions to make sure that the underlying
structure is the same for all the cross-sectional members. In practice, such constraints are likely to not be
respected; one can resolve this problem by using fixed effects to allow for individual heterogeneity in all
the variables. However, the conventional approach of average differentiation, commonly used to remove
fixed effects, can lead to biased coefficients because the fixed effect hypothesis means that the individual
specific effect is correlated with the independent variables. Therefore, to overcome this problem we use
forward mean-differencing, also known as the Helmert procedure (Arellano and Bover, 1995). In this
procedure, to remove the fixed effects, all variables in the model are transformed in deviations from
 92

forward means. This procedure gives more weight to data from the early period and allows no
transformation on the last one since no future observation is available. The same transformation is applied
on the error vector; In fact, with the hypothesis of neither autocorrelation nor homoscedasticity, the
procedure does not modify its features.
Moreover, the differencing could also result from a simultaneity problem since the lagged regressors are
correlated with the differenced error term. In addition, heteroscedasticity may also exist due to the pres-
ence of heterogeneous errors with different countries in the panel. Accordingly, after eliminating the fixed
effects by differencing, the generalized method of moments estimator (GMM) using lagged regressors as
instruments is applied to estimate the coefficients more consistently.
In our model, we assumed that the residuals vector was independent and identically distributed. However,
this assumption typically fails in practice, as the concrete variance-covariance matrix of errors is unlikely
to be diagonal. Thus, to isolate the shocks on one of the VAR errors, it is necessary to decompose the
residuals so that they become orthogonal. According to Sims (1980), the variables in VAR should have a
recurrent causal order based on their degree of exogeneity. This procedure is also known as the Cholesky
decomposition of the variance-covariance matrix of residuals and ensures the orthogonalization of the
shocks. In other words, the variables that come earlier in the order affect the subsequent variables at the
same time and with a lag, whereas the variables that come later only affect the previous variables with a
lag (Love and Zicchino, 2006).
Based on the model selection criteria, we specified the correct lag length and we made an estimation using
the PVAR with the selected lag length. Once all the coefficients of the panel VAR are estimated, we
compute the impulse response functions (IRFs) and the variance decompositions (VDCs). The special
feature of the use of VAR is that it allows to draw the impulse response function (IRF) and the variance
decomposition of the error (FEVD) to identify the shocks.
B) Impulse response functions
An impulse response function (IRF) is simply an illustrative procedure as to show how a stable model in
equilibrium react to an innovated chock to any of the included regressors. This impulse, be it temporary
or permanent dissipates through the model and shows how the response variable returns to equilibrium
after the disturbance. This is simply a method of making the output of the regressions intuitively
understandable and furthermore enabling us to calculate the long run multiplier effect.
The IRF has the convenient interpretation such that the first effect of an impulse, the impulse at time t is
just simply the impact multiplier of that variable. The autoregressive model runs the impact of the impulse
over the horizon and shows how the response variable returns to equilibrium. All the individual effects
that occurs at each time is summed up and forms a cumulative response of an impact that occurs at time t.
This cumulative response is simply the long run multiplier of an impulse that occurred at time t. The
impulse response functions (IRFs) are calculated by counting on the Cholesky decomposition. They de-
scribe one variable's reaction in reply to changes in another variable in the system, as all other shocks are
 93

held equal to zero. However, to isolate shocks to one of the variables in the system, it is essential to
decompose the residuals using a method by which they turn into orthogonal, because the actual variance-
covariance matrix of the errors is unlikely to be diagonal. The usual convention is adopting a particular
ordering, and then any correlation between the residuals of any two elements is allocated to the variable
that is coming first. The identifying assumption is that the following variables are affected simultaneously
by the variables that appear earlier in the ordering, as well as with a lag, whereas the variables coming
later affect the preceding variables only with a lag. That is to say, in the system, the variables coming
earlier are more exogenous, while the variables coming later are more endogenous. Finally, to analyze the
impulse response functions an estimation of their confidence intervals is needed. The standard errors of
the impulse response functions need to be considered, because the matrix of IRFs is built from the esti-
mated VAR coefficients. Consequently, the standard errors of the impulse response functions and the
confidence intervals are generated by use of Monte Carlo simulations (Garita, 2011).
C)The variance decomposition
The variance decomposition shows how much of the error variance of each of the variables can be ex-
plained by shocks to the other variables. Thus, the variance decomposition provides information about the
relative importance of each random innovation in affecting the variables in the system.

D) Granger Causality test

Since the general output from running VAR models does not include the common F-test or any R2 value,
we need a different method of determining whether the included variables have any explanatory power in
explaining the model. Granger causality is a method of determining whether a variable, say xt has any
information on yt when regressing yt on its own lagged values and on lagged values of xt. Using the VAR
framework, we can test if the included variables have any explanatory power on all other variables in the
model. If for example, lagged values of xt has no such explanatory power in the system, then xt is weakly
exogenous and should be determined outside the system. It is however important to note that this testing
procedure does not imply causality in the general sense, it does only consider whether a variable Granger-
causes another variable.
E) Eigenvalue stability condition
To state that a VAR model is stable it is necessary to test if all eigenvalues from the resulting output has
a modulus less than 1 (Pattersson, 2000). This essentially means that the model is stable in the sense that
there exists a point that the dynamics of the model converges against. If the any of the modulus on the
eigenvalues would be greater than 1, then consequently there would be no long run equilibrium and the
values in the future would just continue to increase. This is something that is testable in the statistical
software used throughout the paper and as such I will generate a series of eigenvalue stability tests.
 94

F) Stationarity
One of the prerequisites of running VAR models is that the variables must be stationary. Going by defini-
tion in Greene (2008), a stochastic process is weakly stationary if all of the following requirements are
satisfied.
G) The Relative Impact of the Resource Gaps on Output Gap.
To measure the relative impact of the three resource gaps: Investment-Saving gap (ISGAP), Exports-Im-
ports gap (EMGAP) and Budgetary gap (BGAP), on output gap, we estimate the beta (𝛽ⅈ ) coefficients for
the gaps. We employed the formula below:
𝑍𝑖 −𝑆𝐸𝐺
𝛽ⅈ = ....................................................... (5.33)
𝑆𝐸𝐷

where
𝛽ⅈ ; = the estimated beta coefficient of the resource gap i.
𝑍ⅈ ;_ = the estimated coefficient of ith resource gap.
𝑆𝐸𝐺 ; = the standard deviation of the ith resource gap.
𝑆𝐸𝐷 = the standard deviation of the dependent variable.
Thus, 𝛽1 , 𝛽2 and 𝛽3 represent the beta coefficients of the Investment-Savings gap, Exports-Imports gap
and Budgetary gap respectively. It is worth noting that for the purpose of this analysis the values are taken
in absolute terms that is we disregard the accompanying signs.

5.4.2.3 Dynamic Common Correlated Effects (DCCE)estimation techniques

we employ the Dynamic Common Correlated Effects (DCCE) estimator technique to check the robustness
of the result we found using PVAR. Given the nature of our dataset, we resort to the Dynamic Common
Correlated Effects (DCCE) estimator of Chudik and Pesaran (2015). While the analysis of macro panel data
is still dominated by estimators developed for micro datasets (primarily the estimators by Arellano and Bond
(1991) and Blundell and Bond (1998), devised for panels where T is small relative to N ), the DCCE
estimator is particularly suitable when both the cross-section and the time series dimensions are sufficiently
large. Indeed, our sampling period spans 39 years, which allows us to exploit temporal variation in addition
to cross-country heterogeneity.
Unlike standard estimators, a further advantage of the DCCE estimator is that it is robust to unknown types
of error cross section dependence, which is likely to feature due to the presence of common shocks and
unobserved components. This is highly relevant in our case, as the last few decades have witnessed
increased economic and financial integration that generates strong interdependencies amongst the cross-
sectional units in our sample. Indeed, this period captures several macroeconomic and financial cycles across
all countries in our sample, as well as common shocks. unaccounted for cross-sectional dependence can
lead to severe biases and this problem becomes more acute in dynamic panel settings, as discussed in
(Phillips & Sul, 2007)).
 95

Moreover, the DCCE estimator addresses another potential source of inconsistencies that may arise if the
slope parameters are falsely assumed to be identical across countries (see Pesaran and Smith, 1995). Thus,
we control for heterogeneity by first estimating country-specific effects, which are subsequently combined
through a mean-group (MG) estimator to obtain estimates of the average effects.
Our choice of a dynamic framework is motivated by the literature on current account dynamics, which
suggests that there is considerable persistence in current account balance. In this vein, in order to estimate
the relationship between current account deficit, budget deficit and private saving-investment deficit, while
controlling for variables that are known to affect current account deficit, we adopt as our baseline
specification the following heterogeneous dynamic panel model with a multifactor error structure:

After re-estimating and comparing the baseline model replicating the specifications in previous
studies, we also consider a dynamic version of Eq.(7) below, which includes one lag of the
dependent variable (𝑅𝐺𝐷𝑃𝐺ⅈ𝑡−1 ).

𝑅𝐺𝐷𝑃𝐺ⅈ𝑡 = 𝛽0 + 𝛽1 𝑅𝐺𝐷𝑃𝐺ⅈ𝑡−1 + 𝛽2 𝑙𝑛𝐶𝐴𝐵ⅈ𝑡 + 𝛽3 𝑙𝑛𝐵𝐵ⅈ𝑡 + 𝛽24 𝑙𝑛𝑆𝐼𝐵ⅈ𝑡 + 𝜂ⅈ𝑡 + 𝜀ⅈ𝑡

(5.34)
𝜂ⅈ𝑡 = 𝛼ⅈ + 𝜆′ⅈ 𝑓𝑡 + 𝜀ⅈ𝑡 (5.35)

where RGDPG is the real gross domestic product growth rate (proxy measure of economic growth),
the macroeconomic imbalances are measured by Current USD and transformed into natural logarithm,
CABi,t is the current account deficit for country i in year t, BB i,t is a government budget deficit for country
i in year t and SIB is private saving-investment gap for country i in year t. otherwise, xi,t is a k-dimension
vector of control variables as described in the previous subsection and assumed to be weakly exogenous,
αi accounts for time-invariant unobserved country specific effects, ft is an m×1 vector of unobserved
common factors (capturing common business cycles or exposure to global economic, political or financial
shocks, for example) with corresponding country-specific factor loadings λJi and ei,t represents the
idiosyncratic errors, possibly correlated across countries.

This is an extremely flexible specification that, with suitable restrictions on the parameters, encompasses
several approaches used in empirical practice, e.g. static and/or (partially) pooled panels, some of which
will be considered below. However, these frameworks can lead to biased estimates, particularly in the
presence of common unobserved factors, which is likely to be the case in our application.
Consistent estimation is carried out with the Dynamic Common Correlated Effects estimator of Chudik and
Pesaran (2015), which approximates the unobserved common factors by augmenting the estimation equation
with additional terms containing cross-section averages. Mean Group (MG) estimates can then be obtained
by averaging estimated coefficients across countries, with the corresponding standard errors computed non-
parametrically following Pesaran and Smith (1995). Although MG-type estimators are likely to produce
 96

somewhat larger standard errors than pooled estimators, as a much larger number of parameters is
estimated, they are consistent both if slope parameters are homogeneous or if there is slope heterogeneity
across countries.
In addition, we also consider an IV extension of the DCCE estimator that accommodates the possibility of
endogenous regressors ( Everaert and Pozzi, 2014), as well as the "half-panel jacknife" bias correction
method of ), in which the bias-corrected estimates are obtained .

The error-correction mechanism for the long run equations is specified as:
𝑅𝐺𝐷𝑃𝐺ⅈ𝑡 =𝜆ⅈ + 𝜆ⅈ 𝑅𝐺𝐷𝑃𝐺ⅈ𝑡−1 + 𝜆ⅈ 𝑙𝑛𝐶𝐴𝐵ⅈ𝑡 + 𝜆ⅈ 𝑙𝑛𝑆𝐼𝐵ⅈ𝑡 + 𝜆ⅈ 𝑙𝑛𝐵𝐵ⅈ𝑡 + 𝜀ⅈ𝑡 (5.36)
We can proceed to run regressions in first difference provided the series of interest are I(1). Although we
may well lose the long-run relationship inherent in the data. There is a need to use variables in their levels
as well in the regressions. The Error Correction Model is designed to fit in variables both in their levels
and first differences and thus captures both the short run disequilibrium and long run equilibrium adjust-
ments between variables. The Error Correction Model showing the relationship between the
𝑅𝐺𝐷𝑃𝐺𝑡 ,CABt, SIB and BBt is specified as follows:
∆𝑅𝐺𝐷𝑃𝐺ⅈ𝑡 =𝜆0 + 𝜆1 𝑅𝐺𝐷𝑃𝐺ⅈ𝑡−1 − 𝜆2 𝐶𝐴𝐵ⅈ𝑡−1 − 𝜆2 𝑆𝐼𝐵ⅈ𝑡−1 − 𝜆3 𝐵𝐵ⅈ𝑡−1 − 𝜆4 𝐹ⅈ𝑡 + 𝜆∗ 5 ∆𝑆𝐼𝐵ⅈ𝑡−𝑗 +
𝜆∗ 5 ∆𝐶𝐴𝐵ⅈ𝑡−𝑗 + 𝜆∗ 6 ∆𝐵𝐵ⅈ𝑡−𝑗 + 𝜀ⅈ𝑡 (5.37)
For the estimation of the short run dynamics (𝜆∗ 5 ,, 𝜆∗ 6 the transformation of DCCE model into Error
Correction Representation is required. Error correction term (−𝜆4 ) is the rate of adjustment which
indicates that how quickly variables adjust towards equilibrium and its negative sign represents the
convergence in the short run. This term should be negative and statistically significant to establish the
long run relationship among variables.

5.5. ESTIMATION AND INTERPRETATION OF RESULTS

5.5.1. INTRODUCTION
This chapter presents the results of the empirical estimation and gives an economic interpretation of the
results. We start with data description, test for cross sectional dependency, non-stationarity and go on
to estimate PVAR and DCCE as well as Granger causality test follows.
5.5.2. PRELIMINARY ANALYSIS

5.5.2.1. DATA DESCRIPTION

To investigate the effect of macroeconomic imbalances on economic growth, we use annual observations
from 1980 to 2018 for 35 Sub-Saharan African countries. In the study, we focus on three kinds on macro-
economic imbalances variables which is measured in current Billion US$ Billion (current account bal-
ance(CABB), fiscal balance(BBB) and private saving-investment balance(SIBB)) because according to
the literature, these macroeconomic imbalances expenditures account for a major part of gaps in different
countries. Also, their growth impact has been extensively debated. To carry out the study, we analyze the
 97

effect of these imbalances on real GDP growth. All macroeconomic imbalance variables are expressed in
natural logarithms. The variables used in this study and their descriptive statistics are shown in Table 5.1.
These variables are from World Development Indicators (World Bank) and World Economic Outlook
Database (IMF) and Africa Development Bank (AfDB).
Therefore, Table 5.1 indicates that the pooled average annual real GDP growth(%), current account
balance, budget balance, and private saving-investment balance in 35 SSA countries between 1980 and
2018 stood at 3.57%, -0.37, -0.53 , and -0.27, respectively. This reflects the high current account balance
and low budget balance and private saving-investment balance in 35 SSA countries as compared to other
low-income countries.
Table 5.1: Description Statistics for variables
Variable Obs Mean Std. Dev. Min Max
RGDPG 1365 3.57 5.08 -41.90 33.60
BBB 1365 -0.53 1.99 -20.36 13.83
CABB 1365 -0.37 2.70 -21.22 36.52
SIBB 1365 -0.27 6.15 -62.76 54.71
Source: Author calculation

The descriptive analysis of SSA’s data highlights two econometric issues. First, it identifies a concern
that the individual country series are characterized by cross-sectional correlation or dependence. Figure
5.1. indicates a high degree of correlation between country- level explanatory variables. This is to be
expected as these variables are linked by a country’s trade and economic relations, and financial
institution status. One would expect that countries that share these common factors would likewise have
systematic error correlations.
 98

Figure 5.1: Scatter Correlation Matrix Plot of Variables

Second, as seen in Figure 4.2, while many countries align along a linear relationship between RGDPG
and the other variable such as CABB ,BBB, and SIBB, there are a disproportionate number of countries that lie
outside the 95% confidence interval for this linear relationship. This suggests that there may be
heterogeneity in the RGDPG slope parameter across countries.
Figure 5.2A: Scatter Plot of Real GDP growth vs Current Account Balance Variables
 99

Figure 5.2B: Scatter Plot of Real GDP growth vs. Budget balance Variables

Figure 5.2C: Scatter Plot of Real GDP growth vs Private saving-investment Balance Variables

These two econometric issues have potential implications for model specification and efficiency of
standard panel data estimators. In subsequent sections, the procedures adopted in addressing the
aforementioned issues are discussed as follows.
 100

From the descriptive stats output we can see the mean and median are different, especially considering the
standard error. We also see from the graphical output, a bunch of outliers, and a heavily skewed distribu-
tion. Given the skewness of the data and the presence of outliers, log transforming is at least worth trying.
Well, transforming data sets works most effectively for data distributions that are skewed to the right by
the presence of outliers. The possibility of transforming data to an alternative ratio scale is particularly
useful with skewed data, as in some cases the transformation will normalize the data distribution.
Therefore, it is important to use logarithmic transformation of the macroeconomic imbalance (current
account balance(CAB), budget balance(BB) and private saving-investment balance(SIB) data series to
make better predicted outcomes from the PVAR and Dynamic Common Correlated effect model.
These models will have unit changes between the x and y variables, where a single unit change in x will
coincide with a constant change in y. Taking the log of one or both variables will effectively change the
case from a unit change to a percent change.. Another way to think about it is when taking a log of a
dataset is transforming our model(s) to take advantage of statistical tools that improve on features that are
normally distributed. Because, Logarithmic transformation is a convenient means of transforming a highly
skewed variable into a more normalized dataset. When modeling variables with non-linear relationships,
the chances of producing errors may also be skewed negatively. In theory, we want to produce the smallest
error possible when making a prediction, while also considering that we should not be overfitting the
model. Using the logarithm of one or more variables improves the fit of the model by transforming the
distribution of the features to a more normally shaped bell curve.

5.5.2.2. TEST OF CROSS-SECTION INDEPENDENCE

The empirical work in this study is based on the annual data of 1980 up to 2018 and the macro-
economic imbalances, that is, current account balance, budget balance and private saving-in-
vestment balance are transformed into natural logarithm. To begin the analysis with, first, we start by
examining cross-sectional dependence among the series in our panels. Most of the recent theoretical and
applied panel data econometric studies have emphasized the need to address the methodological issue
related to cross-section or “between groups” dependence in error terms when dealing with panel data
models.
A) Panel Time Series Cross-Sectional Dependence Test Results
Cross-sectional correlation often emanates from unobserved common “shocks” and unobserved, time-in-
variant heterogeneous error components (Eberhardt & Teal, 2011, 2014; Pesaran & Tosetti, 2011;Sarafidis
& Wansbeck,2012 ). This error component is a sub-component of the error term, incorporating spatial
dependence and idiosyncratic pairwise dependence in the disturbance (De Hoyos & Sarafidis, 2006). Alt-
hough the notion of cross-sectional dependence has been in existence since the 1930s, it is often ignored
by researchers in panel model estimation.
 101

The existence of cross-sectional correlation between error terms can have severe implications for the es-
timation of both coefficients and standard errors using standard panel data estimators. This can lead to
poor policy decisions based on biased parameter estimates. For this purpose, the three most often used
cross-sectional dependence test procedures- Pesaran (2004), Friedman (1937), and Frees (2004) cross-
sectional dependence (CD) tests- were employed to examine the between-group correlation in error terms
(as a post-estimation diagnostic test) and panel time series variables (as a pre-estimation diagnostic test).
It should be noted that Frees and Friedman’s tests were originally designed for static models, unlike Pe-
saran’s CD test for static and dynamic models. Pesaran’s cross-sectional dependence test is more appli-
cable for pre- and post-estimation testing, unlike other tests that are more appropriate as post-estimation
tests (De Hoyos & Sarafidis, 2006).
Accordingly, I test for cross-sectional independence in the data used to estimate models as reported in
Tables 4.3 below. Table 4.3 presents the test results for cross-sectional correlation. It shows the average,
country-specific correlation coefficients for the panel series full matrix and off-diagonal matrix elements,
as well as Pesaran’s cross-sectional dependence test statistics. The results indicate high positive, pairwise
cross-sectional correlation of panel time series for real GDP growth, current Account balance, Budget
balance, and Private saving-investment balance. The results further reveal the presence of cross-sectional
dependence based on Pesaran’s CD test statistics for each variable. The null hypothesis of cross-sectional
independence is rejected at the 1% significance level.
Table 5.3: Panel Time Series Cross-Sectional Dependence Test Results
Variable CD-test p-value corr abs(corr)
RGDPG 10.77 0.00 0.07 0.16
CABB 44.92 0.00 0.30 0.45
BBB 38.86 0.00 0.26 0.36
SIBB 4.51 0.00 0.03 0.33
Notes: Under the null hypothesis of cross-section independence CD ~ N(0,1)

As a result of the tests statistics above, the null hypothesis of cross-sectional independence is rejected for
all variables under consideration. This indicates that the individual country, panel data series employed in
this study are cross sectionally dependent and correlated, likely due to similar patterns of common mac-
roeconomic shocks. The standard (or parametric) average absolute correlation indicates positive pairwise
correlation coefficients of all the estimated residuals from replicated models. Also, the pairwise average
Spearman rank correlation estimates from the models are found to be positive and low below 0.5. This
indicates that the upper-diagonal has low positive and negative elements of country-specific pairwise cor-
relations coefficients, which cancel each other out during averaging. This problem invalidates Friedman’s
cross-sectional dependence (CD) test. As a result, I do not place much weight on the finding that Fried-
man’s CD test does reject the null of cross-sectional independence. In contrast, Frees’ CD test, based on
 102

the average sum of squares of the rank of pairwise correlations, rejects the null hypothesis of cross-sec-
tional independence at the 1% significance level. Similar results are obtained using Pesaran’s CD test. As
a result, I conclude that the models’ error terms are characterized by significant cross-sectional depend-
ence.
Table 5.4 Estimated residual Cross sectional dependence test results
CD tests P-value average absolute value of the off-diagonal elements
Pesaran's test 10.86 0.0000 0.16
Friedman's test 147.53 0.0000 0.16
Frees' test 0.67 0.0000 0.16
NB. Under the null hypothesis of cross-section independence, and a normal distribution had been used
to approximate Frees' Q distribution.

Taking into account cross-sectional dependence and country-specific heterogeneity in empirical analyses
is essential as our sample countries are highly integrated and highly globalized in their economic relations.
If cross-sectional dependency does exist, the use of dynamic PVAR and Dynamic common correlated
effect estimation approach should be more efficient than an ordinary least-squares (OLS) approach in
estimating panel data causality. Moreover, the causality results obtained from the estimator developed by
Zellner (1962) should be more reliable than those obtained from county specific OLS estimations.
A further issue to decide is whether to treat slope coefficients as homogenous to impose the causality
restriction on the estimated parameters. The causality from one variable to another variable by imposing
the joint restriction for the panel is the strong null hypothesis and the homogeneity assumption for the
parameters is unable to capture heterogeneity due to country-specific characteristics.

B) Heterogeneous Slope Estimators

The preceding exploratory data analysis has determined that the relationship between real GDP growth,
current account balance, budget balance, and private saving-investment balance in SSA countries is likely
heterogeneous due to differential macroeconomic policies, and prices across countries. To account for
heterogeneous effects, I next consider some recent panel data estimators that are designed to address these
econometric issues A test of slope homogeneity was performed as a robustness check using the test statistic
suggested by Swamy (1970). A test of slope homogeneity in panels with a large number of observations
of the cross-sectional (N) and time (T) dimension, which is based on Pesaran, Yamagata (2008) and
Blomquist, Westerlund (2013) is performed. Thus. the null hypothesis of the test is of homogenous
slopes, implying that all slope coefficients are identical across cross-sectional units. The test allows for
non-normally distributed errors, such as serial correlated errors. The test results are reported as follows.
 103

Table 5.5 Test for slope homogeneity (H0: slope coefficients are homogenous)
Delta P value
-5.34 0.00
adj. -3.48 0.00
The result of homogeneity test indicates that the null hypothesis of slope homogeneity is rejected, and
thus there is an evidence on heterogeneous slope.
5.5.3-NON-STATIONARITY TEST
The CADF test is a panel unit-root test that takes cross-sectional dependencies into account. The Fisher’s
combined p-values test proposed by Maddala and Wu (1999) and the Pesaran (2007) cross-sectionally
augmented Im, Pesaran and Shin (2003) (henceforth, CIPS) MultiPurt employed in this study. The test
is based on the augmented individual cross-section ADF (CADF) regressions of with cross-section
averages of lagged levels and first difference.
The Multipurt panel unit root diagnostic results in Table 5.6 indicates that indicate that all variables are
stationary in levels, and in first difference. This also confirms the non-stationarity of those series at levels
and first difference, i.e. for all the panel units, therefore, the variables are integrated of order one.
Table 5. 6 Panel Unit root test
A) Maddala and Wu B) Pesaran (2007) (A) Maddala and Wu B) Pesaran (2007)
(1999) Panel Unit Root Panel Unit Root (1999) Panel Unit Root Panel Unit Root
test (MW) test (CIPS test (MW) test (CIPS

Specifi- level difference

Variable cation

without With without With without With without With

trend trend trend trend trend trend trend trend

lags chi_sq chi_sq chi_sq chi_sq chi_sq chi_sq chi_sq

- - -
RGDPG 0 770.12*** 730.12*** -19.07 2282.48*** 1938.59***
18.13*** 27.81*** 27.31***
- - - -
RGDPG 1 459.95*** 424.53*** 1476.85*** 1242.02***
14.15*** 13.20*** 26.12*** 24.88***
- -
CAB 0 109.71*** 141.52*** -0.68 -3.06*** 1923.64*** 1682.75***
26.55*** 25.69***
- -
CAB 1 107.10*** 140.11*** 1.90 -0.63 886.46*** 765.34***
21.02*** 19.88***
- - -
BB 0 263.81*** 250.06*** -9.72*** 1357.58*** 1159.22***
10.78*** 24.42*** 23.03***
- -
BB 1 158.80*** 155.76*** -6.80*** -5.94*** 858.68*** 724.13***
18.43*** 15.88***
- -
SIB 0 183.71*** 231.40*** -4.89*** -6.15*** 1665.53*** 1418.15***
25.87*** 24.74***
- -
SIB 1 113.79*** 132.93*** -1.60** -2.99*** 836.6***2 688.21***
18.30*** 16.15***
NB> Null for MW and CIPS tests: series is I(1). MW test assumes cross-section independence.
CIPS test assumes cross-section dependence is in form of a single unobserved common factor -mul-
tipart- uses Scott Merryman's -xtfisher- and Piotr Lewandowski's -pescadf-. ***-p<1%, **-p<5% and
*-p<10%
 104

After testing the stationarity of the variables, the possible existence of a long-term relationship between
the variables was analyzed in the following section using panel cointegration test.

5.5.4. COINTEGRATION TEST

Once the stationarity order is defined, our next step is to apply panel cointegration test. Granger (1988)
showed that when the time series become stationary only after being differentiated once, they might have
linear combinations that are stationary without differencing. These series are generally called cointegrated.
If the integration of order one is implied, the next step is to use the cointegration analysis to determine
whether there is a long-term relationship between the set of integrated variables. In this investigation,
Westerlund (2007) cointegration test is deployed to test the existence of long run equilibrium relationship
between Real GDP growth, current account balance, fiscal balance and private saving-investment balance.
Westerlund (2007) implements the four error-correction-based panel cointegration tests which are general
enough to allow a large degree of heterogeneity, both in the short-run dynamics and, in the long-run coin-
tegration relationship and dependence across as well as within the cross- sectional units. The fundamental
idea is to test the absence of cointegration by establishing if the individual members of panel are error-
correcting or not. After testing the stationarity of the variables, the possible existence of a long-term rela-
tionship between the variables was analyzed in the following section using panel cointegration test.

Once the time series properties of the data are evaluated then Error based cointegration test is conducted
to test whether there exists any long run relationship among the variables. Since our sample data size is
small, we construct and test the restricted Westerlund error correction test with short‐run dynamics for
all series with a single lag and lead. Results of the restricted ECM model are reported in table 5.7.

Table 5.7: Panel cointegration tests

restricted case with single lag and lead
constant constant and trend
Statis- z- P- Robust Z- P- Robust
tic Value value value P-value Value value value P-value
Gt -3.60 -8.58 0.00 0.00 -3.78 7.34 0.00 0.00
Ga -16.68 -4.81 0.00 0.00 -18.26 2.02 0.02 0.00
Pt -22.45 -10.49 0.00 0.00 -22.91 8.86 0.00 0.00
Pa -17.29 -8.87 0.00 0.00 -17.68 4.29 0.00 0.00

Tables 5.7 present the results from the corresponding cointegration tests for the periods 1980-2018.
According to these results, in the unrestricted case without trend and with trend the null hypothesis of no
 105

cointegration of the group‐mean tests (Ga and Gt)) and the panel tests (Pt and Pa). is rejected at 1%
significant level by simple and robust p‐values (Table 4). Thus, we conclude that almost all variables are
cointegrated in group mean and panel tests. For this reason, it is concluded that there is a long-term
relationship among the variables.
5.5.5 EMPIRICAL RESULTS OF HOMOGENEOUS PANEL VAR
We start with a homogeneous panel VAR setup in order to compare our results with the previous literature.
We relax this assumption later by using a more appropriate setup with heterogeneous coefficients and
corrected for cross-sectional dependence. For the homogeneous VAR, we apply GMM-style estimators to
deal with the Nickell (1981) bias as in Abrigo and Love (2015). The main results of the baseline PVAR
model are given in Table 2. We report estimates of the coefficients given in equation (1), where the fixed
effects and country-time dummies have been removed.

A) Model Selection
The procedure of choosing the length of the lag on the pVAR should be done using a set of moment
selection criterions. The results of choosing lag length are given in Table 1 below. As a rule of thumb, I
begin by choosing the lag length that minimizes the MBIC, MAIC and MQIC. Most often the lag length
is chosen by focusing on the MAIC criterion, supported by Serena and Perron (2001)
Table 5.8: Lag-order selection statistics for panel VAR estimated
lag CD J J pvalue MBIC MAIC MQIC
1 0.95 72.07 0.23 -379.25 -55.93 -177.96
2 0.95 44.25 0.63 -294.24 -51.75 -143.26
3 0.96 23.45 0.86 -202.21 -40.55 -101.56
4 0.96 13.24 0.66 -99.59 -18.76 -49.27

As in Abrigo and Love (2015), in our case the first-order panel VAR is the preferred model, since this has
the smallest MBIC, MAIC and MQIC. However, the over-all coefficient of determination suggests apply-
ing a model with more than 1 lag. Balancing the need of allowing for a sufficient number of lags given
the nature of the data and trying to avoid overparameterization, we set the number of lags to 5, opting for
a fourth-order panel VAR model.
The evidence shown in Table 1 above is supportive to the choice of first-order panel VAR (one lag) since
this has the smallest MBIC, MAIC and MQIC.
The following tables will present estimation output using the same sample-size. including instrumental
setups through the fifth lagged dependent variable as instrument.

Table 5.9. Coefficients of the PVAR (1) model

Variables ∆RGDPGt-1 ∆lnCABt-1 ∆lnBB t-1 ∆lnSIB t-1
 106

∆RGDPGt 0.21*** - 3.11*** -2.59*** -3.52***

∆lnCABt -0.0002 0.595*** 0.402*** 0.043
∆lnBBt 0.0001 -0.003 0.933*** 0.119***
∆lnSIBt 0.001 0.089* -0.240*** 0.563***

Table 5.9 presents the first output from running a panel VAR model on GDP growth and Δ macroeconomic
imbalances. The first section of table 5 shows how GDP growth is affected by its own lagged levels and
the lagged levels of Δ macroeconomic imbalances using GMM instruments of lag one up through lag five
of the dependent variables.
From the above model results, where all the variables are considered endogenous, the one that represents
the core of our research is the first one, RGDP model equation. In this model specification, the impact
macroeconomic imbalances (or short run impact) is simply the coefficient on the 1-period lagged itself,
current account balance, fiscal balance and net private saving balance. This specification also produced
statistically significant coefficient on past values of real GDP growth itself, current account balance, fiscal
balance and net private saving balance on real GDP growth.
Although we are aware that the interpretation of these coefficients may be meaningless due to the a-theo-
retical nature of the VAR models, nonetheless, we opted to report them in Table 2 and briefly discuss the
significance of the estimated coefficients.
The way to think about the output in table 5.9 is essentially to consider what happens to any variable of
interest if we introduce a chock into the other variable. Here we can think that if we chock the real GDP
growth itself, current account balance, fiscal balance, or net private saving balance in the previous period
by one unit, ceteris paribus, real GDP is supposed to increase today by 0.21, or by 3.11 or decrease by
2.59 or 3.52 respectively. This procedure is continuing until the effect dissipates and we can interpret this
simple procedure of following up on a change in one period as the response on current account balance
following the impulse on macroeconomic imbalance, and impulse response function. As said earlier, more
insightful interpretations are provided with the results obtained from the impulse response functions (IRF)
displayed in Figures 5.4 and 5.5.

B) Stability test
Prior to estimating impulse-response functions (IRF) and forecast-error variance decompositions (FEVD),
however, we first check the stability properties of the estimated panel VAR model. The stability of the
panel VAR requires the modulus of the eigen-values of the dynamic matrix to lie within the unit circle.
The resulting table and graph of eigenvalues reported in Annex (2) confirms that the estimate is stable.
 107

Table 5.10 stability test

Eigenvalue

Real Imaginary Modulus

0.887 0.000 0.887
0.596 0.105 0.605
0.596 -0.105 0.605
0.224 0.000 0.224

Figure 5.3 stability

C)Granger causality tests

Complementing the IRF analysis, we also examine potential causal linkages among the variables via
Granger causality tests. First, we carry out Granger causality tests based on the homogeneous PVAR
model in line with Abrigo and Love (2016). The results, reported in Table 6, strongly indicate in favor of
rejecting the hypothesis of non-causal relationship from macroeconomic imbalances (current account, fis-
cal and net private saving balances) to real GDP growth at the 1% levels of statistical significance.
Table 5.11: Granger causality test results
chi2 chi2 chi2 chi2
RGDPG lnCAB lnBB lnSIB
lnCAB 14.12*** 0.05 0.03 0.75
lnBB 6.79*** 18.67*** 0.01 2.69*
lnSIB 21.85*** 0.59 9.44*** 9.61***
ALL 24.01*** 19.10*** 10.21** 9.72**

Note: The table shows the results of the Granger causality Wald test based on the baseline GMM PVAR
specification in line with Abrigo and Love (2016). Null-hypothesis: variable X (row) does not Granger-
cause variable Y (column).
 108

D)Impulse Response Function

Also, the stability condition of the PVAR model, as Abrigo and Love (2015) are describing it, implies the
model is invertible and has an infinite-order vector moving-average (VMA) representation, “providing
known interpretation to estimated impulse-response functions and forecast-error variance decompositions.
In order to obtain the needed results, we need to multiply the current account balance, the fiscal balance
and the net private saving balance variables with the (-1) constant, because, in the original model, when a
positive shock happens in this variable, it really means the deficit is decreasing. This is the reason for
which we have multiplied variables by (-1).

Following the estimation of the PVAR model we compute orthogonalized impulse response functions
(IRFs), cumulative impulse response functions (IRFs) and forecast error variance decomposition (FEVD)
to track the relationship between current account, balance, budget balance and private saving-investment
balance.
Figures 5.4 display the impulse response functions respectively for the Real GDP growth and for all the
endogenous variable of the panel VAR model. The accumulated impulse responses (solid line) are pre-
sented over time 10. The confidence bands, given by 2.5th and 97.5th percentiles of the 1000 simulated
impulse responses are presented by the dash lines.
It is clearly, as figure 5.4 can tell us, that when the current account deficit increases in the defined block
of countries, the real GDP growth follows an upward curve until 2-year horizon and then downward till
4-year horizon, then after increases. From this, it is interesting to note the positive non-linear relation
between the current account balance and real GDP growth. In addition, when the budget deficit increases
in the defined block of countries, the real GDP growth follows an upward curve until 5-year horizon and
then moves downward. It also shows that there is a positive non-linear relation between the budget deficit
and real GDP growth. Furthermore, when the lagged real GDP growth increases in the defined block of
countries, the real GDP growth follows a sharp downward curve until 3-year horizon and then become
stable when lagged real GDP growth decreases. It also shows that there is a temporal positive linear rela-
tion between the lagged Real GDP growth and real GDP growth.

While, when net private saving deficit increases, the real GDP growth temporarily decreases until the first
year, then it permanently increases. Thus, we do find an evidence that there is a positive-linear relation
between the private saving-investment balance and real GDP growth in the long run.

Table 5.12 Orthogonalized IRF

Impulse variable
Response variable and Forecast horizon
 109

Forecast horizon
RGDPG RGDPG lnBB lnCAB lnSIB
0 4.862 0 0 0
1 1.009 -0.188 0.201 -0.752
2 0.181 0.069 0.171 -0.619
3 0.015 0.255 0.147 -0.380
4 -0.010 0.336 0.145 -0.190
5 -0.009 0.350 0.151 -0.062
6 -0.005 0.331 0.156 0.015
7 -0.003 0.298 0.157 0.058
8 -0.002 0.262 0.153 0.078
9 -0.001 0.228 0.146 0.086
10 -0.001 0.198 0.136 0.085

The response of the real GDP growth to macroeconomic imbalances shocks seems to be statistically sig-
nificant. This seems to be consistent with the conclusions pointed out by Ferraro et al. (2010).
These results are not surprising that current account deficit and real GDP growth have an inverse relation-
ship and in line with theoretical expectations that positive domestic economic output boosts demand for
foreign goods and services and consequently deteriorates the current account balance (see Calderon et al.,
2002; Aristovnik, 2008). Although an increase in domestic output can be associated with a greater savings
rate, it seems that the rise in consumption and investment together are somewhat greater, thus leading to
an expansion of the current account deficit (Backus et al., 1994). This is also consistent with the findings
of Aguiar and Gopinath 2007, who report that emerging markets are characterized by strongly counter-
cyclical current accounts.
An improvement of the fiscal balance tends to have a positive effect on the real GDP growth. A variety of
models predict a positive relationship between macroeconomic imbalances and real GDP growth over the
medium term. It seems also important to report how an improvement of fiscal balance has a positive and
significant impact on the GDP growth of this group of countries, consistent with standard theory.
As already mentioned, the impulse response functions capture the time profile of the effect of shocks at a
given point in time on the expected future values of variables in a dynamic system (Simo-Kengne, 2012).
We present them in Figure 1. The sign of estimated coefficients given in Table 2 are in line with our
expectations, thereby producing theoretically consistent impulse response functions. According to the es-
timated PVAR model with RGDPG (as a proxy variable for economic activity), it can be concluded that
responses of real GDP growth to current account deficit, fiscal deficit, and net private saving gap in a 1-
lag model are statistically significant at the 1% significance level.
 110

Figure 5.4 plots the responses of Real GDP growth to a shock in the 1-lag PVAR model.

The way to calculate the long run impact is, as presented in the table 5.13 and figure 5.5, through adding
together each impact on each step of the horizon in the impulse response function. This produces a cumu-
lative response function that tends towards a long run equilibrium if the model is stable.
We observe that a fiscal balance or current account balance shock of one standard deviation results in an
increase in real GDP growth by about 2.14% or 1.58% after ten years, respectively. Thus , budget balance
or current account balance and real GDP growth have a positive linear long run relationship. The impulse
response functions in Figure 6 show that real GDP growth responds negatively to private saving-invest-
ment balance in the long run and thus an increase in private saving-investment balance causes a decline
in real GDP growth by about 1.68%.
Table 5.13 Cumulative Orthogonalized IRF
Response variable and Forecast Impulse variable
horizon lnBB lnCAB lnSIB
RGDPG
RGDPG
0.00 4.86 0.00 0.00 0.00
1.00 5.87 -0.19 0.20 -0.75
2.00 6.05 -0.12 0.37 -1.37
3.00 6.07 0.14 0.52 -1.75
4.00 6.06 0.47 0.66 -1.94
5.00 6.05 0.82 0.81 -2.00
6.00 6.04 1.15 0.97 -1.99
7.00 6.04 1.45 1.13 -1.93
8.00 6.04 1.71 1.28 -1.85
9.00 6.04 1.94 1.43 -1.77
10.00 6.04 2.14 1.56 -1.68
 111

Figure 5.5. long run equilibrium

E) Variance Decomposition
Although the impulsive responses provide information about the effect of changes in one variable on an-
other, it is important to note that they do not show the proportions in which shocks on a variable explain
the fluctuations of other variables. To estimate the extent of changes in one variable in explaining the
shifts in other variables, we perform a variance decomposition. The variance decompositions display the
proportion of movements in the dependent variables that are due to their own shocks versus shocks to the
other variables. The variance decomposition in Table (6) clarifies how macroeconomic imbalances affect
economic activity in the sample used in this study.
These results seem to be confirmed by the forecast-error variance decomposition (FEVD) based on a
Cholesky decomposition of the residual covariance matrix of the underlying panel VAR model (see Table
7). The results contained in Table 7 show that the 92.4%, variation in Real GDP is explained by about
0.9%,2.4% and 4.3% by current account deficit, fiscal deficit and private saving-investment deficit, re-
spectively just after ten quarter. It finds that private saving-investment balance seems to play especially
important role in explaining the variation of Real GDP growth. Next to private saving-investment bal-
ance, Budget balance also plays an important role in explaining the real GDP growth variation.
Table 5.14: Variance decomposition analysis Variation in the row variable explained by column variable
(in %, 10 periods ahead)
 112

Response variable and Fore- Impulse variable

cast horizon lnCAB lnBB lnSIB
RGDPG
RGDPG
0 0 0 0 0
1 1 0 0 0
2 0.975 0.002 0.001 0.022
3 0.959 0.003 0.002 0.037
4 0.950 0.004 0.004 0.042
5 0.944 0.004 0.008 0.043
6 0.939 0.005 0.013 0.043
7 0.934 0.006 0.017 0.043
8 0.930 0.007 0.020 0.043
9 0.927 0.008 0.023 0.043
10 0.924 0.009 0.025 0.043

As indicated in table 5.7, most of the forecast error variance is attributed to own innovations, yet fluctua-
tions in other variables do have notable explanatory power, and the patterns are consistent with the insights
from the IRFs. Focusing on the impact of current account balance, fiscal balance and private net saving
balance on real GDP growth (Model 1), budget balance and private saving-investment balance fluctuations
contribute notably to the variance of real GDP growth, whereas current account balance explained less.
Most of the impact on real GDP growth manifests itself in the first year after the initial shock with 100%
of its total variance explained, and remains persistent gradually decreasing over the following decade,
eventually reaching 92.4 %.

To summarize, from impulsive response function and variance decomposition results, we find evidence
that government deficit financing and higher private investment against saving has a significant impact on
economic growth in long term. This effect could be observed by an increase in GDP growth. However,
private deficit spending has a higher increase in growth and reduce output gap than government deficit
spending. On the other hand, our results find evidence that current account imbalance has negligible effect
on the growth in Sub- Saharan Africa countries.
Firstly, these results can be explained by the fact that in many developing countries, a large part of public
budgets is allocated to infrastructure, usually crowding in effect to the private investments. In addition,
Nowadays, many SSA countries attracts a lot of foreign direct investment that boosts growth and reduces
output gap. Besides, good infrastructure is important in attracting private investment, and returns on
 113

investment. Consequently, the government deficit financing may be good enough to crowd in private in-
vestment.
5.5. 6. ESTIMATES OF THE RELATIVE STRENGTHS OF THE RESOURCE GAPS
We used the formula given earlier above in chapter four, to estimate the relative strengths of the resource
gaps. As specified in the formula, 𝛽1 , 𝛽2 and 𝛽3 , represented the relative strength for Investment-Savings
gap, Export-Import gap and Budgetary gap respectively.
𝑍𝑖 −𝑆𝐸𝐺
We computed and obtained the following results using this formula,𝛽ⅈ = .
𝑆𝐸𝐷

where
𝛽ⅈ ; = the estimated beta coefficient of the resource gap i.
𝑍ⅈ ;_ = the estimated coefficient of ith resource gap.
𝑆𝐸𝐺 ; = the standard deviation of the ith resource gap.
𝑆𝐸𝐷 = the standard deviation of the dependent variable.
Thus,
−3.519004−.75286988
𝛽1 , Investment-Savings imbalance= =-72.44
.0589702

3.1107774−.8278115
𝛽2 , current account imbalance= =38.71
.0589702

−2.594494−.9960744
𝛽3 , budget imbalance = =-60.89
.0589702

It is worth noting that for the purpose of this analysis the values are taken in absolute terms that is we
disregard the accompanying signs.
From the indexes obtained above, we concluded that private saving-investment imbalance was the most
binding constraint on economic growth in SSA during the period under review, while Fiscal gap and
current account imbalance followed in that order of importance. Our results differed from those of some
past authors already reviewed. For example, Chenery and Bruno (1962) found similarities between Invest-
ment-Saving imbalance and current account imbalance. Our results showed that the two deficits were not
similar, given the magnitude of the indexes of the deficits. Also, while EJ-Shibly and Thirwal (1981)
reported that the indexes of Investment - Saving deficit and current account deficit were roughly of the
same magnitude, our estimation has shown that significant difference exists between the indexes of the
two deficits.
Our results also validated the original Chenery hypothesis. Chenery and Bruno(1962) Adelman and
Chenery (1966), Chenery and MacEwan (1966), Chenery and Strout(1966) and Chenery and Eckstein
(1970) had argued that for countries in their pre-take-off stage of development Investment-Saving gap
could predominate. Thus, suggesting that such countries are suffering from low monetized savings. The
implication from our results is that, although many of SSA economies are less developed, the economies
 114

have been in the pre-take-off-stage of development and requires higher public and private investment. Put
differently, SSA economies showed a characteristic of suffering from low export due to primary product
concentration, monetized low savings, and thus those domestic investments require higher import of cap-
ital and raw material inputs for their investments, leading higher current account deficit as well as higher
budget deficit. Hence, the fact that Investment-Savings imbalance) exerts the greatest relative impact fol-
lowed by budget imbalance (BGAP) suggests that it was very important and should deserve policy atten-
tion. It should be noted from our earlier argument that availability of savings does not automatically sug-
gest that such savings would be invested.
In other words, the enabling environment for transforming savings into investment is very crucial. In SSA
the enabling environment has not been present in an encouraging manner. In fact, it had been argued that
the absence of the enabling environment has exacerbated capital flight, (Ajayi 1991). This might have
been largely responsible for the relatively wide Investment-Savings gap.
The relatively higher index for Saving-investment gap from our results deserves some explanation. We
know that SSA economy is relying heavily on the inflow of fund and foreign direct investment will spend
on imports, as it normally happens during higher government spending and private investment, its impact
is usually felt the more on current account deficit and next to the budgetary activities of the government
than on the private sector balance of SSA economies. This is because the government first spends the
money.

5.5.7 ROBUSTNESS CHECK USING DYNAMIC COMMON CORRELATED EFFECT

ESTIMATION
After examining the impact of the three macroeconomic imbalances on economic growth using PVAR,
the next step is to check the robustness and consistency of the findings of PVAR by estimating the long
run and short run effect of these imbalances on economic growth using the Dynamic common correlated
effect estimation techniques. The estimates of long run coefficients reported in Table 5. As revealed by
the table all three estimators indicate a positive and statistically significant impact of BB and SIB on CAB.
According to the results of the pooled mean group and Dynamic common corrected effect estimator, both
the budget deficit and the net savings gap are positive and statistically significant in all models. The error
correction term (ECT) estimated by inserting the long run coefficients in the short run dynamic specifica-
tion of the model turns out to be negative and statistically significant under all estimation techniques. The
negative ECT shows that the system is driven to its long run cointegration path. The coefficient of ECT
reflecting the speed of adjustment is estimated to be around 75-83% per year. These statistics (ECT) sug-
gest that all the series (RGDPG, lnBB; lnSIB and lnCAB) are co-integrated in the long run and hence,
there is strong evidence in support of the three-gap constrained growth hypothesis in SSA. This suggests
that the three deficits have a long run relationship.
The PMG1 estimator suggests that, on average, a strengthening (deterioration) in BB-to-GDP ratio of 1%
point is associated with an improvement (deterioration) in real GDP growth of about 3.2% in the long
 115

run,but not statistically significant. The impact of a 1%-point strengthening (deterioration) of BB on the
RGDOG is of order 18.1%,47.9% and 3.2%, in the long run under the dynamic common correlated
estimators of MG1, MG2 and PMG2 estimators respectively. The coefficient of budget balance is positive
and statistically significant under MG1 and MG2, whereas it is not statistically significant under PMG2.
The coefficient of saving gap is consistently positive and significant only under MG1 estimator. The pri-
vate saving gap exerts a positive effect on RGDPG though the impact is weaker compared to BB and
CAB. A strengthening (deterioration) in saving gap ratio of 1% point is associated with an improvement
(deterioration) on real GDPG of 16.2% in the long run using MG1 estimation model.
Based on these results, it can be said that the triple deficit constrained/effects economic growth in SSA
countries in the long run. However, only budget balance has statistically significant at 10%and an increase
of 1% in the budget deficit decreases the real GDP growth on average by a rate of 28.7% in the short run
using MG2 estimation model. Other macroeconomic imbalances has no impact on economic growth in
the short run.

Table 5.15. Estimated long-run relationship and short run adjustment

(1) (2) (3) (4)

D.RGDPG D.RGDPG D.RGDPG D.RGDPG

D.lnCAB -12.291* -8.609 -1.232 -12.291*
(7.184) (5.451) (8.560) (7.184)
D.lnBB -4.590 -12.389** -28.668* -4.590
(10.696) (6.205) (15.255) (10.696)
D.lnSIB -0.865 -2.274 -5.770 -0.865
(4.018) (2.552) (5.693) (4.018)
ECT -0.752*** -0.834*** -0.824*** -0.752***
(0.094) (0.046) (0.060) (0.094)
lnCAB 0.038 -2.469 -23.810** 0.038
(13.658) (7.522) (11.792) (13.658)
lnBB 3.244 18.133** 47.893** 3.244
(18.326) (8.144) (18.955) (18.326)
lnSIB -0.559 5.516 16.234* -0.559
(8.803) (4.734) (8.947) (8.803)
Obs. 1225 1330 1225 1225
R-squared 0.518 0.529 0.666 0.518
RMSE 4.56 4.31 4.49
N 1330 1330 1225. 1225.
 116

cd 3.64 -2.38 -0.95

cdp 0.00 0.02 0.34
Standard errors are in parenthesis
*** p<0.01, ** p<0.05, * p<0.1

As stated earlier, in order to conclude that there is long run relationship between the three macroeconomic
imbalances and economic growth, the error-correction term should be negative and less than one as well
as statistically significant. As reported in the above Table 7, the error-correction terms in all model are
negative and statistically significant at 1% . This result indicates that there is a long run linear
combination among Real GDP growth, current account deficit, budget deficit and private
saving-investment deficit and thus triple deficits has an impact on economic growth for the
whole sample in SSA countries.

Regarding the impact of these three macroeconomic imbalances on economic growth in each country
under consideration, the error correction coefficient is negative but statistically insignificant for 33 coun-
tries out of 35 SSA countries, but it is statistically significant, negative in sign and less than one in value
for two countries (kenya, 10). Hence, for Kenya, and 10, our results indicate that long run relationship
between real GDP growth, current account, net government saving, net private saving.

Table 5.16. Dynamic Common Correlated Effect Mean Group Model (DCCEMG
Short run Error Long run
correc-
D.lnSIB lnCAB lnBB lnSIB
Estimator D.lnCAB D.lnBB tion term
-
-24.94** -5.74*** -0.95 -22.95 46.82 22.55
Benin 23.99***
Botswana -3.19*** 13.12 5.41** -0.77 -7.33*** 12.36 -6.15
Burkina Faso 13.67 -2.88*** -9.60 -1.05 -23.15 -3.82*** 8.94**
-
-34.85** -29.15** -0.41 -139.15 217.36 29.46**
Burundi 11.07***
-
-57.37 -52.14*** 0.66*** -0.49 -30.12*** 66.49***
Cameroon 32.92***
Cape Verde -0.27*** -15.33 -10.58 -0.51 -7.54*** 17.41 14.93
Central African Republic 106.27 -196.50 -140.60 -0.72 -327.23 543.38 268.02
Chad 30.63 -64.50 2.6.4*** -1.07 19.55 29.07 -30.50
-
-7.43*** 11.65*** -1.40 18.05*** 57.07*** 0.83***
Comoros 57.55***
Congo, Dem. Rep. 4.10** -0.27 -8.83 -0.14* -15.85*** 97.13* 43.71**
Congo, Rep. -0.64*** 3.49 -4.79 -0.49 -4.69*** 18.12 23.17
Cote d'Ivoire -3.01*** 8.79* -5.39*** -0.29 -8.36*** 1.39*** 19.28
 117

Eswatini -62.61 32.20 22.75 -0.46 73.28 -75.45* -61.09

Ethiopia -15.42 -34.78 -0.61*** -0.90 -25.23 53.24 7.35
Gabon -17.50 17.45 3.51*** -0.83 30.40 -39.38 -12.93
Gambia 98.23 -262.26 -55.16 -1.37 -71.20** 118.04 13.78***
Ghana 18.24 -15.22 -8.57 -1.34 -11.33 9.04 4.39
Kenya -2.94*** -4.58*** 2.83*** -0.24*** -14.57*** 24.40*** 22.89***
-
10.47*** -41.68 -0.63 5.72*** 13.88*** 8.38***
Lesotho 14.75***
Madagascar 23.02 -45.18 -19.83 -1.40 -10.26* -3.66*** 5.06***
Malawi 67.59 -29.85 3.87*** -0.96 -25.92 28.91 11.26
Mali -66.41 22.02 11.23 -1.00 23.55 -8.53** -12.79
Mauritius -8.35*** 8.64*** 2.12*** -0.84 -9.97*** 17.11*** 0.91***
Niger -22.81* 58.86 58.83 -1.19 13.94 -58.23 -38.44
Nigeria -2.90 2.29 1.85 -0.18 3.77*** 5.05*** -3.26***
Rwanda -145.78 -58.78 52.75 -1.37 -7.46*** 65.41 -26.11
Senegal -3.24*** -4.32*** -3.90*** -1.28 11.61 -19.78 -2.31***
Seychelles 20.69*** -198.56 -32.76* -0.90 -146.23 265.61 103.16
Sierra Leone 135.95 -298.99 -83.55 -0.65 -140.61 194.23 62.45
South Africa -3.90 2.25* 6.93 -0.96 -0.04*** 2.60 1.10***
Sudan -1.81*** 7.54 3.22 -0.77 0.39*** -0.58*** -3.75
Tanzania -2.24*** 4.91*** -0.12*** -0.79 1.47*** -6.94*** -0.83***
Togo -36.85 212.54 19.88* -0.90 -2.18*** 69.23 30.72
Uganda 10.15** -17.04 -5.90 -0.88 -2.95*** 20.39 -0.80***
Zambia -8.46 6.92** 9.67 -0.70 19.25 -1.71*** -1.70***

Notes: ***, **, * denote significance at 1%, 5% and 10% respectively

In addition, there are policy implications in these findings: the three gaps have a statistically significant
long run effect on economic growth (RGDPG). More importantly, authorities of this economy must put
in place a fully disciplined fiscal policy that should ensure the drastic curtailment of fiscal deficits and, at
the same time, create a conducive environment to attract foreign remittances and foreign investment,
which would help to generate healthy external balances. In addition, exchange rate stability can promote
the exports sector, minimizing external imbalances through creating critical surpluses in current accounts
and including related comprehensive discipline policies that may be pursued, which would enable the
external, private saving-investment, fiscal, and monetary sectors to perform without creating adverse im-
balances in this economy.
This study broadly relates to possible explanations for the divergent results among the many empirical
papers. The different findings may largely arise from the differences in methodology and data. In some
previous studies, the possibility of cross-sectional dependence and slope heterogeneity was ignored in the
 118

series. Further, analysis of data sets that focus on a short period of time may not yield reliable evidence.
Lack of longer-term data for countries, limits the possibility for clear-cut, differentiated results. Not to put
a fine point on it, but differences in econometric techniques, data measures, samples employed, etc. yield
different results.
All in all, based on our empirical findings, from a policy standpoint, such an evidence implies that the
causes of large and persistent external deficits must be sought somewhere else than the budget side of the
economy. Behind this, there might be several reasons, such as the structure of foreign trade, the exchange
rate regime pursued, and the international competitiveness of the particular country in question, the degree
of capital mobility. Nevertheless, it is obvious that the case for the twin or triple deficits hypotheses is
more likely to be seen in countries with economies that are highly integrated with international markets,
open to capital movements, and experience intensive international competitiveness.

5.6. SUMMARY, POLICY RECOMMENDATION, AND CONCLUSSION

The study examined the impact of macroeconomic imbalances (external and internal imbalances) on
economic growth and which had been the most significant and binding constraint on economic growth in
SSA countries between 1980 and 2018. A three-gap model was specified. Some of our equations were
estimated using the PVAR GMM estimating technique and examined the robustness of this results using
dynamic common correlated effect estimators. The Granger causality tests conducted found that the three
macroeconomic imbalances combined together to impose effects on output.
5.6.1. SUMMARY
According to the findings of PVAR GMM analysis, from the impulsive response function results, we find
evidence that a fiscal balance or current account balance shock of one standard deviation results in an
increase in real GDP growth by about 2.14% or 1.58% after ten years, respectively. Thus, budget balance
or current account balance and real GDP growth have a positive linear long run relationship. On the other
hand, real GDP growth responds negatively to private saving-investment balance in the long run and thus
an increase in private saving-investment balance causes a decline in real GDP growth by about 1.68%.
The variance decomposition results also shows that the 92.4%, variation in Real GDP is explained by
about 0.9%,2.4% and 4.3% by current account deficit, fiscal deficit and private saving-investment deficit)
, respectively just after ten quarter. It finds that private saving-investment balance seems to play especially
important role in explaining the variation of Real GDP growth. Next to private saving-investment bal-
ance, Budget balance also plays an important role in explaining the real GDP growth variation.
Based on the PVAR, the estimated strength of the impact of the macroeconomic imbalances on economic
growth is calculated and the values are taken in absolute terms. The indexes obtained shows that private
saving-investment gap was the most binding constraint on economic growth in SSA during the period
under review, while Fiscal gap and current account Gap followed in that order of importance. these results
are consistent with the results of variance decomposition function.
 119

According to the results of the robustness check using dynamic common correlated mean group estimator,
the long run coefficients of the variables included in this study reveal that one unit increase in Investment-
Savings gap will increase the real GDP growth by 16.2%. Also, a unit increases in Budgetary gap will
increase real GDP growth by 47.8%. However, a unit increases in Export-Import gap will decrease the real
GDP growth by 23.8%. from these results, we can conclude that these findings are not consistent with the
results of PVAR GMM.
Thus, suggesting that Investment-Savings gap had been the most binding constraint on economic growth
in SSA countries, while fiscal imbalance and current account imbalance followed in order of importance.
This result was consistent with what was reported by Chenery hypothesis, Chenery and Bruno (1962),
Adelman and Chenery (1966), Chenery and MacEwan (1966), Chenery and Stroiit (1966) Chenery and
Eckstein (1970) to the effect that countries in their pre-tame-off stage of development usually have
Investment-Savings gap predominance. This would suggest that such economies are suffering from low
monetized savings. Furthermore, our result invalidated the original hypothesis of Chenery and Bruno
(1962), Mckinnon (1964), Chenery (1967), El- Shibly and Thirwall (1981), and Mwega et al (1994), who
reported the predominance of current account imbalance. Thus, the SSA result suggested that although it
is a less developed economy, it has the pre-take off stage of development and it is suffering from low
monetized savings. To finance the gaps, SSA countries had been incurring both internal and external debts
and at times relying on meagre grants, though not reliable. The huge public debts with its burdens had
been posing threats to fiscal stability. We obtained that external debt burdens had been more severe than
internal debts, as the latter has been impairing capital accumulation, economic growth and fiscal stability.
5.6.2. POLICY IMPLICATIONS AND RECOMMENDATIONS
It is very important to put in place policy or support measures to address the economic problems of
macroeconomic imbalances and the attendant impacts on economic growth. In this regard we like to
highlight some policy implications and suggest the following recommendations. One, given the
desirability of ensuring that resources (savings from the government, private and foreign) are adequate to
increase real GDP growth and thus reduce output gap and promote sustainable economic growth, it is
imperative that the authorities cultivate the habit of monitoring the relationships between desired and
available resources. This would be a positive change away from the hitherto practice whereby policy
makers usually attempted to fill some gaps the depth and magnitude they did not have any quantitative
measure, the predictable end result of which had been failure and its attendant consequences. If the
relationships between desired and available resources as well as potential and actual output were well
monitored, more optimal utilization of resources would be guaranteed. The advantages inherent in the
above are numerous. Prominent is that the economy would by this set the stage for meaningful economic
development that would increase capacity utilization and total output. The fact that current account
imbalance is the most binding constraint on economic growth calls for policy action. The severity of the
imbalances is further underscored by the SSA countries. It is therefore imperative to diversify the economy
 120

away from the over reliance on primary products (agricultural and natural resources) export are
exogenously determined therefore making the revenue from it highly unreliable. SSA should put in place
efforts to diversify and adding values on this product and generate more foreign exchange revenue, for
example, introduction of agro-processesing and other manufacturing industries that promote export and
import substitutions. Also, the country should take another look at development of agriculture, which is a
prerequisite to sustainable industrial development. We recognize the fact that inflow of fund may not
correspond with outflow of fund all the time. Thus, the need to run deficit could not be totally ruled out.
In running fiscal deficits, therefore, high degree of caution should be exercised such that two issues are
given considerations. One, fiscal deficit should be run on productive ventures. Put differently, investment
projects that are self-liquidating should be considered. Two, utmost restraint should be exercised in
borrowing from outside the country. External borrowing, it has been established carries much more severe
burdens than internal borrowing. The relatively large size of the investment savings gap indexes no doubt
calls for policy action. The problem with saving and investment as argued earlier is that different economic
agents usually undertake each. This has an important policy implication. This is to the effect that the
measures to address savings mobilization are not necessarily the same measures to stimulate would-be-
investors to invest. It should be noted that a saving is a function of two main variables, namely, level of
disposable income and the marginal propensity to save. The latter could be further influenced by factors
namely interest rate, resource distribution and institutional mobilization. Policies should be put in place to
enable more labour to be employed, incentives should be given to labour to produce more and earn more.
In fact, at a higher level of income, people would be willing to save, as the opportunity cost of benefits
forgone would be lower if basic consumption needs could be easily satisfied.
Similarly, whether or not to invest in new physical capital, such as machinery, equipment, factories, stores,
and warehouses, depends on an important factor. The factor is whether the expected rate of profit on the
new investment is greater or less than the interest rate that must be paid on the funds that need to be
borrowed to acquire these assets. In fact, if the funds were readily available, a decision would have to be
taken between the alternatives of using the funds to purchase the new physical asset and of lending the
funds to someone else at the existing market rate of interest. Thus, policies should be directed at creating
investment friendly environment notably by ensuring stable and relatively low interest rate, reduce income
inequality and unemployment. Also, effort should be made to ensure political stability; this is with a view
to ensuring the sustainability of the investment friendly environment. The above policy implications and
recommendations have no doubt reiterated the imperativeness for the economic management authorities
to always attempt to pursue two issues vigorously. The first is economic growth, while the second is the
sustainability of the growth. During the period of the review, we could observe that SSA economy recorded
some positive economic growth, but the growth could not be sustained. The reason for the above was
largely due to inability of the economic management authorities to maintain the variables of economic
 121

growth in their desired quantity and quality. The consequences have manifested in terms of increased
poverty as observed in the low per capita income.
5.6.3. CONCLUSION
The study found that the internal deficit has a positive impact on economic growth, while external deficits
negatively affects the economic growth in SSA countries between 1980 and 2018. This suggests that the
economy was constrained by these macroeconomic imbalances and thus producing below what it could
produce, if all resources were in their desired state (adequate). If the economic problem of resource gaps
is addressed, output will increase and this will help to alleviate poverty, reduce unemployment and income
inequality.
 122

REFERENCES

Abell, J. D. (1990). Twin deficits during the 1980s: An empirical investigation. Journal of
Macroeconomics, 12(1), 81-96.
Acaravci, A., & Ozturk, I. (2008). Twin deficits phenomenon: Empirical evidence from the ARDL bound
test approach for Turkey. Bulletin of Statistics and Economics, 2(A08), 57-64.
Adams, S., Sakyi, D., & Opoku, E. E. O. (2016). Capital inflows and domestic investment in sub-Saharan
Africa: evidence from pooled mean group (PMG) estimation approach. Foreign Trade Review,
51(4), 328-343.
Adelman, I., & Chenery, H. B. (1966). The foreign aid and economic development: the case of Greece.
The review of Economics and Statistics, 1-19.
AKBAŞ, Y. E., Fuat, L., & ZEREN, F. (2014). Testing the validity of the triplet deficit hypothesis for
Turkey: Asymmetric causality analysis. Eurasian Journal of Business and Economics, 7(14), 137-
154.
Akbostanci, E., & Tunc, G. (2002). Turkish Twin Deficits: An Error Correction Model of Trade Balance.
No. 106. Retrieved from
AlENcAr, D. A., & Strachman, E. (2014). Balance-of-payments-constrained growth in Brazil: 1951-2008.
Journal of Post Keynesian Economics, 36(4), 673-698.
Alonso, J. A. (1999). Growth and the external constraint: lessons from the Spanish case. Applied
Economics, 31(2), 245-253.
Anoruo, E., & Ramchander, S. (1998). Current account and fiscal deficits: Evidence from five developing
economies of Asia. Journal of Asian Economics, 9(3), 487-501.
Araujo, R. A., & Lima, G. T. (2007). A structural economic dynamics approach to balance-of-payments-
constrained growth. Cambridge Journal of Economics, 31(5), 755-774.
Arezki, R., & Ismail, K. (2013). Boom–bust cycle, asymmetrical fiscal response and the Dutch disease.
Journal of Development Economics, 101, 256-267.
Arize, A. C. (2002). Imports and exports in 50 countries: tests of cointegration and structural breaks.
International Review of Economics & Finance, 11(1), 101-115.
Atesoglu, H. S. (1993a). Balance-of-payments-constrained growth: Evidence from the United States.
Journal of Post Keynesian Economics, 15(4), 507-514.
Atesoglu, H. S. (1993b). Exports, capital flows, relative prices, and economic growth in Canada. Journal
of Post Keynesian Economics, 16(2), 289-297.
Atesoglu, H. S. (1993c). Manufacturing and economic growth in the United States. Applied Economics,
25(1), 67-69.
 123

Bagnai, A. (2016). Italy’s decline and the balance-of-payments constraint: a multicountry analysis.
International Review of Applied Economics, 30(1), 1-26.
Bagnai, A., Rieber, A., & Tran, T. A.-D. (2016). Sub-Saharan Africa’s growth, South–South trade and the
generalised balance-of-payments constraint. Cambridge Journal of Economics, 40(3), 797-820.
Bairam, E. (1988). Government expenditure and economic growth: reflections on Professor Ram's
approach, a new framework and some evidence from New Zealand time-series data. Keio
Economic Studies, 25(1), 59-66.
Bajo-Rubio, O. (2012). The balance-of-payments constraint on economic growth in a long-term
perspective: Spain, 1850–2000. Explorations in Economic History, 49(1), 105-117.
Barbosa-Filho, N. H. (2001). Effective demand and growth: An analysis of the alternative closures of
Keynesian models. Center for Economic Policy Analysis New School University.
Barro, R. J. (1974). Are government bonds net wealth? Journal of Political Economy, 82(6), 1095-1117.
Barro, R. J. (1979). On the determination of the public debt. Journal of Political Economy, 87(5, Part 1),
940-971.
Barro, R. J. (1989). The Ricardian approach to budget deficits. Journal of Economic Perspectives, 3(2),
37-54.
Bartoli, G. (1989). Fiscal expansion and external current account imbalances. Fiscal Policy, Stabilization,
and Growth in Developing Countries, 171.
Basu, S., & Datta, D. (2005). Does fiscal deficit influence trade deficit?: An econometric enquiry.
Economic and Political Weekly, 3311-3318.
Berben, R.-P., & Brosens, T. (2007). The impact of government debt on private consumption in OECD
countries. Economics Letters, 94(2), 220-225.
Bernheim, B. D. (1987). Ricardian equivalence: An evaluation of theory and evidence. NBER
macroeconomics annual, 2, 263-304.
Bértola, L., Higachi, H., & Porcile, G. (2002). Balance-of-payments-constrained growth in Brazil: a test
of Thirlwall's Law, 1890-1973. Journal of Post Keynesian Economics, 25(1), 123-140.
Blecker, R. A., & Ibarra, C. A. (2013). Trade liberalization and the balance of payments constraint with
intermediate imports: the case of Mexico revisited. Structural Change and Economic Dynamics,
25, 33-47.
Blomqvist, Å. (1976). Empirical evidence on the two-gap hypothesis: A revised analysis. Journal of
Development Economics, 3(2), 181-193.
Bolat, S., Değirmen, S., & Şengönül, A. (2014). Does triple deficits have (UN) stable causality for the EU
members? Evidence from bootstrap-corrected causality tests. Procedia Economics and Finance,
16, 603-612.
Britto, G., & McCombie, J. S. (2009). Thirlwall's law and the long-term equilibrium growth rate: an
application to Brazil. Journal of Post Keynesian Economics, 32(1), 115-136.
 124

Britto, G., & McCombie, J. S. (2013). Thirlwall's law and the long-term equilibrium growth rate: an
application to Brazil. An addendum. Journal of Post Keynesian Economics, 35(4), 695-696.
Buchanan, J. M. (1976). Barro on the Ricardian equivalence theorem. Journal of Political Economy, 84(2),
337-342.

Catık, A. N., Gok, B., & Akseki, U. (2015). A Nonlinear Investigation of the Twin Deficits Hypothesis
over the Business Cycle:Evidence from Turkey. Economic Systems, 39(1), 181–196.

Chenery, H. B. (1961). Comparative advantage and development policy. The American Economic Review,
51(1), 18-51.
Chenery, H. B., & Eckstein, P. (1970). Development Alternatives for Latin America. Journal of Political
Economy, 78(4, Part 2), 966-1006.
Chenery, H. B., & MacEwan, A. (1966). Optimal patterns of growth and aid the case of Pakistan. The
Pakistan Development Review, 6(2), 209-242.
Chowdhury, K., & Saleh, A. S. (2007). Testing the Keynesian proposition of twin deficits in the presence
of trade liberalisation: evidence from Sri Lanka.
Christopher Walker, W. (2002). Ricardian equivalence and fiscal policy effectiveness in Japan. Asian
Economic Journal, 16(3), 285-302.
Chu, Y.-h. (1989). State structure and economic adjustment of the East Asian newly industrializing
countries. International Organization, 43(4), 647-672.
Cooper, R.N.(2001).“Is the U.S. Current Account Deficit Sustainable? Will it be Sustained?” Brookings
Papers on Economic Activity 2001(1): 217-226.
Crisp, B. F., & Kelly, M. J. (1999). The socioeconomic impacts of structural adjustment. International
Studies Quarterly, 43(3), 533-552.
Darrat, A. F. (1988). Have large budget deficits caused rising trade deficits? Southern Economic Journal,
879-887.
De la Escosura, L. P., & Rosés, J. R. (2010). Capital accumulation in the long run: The case of Spain,
1850–2000. Res. Econ. Hist, 27, 141-200.
Dhaene, G., & Jochmans, K. (2015). Split-panel jackknife estimation of fixed-effect models. The Review
of Economic Studies, 82(3), 991-1030.
Ditzen, J. (2018). Estimating dynamic common-correlated effects in Stata. The Stata Journal, 18(3), 585-
617.
Dumitrescu, E.-I., & Hurlin, C. (2012). Testing for Granger non-causality in heterogeneous panels.
Economic Modelling, 29(4), 1450-1460.
Easterly, W. (2003). IMF and World Bank structural adjustment programs and poverty. In Managing
currency crises in emerging markets (pp. 361-392): University of Chicago Press.
Eisner, R. (1989). Budget deficits: rhetoric and reality. Journal of Economic Perspectives, 3(2), 73-93.
 125

Eisner, R. (1994). “National Saving and Budget Deficits.” The Review of Economics and Statistics 76(1):
181-186.
El-Shibly, M., & Thirlwall, A. (1981). Dual-Gap analysis for the Sudan. World Development, 9(2), 193-
200.
Elliott, D. R., & Rhodd, R. (1999). Explaining growth rate differences in highly indebted countries: an
extension to Thirlwall and Hussain. Applied Economics, 31(9), 1145-1148.
Elmendorf, D. W., & Mankiw, N. G. (1999). Government debt. Handbook of macroeconomics, 1, 1615-
1669.
Elwell, C.K. (2010). “The U.S. Trade Deficit: Causes, Consequences and Policy Options”, CRS Report
for Congress.
Epaphra, M. (2017). Analysis of budget deficits and macroeconomic fundamentals: A VAR-VECM
approach. Journal of Economics & Management, 30, 20-57.
Fabayo, J. A. (1981). Industrial Capacity Utilisation in a Developing Economy: A case of the Nigerian Cotton Textile
Industry. The Journal of Business and Social Studies, 4(22), 53-64.
Fidrmuc, J. (2003). The Feldstein–Horioka puzzle and twin deficits in selected countries. Economics of
Planning, 36(2), 135-152.
Filho, S. F. M., Jayme, F. G., & Libânio, G. (2013). Balance-of-payments constrained growth: a post
Keynesian model with capital inflows. Journal of Post Keynesian Economics, 35(3), 373-398.
Gabrisch, H. (2015). On the twin deficits hypothesis and the import intensity in transition countries.
International Economics and Economic Policy, 12(2), 205-220.
García-Molina, M., & Ruíz-Tavera, J. K. (2009). Thirlwall's law and the two-gap model: toward a unified"
dynamic gap" model. Journal of Post Keynesian Economics, 32(2), 269-290.
Garcimartín, C., Alonso, J. A., & Rivas, L. (2014). Balance-of-payments-constrained growth and
convergence: one more piece of the jigsaw. Journal of Post Keynesian Economics, 36(3), 555-
585.
Giorgioni, G., & Holden, K. (2003). Does the Ricardian equivalence proposition hold in less developed
countries? International Review of Applied Economics, 17(2), 209-221.
Gould, D. M., & Ruffin, R. J. (1993). What determines economic growth. Economic Review, 2.
Gruber, J. W., & Kamin, S. B. (2007). Explaining the global pattern of current account imbalances.
Journal of International Money and Finance, 26(4), 500-522.
Guarini, G., & Porcile, G. (2016). Sustainability in a post-Keynesian growth model for an open economy.
Ecological Economics, 126, 14-22.
Gurgul, H., Lach, Ł., & Mestel, R. (2012). The relationship between budgetary expenditure and economic
growth in Poland. Central European Journal of Operations Research, 20(1), 161-182.
Hakkio, C.S. (1995). “The U.S. CurrentAccount: The Other Deficit.” Economic Review 80:1124.
Harrod, R. F. (1933). The Theory of Monopolistic Competition. In: JSTOR.
 126

Harrod, R. F. (1939). An essay in dynamic theory. The Economic Journal, 49(193), 14-33.
Hatsopoulos, G.N., Krugman, P.R. & Summers, L.H. (1988). “U.S. Competitiveness: Beyond the Trade
Deficit.” Science 241(4863): 299-307.
Hieke, H. (1997). Balance-of-payments–Constrained Growth: A Reconsideration of the Evidence for the
US Economy. Journal of Post Keynesian Economics, 19(3), 313-325.
Higgins, M. & Klitgaard, T. (1998). “Viewing the Current Account Deficit as a Capital Inflow.” Current
Issues in Economics and Finance 4(13): 1-6.
Hubbard, R. G. (2006). The US current account deficit and public policy. Journal of Policy Modeling,
28(6), 665-671.
IMF. (2018). World economic outlook. In: IMF Washington, DC.
Iram, M., Shadid, A., Mahpara, S., & Fazli, R. (2011). Old wine in new bottles: testing the keynesian
preposition of twin deficit in case of Pakistan. International Journal of Business and Social
Science, 2(5), 209.
Jansen, K. (1989). Financial development and the intersectoral transfer of resources: the case of Thailand.
Development and Change, 20(1), 5-34.
Jayaraman, T. K., Choong, C.-K., & Law, S.-H. (2010). Testing the validity of twin deficit hypothesis in
pacific Island countries: An empirical investigation. Economics Bulletin, 30(2), 1233-1248.
Kaldor, N. (1970). Some fallacies in the interpretation of Kaldor. The Review of Economic Studies, 37(1),
1-7.
Kayhan, S., Bayat, T., & Yüzbaşı, B. (2013). Government expenditures and trade deficits in Turkey: Time
domain and frequency domain analyses. Economic Modelling, 35, 153-158.
Khasawneh, A. Y., Magableh, I. K., Khrisat, F. A., & Massadeh, D. D. (2012). Validity of Thirlwall’s
Law in MENA Countries. Journal of Economic Cooperation and Development, 33(3), 97-114.
Krugman, P. (1989). The case for stabilizing exchange rates. Oxford Review of Economic Policy, 5(3), 61-
72.
Kuijs, L. (2006). How will China's saving-investment balance evolve? : The World Bank.
Labonte, M. (2005). “Is the U.S. Trade Deficit Caused by a Global Saving Glut?”, CRS Report for
Congress.
Lam, N. X. (2012). Twin deficits hypothesis and Feldstein-Horioka puzzle in Vietnam. International
Research Journal of Finance and Economics, 101, 169-179.
Landau, L. (1971). Saving Functions for Latin America: Differences in Savings Ratios among Latin
American Countries. Studies in Development Planning, Cambridge.
Lanzafame, M. (2014). The balance of payments-constrained growth rate and the natural rate of growth:
new empirical evidence. Cambridge Journal of Economics, 38(4), 817-838.
Leachman, L. L. (1996). New evidence on the Ricardian equivalence theorem: a multicointegration
approach. Applied Economics, 28(6), 695-704.
 127

Leiderman, L., & Blejer, M. I. (1988). Modeling and testing Ricardian equivalence: a survey. Staff Papers,
35(1), 1-35.
Leon-Ledesma, M. A. (1999). An application of Thirlwall’s Law to the Spanish economy. Journal of Post
Keynesian Economics, 21(3), 431-439.
Lesiit, M. (1990). Determinants of the current account of the balance of payments: an analysis for Kenya,
1973-88'. MA research paper, University of Nairobi.
Levy, V. (1984). The savings gap and the productivity of foreign aid to a developing economy: Egypt.
The Journal of Developing Areas, 19(1), 21-34.
M. Ghani, G. (2006). Balance of payments constrained growth model: an examination of Thirlwall's
Hypothesis using McCombie's Individual Country Method. Applied Economics Letters, 13(12),
763-768.
Magazzino, C. (2012). Wagner versus Keynes: Public spending and national income in Italy. Journal of
Policy Modeling, 34(6), 890-905.
Mann, C.L. (2002). “Perspectives on the U.S. Current Account Deficit and Sustainability.” Journal of
Economic Perspectives 16(3): 131-152.
Marinheiro, C. F. (2008). Ricardian equivalence, twin deficits, and the Feldstein–Horioka puzzle in Egypt.
Journal of Policy Modeling, 30(6), 1041-1056.
McCombie, J. (1981). Are international growth rates constrained by the balance of payments? A comment
on professor THIRLWALL. PSL Quarterly Review, 34(139).
McCombie, J., & Thirlwall, T. (2004). Essays on balance of payments constrained growth: theory and
evidence: Routledge.
McCombie, J. S. (1993). Economic growth, trade interlinkages, and the balance-of-payments constraint.
Journal of Post Keynesian Economics, 15(4), 471-505.
McCombie, J. S., & Thirlwall, A. P. (1997). The dynamic Harrod foreign trade multiplier and the demand-
orientated approach to economic growth: an evaluation. International Review of Applied
Economics, 11(1), 5-26.
McCombie, J. S. (2012). Balance-of-payments-constrained economic growth. Edward Elgar Publishing
Limited.

McGregor, P. G., & Swales, J. K. (1991). Thirlwall's law and balance of payments constrained growth:
further comment on the debate. Applied Economics, 23(1), 9-20.
McKinnon, R. I. (1964). Foreign exchange constraints in economic development and efficient aid
allocation. The Economic Journal, 74(294), 388-409.
Milesi‐Ferretti, G. M., & Razin, A. (1996). Persistent current account deficits: a warning signal?
International Journal of Finance & Economics, 1(3), 161-181.
 128

Moran, C. (1989). Imports under a foreign exchange constraint. The World Bank Economic Review, 3(2),
279-295.
Moreno-Brid, J. C. (1998). On capital flows and the balance-of-payments-constrained growth model.
Journal of Post Keynesian Economics, 21(2), 283-298.
Mukhtar, T., Zakaria, M., & Ahmed, M. (2007). AN EMPIRICAL INVESTIGATION FOR THE TWIN DEFICITS
HYPOTHESIS IN PAKISTAN. Journal of Economic Cooperation Among Islamic Countries, 28(4).
Mwega, F., Mwangi, N., & Olewe-Ochilo, F. (1994). Macroeconomic Constraints and Medium Term
Growth in Kenya: A Three-Gap Model. African Economic Research Consortium. Nairobi.
Narayan, P. K., Prasad, R. N., & Prasad, A. (2009). The monetary approach to understanding the balance
of payments for the Fiji Islands. Pacific Economic Bulletin, 24(2), 113-121.
Narayan, S. (2013). A structural VAR model of the Fiji Islands. Economic Modelling, 31, 238-244.
Ndulu, B. J. (1986). Investment, output growth and capacity utilization in an African economy: The case
of manufacturing sector in Tanzania. Eastern Africa Economic Review, 2(1), 14-30.
Ndulu, B. J. (1991). Growth and adjustment in sub-Saharan Africa. Economic Reform in Sub-Saharan
Africa, Washington, DC: World Bank, 287-302.
Onafowora, O. A., & Owoye, O. (2006). An empirical investigation of budget and trade deficits: the case
of Nigeria. The Journal of Developing Areas, 153-174.
Otadi, N., Hassani, A., Javid, A., & Khiabani, F. (2010). Oily compounds removal in wastewater treatment
system of pars oil refinery to improve its efficiency in a lab scale pilot. Journal of water chemistry
and technology, 32(6), 370-377.
Oyejide, T., & Raheem, M. (1990). Macro-economic constraints and Growth Programming: Empirical
Evidence from Nigeria. Revised draft of the paper prepared for a Research Project on Medium
Term Adjustment Strategy sponsored by WIDER Helsinki.
Palley, T. I. (2003). Pitfalls in the theory of growth: an application to the balance of payments constrained
growth model. Review of Political Economy, 15(1), 75-84.
Perraton, J. (2003). Balance of payments constrained growth and developing countries: an examination of
Thirlwall's hypothesis. International Review of Applied Economics, 17(1), 1-22.
Phillips, P. C., & Sul, D. (2007). Transition modeling and econometric convergence tests. Econometrica,
75(6), 1771-1855.
Podkaminer, L. (2017). “Thirlwall’s Law” reconsidered. Empirica, 44(1), 29-57.
Prakash, K. A., & Maiti, D. S. (2013). Oil price shocks and effectiveness of devaluation in Fiji.
Prebisch, R. (1950). The economi development of Latin America and it principal problems. Economic
Bulletin for Latin America, 7(1), 385-421.
Prebisch, R. (1959). Commercial policy in the underdeveloped countries. The American Economic
Review, 49(2), 251-273.
 129

Rattsø, J. (1994). Medium-run adjustment under import compression: Macroeconomic analysis relevant
for sub-Saharan Africa. Journal of Development Economics, 45(1), 35-54.
Ravinthirakumaran, N., Selvanathan, S., & Selvanathan, E. A. (2016). The twin deficits hypothesis in the
SAARC countries: an empirical investigation. Journal of the Asia Pacific Economy, 21(1), 77-90.
Razmi, A. (2011). Exploring the robustness of the balance of payments-constrained growth idea in a
multiple good framework. Cambridge Journal of Economics, 35(3), 545-567.
Reinhart, C. M., & Trebesch, C. (2016). The international monetary fund: 70 years of reinvention. Journal
of Economic Perspectives, 30(1), 3-28.
Ricardo, D. (1966). Economic essays: New York: Kelley.
Romero, J. P., & McCombie, J. S. (2016). The Multi-Sectoral Thirlwall’s Law: evidence from 14
developed European countries using product-level data. International Review of Applied
Economics, 30(3), 301-325.
Roubini, N. (1988). “Current Account and Budget Deficits in an Intertemporal Model of Consumption
and Taxation Smoothing: A Solution to the Feldstein- Horioka Puzzle?” NBER Working Paper,
No. 2773, Camberidge, MA.
Roubini, N., & Wachtel, P. (1999). Current-account sustainability in transition economies. In Balance of
payments, exchange rates, and competitiveness in transition economies (pp. 19-93): Springer.
Roy, A., & Gupta, S. D. (2013). An empirical investigation of budget and trade deficits: the case of Bang-
ladesh. International Journal of Economics and Financial Issues, 3(3), 570-579.

Saleh, A. S. (2006). Long-run linkage between budget deficit and trade deficit in Lebanon: Results from
the UECM and Bounds tests. International Journal of Economics, Management and Accounting,
14(1).
Samadi, A. H. (2006). The twin deficits phenomenon in some MENA countries. Iranian Economic
Review, 10(16), 129-140.
Sarafidis, V., & Wansbeek, T. (2012). Cross-sectional dependence in panel data analysis. Econometric
Reviews, 31(5), 483-531.
Seater, J. J. (1993). Ricardian equivalence. Journal of Economic Literature, 31(1), 142-190.
Solimano, A. (1990). Macroeconomic constraints for medium-term growth and distribution: A model for
Chile: World Bank Publications.
Soukiazis, E., & Antunes, M. (2011). Is foreign trade important for regional growth? Empirical evidence
from Portugal. Economic Modelling, 28(3), 1363-1373.
Soukiazis, E., Cerqueira, P. A., & Antunes, M. (2012). Modelling economic growth with internal and
external imbalances: empirical evidence from Portugal. Economic Modelling, 29(2), 478-486.
 130

Soukiazis, E., Cerqueira, P. A., & Antunes, M. (2013). Growth rates constrained by internal and external
imbalances and the role of relative prices: empirical evidence from Portugal. Journal of Post
Keynesian Economics, 36(2), 275-298.
Soukiazis, E., Cerqueira, P. A., & Antunes, M. (2014). Explaining Italy's economic growth: A balance-
of-payments approach with internal and external imbalances and non-neutral relative prices.
Economic Modelling, 40, 334-341.
Taylor, L. (1990). Foreign resource flows and developing country growth: a three-gap model. Problems
of Developing Countries in 1990s, World Bank Discussion Paper, 97.
Taylor, L. (1990). A three-gap model. Problems of Developing Countries in the 1990s. Washington, DC:
World Bank, 55-90.
Taylor, L. (1992). A Gap Model Approach. Paper presented at the Nonlinearities, Disequilibria and
Simulation: Proceedings of the Arne Ryde Symposium on Quantitative Methods in the
Stabilization of Macrodynamic Systems.
Taylor, L. (1993). A three-gap analysis of foreign resource flows and developing country growth. The
Rocky Road to Reform: Adjustment, Income Distribution, and Growth in the Developing World,
9-37.
Taylor, L. (1994). Gap models. Journal of Development Economics, 45(1), 17-34.
Theil, H. (1958). Economist Forecasts and Policy. In: North-Holland Publishing Company, Amsterdam-
London.
Thirlwall, A. P. (1986). A general model of growth and development on Kaldorian lines. Oxford Economic
Papers, 38(2), 199-219.
Thirlwall, A. P. (1991). Professor Krugman’s 45-degree rule. Journal of Post Keynesian Economics,
14(1), 23-28.
Thirlwall, A. P. (1997). Reflections on the Concept of Balance-Of-Payments–Constrained Growth.
Journal of Post Keynesian Economics, 19(3), 377-385.
Thirlwall, A. P. (2001). The relation between the warranted growth rate, the natural rate, and the balance
of payments equilibrium growth rate. Journal of Post Keynesian Economics, 24(1), 81-88.
Thirlwall, A. P. (2012). Balance of payments constrained growth models: history and overview. In Models
of balance of payments constrained growth (pp. 11-49): Springer.
Thirlwall, A. P. (2013). Economic growth in an open developing economy: the role of structure and
demand: Edward Elgar Publishing.
Thirlwall, A. P., & Hussain, M. N. (1982). The balance of payments constraint, capital flows and growth
rate differences between developing countries. Oxford Economic Papers, 34(3), 498-510.
Tinbergen, J. (1956). On the theory of income distribution. Weltwirtschaftliches Archiv, 155-175.
Vamvoukas, G. A. (1999). The twin deficits phenomenon: evidence from Greece. Applied Economics,
31(9), 1093-1100.
 131

Vanek, J. (1967). Estimating foreign resource needs for economic development: theory, method, and a
case study of Colombia: New York: McGraw-Hill.
Verma, R., & Jayaraman, T. K. (2010). An empirical analysis of sustainability of Fiji’s trade deficits. The
Asia-Pacific Economic Journal, 8(1), 29-40.
Vidal Alejandro, P. (2016). Cuba’s reform and economic growth: a comparative perspective with Vietnam.
Journal of Economic Policy Reform, 19(2), 148-165.
Waelbroeck, J. (1984). Capital, foreign exchange, and growth: the two-gap and labor-income-floor views.
In Economic structure and performance (pp. 281-294): Elsevier.
Wawire, N. (2006). Determinants of tax revenues in Kenya. Unpublished Ph. D Thesis, Kenyatta
University.
Weisskopf, T. E. (1972). An econometric test of alternative constraints on the growth of underdeveloped
countries. The review of Economics and Statistics, 67-78.
World Bank, T. (2018). World Bank open data. World Bank Web site (INTERNET).
Yi, C.-D. (2003). An empirical analysis of Ricardian equivalence on real exchange rate and current
account: Korea. International Economic Journal, 17(4), 61-83.
 132

Appendix

Table 4.2: De-

scriptive Statistics
CAB_GDP BB_GDP SIB_GDP
By CountrySam-
ple
Benin -8.60 -3.26 7.18
Botswana 3.68 5.91 -4.13
Burkina Faso -7.77 -3.86 3.44
Burundi -10.00 -7.96 20.26
Cameroon -2.46 -1.33 -0.35
Cape Verde -10.55 -8.58 11.42
Central A.R -6.93 -2.50 3.68
Chad 13.82 -3.70 12.31
Comoros -5.66 -1.43 1.86
Congo, D.Rep. -4.04 -4.23 19.16
Congo, Rep. -4.16 -2.17 6.44
Cote d'Ivoire -0.56 -5.44 -4.73
Eswatini -2.52 -1.45 15.84
Ethiopia -4.48 -4.51 0.38
Gabon 8.09 0.84 -3.29
Gambia -7.30 -4.50 -1.11
Ghana -5.96 -8.64 -3.54
Kenya -4.23 -4.73 -0.01
Lesotho -11.35 -3.63 1.07
Madagascar -8.26 -5.25 3.23
Malawi -9.51 -4.83 -0.82
Mali -5.82 -3.57 39.79
Mauritius -3.76 -4.95 -1.50
Niger -10.75 -4.12 6.33
Nigeria 1.86 -1.50 -13.16
Rwanda -7.89 -4.54 1.95
 133

Senegal -7.87 -3.09 7.00

Seychelles -14.85 -2.23 13.48
Sierra Leone -10.66 -8.73 2.90
South Africa -1.49 -3.79 -1.26
Sudan -14.25 -5.09 -4.68
Tanzania -5.90 -3.46 5.16
Togo -11.47 -4.42 6.89
Uganda -4.87 -4.54 -0.16
Zambia -7.21 -7.58 14.20

Lag 1 Lag 2
CAB_
GDP BB_GDP
BB_G CAB_G BB_GDP SIB_GDP SIB_GDP CAB_GDP CAB_GDP BB_GDP BB_GDP
DP DP SIB_GDP BB_GDP CAB_GDP SIB_GDP BB_GDP CAB_GDP SIB_GDP
country Wi Wi Wi Wi Wi Wi Wi Wi Wi
Benin 0.65 7.33*** 0.76 12.13*** 0.74 4.15 1.55 8.62*** 1.38
Botswana 0.34 1.55 3.24* 0.82 1.23 0.08 3.94 11.89*** 6.89**
Burkina Faso 0.01 1.27 1.41 0.21 0.42 0.08 0.19 3.55 2.30
Burundi 0.15 0.06 0.84 0.72 0.64 1.07 0.68 0.20 0.88
Cameroon 1.32 19.83*** 6.34*** 8.47*** 7.21*** 1.83 0.70 3.93 0.36
Cape Verde 1.00 0.24 0.30 2.22 0.10 1.59 0.99 6.28* 1.69
Central African Re-
public 0.24 2.82* 1.97 1.02 2.58 0.24 0.72 2.76 2.77
Chad 0.60 0.02 0.04 1.18 0.34 0.00 1.29 0.46 1.36
Comoros 2.48 0.00 1.13 0.39 0.30 4.12 5.98* 0.68 4.87*
Congo, Dem. Rep. 0.01 3.53* 0.03 0.76 2.86* 1.07 1.19 15.62 0.04
Congo, Rep. 3.78 0.27 3.85* 1.94 0.43 5.21 4.61 0.86 8.42
Cote d'Ivoire 0.06 0.02 0.72 0.74 0.31 4.31 2.29 3.86 3.33
Eswatini 6.05 2.15 0.13 1.24 0.33 6.55*** 4.29 0.57 0.97
Ethiopia 1.05 0.02 0.64 1.32 1.82 0.70 1.56 4.58 1.78
Gabon 0.89 0.03 0.13 0.73 0.02 1.60 2.02 1.73 1.02
Gambia 0.01 5.24 10.85*** 0.00 0.39 10.44*** 0.06 4.66 20.82***
Ghana 0.01 2.15 1.04 0.13 2.31 0.27 6.66 4.00 2.22
Kenya 0.00 4.32 1.46 0.53 0.20 3.09* 1.04 6.93 4.52
Lesotho 0.02 0.26 0.06 0.15 0.00 1.08 0.01 0.27 3.84
Madagascar 0.09 0.00 0.05 0.00 0.11 0.00 0.62 0.26 0.13
Malawi 1.70 0.19 1.25 0.02 4.97** 0.13 1.75 0.23 1.48
 134

Mali 1.50 1.17 0.11 2.39 5.44** 1.94 5.50* 1.58 0.45
Mauritius 0.00 1.97 0.52 1.62 2.19 0.00 0.98 1.34 0.68
Niger 2.06 0.24 1.67 1.82 2.03 0.00 3.26 0.72 2.91
Nigeria 0.10 0.30 0.00 3.06* 1.01 0.02 0.71 1.57 1.71
Rwanda 0.00 5.46 1.66 0.00 1.08 0.90 0.67 6.19* 2.99
Senegal 3.47 4.90 2.00 0.86 1.17 2.66 7.21 6.00* 8.23
Seychelles 1.30 0.43 0.00 0.09 1.06 1.16 1.97 1.39 4.73
Sierra Leone 0.71 0.43 0.42 1.07 0.44 0.63 4.25 0.54 0.49
South Africa 2.98 0.06 0.79 0.20 0.53 0.01 2.81 0.54 0.82
Sudan 1.57 6.37*** 4.86 15.31*** 14.79*** 1.45 11.33 3.70 4.92*
Tanzania 0.57 1.68 0.06 2.14 0.10 0.24 0.98 2.47 0.66
Togo 0.06 0.07 0.23 1.02 0.05 0.18 0.73 0.87 0.44
Uganda 6.30 0.65 5.39 0.30 0.01 2.44 6.68 0.30 7.09
Zambia 0.44 0.47 0.03 0.01 0.18 1.76 3.02 2.97 1.18