P a g e | 1
NKALU, Chigozie Nelson PG/MSc/14/69634
THE EFFECTS OF BUDGET DEFICITS ON SELECTED MACROECONOMIC VARIABLES IN
NIGERIA AND GHANA
DEPARTMENT OF ECONOMICS
FACULTY OF THE SOCIAL SCIENCES
Ebere Omeje Digitally Signed by: Content manager’s Name DN : CN = Webmaster’s name O= University of Nigeria, Nsukka OU = Innovation Centre
P a g e | 2
THE EFFECTS OF BUDGET DEFICITS ON SELECTED MACROECONOMIC
VARIABLES IN NIGERIA AND GHANA
By
NKALU, Chigozie Nelson
PG/MSc/14/69634
DEPARTMENT OF ECONOMICS
FACULTY OF THE SOCIAL SCIENCES
UNIVERSITY OF NIGERIA, NSUKKA
SEPTEMBER, 2015
P a g e | 3
THE EFFECTS OF BUDGET DEFICITS ON SELECTED MACROECONOMIC
VARIABLES IN NIGERIA AND GHANA
AnMSc. Thesis
By
NKALU, Chigozie Nelson
PG/MSc/14/69634
DEPARTMENT OF ECONOMICS
FACULTY OF THE SOCIAL SCIENCES
UNIVERSITY OF NIGERIA, NSUKKA
SUPERVISOR: PROF. C. C. AGU
SEPTEMBER, 2015
P a g e | 4
TITLE PAGE
The Effects of Budget Deficits on Selected Macroeconomic Variables in
Nigeria and Ghana
P a g e | 5
CERTIFICATION
This is to certify that NKALU, Chigozie Nelson, a post-graduate student of the Department
of Economics, University of Nigeria, Nsukka with Registration Number: PG/MSc/14/69634
has satisfactorily completed the requirements for the award of Master of Science (MSc.) in
Economics
____________________________ ____________________
Nkalu, Chigozie Nelson Date
PG/MSc/14/69634
(The Researcher)
________________ ___________________
Prof. C. C. Agu Date
(Supervisor)
________________ ___________________
Prof. S. I. Madueme Date
(Head of Department)
P a g e | 6
APPROVAL PAGE
This research work titled: “The Effects of Budget Deficits on Selected Macroeconomic
Variables in Nigeria and Ghana” has followed due process and has been approved to have
met the minimum requirement for the award of the Master of Science degree in the
Department of Economics, Faculty of the Social Sciences, University of Nigeria, Nsukka.
________________ ___________________
Prof. C. C. Agu Date
(Supervisor)
________________ ___________________
Prof. S. I. Madueme Date
(Head of Department, Economics)
________________ ____________________
Prof. A. I. Madu Date
(Dean, Faculty of the Social Sciences)
________________ ____________________
Date
(External Examiner)
P a g e | 8
ACKNOWLEDGMENT
This study would not have been a huge success without the inputs of some men of goodwill.
Let me first and foremost express my deep gratitude to my family for whom I derive
unquenchable zeal to surmount some constraints in the course of this study. Special thanks to
my mother, brothers and sisters for the much love and encouragements which indeed has
reshaped my entire being. My most and sincere thanksgiving goes to the Almighty God for
His infinite mercies throughout my existence.
I must as a matter of fact recognize my able and capable supervisor and academic adviser –
Prof. C. C. Agu for his ever and relentless guidance and mentorship. My heartfelt gratitude
goes to my brothers, lecturers and senior colleagues in the Department of Economics,
University of Nigeria, Nsukka for their immeasurable contributions towards making this
study a colossal success. Special thanks to Dr. Jude O. Chukwu, Dr. Richard K. Edeme, Dr,
Emmanuel Nwosu, Dr. Ezebuilo R. Ukwueze, Dr. I. Ifelunini, Dr. (Mrs) Gladys Aneke, Prof.
(Mrs) S. I. Madueme, Miss Anaduaka Uchechi S., Emecheta Chisom, Nnetu Vivian, Mr.
Ekene and Mr. Nchege Johnson. Regrettably enough, others whose names cannot contain on
this page due to time and space constraints should bear with me.
Thank you all, and may our good Lord shower you with abundant blessings - Amen!
P a g e | 9
TABLE OF CONTENTS
Title Page - - - - - - - - - - i Certification - - - - - - - - - - ii Approval Page - - - - - - - - - iii Dedication - - - - - - - - - - iv Acknowledgement - - - - - - - - - v Table of Contents - - - - - - - - - vi List of Figures - - - - - - - - - ix List of Tables - - - - - - - - - - x Abstract - - - - - - - - - - xi CHAPTER ONE: INTRODUCTION
1.1 Background of the Study - - - - - - - 1
1.2 Statement of the Problem - - - - - - - 4
1.3 Research Questions - - - - - - - 7
1.4 Objectives of the Study - - - - - - - 8
1.5 Hypotheses of the Study - - - - - - - 8
1.6 Policy Relevance of the Study- - - - - - - 8
1.7 Scope of the study - - - - - - - 9
1.8 Conceptual Framework - - - - - - - 9
1.9 Structure of the Study - - - - - - - 13
CHAPTER TWO: REVIEW OF RELATED LITERATURE
2.1 Review of Theoretical Literature - - - - - - 14
2.1.1 Budget Deficits, Crowding In and Crowding Out Effects Schools of Thought- 14
2.2 Thematic Issues - - - - - - - - 17
2.2.1 Budget Deficits in Ghana: An Overview - - - - - 17
2.2.2 Determinants of Budget Deficit Growth in Ghana - - - - 18
2.2.3 Fiscal Policy in Ghana - - - - - - - 19
2.2.4 Interest Rate Policy in Ghana - - - - - - - 20
2.2.5 Fiscal Policy in Nigeria - - - - - - - 20
2.2.6 Macroeconomic Environments & Analytical Comparison between Nigeria & Ghana-
22
2.3 Empirical Literature - - - - - - - - 24
2.3.1 International Evidence - - - - - - - 25
2.3.2 Nigerian–Ghanaian Evidence - - - - - - - 31
2.4 Limitation of Previous Studies and Value-To-Be-Added - - - 46
P a g e | 10
CHAPTER THREE: RESEARCH METHODOLOGY
3.1 Theoretical Framework- - - - - - - - 48
3.2 Model Specification - - - - - - - - 51
3.3 Estimation Procedure - - - - - - - - 53
3.3.1 Empirical Models: Seemingly Unrelated Regression (SUR) - - - 53
3.3.1.1 Two Stage Least Squares and Instrumental Variables - - - 54
3.3.2 Lag/Length/Bandwidth Selection - - - - - 55
3.3.3 Stationarity/Unit Rood Test - - - - - - - 55
3.3.4 Cointegration Test - - - - - - - - 56
3.4 Model Justification - - - - - - - - 57
3.5 Data Sources - - - - - - - - - 58
3.6 Econometric Software for Analyses - - - - - - 58
CHAPTER FOUR: DATA ANALYSES AND EMPIRICAL RESULTS
4.1 Presentation of Results - - - - - - - 59
4.1.1 Presentation of Nigerian Data - - - - - - 59
4.1.1.1 Descriptive Statistics for all Variables - - - - - 59
4.1.1.2 Analysis of the Correlation Matrix - - - - - - 60
4.1.2 Presentation of Ghanaian Data - - - - - - 61
4.1.2.1 Descriptive Statistics for all Variables - - - - - 61
4.1.2.2 Analysis of the Correlation Matrix - - - - - - 62
4.2 Stationary (Unit Root) Tests Results - - - - - - 62
4.2.1 Lag Length/Bandwidth Selections - - - - - - 64
4.3 Cointegration Tests - - - - - - - - 64
4.4 Analysis of the SUR Models and Two-Stage Least Squares Estimation Results 65
4.4.1 Analysis of the SUR Estimation based on Economic Criteria - - 69
4.4.2 Analysis of the SUR Results based on Statistical Criteria - - - 70
4.4.3 Analysis of the SUR Estimation based on Econometric Criteria - - 70
4.4.3.1 Autocorrelation - - - - - - - - 70
4.4.3.2 Other Diagnostic Tests - - - - - - - 72
4.4.3.3 Functional Form Specification (Ramsey Reset) Test - - - - 72
4.5 Evaluation of Hypotheses - - - - - - - 71
CHAPTER FIVE: SUMMARY, CONCLUSION AND POLICY IMPLIC ATIONS
5.1 Summary of Findings - - - - - - - - 76
5.2 Policy Implications of Findings - - - - - - 76
P a g e | 11
5.3 Policy Recommendations - - - - - - - 77
5.4 Conclusion - - - - - - - - - 79
REFERENCES - - - - - - - - - 80
APPENDIX - - - - - - - - - - 90
LIST OF FIGURES
Figure 1.1 Graph showing Inflation and Budget Deficit interactions in Nigeria (1970 -2013) -
5
Figure 1.2 Graph showing, Inflation, and Budget Deficit interactions in Ghana (1970 -2013) -
6
P a g e | 12
LIST OF TABLES
Table 1.0: Tabular Summary of Empirical Literature Reviewed - - - 38
Table 1.1: Nigeria: Descriptive Statistics - - - - - - 60
Table 1.2: Nigeria Correlation Matrix - - - - - - 61
Table 1.3: Ghana: Descriptive Statistics - - - - - - 61
Table 1.4: Ghana Correlation Matrix - - - - - - - 62
Table 1.5: Summary of ADF and PP Stationary (Unit Root) Tests for the Variables in the
Models, 1970 – 2013 - - - - - - - - - 63
Table 1.6: Cointegration (Augmented Engle-Granger) Test Results - - - 65
Table 1.7: SUR Estimation Results - - - - - - - 67
Table 1.8: Other Diagnostic Tests - - - - - - - 72
Table 1.9: Ramsey Reset Specification Test - - - - - - 73
P a g e | 13
ABSTRACT
This study investigates the effects of budget deficits on selected macroeconomic variables in Nigeria and Ghana using annual time-series data of both economies covering from 1970 to 2013; and taking previous empirical studies as its point of departure. The specific objectives of the study include: to examine the effects of budget deficits on interest rates, inflation, and economic growth in Nigeria and Ghana within the methodological framework of Seemingly Unrelated Regression (SUR) model and Two-Stage Least Squares (2SLS). The study employs Eagle-Granger Cointegration test, Augmented Dickey Fuller (ADF) and Phillips-Perron (PP) tests in estimating the systems equations. Data sourced from World Bank, IMF - World Economic Outlook, Central Bank of Nigeria, Bank of Ghana and others, were analyzed using SUR model with several diagnostic and specification tests to examine the objectives of the study. From the perspective of this study, the empirical findings demonstrated that budget deficit has statistically negative effects on interest rate, inflation, and economic growth for both economies thereby supporting the neoclassical argument in the literature that budget deficit slows growth of the economy through resources crowding-out. Based on the empirical findings, many recommendations were made for both Nigeria and Ghana economies one of which stated that the government of Nigeria and Ghana should be mindful of the sources of financing the budget deficits so as to effectively manage the economic fluctuations and increase activities in the real sector. Also, it was recommended that both economies should pursue policies that will boost production of goods for both domestic consumption and exports in the long run through a combination of import substitution and export promotion strategies.
P a g e | 14
CHAPTER ONE
INTRODUCTION
1.1 Background of the Study
Budget deficit and its effects on macroeconomic variables is one of the most discussed issues
amongst economists and policy makers in both developed and developing countries (Saleh,
2003; Aisen & Hauner, 2008; Georgantopoulos & Tsamis, 2011). Intuitively, it is a
commonplace to construe that huge budget deficits have adverse macroeconomic effects such
as high interest rates, current account deficits, inflation, exchange rates volatility, with
implications on growth and development (Bernheim, 1989).
The budget deficit effects could either be negative, positive or a no positive or negative
relationship on macroeconomic variables. Budget deficit and its effects on any given
economy could be attributable to different methodologies countries employed and the nature
of data used by different researchers as most of the studies regress the macroeconomic
variable(s) on the fiscal deficit or the deficit on the macroeconomic variable(s)(Anyanwu,
1997).
Budget deficit refers to government expenditure exceeding government revenue over a period
of time (Anyanwu, 1997). When a deficit occurs in a country, it becomesimperative to find
remedy for financing such deficits so as to eradicate its negative implications. Nigeria and
Ghana as a developing economies have blamed prolonged economic crisis as one of the
major causes of budget deficit(s) in both economies as it has resulted in over indebtedness
and debt crisis, high inflation, poor investment performance and growth (Ezeabasili,
Mojekwu & Herbert, 2012). In Nigeria, public expenditure has led to increase in the fiscal
imbalances that siphon funds from the private sector investment, retarding growth and
reducing standard of living (Mpia & Ogrike, 2014). Fiscal imbalances create potential large
P a g e | 15
burden on future generations as workers may be forced to finance unfunded social
programmes. Budget deficits, therefore, lead to incurring debts which is a stock of liabilities
of the government (Udu & Agu, 2000). Budget deficit is generally associated with recession
because of the effect on revenues and expenditures (Dernberg, 1985).
The Ghanaian economy embarked recently, on the second leg of its centenary of
independence and democracy, announcing bold objectives that included an accelerated gross
domestic product (GDP) growth rate of 8% in 2009 and 10% before 2015 and achievement of
middle-income status of US$1000 dollars per capita by 2015 (Ackah, Aryeetey & Aryeetey,
2009). In the five decades since her independence from British rule in 1957, Ghana has gone
through different cycles of growth, marked by poor economic performance and military coup
d’états through to the 1980s. National economic policies during this period were often devoid
of market principles, and characterized by frequent price and income controls. At best, the
economy muddled through, with low productivity, high and volatile prices, an overvalued
currency and high interest rates (Ndulu & Connell, 2008).
The choice of this study which brought the economies of Nigeria and Ghana into focal point
for empirical investigation is formed by a number of reasons. Besides the obvious reason that
both economies share similarities in political and economic structures, the economies have
experienced very large fluctuations in the government budget deficits and high accumulation
of foreign debt, poor export performance, huge service account deficits, external debt
amortization, low inflow of foreign direct investment, misappropriation of external funding
support, excessive domestic monetary and credit expansion; price distortions and a
deterioration in the terms of trade (Ogiogio, 1996; & Obioma,1998).
In Nigeria, available data from the CBN (2012) statistical bulletin, show that deficit of -
8.62% of GDP was recorded in 1970 which rose to a surplus of 2.58% of GDP in 1971 and
P a g e | 16
declined to -0.82% of GDP in 1972. In 1974, Nigeria experienced a remarkable improvement
in the overall fiscal balances from 1970 to 2013 as surplus rose from 1.92% of GDP in 1973
to a surplus unit of about 9.54% of GDP. The Nigeria overall fiscal balance deteriorated
between 1980 and 1994 and recorded greater deficit of about -12.44% of GDP in 1982 on the
average. However, between 1995 and 2013, the Nigerian economy recorded a surplus of
about 1.19% of GDP on the average in 1996 with other years experiencing different deficit
percentages to GDP.
In Ghana, there has been huge and continuous deterioration in government fiscal position.
The economy has been in a persistent tendency towards budget deficit since independence as
a result of over expanding government expenditure, inadequate revenue generation capacity
of government and increasing debt levels (Pomeyie, 2001). The available statistics from
World Bank (2014) show that the Ghanaian economy has not recorded any surplus since their
independence and between 1970 and 2013.It is evidenced that the trends of the overall fiscal
balances of Ghanaian economy between 1970 and 2012 has been on the deficit side with a
huge deficit records of about -10.79% of GDP and -12.12% of GDP in 1982 and 2012
respectively. Apart from the period between 1981 and 1990 when there was remarkable fiscal
discipline, the government budget was consistently in deficit in the 1990s. On average, the
deficits was more than 5% of GDP in 1993.
As the economy of Ghana grows, policy makers have been concerned with the extent to
which the budget deficit is sustainable, and its effects on macroeconomic variables. However,
a deficit policy plays a vital role in assisting countries to achieve macroeconomic stability,
poverty reduction, income redistribution and sustainable growth. For this reason, most
governments use the budget as effective tool in achieving their economic objectives. This
means that large and accumulating budget deficit may not necessarily be a bad policy
objective if such deficits are effectively utilized to enhance economic growth. It is in line
P a g e | 17
with this that an appropriate operational definition and measure of budget deficit must be
clearly stated. Otherwise, the occurrence of large nominal budget deficit may be misleading
depending on the operational measure adopted by a particular country (Antwi, & Mills,
2013).
In Nigeria, the economy was caught in the deficit trap since early 1980s when the world oil
market collapsed, and since then, there have been frantic efforts to exit the trap but all to no
avail(Wosowei, 2013). Nevertheless, the fiscal policy adoption of Nigeria and Ghana in
financing deficits are attributable to major factors causing rapid monetary growth, exchange
rate depreciation and rising inflation. Thus the motivation for this study is to examine the
short and long run effect of budget deficit on interest rate, inflation and economic growth in
Nigeria and Ghana.
1.2 Statement of the Problem
Different schools of thought have demonstrated their opinions on budget deficits. Most
common are the Keynesian and the Ricardian Schools of Thought. While the Keynesian
posits that budget deficit affects mainly macroeconomic variables, Ricardian School refutes
the proposition (by the Keynesian school) and posits that budget deficits do not affect mainly
macroeconomic variables. However, budget deficit in developing economy like Nigeria and
Ghana plays a pivotal role in achieving economic and social objectives including
macroeconomic stability, sustainable growth and poverty reduction.
In recent times, the deficit positions of the Ghanaian budgets have worsened, drawing
attention to its long term sustainability (Bank of Ghana, 2005). As the two countries (Ghana
and Nigeria) consistently operate budget deficits, this lead to accumulation of government
debts. As past deficit adds up to current borrowings, it creates higher interest payments,
raising inflation with high volatility in real interest rate, depreciates exchange rate and retards
P a g e | 18
growth. This calls for further borrowing to cover the interest payment and the increasing
primary deficit which affects the rate of future borrowing.
Figure 1.1 Graph showing Inflation and Budget Deficit interactions in Nigeria (1970 -
2013)
Source: Researcher’s computation using data from CBN Bulletin of various years As seen in figure 1.1 above, the greater percentage of the overall budget balance in Nigeria
were on the deficit side with high inflation rate and unstable real interest rate. In 1990s the
average inflation rate soared to 72.8%. But, by 2007, the economy experienced a sharp
average fall of 6.57% in the inflationary trend.The interactions of real interest rate and budget
deficits in the figure 1.3 clearly distinguish between inflation and real interest rate.
In Ghana, the stock of government debt to GDP has been rising steadily from 17.2% in 2006
to 24.9% in 2007 and to 28.1% in 2008 (Bank of Ghana, 2007). Clearly, Ghana cannot use
new borrowing indefinitely to finance interest payments since changes in taxes and
government spending is followed by adjustment in future taxation and spending (Luporini,
1999).
-20.00
0.00
20.00
40.00
60.00
80.00
% of GDP
Year
Budget Deficit (% of GDP), and Inflation in Nigeria
BD(% of GDP) INFL
P a g e | 19
Figure 1.2 Graph showing, Inflation, and Budget Deficit interactions in Ghana (1970 -
2013)
Source: Researcher’s computation using data from World Bank, WEO Database, (2014)
In figure 1.2 above, Ghana recorded a deficit of -7.05% of GDP on the average in 1981 with
inflation soaring to about 117.8%. In 1982, budget deficit worsen on the average of -10.79%
of GDP with low inflation of about 21.3%. This shows that even if Ghana meets its interest
payment on debt by borrowing more, it must roll over its debt indefinitely because tax
revenue is not enough to pay for other expenditure. This has led to growing debt and
increasing tax rate.
Generally, in the case of Nigeria and Ghana, it has been claimed thatthe main causes of these
high rates of inflation were the widening fiscal imbalances, sources of deficit financing,
economic growth and the depreciation of the exchange rate. Nonetheless, the transition to
high inflation rates over the period resulted in substantial real cost and large losses in income,
at the same time as the performance of the economy as a whole declined as a result of
widening fiscal deficits and exacerbated by poor macroeconomic management and political
uncertainty(Arestis & Sawyer, 2006).
P a g e | 20
Nevertheless, empirical studies on the effects of budget deficit on macroeconomic variables
such as interest rate, inflation and growth seem not to lay credence on Keynesian proposition
or Ricardian Equivalence Hypothesis (REH). The major trust of this study is to examine the
short and long-run effect of budget deficits on interest rate, inflation and economic growth in
Nigerian and Ghana. This is necessitated by the fact that the previous empirical studies in this
area have no conclusive evidence in support of Keynesian-Ricardian paradigms in Nigeria
and Ghana. This is because, none of the previous studies conducted on the effects of budget
deficits in both economies used all the relevant variables and a few do not employ the
appropriate methodology. This is one of the motivations behind this study.Hence, this study
departs from previous studies due to the inclusion of relevant variables and thus employs the
most appropriate and robust methodology to explore the budget deficits and how it affect the
selected macroeconomic variables of these economies in both short and long run.
Against this background, the study contributes to the large body of the existing literature on
fiscal policies in Nigeria and Ghana in two ways. First, unlike many empirical studies, the
study employs a more vigorous and robust approach; the Seemingly Unrelated Regression
(SUR) model to analyze the effect of budget deficit on the selected macroeconomic variables
in Nigeria and Ghana over the period covering from 1970 to 2013. Second, the study
provides empirical evidence for the economies with recent fiscal policies for which
researches have not been conducted recently. Besides, previous studies have advanced in
characterizing the implications of alternative sources and composition of deficits spending
without investigating the effects of budget deficits on the selected macroeconomic variables
in the two economies.
1.3 Research Questions
In the light of the above discussions, the following research questions shall be addressed:
i. What are the effects of budget deficiton interest rate?
P a g e | 21
ii. What are the effects of budget deficit on inflation?
iii. Does budget deficit have any impact on economic growth?
1.4 Objectives of the Study
The broad objective of this study is to investigatethe short and long run effects of budget
deficits on selected macroeconomic variables in Nigeria and Ghana. Specific objectives of
the study are:
i. To examine the effects of budget deficits on interest rates in Nigeria and Ghana.
ii. To ascertain the effects of budget deficits on inflation in Nigeria and Ghana.
iii. To evaluate the effects of budget deficits on economic growth in Nigeria and Ghana.
1.5 Hypotheses of the Study
In view of the above research questions, the following hypotheses are formulated in order to
ascertain the answers to the questions:
H01: Budget deficitshavenosignificant effects on interest rate in Nigeria and Ghana.
H02: Budget deficitshavenosignificant effects on inflation in Nigeria andGhana
H03: Budget deficits have no significant effects on economic growth in Nigeria and Ghana.
1.6 Policy Relevance of the Study
Developing economies like Nigeria and Ghana are in search of long-term policies not only
for macroeconomic stabilization but also for sustainable economic growth and development.
The private sector has always mourned that the fiscal deficit and its inflationary financing has
led to a slow-down in economic activity. There is need for a comprehensive study which
exposes the dynamics of the government fiscal deficit and its effects on the economy. This
study is designed to investigatethe short and long run effects of budget deficits on some
selected macroeconomic variables in Nigeria and Ghana. An academic study such as this is
P a g e | 22
crucial to both the Nigerian and Ghanaian economy in ascertaining the extent to which
budget deficit affects some macroeconomic variables in both short and long run. However,
the justification of this study is that it will add value to policy making especially in fiscal
adjustment and macroeconomic management. The findings of the study will also assist policy
makers to understand the centrality of the government fiscal deficit in policy formulation. It
is further hoped that the findings of the study will assist policy makers to place the right
emphasis on the role of fiscal adjustment in economic policy.
Furthermore, given the Keynesian-Ricardian dichotomy on budget deficit, it is needful to
investigate the effects of budget deficits on some selected macroeconomic variables in
Nigeria and Ghana. As a result, this research work seeks to contribute to the ongoing debate
on the relationship of budget deficits and some selected macroeconomic variables in Nigeria
and Ghana. Therefore, policy makers, experts, institutions, government agencies, researchers
and students in areas of public sector economics, development economics, international
economics, and econometrics are likely to find the outcome of this study very useful.
1.7 Scope of the Study
The study, as stated abinitio tends to investigate the short and long run effects of budget
deficits on selected macroeconomic variables in Nigeria and Ghana. It is pertinent to note that
the nexus between budget deficits and macroeconomic variables are still not well ascertained
especially when viewed in both short and long run. However, the study covers from 1970 to
2013. The range is chosen to ensure high level of degree of validity and precision in the
study. The variables of interest which are used to investigate the short and long run effects of
budget deficits on the selected macroeconomic variables in Nigeria and Ghana are: budget
deficit, interest rates, inflation,exchange rate depreciation, government expenditure, money
supply (M2), total savings, and real GDP.
1.8 Conceptual Framework
P a g e | 23
Awe and Funlay (2014) state that budget deficit occurs when government expenditures
exceed its revenues, thus the level of public savings is negative. This may harm the economic
growth of a country. Budget deficits lead to incurring debts which is a stock of liabilities of
the government (Keating, 2000). Budget deficit is a situation where the government budgets
to spend more than it intends to collect as revenue (Udu and Agu, 2000). Dernberg (1985)
posits that budget deficits are generally associated with recession because of the effect on
revenues and expenditures.
Fiscal deficit has gathered substantial attention in the literature in the area of macroeconomic
theory due to its impact on the macroeconomic variables. When budget deficit is financed by
borrowing, it expands government’s demand for credit through competition with households
and business firms (Hyman, 1994). This puts upward pressure on interest rate and slows
down the rate of capital formation. In Keynesian models, this occurs through a rise in real
interest rate which reduces investment purchases through the transmission mechanism.
However, this depends on the responsiveness of interest rate to increased demand for credit
and the reaction of private investors to higher interest rate. For instance, if investment
demand is unresponsive to changes in interest rate, the effect on investment will be very
small. Also, if the economy is in deep recession, any extra borrowing by government may put
little upward pressure on interest rate because the supply curve for funds will be quiet flat.
Yet, if the return on private investment exceeds the return on government investment
particularly on infrastructure, the rise in public investment would crowd in private investment
(Barua, 2005).
Also, large budget deficit increases the debt crisis in terms of its services and levels. As the
national debt grows, interest payment also grows which serves as tax on investment. This
reduces private investment, increases unemployment, lowers tax revenue and leads to higher
future deficits. Hence, the economy continues to incur mounting debts which may lead to its
P a g e | 24
collapse. Most countries trapped in debt servicing difficulties did run huge budget deficits at
some point in time. Yet, it is argued that public debt growth compels government to target
higher economic growth and revenue in order to finance any rising debt obligations.
Otherwise it will force the economy into a deficit trap (Barua, 2005).
On the burden of future generations, most economists agree that financing budget deficit
through external debt means the postponement of tax increases. However, it is asserted that
such burden depends on how the contracted loan is utilized. If the funds are spent on current
consumption expenditure, future generations are likely to be worse-off but if spent on
productive activities such as education and health then future generations are likely to be
better-off (Mankiw, 2003).
In addition, if budget deficit is monetized, it increases aggregate demand through increase in
government purchases without a corresponding increase in taxes. Hence, governments need
to run fiscal deficit particularly in the early stages of development to lead the economy in the
path of growth and development (Xiomara & Greenidge, 2003). Secondly, it increases money
supply. This exerts downward pressure on interest rate and upward pressure on equilibrium
money stock and price level unless the economy is in deep recession. This leads to higher
inflation, uncertainty and instability of real interest rate which tends to lower real tax revenue.
Hence, monetized deficit should be kept low and effectively managed in the short-to-medium
term (Bebi, 2000; Turnovsky, 2000).
Causes and Determinants of Budget Deficit Growth
In general, changes in budget deficit is attributed to changes in government spending or tax
revenue or both. Government receives revenue in its daily transactions and on capital items in
the form of taxes and interests. On the other hand, government pays for daily activities and
capital items such as administrative expenses, loans and grants. Thus, budget deficit increases
when government spending persistently exceeds its revenue. If expenditure continue to mount
P a g e | 25
up throughout the years whereas revenues especially taxes are poorly collected, it widens the
budget deficit position of the country. In this case, the accumulated value of past deficit
creates increase debts which must be financed together with the accompanying interest
payments.
With reference to political-economic models of government behaviour, it is recognized that
incumbent administrations tend to stimulate their economies on the eve of political elections
through tax cuts, increase spending and transfer payments. This occurs in countries where
political power changes frequently between rival parties. In such cases, each rival
administration spends over and above its budget and deliberately wait until after election
before implementing policies to reduce the deficit. These ad hoc policies tend to widen the
overall budget deficit and debt levels of the countries (Sachs & Larrain, 1993).
However, the extent of the impact of budget deficit on an economy is blamed by
macroeconomic factors such as expected inflation, cyclical position of the economy which
influences tax revenues and changes in expenditure. Theory predicts that cyclical fluctuations
in output which is caused by economic boom and/or recession impact significantly on budget
deficit. In periods of recession when output is low, budgets tend to be in deficit because direct
taxes fall sharply due to contraction in tax base. Also, certain categories of government
spending become countercyclical and rise during business cycle downturn. Yet, such
fluctuations in output growth are endemic in free market economies (Gebhard & Silika,
2006).
Budget Deficit Growth and Economic Sustainability
Financing of budget deficit in Ghana and other developing economies like Nigeria have had
diverse macroeconomic burden on the economy (Antwi et al, 2013). For instance central
bank financing of budget deficit have expanded the monetary base and money supply.
P a g e | 26
According to Wetzel and Roumeen (1991) central bank financing of budget deficit in Ghana
has distorted the distinction between monetary and fiscal policies whereas the sale of
domestic bond increased interest rate and this has led to increase in the net domestic
financing from 0.49 percent of GDP in 2004 to 4.15 percent in 2006. As a result, money
supply (currency and deposits) increased from 6.84 percent in 2005 to 34.4 percent in 2006,
and thus the domestic interest and bank rate reduced due to low demand for bonds (ISSER,
2007). Also, Ghana has accumulated large external debt and so borrow externally only on
short term bases at high interest rate. This is because foreign financing raises the cost of
servicing external debt. For this reason, Ghana’s access to external borrowing prior to 1984
had been limited, ranging between -0.74 and 1.62 percent of GDP. In recent times however,
debt levels have been falling. External debt fell from 72.5 percent in 2004 to 26.9 percent in
2006 with debt service to GDP reducing from 6.8 percent in 2004 to 6.0 percent in 2006
(Wetzel and Roumeen, 1991: 48; ISSER, 2007)
1.9 Structure of the Study
This study will be organized in five chapters. Following this introduction as Chapter 1,
Chapter 2 presents a review of literature on both thetheoretical and empirical evidences.
Chapter 3 discusses the methodology used in the study, Chapter 4 presents the analysis and
interpretation of the results obtained using the methodologies in previous chapter. Finally,
Chapter 5 highlights the conclusion and recommendations on fiscal policy management in the
Nigeria and Ghana.
P a g e | 27
CHAPTER TWO
REVIEW OF RELATED LITERATURE 2.1 Review of Theoretical Literature
This section reveals the theoretical framework related to budget deficit and macroeconomic
variables. Some of the economist such as Keynesian, Neoclassical and Ricardian Schools of
Thought gave either positive or negative support to the relationship between budget deficits
and macroeconomic variables.
2.1.1 Budget Deficits, Crowding In and Crowding Out Effects Schools of Thought
In analyzing the literatures on the relationship between budget deficits and macroeconomic
variables, Bernhein (1989) provides a brief summary of the three paradigms, and the cursory
of the paradigms are presented below.
The Neoclassical School
The neoclassical school proposes an adverse relationship between budget deficits and
macroeconomic variables. They argue that budget deficits lead to higher interest rates,
discourages the issue of private bonds, private investments, and private spending, increases
inflation level, and cause a similar increase in the current account deficits and finally slows
the growth rate of the economy through resources crowding out.
P a g e | 28
The Neoclassical school considers individuals planning their consumption over their entire
cycle. By shifting taxes to future generations, budget deficits increase current consumption.
By assuming full employment of resources the neoclassical school argues that increased
consumption implies a decrease in savings. Interest rate must rise to bring equilibrium in the
capital markets. Higher interest rates, in turn, result in a decline in private investment,
domestic production and an increase in the aggregate price level. Furthermore, Yellen (1989)
in Onuorah, and Ogomegbunam (2013) argues that in standard Neoclassical Macroeconomic
models, if resources are fully employed, so that output is fixed, higher current consumption
implies an equal and offsetting reduction in other forms of spending. Therefore, there will be
fully crowding-out of investment and/or net exports.
It is worth noting that it is important to distinguish between “financial” crowding out and
“resource” crowding-out. “Resource” crowding out occurs when the government competes
with the private sector on purchasing certain resources (skilled labour, raw materials and so
on). When the government sector expands, the private sector will contract because of the
increase in prices on these resources due to an excess demand by the government, hence this
leads to a fall in investment and consumption by the private sector. Thus the government
sector’s expansion crowds out the private sector. It is worth noting here as well that resource
crowding out is an important issue to take into account especially in developing countries
where resources are scarce even sometimes to the private sector, so any excess demand for
these resources by the government will severely impinge on private sector productivity.
The Keynesian School
The Keynesian economists propose a positive relationship between budget deficits and
macroeconomic variables. They argue that usually budget deficits result in an increase in
domestic production, increases aggregate demand, increases savings and private investment
at any given level of interest rate. The Keynesian absorptive theory suggests that an increase
P a g e | 29
in the budget deficits would induce domestic absorption and thus, import expansion, causing
current account deficit. In the Mundell-Fleming framework, an increase in the budget deficit
would induce an upward pressure on interest rate, causing capital inflows and an appreciation
of the exchange rate that will increase the current account balance.
The Keynesians provide a counter argument to the crowd-out effect by making reference to
the expansionary effects of budget deficits. They argue that usually budget deficits result in
an increase in domestic production, which makes private investors more optimistic about the
future course of the economy resulting in them investing more. This is known as the
“crowding-in” effect. It is worth noting here that the traditional Keynesian view differs from
the standard neoclassical paradigm in two fundamental ways. First, it permits the possibility
that some economic resources are unemployed. Second, it presupposes the existence of a
large number of liquidity-constrained individuals. This second assumption guarantees that
aggregate consumption is very sensitive to changes in disposable income. Many traditional
Keynesians argue that deficits need not crowd out private investment. Eisner (1989) suggests
that increased aggregate demand enhances the profitability of private investments and leads
to a higher level of investment at any given rate of interest. Hence deficits may stimulate
aggregate savings and investment, despite the fact that they raise interest rates. He concludes
that “evidence is thus that deficits have not crowded-out investment. There has rather been
crowding-in”. Heng (1997) utilized an overlapping-generations (OLG) model to provide a
theoretical framework to analyze the “crowding-in” issue of private capital by public capital.
He shows that public capital crowds-in private capital through two channels, namely, via its
impact on the marginal productivity of labour and savings, and via (gross)
complementarity/substitutability between public and private capital.
The Ricardian School
P a g e | 30
Finally, there is another contrary approach advanced by Barro (1989) known as the Ricardian
Equivalence Hypothesis (REH). Ricardian equivalence, or the Barro-Ricardo equivalence
proposition, is an economic theory which suggests that government budget deficits do not
affect the total level of demand in an economy. It was initially proposed by the 19th century
economist David Ricardo. In simple terms, the theory can be described as follows.
Governments may either finance their spending by taxing current taxpayers, or they may
borrow money. However, they must eventually repay this borrowing by raising taxes above
what they would otherwise have been in future. The choice is therefore between "tax now"
and "tax later". Suppose that the government finances some extra spending through deficits -
i.e. tax later. Ricardo argued that although taxpayers would have more money now, they
would realize that they would have to pay higher tax in future and therefore save the extra
money in order to pay the future tax. The extra saving by consumers would exactly offset the
extra spending by government, so overall demand would remain unchanged.
More recently, economists such as Robert Barro have developed more sophisticated
variations on the same idea, particularly using the theory of rational expectations. Ricardian
Equivalence suggests that government attempts to influence demand using fiscal policy will
prove fruitless. He argues that an increase in budget deficits, due to an increase in
government spending, must be paid for either now or later, with total present value of receipts
fixed by the total present value of spending. Thus, a cut in today’s taxes must be matched by
an increase in future taxes, leaving real interest rates, and thus private investment, and the
current account balance, exchange rate and domestic production unchanged. Therefore,
budget deficits do not crowd-in nor crowd out macroeconomic variables i.e. no positive or
negative relationship exists.
2.2 Thematic Issues
2.2.1 Budget Deficits in Ghana: An Overview
P a g e | 31
Ghana’s economy has maintained commendable growth trajectory with an average annual
growth of about 6.0% over the past six years. In 2013 growth decelerated to 4.4%,
considerably lower than the growth of 7.9% achieved in 2012. Growth has, however, been
broad-based, driven largely by service-oriented sectors and industry, which on average have
been growing at a rate of 9.0% over the five years up to 2013. Over the medium term to 2015,
the economy is expected to register robust growth of around 8%, bolstered by improved oil
and gas production, increased private-sector investment, improved public infrastructure
development and sustained political stability (African Economic Outlook, 2014).
According to AEO (2014), the continued widening budget deficit has been a major constraint
to fiscal and debt sustainability. Following an expenditure overrun in 2012, marked by an
unprecedented budget deficit of around 12% of GDP, the situation persisted in 2013, with
about the same level of budget deficit. Revenue enhancing and expenditure consolidation
measures underway in 2014 are expected to ease the fiscal deficit to 9%. In conjunction with
fiscal constraints, inflation has been on the rise resulting from a number of factors including
the removal of subsidies on petroleum prices and a gradual rise in electricity and water tariffs.
It is also worth noting the rise in public debt from 43% of GDP in 2011 to 48% in 2012, and
further to 53.5% in September 2013, resulting from a widened budget deficit. The external
sector will continue to experience a widened current-account deficit of around 12% of GDP
in 2014, exacerbated by a decline in commodity prices of major export commodities,
particularly on gold and cocoa.
2.2.2 Determinants of Budget Deficit Growth in Ghana
A model involving variation in inflation, government expenditure during wartime, cyclical
fluctuation in output during economic boom and recession in the postwar period was tested to
ascertain if it differs significantly from those during the world wars in the Swiss federal state.
The estimate showed some cyclical fluctuation in the world war periods. This supports the
P a g e | 32
assertion that significant determinant of budget deficit is increase in state expenditure during
wartime. In this case, civilian expenditure was reduced and/or taxes increased to finance
military expenditure during the war (Gebhard and Silika, 2006). In Ghana, changes in
inflation, interest rate and real GDP have reacted negatively to changes in budget deficit. For
instance, Antwi and Mills (2013) observed that high inflation in 1983 caused budget deficit to
increase by 35.8 percent due to decline in direct tax revenue. Also, changes in real interest
rate increased budget deficit by 11.3 percent of GDP in 1984. Again, high wage bill increased
the deficit by 2.5 percent in 1985. Thus, changes in macroeconomic variables have had strong
impact on the fiscal deficit in Ghana. However, these effects have become less pronounced
over the past years as the Ghanaian economy has grown more stable (Wetzel & Roumeen,
1991).
2.2.3 Fiscal Policy in Ghana
The government of Ghana is committed to fiscal consolidation with the ultimate objective of
reducing the budget deficit to around 5% of GDP by 2016. However, trend performance of
government operations continues to register widened budget deficit. Following an
expenditure overrun in 2012, marked by a significant budget deficit of around 6% of GDP,
the situation persisted in 2013 with a deficit of 7.8%. Fiscal measures implemented in the
second half of 2013 are expected to yield dividends in 2014. Key contributors to widened
budget deficit have been increased spending on wages and salaries, interest payments,
subsidies and arrear payments (African Economic Outlook, 2014).
According to AEO (2014), for the government of Ghana to address fiscal constraints
effectively, efforts should aim to raise tax revenue, in view of its substantial share (80%) of
total domestic revenue. Oil revenue is still low, accounting for just 0.2% of total revenue.
Grants from development partners are marginal and have maintained a diminishing trend,
accounting for only 7% of total revenues in 2013, down from around 14% in 2010. Despite
P a g e | 33
the marginal contribution of oil receipts, it is worth noting the distribution formula of such
receipts. In compliance with the Ghana Petroleum Revenue Management Act, about 30% of
total oil revenue to government is retained by the Ghana National Petroleum Commission
(GNPC) for the development of the oil and gas industry, while the remaining 70% is
appropriated through the Annual Budget Funding Amount (ABFA) and Ghana Petroleum
Funds (GPF) at around 40% and 60% respectively. While resources under ABFA are
reserved for funding priority projects, petroleum funds (GPF) are partly invested for future
generations through the established Ghana Heritage Fund.
2.2.4 Interest Rate Policy in Ghana
Interest rates were administratively controlled by the Bank of Ghana (BOG). The rationale
for the controls was that credit had to be cheap so as to promote investment and support that
favors borrowers (Daumont, Le Gall, & Leroux, 2004). It was the BOG that determined the
structure of the bank interest rates, which include the minimum interest rates for deposits and
maximum lending rates. Preferential lending rates were given to priority sectors such as
agriculture. The structure of interest rates determined by the BOG made no allowance for
loan maturity or risk; indeed, incentives for banks to extend credit were often perverse
because riskier sectors such as agriculture were accorded preferential inflation rates. In most
of the years, nominal interest rates were held below the prevailing inflation rates. However,
when inflation escalated in the middle of 1970s and the early 1980s, real interest rates were
highly negative (Antwi, 2009)
2.2.5 Fiscal Policy in Nigeria
Fiscal policy Management in 2012 and 2013 has centred on consolidation in order to ensure
macroeconomic stability. The fiscal deficit as a percentage of GDP has been estimated at -
1.8% in 2013, up from -1.4% in 2012, but well below the fiscal stance of a maximum of 3.0%
deficit enshrined in the Fiscal Responsibility Act. The Medium Term Expenditure
P a g e | 34
Framework (2014-2016)and Fiscal Strategy Paper proposed benchmark oil prices of USD 74,
USD 75 and USD 76 per barrel (pb) for 2014, 2015 and 2016 respectively. A benchmark oil
price of USD 77.5 pb was however set in the 2014 budget presentation to the national
assembly.
The 2012 budget was signed into law by the president of the Federal Republic of Nigeria in
April of the year following its passage into law by the legislative arm of government. The
level of implementation of the budget was 71.6%. The 2013 budget was signed into law by
the president in February, which was two months earlier than the preceding year as the
disagreement between the executive and the legislature over appropriation were resolved
early. The eventual implementation rate was around 70.0%. The ratio of capital expenditure
to total expenditure diminished to an estimated 23.9% in 2013 from 24.3% in 2012. The share
of capital expenditure on social community services (Health, Education and other allied
services) in the total rose from 10.0% in 2011 to 11.1% in 2012 while economic services
(agriculture and infrastructures) declined from 42.1% to 36.7%, respectively. The capital
component of Subsidy Reinvestment and Empowerment Programme (SURE-P) contributed
about NGN 272.5 billion, or USD 1.72 billion, thus raising the total capital expenditure to
NGN 2 059 billion (USD 13.03 billion) in 2013 (AEO, 2014).
In pursuance of fiscal prudence and the burgeoning debt profile, the government limits its
borrowing requirements in compliance with the Fiscal Responsibility Act (2007). Figures
from the Debt Management Office as at 31 December 2013 showed that Nigeria’s public debt
stock was USD 64.51 billion. Of this amount, the external debt of both the federal and state
governments was only USD 8.82 billion, of which the state governments constituted about
38.2%. The balance of USD 55.69 billion (about 86.3% of the total) drawn by both the
federal and state governments makes up the domestic debt component. In this regard, new
borrowing in 2014 is estimated to be NGN 572 billion (USD 3.62 billion), slightly down
from NGN 577 billion (USD 3.65 billion) in 2013 (AEO, 2014).
P a g e | 35
The observed persistent decline in oil revenues portends risk for fiscal policy and will shape
the trajectory of the medium-term fiscal outcome. Total revenue was on a declining trend
throughout 2013. It is noteworthy, though, that non-oil revenues have been rising
significantly over the same period, thereby compensating for the shortfall in oil revenues,
albeit inadequately. If the declining oil revenues are not contained and the rise in non-oil
revenues is not sustained, fiscal risks may set in. This could hinder the success of the
country’s ongoing reforms, and an overall negative impact on economic activities may result.
In addition, the government has been adjusting expenditures to accommodate the shortfall in
revenues. Capital expenditure, however, suffers huge downward adjustments because
recurrent expenditures, which are mainly salaries and overhead, can hardly be adjusted
automatically. These downward adjustments in capital expenditure may further slowdown
total economic activities and growth.
2.2.6 Macroeconomic Environmentsand Analytical Comparison between Nigeria and
Ghana
This section provides a review of economic developments in Nigeria and Ghana; and trends
in the major macroeconomic variables in recent years. The variables discussed in this section
include Gross Domestic Product (GDP), Inflation, the Value of Total Trade, Imports and
Exports.
The Nigerian economy faced numerous challenges which impacted overall economic activity
in 2012. Declines in the real growth rates of economic activity were experienced in both the
oil and non-oil sectors. Oil production was less than expected due to security challenges, and
floods which occurred in the latter part of the year, while the non-oil sector (notably
Agriculture, Wholesale & Retail Trade) was mostly affected by the floods and weaker
consumer demand (National Bureau of Statistics, 2013). The revised data for the NBS (2012)
indicates that real GDP grew by 6.34 % in the first quarter and 6.39% in the second quarter of
P a g e | 36
2012. The rate of economic activity was slightly higher than the initial estimates of 6.17%
and 6.28% respectively.
According to the Nigerian National Petroleum Corporation (NNPC), oil production was
estimated at 2.37 million barrels per day (mbpd) during the first half of 2012, as against
2.48mbpd produced in the first half of 2011. The 4.4% decline in crude production levels was
attributed to disruptions in production due to cases of oil theft and vandalization in the oil
producing areas. On the other hand, non-oil sector was affected by the incidence of flooding,
as well as muted consumer demand for the most part of the year, as seen in the Wholesale
and Retail Trade, Telecommunication and Post sectors while infrastructure challenges still
hampered Manufacturing. However, the Manufacturing sector did record a slight uptick in the
second quarter as a result of positivedevelopments in the power sector.
In Ghana, the economy is expected to slow down for the fourth consecutive year to an
estimated 3.9% growth rate in 2015, owing to a severe energy crisis, unsustainable domestic
and external debt burdens, and deteriorated macroeconomic and financial imbalances
(African Economic outlook, 2015). However, the provisional gross domestic product (GDP)
figures issued by the Ghana Statistical Services (GSS) further suggest that the economy
expanded by 4.2% in 2014, less than the growth of 7.3% recorded in 2013. The drivers of
growth continue to be the service sectors, which constitute 50.2% of the economy, followed
by industry and agriculture at 28.4% and 19.9% respectively (AEO, 2015).
High growth rates in Ghana over the recent years have been accompanied by the build-up of
macroeconomic imbalances. In 2014 current account and fiscal deficits widened to 9.2% and
10.4% of GDP respectively, and the rate of inflation averaged 17.0%. By the end of
December 2014, foreign reserves were at 3.2 months of import cover. The domestic currency,
the cedi (GHS) depreciated by over 30% in nominal terms over the first nine months of the
P a g e | 37
year compared to a depreciation of 4.1% during the corresponding period in 2013. The
continued growth in the budget deficit resulted in public debt increasing from 55.8% of GDP
in December 2013 to 67.1% of GDP by the end of December 2014. To address the
increasingly unsustainable fiscal and current account imbalances, the Ghanaian authorities
started negotiations for a stabilization programme with the International Monetary Fund
(IMF) that was expected to begin in early 2015(AEO, 2015).
According to National Bureau of Statistics(NBS) (2013) as at December 2012, the headline
inflation rate showed a general downward trend during the year, despite the economic
challenges that the country witnessed. From the 12.6% recorded in January (year- on-year),
the headline inflation rate reached12.9% in April and June before slowing to 11.3% through
September. It rose further to 12.3% in November before falling slightly to 12.0% in
December. As a result, the average inflation rate for the year stood at 12.2%. On a month-on-
month basis, the headline inflation rate rose sharpest in March (1.6%). The major sources of
inflationary pressure in 2012 still appear to be structural and infrastructural constraints.
Nevertheless, the removal of fuel subsidies early in the year, the devastating flood that
occurred in the third and fourth quarters of 2012 as well as seasonal effects also played major
roles in driving up prices at various times.
In Nigeria, as at the third quarter of 2012, the value of total Merchandise Trade for the
country was estimated at N20,885.4 billion over the first three quarters of the year. Compared
to levels recorded during the first three quarters in 2011, the value of total merchandise trade
had remained roughly unchanged, increasing marginally by 0.4%. The value of total
merchandise trade points to increasing exports over the period while imports have been on
the decline. Specifically, imports have continued to trend downwards since the second quarter
of 2011, while the value of exports which increased substantially in late 2011, dipped in the
P a g e | 38
first quarter of 2012, but picked up in the second and third quarters of 2012(National Bureau
of Statistics (NBS), 2013).
2.3 Empirical Literature
On the empirical front, there is a large body of literature documenting the effects (both short
and long run) of budget deficits on macroeconomic variables in both developed and
developing economies. This large body of literature on the topic could be divided into
international, and Nigerian-Ghanaian evidence for distinctive purposes; and thus logically
arranged in order of objectives of the study and year(s) in which the study was carried out.
2.3.1 International Evidence
Vamvoukas (1998) empirically examines the short and long-run effects of budget deficits on
interest rates for Greece using annual time series data from 1970 to 1990 within the
methodological framework of cointegration, ECM strategy, and several diagnostic and
specification tests. The estimation results support the Keynesian model of a significant and
positive relationship between budget deficits and interest rates
Gupta and Uwilingiye (2009) investigate the direction of temporal causality between budget
deficit and interest rate for South Africa using quarterly and annual data for the period of
1961 to 2005. The results show that budget deficit Granger causes interest rate in the
quarterly data. However, for the annual data, the study finds no causal relationship between
the budget deficit and the Treasury bill rate. The two variables are positively cointegrated for
both data frequency.
Similarly, Bonga-Bonga (2011) investigates the extent of the effects of the systematic and
surprise changes in budget deficits on the long-term interest rate in South Africa between
1960 and 2000 using vector autoregressive (VAR) techniques. The study finds a positive
P a g e | 39
relationship between the budget deficits and long-term interest rates. On the other hand,
Akinboade (2004) uses the LSE approach and Granger-causality methods to investigate the
nexus between budget deficit and interest rate in South Africa, the study finds no relationship
between the budget deficit and interest rates.
Mukhtar and Zakaria (2008) uses Granger Causality test and Error Correction Model (ECM)
to examine long run relationship between budget deficits and interest rates for Pakistan using
quarterly time-series data for the period 1960 to 2005. The regression results show that
budget deficits have no significant effect on nominal interest rates. The results equally reveal
that budget deficit-GDP ratio has significant positive impact on nominal interest rates.
Aisen and Hauner (2008)asses the relationship between budget deficits and interest rates in
60 advanced and emerging economies with a panel dataset spinning from 1970 to 2005.The
result shows a significant and positive relationship between budget deficits and interest rates.
The study also finds that the effects of budget deficits on interest rates varied by country
group and period. The resultequally shows that the effects were larger and more robust in the
emerging markets and in later periods than in the advanced economies and in earlier periods.
The study further reveals that the effect of budget deficits on interest rates depends on
interaction terms and is only significant under one of several conditions such as sources of
deficit financing (mostly domestically financed), and/or interact with high domestic debt; low
financial openness, and/or interest rates liberalization.
Noula (2012) examines the determination of fiscal deficit and nominal interest rate in
Cameroon using annual time series data from 1974 to 2009. The study employs a loanable
funds model to test for fluctuations in the economy budget deficits and nominal lending rates.
Moreover, the empirical assessmentcarried out using ADF test and Error Correction Model
reveals a significant positive association between budget deficits and domestic nominal
P a g e | 40
lending interest rate. Also, the result from the Pairwise Granger Causality test conducted
shows a bi-directional causality between budget deficits and nominal interest rate.
Guess and Koford (1984) utilizes the Granger Causality test to study the causal relationship
between budget deficits and inflation, GNP and private investment using annual time-series
data for seventeen OECD countries for the period 1949 to 1981. The results show that budget
deficits do not exert negative changes in these variables. In the same vein, Darrat (1985)
applies Ordinary Least Square (OLS) technique in examining empirically the linkage
between deficits and inflation in the US during the post-1960 period. The estimation results
show that both monetary growth and federal deficits significantly influenced inflation during
the 1960s and 1970s. In addition, the study concluded that federal deficits bore a stronger and
more reliable relationship to inflation than monetary growth.
Easterly and Schmidt-Hebbel (1993) analyze data from a sample of 10 countries and find
strong evidence that over the medium term, money financing of the deficit leads to higher
inflation, while debt financing leads to higher real interest rates or increased repression of
financial markets. Moreover, Tekin-Kuru and Ozmen (1998) examine the long run
relationship among budget deficits, money supply and inflation in Turkey. The results reveal
that while the endogeneity of supply of money and inflation reject the validity of the
monetarist view, lack of direct relationship between inflation and budget deficit repudiate the
pure fiscal theory explanations.
Darrat (2000) utilizes an Error Correction Model (ECM) to investigate if high budget deficits
have any inflationary consequences in Greece over the period 1957- 1993. The empirical
result shows that the deficit variable exerts a positive and statistically significant impact on
inflation in Greece. More also, Catão and Terrones (2005) find a strong link between fiscal
deficits and inflation using a sample of 107 countries over the period 1960 to 2001. Their
P a g e | 41
results show that, a 1 percent reduction in the ratio of the budget deficit to GDP is associated
with an 8.75 percent lower inflation rate.
Makochekanwa (2008)studies the impact of budget deficit on inflation in Zimbabwe for the
period 1980 to 2005. The study finds a positive and stable long run relationship between the
budget deficit, exchange rate, GDP and inflation. Habibullah, Cheah and Baharom (2011)
examine the relationship between budget deficit and inflation in thirteen Asian developing
countries using annual data for the period 1950 – 1999. The study employs Granger causality
test and their result shows the existence of a long run relationship between budget deficit and
inflation thus concluding that budget deficits are inflationary in Asian developing countries.
Ndashau (2012) adopts Granger causality techniques, augmented by vector error correction
model (VECM) to highlight the existence of a causality effect from inflation to budget
deficits scaled by the money base. However, the effect of budget deficits on inflation was not
statistically significant. On the other hand, Lin and Chu (2013) employ a dynamic panel
quantile regression (DPQR)model following the autoregressive distributive lag (ARDL)
regime to examine the extent to which fiscal deficits are inflationary in 91 countries between
1960 and 2006. The results of the study show that fiscal deficits are inflationary only in high
inflation countries.
Dwyer (1982) applies a Vector Autoregression (VAR) Model to test for the linkage between
government deficits and macroeconomic variables (such as prices, spending, interest rates
and the money stock) in the U.S. over the period 1952 – 1978. The results are consistent with
the hypothesis that there are no perceived wealth effects of predictable changes in
government debt held by the public and as a result, no effects of the debt on inflation. No
evidence is found that larger government deficits increase prices, spending, interest rates, or
P a g e | 42
the money stock. The study also shows that the reason for the decline in inflation rates can be
attributed to the decline of money growth despite borrowing.
Barro (1990; 1991) investigates the effects of tax financed government expenditure on
investment and output in a cross-sectional study of 98 countries over the period 1960-85. The
study reveals that the ratio of real government consumption to real GDP (gc/y) has a negative
association with growth and investment. The result shows that government consumption exert
no direct effect on private productivity, but lowers savings and growth through the distorting
effects from taxation or government-expenditure programs.
Karras (1994) studies the relationship between budget deficits and macroeconomic variables
in a Cross-sectional study involving 32 countries for the period 1950-1980, using OLS and
generalized least square (GLS). The results show that deficits do not lead to inflation, they
are negatively correlated with the rate of growth of real output and increased deficits appear
to retard investment. Similarly, Al-Khedir (1996) investigates the relationship between
budget deficits and macroeconomic performance of the G-7 countries for the period 1964-
1993 using VAR. The study reveals that budget deficits lead to higher short-term interest
rates in the 7 countries. However, the results show that deficits have no impact on the long-
term interest rates. The trade balance has been worsened by the budget deficit and economic
growth in all 7 countries.
Mugume and Obwona (1998) examine the interaction between fiscal deficits and other
macro-level variables for Uganda in the post reform period. The results show that the
unsustainability of the budget deficit has implications for public, external and monetary
sectors. In particular, the study finds a negative relationship between fiscal deficits and
economic growth. Also the study reveals that fiscal deficit is linked to inflation, exchange
rate depreciation and the widening of current account deficit.
P a g e | 43
Vuyyuri and Seshaiah (2004), study the interaction of budget deficit with other
macroeconomic variables (Nominal effective exchange rate, GDP, Consumer Price Index and
money supply) for India, using Cointegration approach and Variance Error Correction
Models (VECM) for the period 1970 – 2002. The study reveals that exchange rate, GDP,
consumer price index and money supply to be cointegrated. Also the study find a bi-
directional causality between budget deficit and nominal effective exchange rates. But the
results find no significant relationship between budget deficit and GDP, Money supply and
consumer price index. The results also show that the GDP Granger causes budget deficit.
Brownbridge and Mutebile (2007) analyze the impact of an increase in the fiscal deficit on
macroeconomic policy management and the fiscal sustainability in Uganda between 1980 and
2005. The study argue that aid funded deficits may have effects akin to the Dutch disease
through the appreciation of the exchange rate with adverse effects for export sector
competitiveness. In addition, Keho (2010) investigates the causal relationship between budget
deficits and economic growth for seven West African countries over the period 1980-2005.
The author finds mixed results with three out of the seven countries showing no evidence of
causality, one showing a unidirectional causality running from deficit to growth and the rest
showing two-way causality between budget deficits and economic growth.
Georgantopoulos and Tsamis, (2011) investigate the casual link between budget deficit and
other macroeconomic variables (Consumer Price Index (CPI), Gross Domestic Product
(GDP) and Nominal Effective Exchange Rate for Greece during the period 1980-2009. Their
findings reveal no link between the budget deficit and CPI but they find casual links between
budget deficit and GDP and Nominal Effective Exchange Rate.
Odhiambo, Momanyi, Othuon and Aila (2013)also study the relationship between budget
deficits and economic growth in Kenya for the period 1970 to 2007. The regression results
P a g e | 44
show a positive relationship between budget deficit and economic growth. Buscemi and
Yallwe, (2012) using GMM technique to test for the nexus between budget deficit and
economic growth using dataset of China, India and South Africa.The study shows that fiscal
deficit results are significant and positively correlated to economic growth and saving in
China, India and South Africa. However, the authors reveal that real interest rates are
negatively and significantly correlated with economic growth and saving. The main
conclusion by the authors is that, fiscal deficit affects the economic growth and saving
through the means financing the deficit.
On the other, Najid (2013) uses Granger-causality test in estimating the relationship between
budget deficit and economic growth of Pakistan using time series data for the period of 1971
to 2007. The result shows that bi-directional causality runs from budget deficit to GDP and
GDP to budget deficit. The study concluded by analyzing that Budget deficit has no role in
bringing back the economy of the country to a stable level of equilibrium.
Lwanga and Mawejje (2014) use a vector autoregressive (VAR) and vector error correction
model (VECM) approach, pairwise granger causality test, and variance decomposition
techniques in investigating the macroeconomic effects of budget deficits in Uganda for the
period 1999 to 2011. The cointegration results indicate that the variables under study are
cointegrated and thus have a long run relationship. Similarly, the results based on the VECM
reveal unidirectional causal relationships running from budget deficits to current account
balance, inflation to budget deficit and budget deficit to lending interest rates. But the results
show no causal relationship between gross domestic product (GDP) and budget deficits in
Uganda. The Pairwise Granger Causality test results reveal unidirectional causal relationships
running from budget deficit to current account, budget deficit to GDP, inflation to budget
deficit, and a bi-directional causal relationship between the current account balance and GDP.
P a g e | 45
The study concluded that budget deficits in Uganda are responsible for widening current
account deficit and raising interest rates
2.3.2 Nigerian-Ghanaian Evidence
Obi and Nurudeen (2008) empirically investigate the effects of fiscal deficits and government
debt on interest rate in Nigeria between 1970 and 2005 using Vector Autoregression
approach (VAR).The empirical findings of the study reveal that the explanatory variables
account for approximately 73.6 percent variation in interest rate in Nigeria. The estimation
also shows that fiscal deficits and government debt are economically and statistically
significant.
Chukwu (2009) employs Johansen and Juselius (2009) multivariate cointegration technique
and Vector Error Correction Model (VECM) to comparatively examine the causal
relationships between internal and external deficits using Nigerian quarterly and annual data
from 1971 to 2006. The cointegration test result suggests a long run stable relationship
between internal deficits and external deficits for both low and high frequency series. Thus
the study supported the Richardian Equivalence hypothesis for quarterly series but validates
the conventional Keynesian view for annual series.
Odionye and Uma (2013) employ augmented Granger causality test approach in examining
the relationship between budget deficit and interest rate in Nigeria using Vector Error
Correction model (VECM) for the period of 1970:1 – 2010:1. The results reveal that in the
long run co-integrating equation, budget deficit exert a positive and significant impact on
interest rate implying that a high budget deficit will increase interest rate in the country. The
result supports the Keynesian proposition. Also, evidence from Johansen co-integration result
indicates that there is a long run relationship between budget deficit and interest rate.
P a g e | 46
Sowa (1994) utilizes Error Correction Model (ECM) in estimating an inflation equation for
Ghana over the period 1963 - 1990. The study shows that inflation in Ghana is influenced
more by output volatility than by monetary factors, both in the long run and in the short run.
In Nigeria, Onwioduokit (1995) employs Granger causality test in investigating the causal
relationship between inflation and fiscal deficits using annual data from 1970 to 1994. The
variables in the empirical model are ratio of fiscal deficit to gross domestic product (GDP),
level of fiscal deficit and inflation rate. The study shows that fiscal deficit causes inflation
without a feedback effect but however feedback exist between inflation and the ratio of fiscal
deficit to gross domestic product.
Omoka and Oruka (2010) employ pair-wise Granger causality test in an attempt to offer
evidence on the causal long term relationship between budget deficit, money growth and
inflation in Nigeria using annual time-series data covering from 1980 to 2005. In considering
the broadest definition of money supply, the study reveal that money supply causes budget
deficit which means that the level of money supply in the Nigerian economy determines
whether there has been or there will be budget deficits. Inflation and budget deficit show a
bilateral or feedback causality proving that the changes that occur in inflation could be
explained by its own lag and also the lag values of budget deficit and in the same vein,
changes that occur in budget deficits are explained by its lagged values and the lagged values
of inflation. The implication of their findings is that both budget deficit and inflation could be
caused by money supply meaning that they are both monetary phenomena.
Oladipo and Akinbobola (2011) employ Granger causality pair-wise test in determining the
causal relationship between budget deficit and inflation in Nigeria. The study find that there
is no causal relationship from inflation to budget deficit but from budget deficit to inflation in
Nigeria. This indicates that budget deficit affects inflation through fluctuations in exchange
rate in the Nigerian economy.
P a g e | 47
Dockery, Ezeabasili and Herbert (2012) examine the long term relationship between fiscal
deficits and inflation for Nigeria using a modelling approach that incorporates the theory of
cointegration and employs vector error correction model. The empirical results show a
positive but insignificant relationship between fiscal deficits and inflation. The analysis of the
Nigerian data also indicate a weak link to previous levels of fiscal deficits with inflation and
provide, moreover, evidence of a positive long-run relationship between money supply
growth and inflation, suggesting therefore that money supply growth is pro-cyclical and tends
to grow at a faster rate than the rate of inflation. Similarly, from the impulse response and
variance decomposition analysis, the study finds that the length of inflation is an important
determinant of the ability of the system to return to its long-run equilibrium following a
shock.
Ozurumba (2012) employs Autoregressive Distributed Lag (ARDL) model and the Granger-
causality test in examining the causal relationship between inflation and fiscal deficits in
Nigeria covering the period 1970-2009. The result of the Granger-causality test shows that
the null hypothesis which says that fiscal deficit does not cause inflation should be rejected
since the result is significant with probability less than 0.05. This implies that fiscal
deficit/GDP causes inflation. However, no feedback mechanism exist between the variables.
The results from the ARDL test confirm a significant negative relationship between growth in
fiscal deficit (% of GDP) and inflation.
Akosah (2013) employs Ordinary Least Squares (OLS) in estimating the threshold effect of
Budget deficits on economic growth in Ghana using quarterly data from 2000 to 2012. The
study finds an inverse long run relationship between budget deficit and economic growth,
especially as the deficits have often been used to finance recurrent expenditures, suggesting
that high budget deficit, driven by recurrent expenditures, slows down economic growth. In
P a g e | 48
the short run, however, the result reveals that budget deficits promote economic growth, but a
deficit beyond the threshold level of 4% of GDP is detrimental to economic growth. This
result is robust and supports the West African Monetary Zone’s (WAMZ) primary fiscal
convergence criterion. The result also shows that fiscal restraint to the level below the
threshold will both stimulate a sustainable economic growth and overall stability in Ghana.
Larbi (2012) investigates the long-run impact of budget deficit on economic growth in Ghana
covering the period of 1980 to 2010 using Johanson cointegration procedure and Granger
Causality test. The empirical evidence from the Johanson cointegration test shows that budget
deficit exerts no significant long-run impact on economic growth. Further evidence from the
Granger Causality test conducted suggests significant positive long-run relationships between
the capital stock, openness, total government expenditure and growth rate with budget deficit
coefficient variable – positive and statistically significant.
Yaya (2010) utilizes the Granger-causality test using a sample of seven West African
countries namely; Benin, Burkian Faso, Cote d’ Ivoire, Mali, Niger, Senegal and Togo to
examine the causal relationship between budget deficits and economic growth in these
countries over a period of 26 years. The study reveals that in three countries, there exist no
causality between budget deficits and economic growth. The above findings indicate a two-
way causality in three countries, deficits having adverse effects on growth. Overall, these
results give support to the West African Economic and Monetary Union budgetary rule
aiming at restricting the size of budget deficit as a prerequisite for sustainable growth and real
convergence. In four other countries, there exist causality evidence between budget deficit
and economic growth implying that deficits retard economic growth rate. These findings have
two main implications. First, they lend support to the control of budget deficits within the
West African Economic and Monetary Union countries in order to increase domestic savings
and finance economic growth. Second, evidence of causality running from economic growth
P a g e | 49
to deficit makes difficult the control of budget deficit to depend on the aggregate
macroeconomic environment. In periods of recession, revenues are expected to decrease,
generating fiscal imbalance. In periods of expansion, revenues increases which leads to
reduction in the size of deficit.
Wosowei (2013) employs Ordinary Least Squares (OLS) in determining the relationship
between fiscal deficit and macroeconomic performance in Nigeria over the period 1980 to
2010. However, the empirical findings show that fiscal deficits even though it met the
economic a prior in terms of its negative coefficients, macroeconomic output is statistically
significant. The result also shows a bilateral causality relationship between government
deficit and gross domestic product, government tax and unemployment, while there is an
independent relationship between government deficit and government expenditure and
inflation.
Antwi and Mills (2013) evaluate budget deficit sustainability of Ghana between 1960 and
2010 using the present value budget constraint approach; employing annual time series data,
the Augmented Dicky Fuller and Philip-Parron tests for unit root rejected the null hypothesis
at 1 percent significance level after first difference. Hence, both government expenditure and
revenue of Ghana are stationary and integrated of order one. The Granger causality test from
the regression result supported a bi-directional causation such that both expenditure and
revenue of Ghana have temporal precedence over each other. This means past and present
values of government revenue provide important information to forecast future values of
expenditure. The test for cointegration favours the sustainability of budget deficit of Ghana at
10 percent significance level in the strong sense. The result equally shows that the Ghanaian
government can continue to service its past accumulated deficits without large future
correction to the balance of income and expenditure. Again, the study achieved the
P a g e | 50
conventional negative sign of the speed of adjustment to long run equilibrium following
shocks to the system at 5% significance level.
Bakare, Adesanya, and Bolarinwa (2014) empirically investigate the long term relationship
between budget deficit, money supply and inflation in Nigeria from 1975 to 2012 using
annual time-series data. The stationary test carried out using Augmented Dickey-Fuller
(ADF) reveals that the variables in the model are stationary at levels. However, the study
tested for a long run relationship among the variables using Johansen co-integration test
which also suggests that there are at least three co-integrating vectors among these variables.
Also, the estimated coefficient of the Error Correction Model shows that about 132% of the
errors in the short run are corrected in the long run. Thus the overall result between inflation
rate and growth of money supply, growth ratio of budget deficit to GDP and growth ratio of
external debt to GDP indicates that the specified model is statistically significant at 5% level.
Awe and Funlayo (2014) investigate the short and long run implications of budget deficit
oneconomic growth in Nigeria using time series data covering period of 1980-2011. The
result from the OLS regression analysis indicates that a negative relationship exist between
budget deficit and economic growth. The study also employs Johansen cointegration
technique in investigating the long run effect of budget deficit. The result shows that there is
a significant long-run relationship between budget deficit and economic growth in Nigeria.
The result from Error Correction Model reveals that budget deficit shows a negative
relationship with gross domestic product while gross capital formation (investment) shows a
positive relationship with GDP.
Oyeleke and Adebisi (2014) analyze the fiscal deficit sustainability of Ghana using Ordinary
Least Squares (OLS), Eagle-Granger and Error Correction Model procedures with annual
time series data covering from 1980 to 2010 to determine whether or not the economy has
achieved feat as a criterion required for membership in proposed West African Monetary
P a g e | 51
Zone (WAMZ). The study reveals that government revenue and expenditure together with
budget deficit have long-run weak relationship which indicate sustainability. Also, the study
reveals that only 29% of disequilibrium between government revenue and expenditure
generated in the economy was restored yearly following shocks in the economy, thus
concluding that the country might not qualify for membership in WAMZ
Osuka and Achinihu (2014) analyze the impact of budget deficits on macroeconomic
variables in Nigeria using annual time-series data from 1981 to 2012. The result from
Johanson cointegration test shows that the variables in the study are all cointegrated of order
one showing the presence of long-run relationship between budget deficits and some selected
macroeconomic variables (GDP, interest rate, nominal exchange rate and inflation rate). The
result from Granger Causality test reveals that there exists no causality between deficits and
interest rate, budget deficits and inflation and budget deficit and nominal exchange rate.
In conclusion, the review of empirical literature on the relationship between budget/fiscal
deficits and other macroeconomic variables gives quite mixed results with some studies
showing no relationship between budget deficits and other macroeconomic variables, some
confirming that indeed budget deficits affect all or some macroeconomic variables. This
implies that the relationship between budget deficits and other macroeconomic variables is
case/country specific and depends on a number of conditions like source of deficit financing
and expenditure pattern, size of the deficit and so on (Lwanga & Mawejje, 2014).
Table 1.0:Tabular Summary of Empirical Literature Reviewed
STUDY COUNTRY SAMPLE METHODOLOGY VARIABLES FINDINGS
Vamvoukas (1998)
Greece 1970 – 1990 OLS, ECM, Johanson Cointegration test
Budget deficits, money supply, government expenditure, unemployment, inflation and
significant and positive relationship between budget deficits and interest
P a g e | 52
RGNP rates
Gupta and Uwilingiye (2009)
South Africa
1961 – 2005 VAR, Granger Causality test
Budget deficits, Treasury bill rate, interest rate, Exchange rate
No causal relationship between budget deficit and treasury bill rate, but positive cointegration exist for both data
Bonga-Bonga (2011)
South Africa
1960 – 2000 VAR Budget deficit, interest rate
Positive relationship between deficits and long-term interest rates
Akinboade (2004)
South Africa
1961 – 2005 LSE, Granger-causality test
Budget deficit, interest rates
No relationship between budget deficitand interest rates
Mukhtar and Zakaria (2008)
Pakistan 1960 – 2005 ECM, Granger-causality test
Budget deficit, nominal interest rates, and GDP
Budget deficits and budget deficit-GDP ratio have no significant effect on nominal interest rates.
Aisen and Hauner (2008)
60 Advanced and emerging economies
1970 - 2005 Panel estimation Budget deficit, interest rate
Effects of budget deficits on interest rates are more robust in the emerging markets and in later periods than in the advanced economies and in earlier periods
P a g e | 53
Noula (2012) Cameroon 1974 – 2009
ADF test, ECM, Pairwise Granger Causality, and Loanable Funds Model
Budget deficit, Money supply, RGDP, exchange rate, and foreign interest rate
Significant positive association between Budget deficit and interest rate
Guess and Koford (1984)
17 OECD countries
1949 – 1981 Panel estimation, Granger Causality test
Budget deficits Inflation, GNP, and Private Investment
Budget deficits do not cause changes in the variables
Darrat (1985) US 1960 – 1970 OLS Budget deficit, Inflation
Monetary growth and federal deficits significantly influence inflation in post 1960 period
Easterly and Schmidt-Hebbel (1993)
10 countries Panel data Panel estimation Budget deficit, inflation, real interest rates
Medium term money financing of the deficit leads to higher inflation, while debt financing leads to higher real interest rates
Tekin-Kuru and Ozman (1998)
Turkey VAR Budget deficit, money supply, Inflation
No direct relationship between budget deficit and inflation
Darrat (2000) Greece 1957 – 1993 ECM Budget deficit, inflation
Statistical significant and positive impact of budget deficit on inflation in short run
Catão and Terrones (2005)
107 countries
1960 – 2001 Panel estimation Budget deficit, inflation, GDP
positive impact of budget deficit on inflation
Makochekanwa (2008)
Zimbabwe 1980 – 2005 VAR Budget deficit, inflation, exchange rate, GDP
Significant positive impact of budget deficit on inflation in the
P a g e | 54
long run Habibullah, Cheah and Baharom (2011)
13 Asian developing countries
1950 – 1999 Panel estimation, Granger-Causality test
Budget deficit, inflation
Long run relationship exist between budget deficit and inflation, and budget deficits are inflationary in Asian developing countries
Ndashau (2012) OLS Budget deficits, inflation, exchange rates
Effect of budget deficits on inflation is not statistically significant
Lin and Chu (2013)
91 Countries
1960 – 2006 DPQR, ARDL Budget deficits, inflation, exchange rate
Fiscal deficits are inflationary only in high inflation countries
Dwyer (1982) US 1952 – 1978 VAR Budget deficits, interest rates, money stock
No effects of budget deficits on inflation, interest rates, and inflation
Barro (1990; 1991)
98 Countries
1960 – 1985 Panel estimation RGDP, Investment
The Budget deficit-RGDP has a negative relationship with GDP and investigate
Karras (1994) 32 Countries
1960 – 1980 OLS, GLS Budget deficit, inflation, investment, and RGDP
Budget deficit do not cause inflation, but negatively correlated RGDP
Al-Khedir (1996)
G-7 Countries
1964 – 1993 VAR Budget deficits, interest rate, trade balance, GDP
Trade balance has been worsen by the budget deficit and GDP in all the 7 Countries
Mugume and Obwona (1998)
Uganda Post reform period
OLS Budget deficit, GDP, inflation, exchange rate depreciation and Current account
Negative relationship between budget deficit and GDP
P a g e | 55
deficit Vuyyuri and Seshaiah (2004)
India 1970 - 2002 VECM Exchange rate, GDP, CPI and money supply
No significant relationship between budget deficit and GDP, Money supply and CPI
Brownbridge and Mutebile (2007)
Uganda 1980 – 2005
Keho (2010) 7 West African countries
1980 – 2005 Panel estimation, Granger-causality test
Budget deficit, GDP
3 countries shows show evidence of no causality, and 4 countries show two-way causality between budget deficits and GDP
Georgantopoulos and Tsamis (2011)
Greece 1980 – 2009 OLS, Granger-causality test
Budget deficit, CPI, GDP, nominal exchange rate
No link between budget deficit and CPI, but causal link exist among budget deficit, GDP and nominal exchange rate
Odhiambo, Momanyi, Othuon L, and Aila (2013)
Kenya 1970 – 2007 OLS, Granger-causality test
Budget deficit, GDP
Positive relationship between budget deficits and economic growth
Buscemi and Yallwe (2012)
China, India, and South Africa
1970 – 2007 GMM Budget deficit, real interest rate, GDP, saving
Budget deficit affects GDP and saving through financing the deficit. Also, real interest rates are negatively and significantly correlated with GDP and saving
P a g e | 56
Najid (2013) Pakistan 1971 – 2007 OLS, Granger-causality test
Budget deficit, GDP
Bi-directional causality Budget deficit to GDP, and GDP to budget deficit
Lwanga and Mawejje (2014)
Uganda 1999 - 2011 VAR-VECM, pair-wise granger-causality, variance decomposition
Budget deficit, inflation, interest rate, current account balance
No causal relationship between GDP and budget deficits. And bi-directional causal relationship between the current account balance and GDP
NIGERIAN-GHANAIAN EVIDENCE
STUDY COUNTRY SAMPLE METHODOLOGY VARIABLES FINDINGS
Obi and Nurudeen (2008)
Nigeria VAR Budget deficit, government debt, interest rate
Budget deficit and government debt are economically and statistically significant
Chukwu (2009) Nigeria 1971 – 2006 VECM, Johanson cointegration test
Internal and External deficits,
Long run stable relationship exist between internal and external deficits for both low and high frequency series
Odionye and Uma (2013)
Nigeria 1970:1 to 2010:4
VECM, Granger-causality test
Budget deficit, interest rate
Budget deficit exert a positive and significant impact on interest rate in the long run
Sowa (1994) Ghana 1963 - 1990 ECM, Johanson Cointegration test
Budget deficit, GDP, Money supply
Budget deficit has effect on GDP in both
P a g e | 57
short and long run
Onwioduokit (1995)
Nigeria 1970 – 1994 Granger-causality test
Budget deficit, GDP, inflation
Budget deficit causes inflation without feedback effect, but feedback effect exist between inflation and ratio of budget deficit-GDP
Omoka and Oruka (2010)
Nigeria 1980 – 2005 Pair-Wise test Budget deficit, inflation, money supply
money supply causes budget deficit; Inflation and budget deficit show a bilateral or feedback causality
Oladipo and Akinbobola (2011)
Nigeria Granger-causality pair-wise test
Budget deficit, inflation, exchange rate
No causal relationship from inflation to budget deficit, but from deficit to inflation
Dockery, Ezeabasili, and Herber (2012)
Nigeria VECM, Johanson Cointegration test
Budget deficit, inflation
Positive but insignificant relationship between budget deficit and inflation
Ozurumba (2012)
Nigeria 1970 – 2009 ARDL, Granger-causality test
Budget deficits, inflation
Significant and negative relationship between growth in fiscal deficit and inflation
Akosah (2013) Ghana 1990 – 2012 OLS Budget deficit, GDP, exchange rate
Negative relationship between budget deficit and GDP in the long run
Larbi (2012) Ghana 1980 – 2010 Johanson Cointegration and
Budget deficit, openness,
Positive long run
P a g e | 58
Granger Causality test
capital stock, government expenditure, growth rate
relationship between budget deficit and economic growth; and significant positive causality exist the capital stock, openness, total government expenditure & growth rate
Yaya (2010) 7 West African countries
26 years Granger-causality Budget deficit, GDP
no causality between budget deficits and GDP in 3 countries; and causality exist between budget deficits and economic growth in 4 countries
Wosowei (2013)
Nigeria 1980 – 2010 OLS, Granger-causality test
Budget deficit, GDP, unemployment, inflation, government expenditure and tax
Bi-lateral causal relationship betweenbudget deficit and GDP, government tax and unemployment
Antwi and Mills (2013)
Ghana 1960 – 2010 Present Value (PV) budget constraint approach, Granger-causality, ADF and PP tests
Budget deficit, interest rate, exchange rate, government expenditure and revenues
Bi-directional causation in both government expenditure and revenues
Awe and Funlayo (2014)
Nigeria 1980 – 2011 OLS, ECM, Johanson cointegration test
Budget deficit, GDP, investment
Long-run relationship exist between budget deficit and GDP, with negative correlation with growth, and positive with
P a g e | 59
investment.
Bakare, Adesanya, and Bolarinwa (2014)
Nigeria 1975 - 2012 ADF test, ECM, and Johanson cointegration test
Budget deficit, inflation, money supply, external debt
Long-run relationship exist among the variables, and thus statistically significant
Oyeleke and Adebisi (2014)
Ghana 1980 – 2010 OLS, Eagle-Granger, and ECM,
Government expenditure, revenue, and deficit
long-run weak relationship between government expenditure and revenue
Osuka and Achinihu (2014)
Nigeria 1981 - 2012 OLS, Johanson cointegration test, and Granger-causality test
Budget deficit, GDP, interest rate, nominal exchange rate and inflation rate
long-run relationship exist between budget deficits and some selected macroeconomic variables
SOURCE: Compiled by the Researcher
2.4 Limitation of Previous Studies and Value-To-Be-Added
There have been a number of studies on the effects of budget deficits on macroeconomic
variable(s) in both developed and developing economies evidenced from the empirical
literature reviewed.Based on the reviewed literature, one can conclude that the earlier studies
made significant contributions on the stock of knowledge as it relates budget deficits and
macroeconomic variable(s). Moreover, the earlier studies such as Sowa (1994), Vamvoukas
(1998), Uwilingiye and Gupta (2007) among others were considered a sine qua non for
generating the contemporary research interest on this topic and in these developing
economies - Nigeria and Ghana.
P a g e | 60
However, the literature discussed above shows that the effects of budget deficits (short or
long run) on macroeconomic variables are mixed and country case; and/or as a result of this,
the methodology and data employed in the empirical literature reviewed are foiled with
limitations. It is pertinent to take into cognizance that almost all the works reviewed were
single economy studies such as Onwioduokit (1995), Chukwu (2009), Bonga-Bonga (2011),
Antwi and Mills (2013), Osuka and Achinihu (2014) among others. Similarly, most of the
studies reviewed only consider either the short or the long run implications of budget deficits
on a particular or single macroeconomic variable; for instance, Vamvoukas (1998), Omoka
and Oruka (2010), Akosah (2013) among others. However, judging from the large body of
literature reviewed, no empirical study was found to have articulated both the short and long
run effects of budget deficits on the selected macroeconomic variables in these West African
countries - Nigeria and Ghana. Therefore, this study tend to fill this lacuna as well as add
more value to the literature by meticulously investigating the short and long run effects of
budget deficits on theselected macroeconomic variables in both Nigerian and Ghanaian
economies while paying particular andkeen attention on the country’s macroeconomic
environment so as to control for factors which was omitted in the literature.
Consequently, the study employs more robust technique – Seemingly Unrelated Regression
(SUR) model in investigating both the broad and specific research objectives.Also, the study
employs a longitudinal annual time-series and quarterly dataset spinning from 1970 to2013.
This is in sharp contrast to country specificNigerian or Ghanaian earlier studies conducted
which employed data ranging a shorter time period. To this objective, the study outlines a
modelling framework that links budget deficits to interest rate, inflation and economic growth
that captures the macroeconomic realities of the Nigerian and Ghanaian economies over the
period reflected by the available data.
P a g e | 61
Another important shortcoming of most previous studies which the current studyseeks to
overcome is that explicit attention was not paid to the time-series characteristics of the data
used. Using recent developments in time series econometrics as provided by Engle and
Granger (1987), Andrew (1991), Phillips and Peron (1988), Dickey and Fuller (1979), Newey
and West (1994), MacKinnon (1996),Johanssen (1991, 1995), Engsted and Bentzen
(2001),this studyis able to derive the relationship between the variables in themodel adopted.
CHAPTER THREE
RESEARCH METHODOLOGY
3.1 Theoretical Framework
To analyze the relationship between fiscal deficits and some selected macroeconomic
variables, it is helpful to begin with some national income accounting identities. The national
income account identity provides the basis for the relationship between budget deficit and
macroeconomic variables. This theoretical framework follows the work of Ezeabasili and
Mojekwu (2011). Under the fiscal approach to balance of payments; the government budget
balance is defined as the gap between revenues and expenditures. The above definition
derives from the national income identity, as:
P a g e | 62
Y = C + I + G + (X – M) …………………………………………………………. (1)
Where Y represents real GDP, C is private consumption, I stands for private investment, G is
government consumption, X and M stand for exports and imports respectively. Assuming the
aggregate demand A = C + I + G then Equation (1) can be rewritten as follows:
Y – A = X – M …………………………………………………………………….. (2)
Equation (2) reflects the behaviour of the external sector of the economy. The direct
interpretation is that, external imbalances always trigger a series of developments in the
economy, which in this case is budget deficit. Therefore, any attempt to restore the balance
must include effort to align revenue with expenditure. In order to isolate the disposable
income, tax (T) and international reserve (R) (the latter is introduced basically on the
assumption of the fixed exchange rate regime) are introduced into the national income
identity. It follows that Equation (1) will become:
Y + R – T = C + I + (G – T) + (R + X – M) ……………………………………… (3)
In the following equation, S (savings) is the disposable income minus private consumption.
That is: S = Y + R – T – C, the private absorption capacity is represented by (C + I), (G – T)
is for budget deficit, while the current account balance (CAB) is represented by (R + X – M),
R represents international transfer receipts and T stands for taxes. Substituting S and CAB by
their respective components, we get:
(S – I) + (T – G) = (R + X – M) ……………………………………………………. (4)
It is often argued that deficit in the current account occurs when aggregate investment
outweighs aggregate savings (Blanchard & Johnson, 2012; Abel & Bernanke, 2001; and
Mankiw, 2003). However, if investments equalssavings and government expenditure is
greater than its revenue then, the current account deficit becomes inevitable. The literature on
the current account is quite obvious when it indicates the degree at which the domestic
economy interacts with its external assets. Thus, (X + R – M) would also be equivalent to the
increase in net official assets plus the rate of capital outflow that is ∆NFA. Hence;
P a g e | 63
CAB = ∆NFA ………………..…………………………………………………….. (5)
The links between net savings of the private sector and the public sector deficit is easily
appreciated through the following illustration.
(S – I) + (T – G) = ∆NFA …………………………………………………………. (6)
The direct interpretation of the above equation assuming S = I is that:
(i) A budget deficit could be financed through a reduction in external net claims, which can
be done through increase in external public debt or reduction of international reserves in the
case of a fixed exchange regime. (ii) Budget deficit could also be financed domestically,
through increase in government debt held by the private economic sector. The relationship in
the banking system provides a clear understanding on how domestic borrowing is applied in
financing a budget deficit and the balance sheet is given as follows:
∆NFAb = ∆M2 - (∆DC g + ∆DC nb) …………………………………………........ (7)
The liability of the banking system is represented by M2 that is the broad money, ∆DC g is
domestic credit of the banking system to government and ∆DC nb is the credit of non-banking
sector (private sector) to the government. Equation (7) expresses the difference between
money expansion and credit expansion and, which works as follows. An increase in money
relative to credit expansion will reflect as an increase in the net foreign asset. In countries
where the capital markets are not advanced (such as Nigeria and Ghana), budget deficit is
usually financed through domestic and external borrowing. This expression can be simplified
as follows:
G – T = ∆DC g - ∆NFA g ………………………………………………………….. (8)
By substituting (8) into (7), the relationship between the financing of the budget deficit and
the banking system is brought to the fore. Thus:
G – T = ∆M2 - ∆DC nb – (∆NFA b + NFA g) ……………………………………… (9)
Equation (9) illustrates the sources through which government deficit can be financed. First,
by an increase in money (∆M2). Second, borrowing from non-banking sector. Lastly, by a
P a g e | 64
…….. (10)
reduction in international reserve or external borrowing. In all, increased budget deficit will
translate into increased current account deficit which precipitates new external borrowing or
draw down of external reserves. However, all the three means of financing a deficit may lead
to appreciation of real and nominal exchange rates under flexible exchange rate regime and
capital mobility.
3.2 Model Specification
Following the theoretical framework in section (3.1) above, the following models were
adopted;
Model I (for objective 1)
Haven laid the required theoretical foundations in the theoretical framework, the study links
the above stated foundations to specify our empirical models, thus the specification of the
interest rate model mirrors the works of Ariyo and Raheem (1991), cited in Tchokote (2004).
The specification of the model considers the following variables: Interest rate (INT) is the
dependent variable; while Government expenditure (GEX), budget deficit (BD), money
supply (MS), and inflation rate (INF) are the independent variable; µt is error term. The
model is represented as:
RIR = ƒ(GEX, BD, MS, INF)
The above relationship can be presented through a system of related equations by considering
a set of t linear equations for each t time points, and thus becomes:
RIR t1Ng = β01 + β1GEXt1
Ng + β2BDt1Ng + β3InMS t1
Ng + β4INF t1Ng + µt1
Ng
RIR t2Gh= β02 + β1GEXt2
Gh + β2BDt2Gh + β3InMS t2
Gh + β4INF t2Gh +µt2
Gh
Where: β0 = the intercept and, β1, β2, β3, β4 are the coefficients of the regression equation.
Ng, Gh = Ng: for Nigeria, Gh: for Ghana respectively
A priori, it is expected that the following relationship will occur; β1, β2, β4> 0; β3 < 0;
P a g e | 65
..... (11)
…. (12)
Model II (for objective II)
The second objective of the study is to examine the short and long run effects of budget
deficit on inflation in Nigeria and Ghana. Following the literature, we present an econometric
model similar to the work of Dockery and Ezeabasili (2012) which essentially is informed by
standard economic theory as evince in both Classical and Keynesian approaches.
Accordingly, a log linear inflation-fiscal deficit econometric model is specified as follows:
INF t1Ng = Φ01 + Φ1BDt1
Ng + Φ2InRGDP t1Ng + Φ3InMS t1
Ng + Φ4EXDEPt1Ng + γt1
Ng
INF t2Gh = Φ02 + Φ1BDt2
Gh + Φ2InRGDPt2Gh + Φ3InMS t2
Gh + Φ4EXDEPt2Gh + γt2
Gh
Where: INF = Inflation rate
BD = Budget deficits
MS = Money supply (M2)
RGDP = Real gross domestic product
EXDEP = Depreciation of the exchange rate
Ng, Gh = Ng: for Nigeria, Gh: for Ghana respectively
γt is a stochastic term.
(A priori, we expect the parameters: Φ1, Φ2, Φ3, Φ4> 0; Φ4> 0 or Φ4< 0).
Model III (for objective III)
The third objective of the study which states – short and long run effects of budget deficit on
economic growth in Nigeria and Ghana can be captured following the adoption of the work
of Awe and Funlayo (2014). The model is specified below:
GDP = ƒ(investment, savings, interest rate, budget deficit).
InRGDP t1Ng = α01 + α1INF t1
Ng + α2InSAV t1Ng + α3RIR t1
Ng + α4BDt1Ng + εt1
Ng
InRGDP t2Gh = α02 + α1INF t1
Gh + α2InSAV t2Gh + α3RIR t2
Gh + α4BDt2Gh + εt2
Gh
Where:
RGDP = Real Gross Domestic Product, a proxy for economic growth
P a g e | 66
INF = Inflation
SAV = Savings
INT = Interest rate
BD = Budget Deficit
α0 = the constant or the intercept
α1 – α4 = the coefficients of the explanatory variables
εt = Stochastic error term
(A priori, we expect the parameters: α1< 0, α2 > 0, α3< 0, α4 < 0 or α4 > 0)
3.3 Estimation Procedure
3.3.1 Empirical Models: Seemingly Unrelated Regression (SUR)
The SUR model is a generalization of multivariate regression using a vectorized parameter
model. The Y matrix is vectorized and vertically concatenated, yv. The design matrix, D, is
formed as a block diagonal with the jth design matrix, X j, is on the jj th block of the matrix.
The model is then expressed as:
E[Y(Nxp)]= { X1(Nxm1) β1 (m1x1) , X2(Nxm2) β2 (m2x1) , Xj(Nxmj) βj (mjx1) ,
Xp(Nxmp) βp (mpx1) ……………………………………………………………………. (13)
Where: mj is the number of parameters estimated (columns) by the j th design matrix, X j.
The SUR model is laid out in metric form as:
E(yv) D B
ŷI (Nx1)
XI 0 0 0 0 0 ᵝ1(mxl)
ŷ2 (Nx1)
(Nxm)
X2 0 0 0 0 ᵝ2 (m2xl) ……………. (14) E(yv) = … (Nxm… 0 0 0 ; (Npx1)
ŷj (Nx1)
x, 0 0 βj(mjx1) (sym) (Nxmj) 0 ŷp (Nx1
) (Nxmp)X
ᵝp(mpx1)
(Npx\ (NpxM) (Mx1) P where:
M is the total number of parameters estimated over the p models, M=∑mj.The (j=1)
P a g e | 67
[I N -Hi][I N - Hj]y j…..........….. (16)
parameter estimates are solved as:
B= [DIQ-ID] [D' Q-lyv]……………………………..… ..….. (15) (MxNp)(NpxNp)(NpxM) (MxNp)(NpxNp)(Npxl)
Q is weight matrix based on the residual covariance matrix of the Y variables and is formed
as:
Q = ∑⊗IN .The ijthelement of ∑is calculated as: (NpxNp) (pxp)
where: H, = Xj(X jXj) -1Xj is the hat matrix for the jth design matrix and df* is the average of
the numerator degrees-of-freedom (df) for the ithand jth models. Thus, a SUR model is an
application of generalized least squares (GLS). In fact, because the residual covariance
matrix is unknown and must be estimated from the data, this application is often called
feasible
generalized least squares (Timm, 2002).
3.3.1.1 Two Stage Least Squares and Instrumental Variables
Two-Stage least squares (2SLS) regression analysis is a statistical technique that is used in
the analysis of structural equations. This technique is the extension of the OLS method. It is
used when the dependent variable’s error terms are correlated with the independent variables.
Additionally, it is useful when there are feedback loops in the model (Angrist & Imbens,
1995).
Many economic models involve endogeneity: that is, a theoretical relationship does not fit
into the framework of y-on-X regression, in which we can assume that the y variable is
determined by (but does not jointly determine) X (Wooldridge, 2006). From a mathematical
standpoint, the difficulties that this endogeneity cause for econometric analysis are identical
to problems associated with omitted variables, and that of errors-in-variables, or
measurement error in the X variables. In each of these three cases, OLS is not capable of
P a g e | 68
delivering consistent parameter estimates. The general concept is that of the instrumental
variables estimator; a popular form of that estimator, often employed in the context of
endogeneity, is known as two stage least squares (2SLS) (Wooldridge, 2006).
Instrumental variable methods allow consistent estimation when the explanatory variables
(covariates) are correlated with the error terms of a regression relationship. Such correlation
may occur when the dependent variable causes at least one of the covariates ("reverse"
causation), when there are relevant explanatory variables which are omitted from the model,
or when the covariates are subject to measurement error. In this situation, ordinary linear
regression generally produces biased and inconsistent estimates(Bullock, Green & Ha, 2010).
However, if an instrument is available, consistent estimates may still be obtained. An
instrument is a variable that does not itself belong in the explanatory equation and is
correlated with the endogenous explanatory variables, conditional on the other covariates. In
linear models, there are two main requirements for using an IV:
• The instrument must be correlated with the endogenous explanatory variables,
conditional on the other covariates.
• The instrument cannot be correlated with the error term in the explanatory equation
(conditional on the other covariates), that is, the instrument cannot suffer from the
same problem as the original predicting variable.
3.3.2 Lag Length/Bandwidth Selection
Appropriate lag length/bandwidth shall be selected using coefficient of determination (R2),
Akaike Information Criterion (AIC), Schwarz information criterion (SC), Hannan-Quinn
information criterion (HQ), and Newey-West method criteria to determine the number of
lags/bandwidths in the series test.
P a g e | 69
3.3.3 Stationarity/Unit Root Test
In regressing a time series variable on another time series variable (s), one might obtain a
high R2 even though there is no meaningful relationship between the variables. This situation
shows the problem of spurious or nonsense regression (Gujarati, 2007). To test for a unit root
in the series, the Augmented Dickey-Fuller (ADF), as specified in Dickey and Fuller (1979),
and Phillips-Peron (Phillips and Peron, 1988) will be employed to ensure more robust results
in testing for the stationarity of the data in our models at level and at differences. However,
the test statistics of the Philips-Peron have the same distribution as those of Dickey-Fuller
with critical levels as provided by MacKinnon (1996). Thus, the model for the ADF is
specified below:
yt = µ + pyt-1 + εt ............................................................................................... (17)
Where µ and p are parameters and εt is assumed to be white noise, y is a stationary series if -
0< p < 1. If p =1, y is a non-stationary series; if the process is started at some point, the
variance of y increases steadily with time and goes to infinity. If the absolute value of p is
greater than one, the series is explosive. Therefore, the hypothesis of a stationary series can
be evaluated by testing whether the absolute value of p is strictly less than one. The simple
unit root test described above is valid because the series is an AR(1) process. If the series is
correlated at higher order lags, the assumption of white noise disturbances is violated. The
Dicky-Fuller tests take the unit root as the null hypothesis H0: p =1. Since explosive series do
not make much economic sense, the null hypothesis will be tested against the one-sided
alternative H1 : p <1. The null hypothesis of a unit root is rejected against the one-sided
alternative if the t-statistic is less than the critical value.
3.3.4 Cointegration Test
To investigate the existence of a long run relationship between budget deficits and some
selected macroeconomic variables, we explore the existence of a long run relationship among
P a g e | 70
the variables in our models. If the selected variables in the models are found to be
cointegrated, it will provide statistical evidence for the existence of a long run relationship.
We employ the Engel-Granger (1987) cointegration test. Basically, the idea of cointegration
is predicated on the thesis that even though two time series may not themselves be stationary,
a linear combination of the two non-stationary time series may be stationary. If this is the
case, the two original non-stationary time series are said to be ‘cointegrated’. Usually, for co-
integration, the two time series have to be stationary after the same number of differencing. If
a given time series becomes stationary after first differencing, it is said to be integrated of
order one I(1). If the time series becomes stationary after second differencing, it is integrated
of order two I(2). If the original time series is stationary, it is integrated of order zero I(0).
When a linear combination of two I(1) series is stationary, then the two time series are
cointegrated.
3.4 Model Justification
There exist many techniques of estimating system equations with heterogeneous units such as
panel data analysis (which is appropriate when the number of units, N, is large but the
number of observations, T, is small), Panel VAR (which assumes that all variables in the
system are endogenous) and Seemingly Unrelated Regression (SUR) model (which takes
right-hand side variables as exogenous and is appropriate when N is small but T is infinite, or
large) (Baltagi, 2008) also cited in (Orji, Akachukwu & Ilori, 2014). Based on the nature of
the system equations in our models, where there exist only two (2) distinct heterogeneous
units (with six equations) and forty four (44) observations, this study shall employ the SUR
model. The choice of the SUR model for analysis of this study is based on the fact that it
relatively takes into consideration the contemporaneous cross-equation error correlation, and
it is more appropriate for our study because of small heterogeneous unit N and Large sample
observation T. Furthermore, the SUR model estimates the parameters of all equations
P a g e | 71
simultaneously, so that the parameters of each single equation also take the information
provided by the other equations into account. This result in greater efficiency of the
parameter estimates, because additional information is used to describe the system.
The choice of the models are necessary so as to estimate the short and long run effects of
budget deficits on selected macroeconomic variables in Nigeria and Ghana for the period
under investigation. Again, because of the endogeneity of real interest rate, real GDP, and
inflation being used as explanatory variables in other equations in the models (model 1 – 3),
Ordinary Least Square (OLS) estimator may be biased;therefore, we shall employ both SUR
model and Two Stage Least Squares (2SLS) thus applying the first to third lags of the
variables as instruments (Wooldridge, 2006). The lags have weak contemporaneous
correlation with the error term. Similarly, Eagle-Grangercointegration test will be carried out
to ascertain if there exist long run effects of budget deficits on the macroeconomic variables
of interest.
3.5 Data Sources
Most of the data needed for this study will be sourced from the World Bank, World
Economic Outlook 2014, World Development Indicators (WDI) 2014, International
Monetary Fund (IMF), Ghanaian Central Bank Database 2013 and the Ghanaian Statistical
Services (GSS) Data. For the Nigerian based data; the study shall explore the Central Bank of
Nigeria (CBN) Statistical Bulletin of various years, CBN Annual Report and Statement of
Account of various years as well as National Bureau of Statistics (NBS).
3.6 Econometric Software for Analyses
The study utilized STATA 13 and EViews version 8 Econometric packages in all
theregression analyses, and Microsoft Excel 2013 for data processing.
P a g e | 72
CHAPTER FOUR
DATA ANALYSES AND EMPIRICAL RESULTS
4.1 Presentation of Results
As stated in the methodology, three models were adopted in line with the objectives of the
study. The results of the various models estimated for both countries are presented
sequentially in this chapter in order to reinforce the robustness of the impact analysis which is
the main focus of this study. First, the results of the descriptive statistics of all variables,
including the correlation matrix and unit root tests are presented. This informs the levels at
which the variables should be specified in the models. Second, the results of the Engle-
Granger cointegration test are presented in order to determine whether the model should be
specified in dynamic or long run forms. Third, the estimation results of the Seemingly
Unrelated Regression (SUR) models on budget deficits and selected macroeconomic
variables in both economies. Finally, the results and discussions of the models and several
diagnostic tests are presented.
4.1.1 Presentation of Nigerian Data
4.1.1.1 Descriptive Statistics for all Variables
The distribution properties of the variables for the model indication that most of the variables
comply with the expectations, and thus exude some theoretically meaningful results with
fairly Jacque-Bera probability values (see Table 1.1). Budget deficit for example has mean
value of -3.89 USD, a median of -3.62 USD, a Jacque-Bera probability value of 1.34 USD
P a g e | 73
RGDP BD EXDEP GEX INF MS RIR SAV
Mean 307225.7 -3.885483 52.58414 882534.1 19.07782 23.08248 -1.674357 11.81713
Median 269457.8 -3.623293 19.14658 79690.90 13.10000 21.83530 -1.863842 10.33859
Maximum 888893.0 9.543593 293.1000 4605320. 72.80000 37.95685 24.26235 23.24536
Minimum 4219.000 -12.43976 0.000000 903.9000 3.500000 9.316832 -41.9836 4.976588
Std. Dev. 252893.1 4.430560 65.68786 1395947. 16.05893 7.042094 14.04725 4.769530
Skewness 0.732471 0.334074 1.428123 1.592740 1.705071 0.227381 -0.517553 0.610069
Kurtosis 2.611511 3.534856 5.210467 4.169281 5.206694 2.131769 3.474337 2.170724
Jarque-Bera 4.211131 1.342904 23.91455 21.10991 30.24738 1.761162 2.376805 3.990135
Probability 0.121777 0.510966 0.000006 0.000026 0.000000 0.414542 0.304708 0.136004 Sum 13517933 -170.9613 2313.702 38831500 839.4240 1015.629 -73.6717 519.9539
Sum Sq. Dev. 2.75E+12 844.0840 185540.5 8.38E+13 11089.24 2132.416 8484.983 978.1819
Observations 44 44 44 44 44 44 44 44
and small standard deviation of 4.43 USD. Also, budget deficit has a minimum of -12.44
USD and a maximum of 9.54 USD.
Table 1.1: Nigeria: Descriptive Statistics
Source: Computed by the Author with EViews 8. The probability of 0.51 for the deficit indicates that it is fairly normally distributed. Real
GDP was not normally distributed with a mean of 307225.7 USD, a median of 269457.8
USD and standard deviation of 252893.1 USD. Real interest rate was negatively skewed with
value of -0.517553 USD, while inflation was positively skewed with value of 1.705071 USD.
4.1.1.2 Analysis of the Correlation Matrix
The results of the correlation are shown in Table 1.2. Budget deficit for example is negatively
correlated to inflation, money supply, real interest rate, and total savings of the economy. The
relationship as indicated in the results is consistent with economic theory in the case of
inflation, money supply real interest rate, but inconsistent in the case of total saving,
particularly in a Keynesian sense. Also shown in Table 1.2 is the positivecorrelation between
budget deficit and real economic growth rate which is also consistent with the theory.
Nevertheless, it should be noted that descriptive statistics merely show the direction of
relationship and not a causation. The strongest level of correlation (0.93) is between
economic growth (RGDP) and government expenditure (GEX), followed by the money
P a g e | 74
RGDP BD EXDEP GEX INF MS RIR SAVRGDP 1.000000BD 0.057620 1.000000EXDEP 0.626227 0.149143 1.000000GEX 0.926623 0.203655 0.498133 1.000000INF -0.14684 -0.15491 0.366486 -0.26292 1.000000MS 0.447005 -0.45899 0.047661 0.399240 -0.07492 1.000000RIR 0.375023 -0.10872 -0.0784 0.365274 -0.49592 0.372618 1.000000SAV 0.453013 -0.35749 -0.01501 0.465272 -0.11647 0.910811 0.317118 1.000000
RGDP BD EXDEP GEX INF MS RIR SAV Mean 1090.470 -5.746432 0.393973 108.7305 31.44955 22.77792 -13.59491 7.816566 Median 988.9500 -5.615 0.040224 108.2054 20.90000 22.76066 -5.307 6.821440
Maximum 3250.700 -1.39 2.023725 125.2731 122.9700 34.10823 28.43000 24.04829 Minimum -1186 -12.12 0.000102 94.51588 3.030000 11.30499 -107.55 1.258312Std. Dev. 884.8847 3.027300 0.576670 7.892617 29.02337 6.289460 29.63427 4.602046 Skewness 0.232518 -0.171343 1.368614 0.171598 1.956847 -0.049047 -2.031623 1.480500 Kurtosis 3.577239 1.865082 3.724895 1.965661 6.298840 1.966814 6.884664 5.655892 Jarque-Bera 1.007348 2.576699 14.69947 2.177342 48.03213 1.974674 57.93438 29.00569 Probability 0.604306 0.275725 0.000643 0.336664 0.000000 0.372567 0.000000 0.000001 Sum 47980.70 -252.843 17.33483 4784.141 1383.780 1002.229 -598.176 343.9289 Sum Sq. Dev. 33669902 394.0755 14.29955 2678.617 36221.30 1700.964 37762.16 910.6897 Observations 44 44 44 44 44 44 44 44
supply (MS) and total savings (SAV) (0.91) while the weakest level of correlation (-0.02) is
between the exchange rate depreciation (EXDEP) and total savings (SAV). In general, the
results of the correlation matrix would be of information value when we embark on empirical
analysis.
Table 1.2: Nigeria Correlation Matrix
Source: Computed by the Author with EViews 8.
4.1.2 Presentation of Ghanaian Data
4.1.2.1 Descriptive Statistics for all Variables
The descriptive statistics of the variables used in the Ghana study is presented in Table 1.3
Table 1.3: Ghana: Descriptive Statistics
Source: Computed by the Author with EViews 8.
The distribution properties of the variables for the model indicate that most of the variables
are fairly well behaved, but. Budget deficit for example has a mean value of -5.746432 USD,
a median of -5.62 USD and relatively small standard deviation (3.03 USD). The probability
P a g e | 75
RGDP BD EXDEP GEX INF MS RIR SAVRGDP 1.000000BD -0.210081 1.000000EXDEP 0.881085 -0.314326 1.000000GEX 0.699617 -0.142803 0.592907 1.000000INF -0.363742 0.113362 -0.376822 -0.377808 1.000000MS 0.465693 -0.37464 0.686128 0.511035 -0.255105 1.000000RIR 0.344141 -0.189115 0.271489 0.460002 -0.935488 0.171016 1.000000SAV 0.132245 -0.465212 0.298932 -0.249874 -0.210841 0.321991 0.142705 1.000000
of 0.28 for the deficit indicates that it is fairly normally distributed. Real GDP was normally
distributed with a mean 1090.47 USD, a median of 988.95 USD, Jacque-Bera probability
value of 2.58 USDand standard deviation of 884.89 USD. Real interest rate negatively
skewed at -2.03 USD while inflation was positively skewed with value 1.96 USD.
4.1.2.2 Analysis of the Correlation Matrix
Table 1.4 present the correlation matrix of the variables applied in this study. The highest
correlation (-0.94) is between inflation (INF) and real interest rate (RIR). This is consistent
with economic theory. The correlation coefficient of (-0.21) was registered between our
variable of interest budget deficit (BD) and real GDP. This not really a problem as the static
correlation is most times not a true reflection of the relationship between the variables when
dynamic models are specified. The weakest correlation (0.11) is between budget deficit (BD)
and inflation (INF).
Table 1.4: Ghana Correlation Matrix
Source: Computed by the Author with EViews 8.
4.2 Stationary (Unit Root) Tests Results
To examine the time series characteristics of the variables in the models, the Augmented
Dickey-Fuller (ADF) and Phillips-Perron (PP) stationarity (unit root) tests were conducted.
Essentially, both the ADF and PP tests are presented in table 1.5 below:
Table 1.5: Summary of ADF and PP Stationary (Unit Root) Tests for the Variables in
the Models, 1970 – 2013.
P a g e | 76
***Significant at 1%, **Significant at 5%, *Significant at 10% Source: Stationarity test results computed using EViews 8 Note: For both ADF and PP, the 5% critical values are given below the statistics in parentheses. Asterisk (**) shows no unit root at 5% critical value.
NIGERIA
Variables Augmented Dick-Fuller (ADF) Test Phillip-Perron (PP) Test At Level At First
Difference Order of
Cointegration
At Level At First Difference
Order of Cointegration
RIR -7.046315** (-3.518090)
- I(0) -3.518090** - I(0)
GEX - -4.561458* (-3.557759)
I(1) - -3.520787** I(1)
MS - -6.331609** (-3.520787)
I(1) - -3.520787** I(1)
INF - -6.587901** (-3.523623)
I(1) - -3.520787** I(1)
BD - -4.099735** (-3.520787)
I(1) -3.518090** - I(0)
RGDP - -5.238360** (-3.520787)
I(1) - -3.520787** I(1)
EXDEP -3.843817** (-3.518090)
- I(0) -3.518090** - I(0)
SAV - -6.001551** (-3.520787)
I(1) - -3.520787** I(1)
***Significant at 1%, **Significant at 5%, *Significant at 10%
GHANA Variables Augmented Dick-Fuller (ADF) Test Phillip-Perron (PP) Test
At Level At First Difference
Order of Cointegration
At Level At First Difference
Order of Cointegration
RIR - -11.30048** (-3.520787)
I(1) - -4.682621** (-3.518090)
I(1)
GEX -4.307310** (-3.518090)
- I(0) -4.261254** (-3.518090)
- I(0)
MS - -6.053898** (-3.520787)
I(1) - -6.057001** (-3.520787)
I(1)
INF -5.084804** (-3.518090)
- I(0) -5.029343** (-3.518090)
- I(0)
BD - -6.249993** (-3.520787)
I(1) - -7.440796** (-3.520787)
I(1)
RGDP - -9.619379** (-3.520787)
I(1) - -5.169553** (-3.518090)
I(1)
EXDEP - -5.463968** (-3.520787)
I(1) - -5.454011** (-3.520787)
I(1)
SAV -3.704077** (-3.518090)
- I(0) -3.682215** (-3.518090)
- I(0)
P a g e | 77
In Nigeria, the result of the unit root tests (ADF) shows that all the variables with the
exception of real interest rate and exchange rate depreciation failed the unit root test at 5%
level of significance in their level form. All variables, however, passed the test for
stationarity in their first difference. Similar results using Phillip-Perron (PP) test were carried
out, and the result also shows that all the variables with the exception of real interest rate,
budget deficit and exchange rate depreciation failed the unit root test at 5% level of
significance in their level form. All variables, however, passed the test for stationarity in their
first difference. In Ghana, both the ADF and PP test present identical results as all the
variables with the exception of government expenditure, inflation and savings failed the unit
root test at 5% level of significance in their level form. All variables passed the test for
stationarity in their first difference.
4.2.1 Lag Length/Bandwidth Selections
Appropriate lag length/Bandwidth was automatically chosen for the variables in the models
as informed by both Schwarz Information Criterion and Bartlett Kernel (See appendix for the
results).
4.3 Cointegration Tests
Having established the fact that some variables in the models are stationary at level I(0) and
others in first difference I(1), it is necessary to further examine if there exist a likelihood of a
long-run relationship amongst the variables. That is to ascertain if the variables are co-
integrated. Once this is done, it implies that although some of the variables exhibit random
walks, there is a stable long-run relationship amongst them and that the randomness will not
make them to diverge from their equilibrium relationship.
However, to do this, Engle-Granger two-step (EGTS) procedure on the variables that are
integrated to order one, that is I(1). The test involves first regressing these variables and
P a g e | 78
obtaining the residuals. Next, the residuals are tested for unit roots by applying ADF
framework. Once the results show a stationary process, it means that the variables are co-
integrated. Furthermore, this was done to know if the variables in the models have a
sustainable long-run relationship and also to avoid the problem of running a spurious
regression. The result for this test is reported in Table 1.6 below:
Table 1.6: Cointegration (Augmented Engle-Granger) Test Results
NIGERIA Variable ADF Test Statistic Test Critical Value at
5% Conclusion
RESID01 -5.936582 -2.931404 Stationary at level
GHANA
Variable ADF Test Statistic Test Critical Value at 5%
Conclusion
RESID01 -3.825501 -3.518090 Stationary at level Source: Author’s computation using EViews 8 The ADF tests on the residuals at level form confirmed that the calculated ADF statistics in
both Nigeria and Ghana are greater (in absolute terms) than the tabulated critical values at 5%
critical value. Thus, the null hypothesis of non-stationarity of the residuals is rejected, thus
concluding that there exist a stable long run relationship amongst them – though there might
be some deviations in the short-run.
4.4 Analysis of the SUR Models and Two-Stage Least Squares Estimation Results
To capture the three objectives of the study, a Seemingly Unrelated Regression (SUR) was
applied using both the Nigerian and Ghanaian data. The estimation of results for the SUR
model (equation 10, 11 and 12) are presented in Table 1.7 below. The equations represent
formulation of the hypotheses on the effects of budget deficits on (1.) interest rate, (2.)
inflation and (3) economic growth in both economies. However, the SUR estimation results
are presented in tables below:
P a g e | 79
Table 1.7: SUR Estimation Results Source: Author’s estimation using STATA 13.
Note: Asterisk (**) shows statistically significant at 5% level of significance. The t-Statistic is given below; the statistics in parentheses.
NIGERIA GHANA Variable RIR INF InRGDP RIR INF InRGDP GEX 1.91e-06
(1.25) -
- 0.7118853** (3.71)
-
BD -0.214025** (-3.42)
-1.457205** (-2.61)
-0.0161941** (-2.47)
-1.23363** (-2.71)
-1.1113837** (-3.08)
-0.0371651** (-3.57)
InRGDP - -2.236352 (-1.09)
- - -0.0064234 (-1.63)
-
InMS 12.40231 (1.66)
-11.90506 (-1.32)
- -19.1024** (-3.60)
-5.406567 (-0.28)
-
RIR - - 0.026369** (2.36)
-0.0079105** (-2.21)
INF -0.4221472** (-3.98)
-
0.0145521** (-3.38)
-0.9134943** (-19.35)
- -0.0140296** (-2.11)
EXDEP - 0.1373776** (3.37)
- - -5.146021** (-2.28)
-
InSAV -
- 0.5001784 (1.31)
-
- 0.0333115 (0.29)
Constant -34.47242 (-1.62)
69.77756** (2.90)
10.67361** (11.79)
-10.42766 (-0.51)
57.80081** (2.99)
7.013945** (26.68)
R2 0.78 0.85 0.82 0.92 0.74 0.70 Adjusted R2 0.76 0.83 0.81 0.90 0.73 0.69 Observation (Obs) 44 44 40 44 44 40 F-stat. (chi2) 34.31 16.85 12.90 496.03 5.60 15.54 P-Value 0.0000 0.0021 0.0118 0.0000 0.2308 0.0037 DW 1.376 1.431 1.576 1.423 1.631 1.391
P a g e | i
4.4.1 Analysis of the SUR Estimation based on Economic Criteria
The estimation results for the Seemingly Unrelated Regression (SUR) equations are presented in
Table 1.7above. The equations represent formulation of the hypotheses that budget deficits exert
effects on interest rate (first study objective), inflation (second objective), and economic growth
(third objective) in Nigeria and Ghana.
The result obtained from the estimation exercise are fairly robust and satisfactory, such that the
variables in the estimation models conformed largely to a priori expectations in terms of
statistical significance. However, as indicated in the SUR equations results above, some
estimated coefficients are consistent with a priori expectations, while others are not. Focusing
our major interest on our core variable which is budget deficit, it is of great interest to note that
the coefficients of the variables in three equations (as shown in Table 1.7) maintain negative
signs in line with our a priori expectations. This suggests that the relationship between budget
deficit; interest rate, inflation and economic growth from the linear-log and log-log form of the
model(s)in Nigeria and Ghana are respectively negative. The t-statistics, that is, the variables in
parentheses in the Table (1.7) confirm that the coefficient of budget deficit is statistically
significant at 5.0 percent. Thus, we can safely reject the null hypotheses that budget deficits do
not have significant effects on interest rate, inflation and economic growth in Nigeria and Ghana.
Furthermore, the coefficients of budget deficit is negatively related to interest rate (RIR) in the
first model; inflation (INF) in the second model, and economic growth (RGDP) in the third
model, but allare statistically significant. This further suggests that, if budget deficit increases by
one percent, the interest rate, inflation and economic growth will decrease by about 0.21%,
2.61%, and 2.47% in Nigeria respectively. Similarly, in Ghana, if budget deficit increases by one
percent, the interest rate, inflation and economic growth will decrease by about 1.23%, 1.11%,
P a g e | ii
and 0.04% respectively. These results support the neoclassical argument in the literature that
budget deficit slows the growth rate of the economy through resources crowding-out. These
findings suggest that the model variables are robust determinants of real interest rate, inflation,
and economic growth in Nigeria and Ghana due to the fact that all their test statistics are
relatively significant.
4.4.2 Analysis of the SUR Results based on Statistical Criteria
4.4.2.1 The t-Test Results:The t-test shows the significance of each variables in the models. Hypothesis: H0: β = 0 (Parameter estimated is statistically insignificant)
H1: β ≠ 0 (Parameter estimated is statistically significant) Decision Rule: Reject H0 if |tcal| > |ttab| Accept otherwise. α = 5% with (n – k)df i.e. 0.05 (44 – 5)df = 0.05 (39)df |tcal| = 1.95 ≈ 2.0
Conclusion: In Nigeria, all the variables in the Model I of the empirical results are statistically
insignificant (with exception of budget deficit and inflation) at 5% level of significance as their
values are not greater than the |ttab|. Also, in Model I of Ghana, all the variables are statistically
significant. In Model II, all the variables in both Nigeria and Ghana are statically significant with
the exception of Real GDP and money supply, likewise the total savings in Model III.
4.4.2.2 The F-tests: This measures the overall significance of the regression models.
Hypothesis:
H0: β0 = β1 = β2 = β3 = β4 = β5 = 0(The model is statistically insignificant)
H1: β0 ≠ β1 ≠ β2 ≠ β3 ≠ β4 ≠ β5 ≠ 0(The model is statistically significant)
Ββ = 5% with k-1/n-kdf
i.e. 0.05 with 4/39df
Decision Rule:
Reject H0 if |tcal| > |ttab|
P a g e | iii
Accept otherwise
F-cal: the values of F-cal in the three (3) models in both economies are all greater than F-tab (see Table 1.7). Where; F-tab = 2.69 Therefore, since F-cal > F-tab in all the models in Nigeria and Ghana, we reject H0 and conclude that the models are statistically significant at 5% level of significance. 4.4.2.3 The Coefficient of Determinations (R2):
In the Table 1.7, the values of R2 in the three (3) models in Nigeria and Ghana hovers between
0.70 and 0.92. However, the R2explains the extent at which the respective dependent variables
(real interest rate, inflation and economic growth respectively) caused the variations in all the
explanatory variables in the respective models. This further suggests that the explanatory
variables explain about specific percentages of the total variations in real interest rate, inflation
and economic growth in the models in both economies respectively.
4.4.3 Analysis of the SUR Estimation based on Econometric Criteria
4.4.3.1 Autocorrelation
The Durbin-Watson (DW) test statistic was employed to check for serial correlation in the
model. From the empirical results presented in Table 1.7, DW ranges from the minimum 1.376
to maximum of 1.631 in the models of the two economies. n = 44; k = 5 (on the average,
excluding the intercept) and from the DW table; dL = 1.336and dU = 1.720 at 0.05 level.
Hypothesis:
H0: No positive autocorrelation
H0: No negative autocorrelation
H1: There is autocorrelation
Decision Rule:
P a g e | iv
Since dL<1.376 < - ... < 1.631< dU, there is inconclusive evidence regarding the presence or
absence of positive first order serial correlation.
4.4.3.2 Other Diagnostic Tests Table 1.8: Other Diagnostic Tests
Source: Author’s estimation using EViews 8 The result of residuals generated the estimated equation was found to be normally distributed for
both Nigeria and Ghana. No serial correlation and heteroscedasticity was observed in the
equation, implying that the estimates are reliable, and result can be relied on for policy
recommendation and formulation.
4.4.3.3 Functional Form Specification (Ramsey Reset) Test
This test is conducted to check whether the models in this study are correctly specified or not.
Here, the F-distribution was used.
NIGERIA
Test Type Statistic Value Probability Remarks Normality Jarqua
Bera 1.684908 0.430654 Normally distributed
residuals Serial Correlation (LM)
F-statistic 6.913589 0.0029 No serial correlation
Heteroscedasticity ( Harvey )
F-statistic 0.821468 0.6475 No heteroscedasticity
GHANA
Test Type Statistic Value Probability Remarks Normality Jarqua
Bera 2.886401 0.236171 Normally distributed
residuals Serial Correlation (LM)
F-statistic 1.136370 0.3319 No serial correlation
Heteroscedasticity (Harvey)
F-statistic 1.339767 0.2464 No heteroscedasticity
P a g e | v
Table 1.9: Ramsey Reset Specification Test
Source: Author’s computation using EViews 8 Hypothesis:
H0: βi = 0 (the model is wrongly specified)
H1: βi ≠ 0 (the model is correctly specified)
Αt α = 5% level
Decision Rule:
Reject H0 if F-stat > F-tab. (k-1/n-k)df.
Accept otherwise.
Where k = number of parameter; n = number of observations Ftab = (5-1/44-5) = (4, 39). Then, from F-Distribution Table = 2.69. Since F-stat (5.39) > F-tab (2.69) in Model I. F-stat (7.46) > F-tab (2.69) in Model II. F-stat
(2.86) > F-tab (2.69) in Model III at 5% levels, we reject H0 and conclude that the Models are
correctly specified.
4.5 Evaluation of Hypotheses
MODEL I
Variables F-Stat. F-tab. Assessment
Fitted^2 5.39 2.69 Well Specified
MODEL II
Variables F-Stat. F-tab. Assessment
Fitted^2 7.46 2.69 Well Specified
MODEL III
Variables F-Stat. F-tab. Assessment
2.86 2.69 Well Specified
P a g e | vi
The study hypotheses stated in Section 1.5 are evaluated as follow:
H01: Budget deficitshas noeffects on interest rate in Nigeria and Ghana.
Decision:
The t-statistic of the slope of the budget deficit on interest rate is -3.42 and -2.71 in Nigeria and
Ghana respectively, hence statistically significant at 5% (α > 1.95 ≈ 2). We reject the null
hypothesis that budget deficits has no effects on interest rate, thereby concluding that there exists
significant negative effect of budget deficits on interest rate in Nigeria and Ghana.
H02: Budget deficitshas noeffects on inflation in Nigeria andGhana
Decision:
The t-value of the slope of the budget deficit on inflation is -2.61 in Nigeria, and -3.08 in Ghana;
hence statistically significant at 5% (α > 1.95 ≈ 2). We reject the null hypothesis that budget
deficits has no effects on inflation, thereby concluding that there exists significant negative effect
of budget deficits on inflation in Nigeria and Ghana.
H03: Budget deficits has no effects on economic growth in Nigeria and Ghana.
Decision:
The t-value of the slope of the budget deficit on economic growth is -2.47 and -3.57 in Nigeria
and Ghana respectively, hence statistically significant at 5% (α > 1.95 ≈ 2). We reject the null
hypothesis that budget deficit has no effects on economic growth in both economies, thereby
concluding that there exists significant negative effect of budget deficits on interest rate in
Nigeria and Ghana. However, these findings support the neoclassical argument in the literature
that budget deficit slows the growth rate of the economy through resources crowding-out. The
P a g e | vii
policy implication of the findings is that the selected macroeconomic variables are negatively
affected by budget deficits in both economies of Nigeria and Ghana.
P a g e | viii
CHAPTER FIVE
SUMMARY, CONCLUSION, AND POLICY RECOMMENDATIONS
5.1 Summary of Findings
The study was carried out to empirically address three research questions on the fiscal deficits in
Nigeria and Ghana for the period from 1970 to 2013 inclusive. The motivation and justification
behind the study and selection of the economies of Nigeria and Ghana were as a result of:first,
the centrality of role of fiscal imbalances in determining the economic growth and stability in
both economies. Second, the similarities in socio-economic and political structures in both
economies. Hence, the study broadly aimed to test for the effects of budget deficits on interest
rates, inflation and economic growth in Nigeria and Ghana.
After testing for these effects of budget deficits on the selected macroeconomic variables in both
economies, the empirical results have shown that there exists negative and statistically
significant effects of budget deficits on interest rate, inflation and economic growth in Nigeria
and Ghana.
5.2 Policy Implications of Findings The results from the study clearly show that budget deficits in Nigeria and Ghana are responsible
for macroeconomic imbalances. An ever rising inflation and interest rates are not desirable and
they could be instrumental for disaster if they reach unsustainable levels. Inflation and interest
rates are unsustainable if they cannot be controlled and their effects are not consistent with
adequate growth, price stability and the countries’ ability to service fully their external debt
P a g e | ix
obligations, (IMF, 1995). Similarly, high interest rates crowd out the private sector and thus
negatively affect national savings and investment. It is therefore necessary for the governments
of Nigeria and Ghana to reduce the size of the budget deficits to a level that will not affect other
macroeconomic variables through fiscal consolidation and boosting domestic production.
Fiscal Consolidations in Nigeria and Ghana
Fiscal consolidation is a policy aimed at reducing government deficits and debt accumulation.
For Nigeria and Ghana the policy should focus on both short term and long term measures. In the
short term the government should aim at gradually reducing the budget deficit by raising
domestic revenue mobilization. In Nigeria and Ghana, revenue from tax (tax to GDP ratio) has
stagnated between 11 and 13 percent over the years yet government spending have continued to
grow (NBS, 2013; BoG, 2014).
5.3 Policy Recommendations
Financing of the growing expenditure has therefore been through foreign aid and government
borrowing both externally and internally; these as noted, have negative consequences. In order to
mitigate the above consequences, governments should institute actions that increase its revenue
collections. Such actions should aim at increasing economies’ tax revenue collections by
adopting efficient and effective methods of tax collections. Such measures include but are not
limited to the following:
i. Reducing the size of the informal sector which has proved hard to tax as this will improve
the revenue collection in the economies that can be used to finance budget deficits rather
than depending on external borrowing for financing of fiscal deficits. Similarly, reducing
the number of unproductive tax exemptions as well as combating tax evasion in the
economies will equally improve the revenue base of the governments.
P a g e | x
ii. Interest rates should be further reduced through appropriate policy and macroeconomic
environments to ensure credit availability and accessibility to the private sectors in both
economies. Furthermore, there is need to support growth in the real sectors of the
economy by encouraging investors to have access to investible funds from banks through
lowering of interest rate.
iii. Appropriate policy to ensure high levels of financial deepening should be pursued in both
economies as there exist the positive nexus between the level of financial deepening and
economic growth in both economies.
iv. Exchange rate depreciation should be discouraged in both economies as it has negative
consequences on growth in Nigeria and Ghana.
v. Appropriate policies should be pursed with vigour by the monetary authorities in both
economies to curtail inflation through suitable money supply. However, the regional
blocks at which the economies of Nigeria and Ghana belong should be mindful of
uniformed policy adoption across country members as this has implication on inflation.
vi. On the expenditure side, government should reduce its overall recurrent expenditure bill,
this could be done by revising the administrative structures created under its
decentralization plan. Also, governments of both economies should maintain the number
of states and local autonomies it had created as rapid growth in the number of districts,
states and local autonomies have contributed to increase in administrative costs which
worsens the budget imbalances.
vii. The economies should pursue policies that will boost production of goods for both
domestic consumption and exports in the long run through a combination of import
P a g e | xi
substitution and export promotion strategies as this will have a positive improvement on
the countries exchange rate, reduce inflation and thus lead to growth.
viii. Finally, the government of Nigeria and Ghana should be mindful of the sources of
financing the budget deficits so as to effectively manage the economic fluctuations and
increase activities in the real sector. Similarly, there is need to entrench fiscal discipline
in government operations at all levels that will ensure management of public finances,
improve budgetary processes, including openness in the budget preparation, execution
and reporting is been advocated.
5.4 Conclusion
The review of the study has shown that while vast growing volumes of research were being
carried out in the developed economies, little attention has been paid to the issue of how the
fiscal deficits affect interest rate, inflation, and economic growth in both economies. Based on
this empirical analysis, appropriate policies can then be drawn given insight to how budget
deficit can perform its roles without necessarily leading to inflation.
In order to achieve high and sustained long-run economic growth when budget deficit is used as
fiscal policy instrument, then, monetary policy, industrial policy and commercial policy must be
strengthened to act as checks and balances in Nigeria and Ghana. Relevant measures to enhance
policy coordination among various arms of government should be put in place. Most especially,
monetary policy should be made to complement fiscal policy measures. Also, fiscal discipline
should be strongly adhered to at every level of government. Furthermore, since inflation has been
established as monetary phenomenon in both Nigerian and Ghanaian economies; for budget
deficit to be effective, some fundamental changes in the productive base of the economy need be
made.
P a g e | xii
REFERENCES
Abel, A. B., & Bernanke, B. S. (2001). Macroeconomics (4thed.). New York: Addison Wesley
Longman, Inc.
Ackah, C. G, Aryeetey, E. B, & Aryeetey, E. (2009). The Impact of Global Economic Crisis on
Ghana. Labour Institute and Labour Research, Discussion Series Paper 5. Retrieved
from: http://www.ghanatuc.org/The-Effects-of-the-Global-Economic-and-Financial-
Crisis-on-the-Ghanaian-Economy-and-Labour-Market.pdf.
African Economic Outlook (2014). Ghana Overview 2014. Retrieved from:
http://www.africaneconomicoutlook.org/fileadmin/uploads/aeo/2014/PDF/CN_Long_EN
/Ghana_ENG.pdf.
African Economic Outlook (AEO) (2015). Regional Development and Spatial Inclusion.
Retrieved from:
http://www.africaneconomicoutlook.org/fileadmin/uploads/aeo/2015/PDF_Chapters/Over
view_AEO2015_EN-web.pdf
African Economic Outlook (AEO) (2013). Structural Transformation of Natural Resources.
Retrieved
from:http://www.undp.org/content/dam/rba/docs/Reports/African%20Economic%20Outl
ook%202013%20En.pdf
P a g e | xiii
Aisen, A., & Hauner, D. (2008). Budget Deficits and Interest Rates: A Fresh Perspective, IMF
Working Paper, Fiscal Affairs Department, International Monetary Fund, WP/08/42.
Akinboade, O. A. (2004). The Relationship between Budget Deficit and Interest Rates in South
Africa: Some Econometric Results. Development Southern Africa, Volume 21, Issue 2.
Akosah, N. K. (2013). Threshold Effect of Budget Deficits on Economic Growth in Ghana: An
Empirical Analysis. Bank of Ghana working Paper. 1-24
Andrew, M. (1991). Negotiating theSingleEuropeanAct: National Interestsand Conventional
Statecraft inthe European Community. World Peace Foundation and the Massachusetts
Institute of Technology (International Organisation), 45(1), 1 – 38.
Angrist, J. D., & Imbens, G. W. (1995). Two-stage least squares estimation of average causal
effects in models with variable treatment intensity. Journal of the American Statistical
Association, 90(430), 431-442.
Antwi, S. Z. X. (2009). Impact of Financial Reforms on the Banking System in Ghana.
International Business & Economics Research Journal, 8(3), 38 – 52.
Antwi, S. Z. X., & Mills, E. F. E. (2013). Consequential Effects of Budget Deficit on Economic
Growth: Empirical Evidence from Ghana. International Journal of Economics and
Finance, 5(3), 28 – 43.
Anyanwu, J. C. (1997). “Fiscal Relations among the Various Tiers of Government in Nigeria”,
Fiscal Federalism and Nigeria’s Economic development: Ibadan. The Nigerian
Economic Society.
Anyanwu, J. C. (1997). Nigerian Public Finance (1sted.). Onitsha: Joane Publishers.
Arestis, P. & Sawyer, M. (2006). The Inter-temporal Budget Constraint and the Sustainability of
Budget Deficits. Leeds: University of Leeds.
P a g e | xiv
Ariyo, A. & Raheen, M. I. (1991). Effects of Fiscal Deficits on Some Macroeconomic
Aggregates in Nigeria cited in Tchokote (2004). Macroeconomics of Fiscal Deficits in
Cameroun, Doctoral Dissertation, University of Ibadan.
Awe, A. A & Funlayo A. K. (2014). The Short and Long-Run Implications of Budget Deficit on
Economic Growth in Nigeria. Canadian Social Science, 10(5), 201-205.
Bakare, I.A.O., Adesanya, O.A and Bolarinwa, S.A. (2014). Empirical investigation between
budget deficit, inflation and money supply in Nigeria. European Journal of Business and
Social Sciences, 2(12), 120-134.
Baltagi, B. H., (2008). Econometric Analysis of Panel Data (4thed.). USA: John Wiley & Sons.
Bank of Ghana (2005).The HIPC Initiative and External Debt: An Empirical Assessment and
Policy Challenges. Accra, Bank of Ghana Research Department. 2-4.
Bank of Ghana (2007). Bank of Ghana Monetary Policy Report. Vol. 1, No. 6/2007. Accra, Bank
of Ghana Research Department. 4.
Bank of Ghana. (2007). Bank of Ghana Monetary Policy Report. Vol. 1, No. 6/2007. Accra,
Bank of Ghana Research Department. 2.
Barro, R. J. (1989). Rational-Expectations. National Review, 41(2).
Barro, R. J. (1990). Government Spending in A Simple Model of Endogenous
Growth, Journal of Political Economy, 98(3).
Barro, R. J. (1991). Economic Growth in A Cross Section of Countries, Quarterly
Journal of Economics, 106(6).
Barua, S. (2005). “An Examination of Revenue and Expenditure Causality in Bangladesh: 1974-
2004” World Bank Working Paper Series WP 0605 Dhaka, Bangladesh Bank Policy
Analysis Unit (PAU).
P a g e | xv
Bebi, H. (2000). Sustainability of Budget Deficits–A Case Study of Namibia (1991- 1999) Nepru
Working Paper No. 73, Windhoek, Namibian Economic Policy Research Unit.
Bernheim, B. D. (1989). A Neoclassical Perspective on Budget Deficits. Journal
of Economic Perspective, 3, 55 – 72.
Blanchard, O., & Johnson, D. R. (2012). Macroeconomics (6thed.). New York: Pearson
Publication, Inc.
Bonga-Bonga, L (2011). “Budget deficit and long-term interest rates in South Africa”.
University of Johannesburg, Department of Economics and Econometrics, Auckland
Park, 2006, South Africa.
Brownbridge, M. & Tumusiime-Mutebile, E. (2007). Aid and Fiscal Deficits: Lessons from
Uganda on the Implications for Macroeconomic Management and Fiscal Sustainability.
Development Policy Review, 25(2), 193 – 213.
Bullock, J. G.,Green, D. P., & Ha, S. E. (2010). Yes, But What’s the Mechanism? (Don’t Expect
an Easy Answer). Journal of Personality and Social Psychology98 (4): 550–558.
Buscemi, A. & Yallwe, A. H. (2012). Fiscal Deficit, National Saving and Sustainability of
Economic Growth in Emerging Economies: A Dynamic GMM Panel Data Approach.
International Journal of Economics and Financial Issues, 2(2), 2146 – 4138.
Catão, L. A.V. & Terrones M. E, (2005). Fiscal Deficits and Inflation. Journal of Monetary
Economics, 52(7), 529 – 554.
Darrat, A. F. (1985). Inflation and Federal Budget Deficits: Some Empirical Results. Public
Finance Quarterly, 13(8), 206-215.
Darrat, A. F. (2000). Are Budget Deficits Inflationary? A Reconsideration of the Evidence.
Applied Economics, 7(10), 633-36.
P a g e | xvi
Daumont, R., Le Gall, F., & Leroux, F. (2004). Banking in Sub-Saharan Africa: What went
wrong? IMF Working Paper. WP/04/55.
Dernberg, T.F. (1988). Macroeconomics Concepts Theories and Policies. USA: McGraw Hill
Publishing Company.
Dickey, D. A. & Fuller, W. A. (1979). Distribution of the Estimators for Autoregressive Time
Series with a Unit Root. Journal of the American Statistical Association, 74(5), 427 –
431.
Dockery, E., & Ezeabasili, V. N. (2012). The Relationship between Fiscal Deficits and Inflation:
Econometric Evidence for Nigeria. Economics and Finance Review, 2(7), 17 – 30.
Dwyer, G. P. (1982). Inflation and Government Deficits. Economic Inquiry,XX(3), 315 – 329.
Easterly, W. & Schmidt-Hebbel, K. (1991). The Macroeconomics of Public Sector Deficits: A
Synthesis. The World Bank Working Paper Series. No. 775.
Easterly, W., Schmidt-Hebbel K. (1993). Fiscal Deficits and Macroeconomic Performance in
Developing Countries. World Bank Research Observer, 8(2) 211 – 237.
Eisner, E. J. (1989). Budget Deficits: Rhetoric and Reality. Journal of Economic
Perspectives, 3(7), 23 – 43.
Engle, R. F., & Granger, C. W. J. (1987). Cointegration and Error Correction: Representation,
Estimation and Testing. Econometric, 55 (2), 251-276.
Ezeabasili V. N., Mojekwu, J. N., & Herbert, W. E. (2012). An Empirical Analysis of Fiscal
Deficits and Inflation in Nigeria. International Business and Management, 4(1), 105 –
120.
Ezeabasili, V. N., & Mojekwu, J. N. (2011). Analysis of Fiscal Deficits and Interest Rates in
Nigeria. Journal of Economics and International Finance, 3(4), 236 – 245.
Gebhard, K., & Silika, P. (2006). Sustainability of Swiss Fiscal Policy. CESIFO Working Paper
1689(1), Switzerland, University of St. Gallen. Retrieved from http://www.cesifo-
P a g e | xvii
group.de/portal/page/portal/DocBase_Content/WP/WP-CESifo_Working_Papers/wp-ces
ifo-2006/wp-cesifo-2006-03/cesifo1_wp1689.pdf
Georgantopoulos, A.G., & Tsamis, A. D. (2011). The Macroeconomic Effects of Budget Deficits
in Greece: A VAR-VECM Approach. International Research Journal of Finance and
Economics, 79(2), 78 – 84.
Guess, G., & Koford, K. (1984). Inflation, Recession and Federal Budget Deficit
(or Blaming Economic Problems on A Statistical Mirage). Policy Sciences, 17(4), 21 – 37
Gujarati, D. N (2007). Basic Econometrics (4thed.). India: Tata McGraw –Hill.
Gupta, R. & Uwilingiye, J. (2009). Measuring the welfare cost of inflation in South Africa: a
reconsideration. South African Journal of Economic and Management Sciences, 12(2),
137 – 146.
Heng, T. K. (1997). Public Capital and Crowding In. The Singapore Economic
Review, 42(2), 32 – 42
Hyman, D. N. (1994). Economics (3rdEd.). Massachusetts: Richard D. Irwin Inc.
Institute of Economic Affairs (2008). Annual 2007 Economic Review and Outlook, Accra, 23 –
29.
Institute of Statistical, Social and Economic Research. (2007). State of the Ghanaian Economy in
2006. Accra-Legon: Wilco Publicity Services Ltd.
Johansen, S. (1991). Estimation and Hypothesis Testing of Cointegrating Vectors in Gaussian
vector Autoregressive Models. Econometrica, 59(3), 39 – 46.
Keating, J.W. (2000). Macroeconomic Modelling with Asymmetric Vector Autoregression.
Journal of Macroeconomics, 21(1).
P a g e | xviii
Keho, Y. (2010), Budget Deficits and Economic Growth: Causality Evidence and Policy
Implications for WAEMU Countries. European Journal of Economics, Finance and
Administrative Sciences, 18.
Larbi, D. A. (2012). The Long Term Impact of Budget Deficits on Economic Growth in
Ghana.Journal of Business Research, 6(2), 22 – 34.
Lin, H.Y., & Chu, H. P. (2013). Are fiscal deficits inflationary? Journal of International Money
and Finance, 32(3), 214 – 233.
Luporini, V. (1999). Sustainability of the Brazilian Fiscal Policy and Central Bank
Independence. Cedeplar Belo Horizonte, the Federal University of Minas Gerais.
Lwanga, M. M & Mawejje, J. (2014). Macroeconomic effects of Budget Deficits in Uganda: A
VAR-VECM Approach. Economic Policy Research Centre. Research Series No. 117.
MacKinnon, J. G. (1996). Numerical Distribution Functions for Unit Root and Cointegration
Test. Journal of Applied Econometrics, 11(3) 601 – 618.
Makochekanwa, A. (2008). The Impact of Budget Deficit on Inflation in Zimbabwe. University
of Pretoria, South Africa.
Mankiw, G. N. (2003). Macroeconomics (5thed.). New York: Worth Publishers Inc.
Mpia, M. N. & Ogrike J. (2014). Budget Deficits and Interest Rates in Nigeria: The VAR
Approach. African Social and Educational Journal, 3(1).
Mugume, A., & Obwona, M. (1998). Public Sector Deficit and Macroeconomic Performance in
Uganda”, Economic Policy Research Centre, EPRC Research Series No. 7.
Mukhtar, T., & Zakaria, M. (2008). Budget Deficits and Interest Rates: An Empirical Analysis
for Pakistan. Journal of Economic Cooperation. 29(2), 1 – 14.
P a g e | xix
Ndulu, B. & Connell, S. (2008). ‘Policy Plus: African Growth Performance’, in B. Ndulu, S.
O’Connell, R. Bates, P. Collier and C. Soludo (Ed) The Political Economy of Economic
Growth in Africa 1960 - 2000, Vol. 1. Cambridge, UK: CUP.
National Bureau of Statistics (NBS) (2013). Retrieved from:
http://www.nsb.gov.sc/statistics/national-accounts.
Newey, W. K., & West, K. D. (1994). Automatic Lag Selection in Covariance Matrix
Estimation. Review of Economic Studies, 61, 631 – 653.
Noula, A. G (2012). Fiscal Deficit and Nominal Interest Rate Determination in Cameroon: An
Application of the Loanable Funds Model. Global Advanced Research Journal of
Management and Business Studies 1(1), 006 – 029.
Obioma, E. C. (1998). Balance of Payments and Exchange Rate Management in Nigeria:
Empirical Evidence from the Monetary Approach. NCEMA Policy Analysis Series. 4 (2).
Odhiambo S. O., Momanyi G., Othuon L., & Aila F. O. (2013). The Relationship between Fiscal
Deficits and Economic Growth in Kenya: An Empirical Investigation. Greener Journal of
Social Sciences, 3 (6), 306 – 323.
Odionye, J. C & Uma, K. E. (2013). The Relationship between Budget Deficit and Interest Rate:
Evidence from Nigeria. European Journal of Business and Social Sciences, 2(1), 158 –
166.
Ogiogio, G. O. (1996). Exchange Rate Policy, Trade Policy Reform and the Balance of
Payments”. In Obadan, M. I. And M. A. Iyoha (Ed.). Macroeconomic Policy Analysis:
Tools, Techniques and Applications to Nigeria. National Center for Economic
Management and Administration.
Oladipo, S. O., & Akinbobola, T. O. (2011). Budget deficit and inflation in Nigeria: A Causal
Relationship. Journal of Emerging Trends in Economics and Management Sciences, 2(1),
1 – 8.
P a g e | xx
Omoka, P. C. & Oruka, L. I. (2010). Budget Deficits, Money Supply and Inflation in Nigeria.
European Journal of Economics, Finance and Administrative Sciences. 19(3).
Onuorah, A. C., & Ogomegbunam, O. A. (2013). Relationship between Macroeconomic
Variables and Budget Deficit. International Journal of Management Sciences, 1(10), 416
– 426.
Onwuoduokit, E (2005). Fiscal Deficit, Inflation and output in Nigeria: A Vector Error
Correction Model Approach. Journal of Economic and Financial Studies, 2 (1), 34 – 54
Orji, A., Akachukwu, S. U., & Ilori, E. A. (2014). Foreign Capital Inflows and Growth: An
Empirical Analysis of WAMZ Experience Using a Sure Model. International Journal of
Economics and Financial Issues, 4(4), 976 – 983.
Osuka, B. O.m & Achinihu, J. C. (2014). The Impact of Budget Deficits on Macro-Economic
Variables in the Nigerian Economy. International Journal for Innovation Education and
Research. 2(11), 164.
Oyeleke, J. O., & Adebisi, D. G. (2014). Econometric Analysis of Fiscal Deficit Sustainability of
Ghana. Journal of Economics and Sustainable Development, 5(28), 34 – 40.
Ozurumba, B. A. (2012). Fiscal Deficits and Inflation in Nigeria: The Causality Approach.
International Journal of Scientific & Technology Research, 1(8).
Phillips, P. C., & Perron, P. (1988). Testing for a Unit Root in Time Series Regression.
Biometrica, 75(2), 335 – 346.
Pomeyie, P. (2001). Macroeconomics: An Introductory Textbook. Accra: Wade Laurel co.
Sachs, J. D., & Larrain, F. B. (1993). Macroeconomics in the Global Economy. New Jersey:
Prentice-Hall Inc.
Saleh, A. S. (2003). Budget Deficit and Economic Performance: A Survey. University of
Wollongong, Economic Working Paper Series.
P a g e | xxi
Sowa, N. K. (1994). Fiscal Deficits, Output Growth and Inflation Targets in Ghana. World
Development, 22(3), 1105 – 1117.
State of the Ghanaian Economy (2009). Institute of Social and Statistical Research Report.
University of Ghana, Legon.
Turnovsky, S. J. (2000). Methods of Macroeconomic Dynamics (2nded.). Massachusetts: MIT
Press.
Udu, E. & Agu, G.A. (2000). Economics. Onitsha: African Fep Publishers Limited Book House
Trust.
Vamvoukas, G. A. (1998). Short and Long-Run Effects of Budget Deficits on Interest Rates.
Journal of Economic Literature, 50(1-2).
Vuyyuri, S., & Sesahiah, S. V., (2004). Budget Deficits and other Macroeconomic variables in
India. Applied Econometrics and International Development. AEEADE. 4-1(37).
Wetzel, D. L., & Roumeen, I. (1991). The Macroeconomics of Public Sector Deficits: The Case
of Ghana. Washington DC, World Bank. Retrieved from http://www-
wds.worldbank.org/servlet/WDSContentServer/WDSP/IB/1991/05/01/000009265_39610
01075 213/Rendered/PDF/multi_page.pdf
Wooldridge, J. M. (2006). Introductory Econometrics: A Modern Approach (5th Ed.). Retrieved
from: http://www.cengage.com
Wosowei, E. (2013). Fiscal Deficits and Macroeconomic Aggregates in Nigeria. Arabian
Journal of Business and Management Review, 2(9).
Xiomara, A. & Greenidge, K. (2003). Debt and Fiscal Sustainability in Barbados. Bridgetown,
Central Bank of Barbados.
P a g e | xxii
Yaya, K. (2010). Budget Deficit and Economic Growth: Causality Evidence and Policy
Implications for WAEMU Countries. European Journal of Economic, Finance and
Administration Sciences, Issue 18.
Yellen, J. L. (1989). Symposium on the Budget Deficit. Journal of Economic
Perspectives, 3(2).
APPENDIX DATA FOR THE REGRESSION (NIGERIA)
NIGERIA
YEAR BD RIR INF RGDP GEX MS EXDEP SAV 1970 -8.62 -29.2695 13.8 4,219.00 903.90 14.95 0.1 7.8 1971 2.58 5.576789 16 4,715.50 997.20 14.61 0.22 7.0 1972 -0.82 3.991658 3.5 4,892.80 1,463.60 14.69 0.2 7.9 1973 1.92 1.569258 5.4 5,310.00 1,529.20 14.67 0 8.4 1974 9.54 -25.6668 12.7 15,919.70 2,740.60 9.32 0.1 6.0
P a g e | xxiii
1975 -1.99 -13.9682 33.9 27,172.02 5,942.60 14.12 0.24 8.5 1976 -4.09 -6.86748 24.3 29,146.51 7,856.70 16.92 0.28 8.5 1977 -2.48 -4.2576 13.8 31,520.34 8,823.80 19.50 0.32 8.2 1978 -8.17 -6.28957 21.7 29,212.35 8,000.00 21.40 0.32 8.7 1979 3.48 -3.31985 11.7 29,947.99 7,406.70 21.88 0.32 9.9 1980 -3.98 -3.54742 10 31,546.76 14,968.50 23.89 0.28 11.6 1981 -8.19 -8.05542 20.8 205,222.06 11,413.70 30.4 0.34 13.78 1982 -12.44 5.355617 7.7 199,685.25 11,923.20 32.2 0.38 15.31 1983 -6.34 -3.58051 23.2 185,598.14 9,636.50 33.3 0.49 17.78 1984 -4.46 -5.4496 39.6 183,562.95 9,927.60 33.7 0.7 18.43 1985 -4.48 5.009445 5.5 201,036.27 13,041.10 32.8 0.83 18.44 1986 -11.94 10.6672 5.4 205,971.44 16,223.70 34.4 1.94 20.15 1987 -5.60 -24.2547 10.2 204,806.54 22,018.70 26.2 4.11 17.75 1988 -8.74 -3.05004 34.5 219,875.63 27,749.50 27.6 6.95 16.72 1989 -6.98 -16.3882 50.5 236,729.58 41,028.30 21.2 16.26 10.98 1990 -8.27 10.00194 7.4 267,549.99 60,268.20 19.8 18.03 11.08 1991 -11.45 7.590991 13 265,379.14 66,584.40 24.2 24.04 12.09 1992 -7.42 -23.9307 44.6 271,365.52 92,797.40 20.9 58.1 10.35 1993 -9.53 4.485998 57.2 274,833.29 191,228.90 24.2 114.7 12.43 1994 -7.81 -8.7724 57 275,450.56 160,893.20 25.6 174.4 12.05 1995 0.05 -41.9836 72.8 281,407.40 248,768.10 15.0 293.1 5.61 1996 1.19 -10.3085 29.3 293,745.38 337,217.60 12.8 19.07426 4.98 1997 -0.18 16.49756 8.5 302,022.48 428,215.20 14.7 19.2189 6.34 1998 -4.92 24.26235 10 310,890.05 487,113.40 18.0 19.87715 7.39 1999 -8.93 3.40836 6.6 312,183.48 947,690.00 19.7 53.7556 8.69 2000 -2.26 -10.25 6.9 329,178.74 701,059.40 19.2 58.24839 8.41 2001 -4.68 22.28285 18.9 356,994.26 1,018,025.60 26.9 70.58229 10.33 2002 -4.36 -12.7334 15.9 433,203.51 1,018,155.80 21.8 85.13346 8.57 2003 -2.39 8.560264 13.2 477,532.98 1,225,965.90 23.0 106.6806 7.73 2004 -1.51 -1.28127 15.8 527,576.04 1,426,200.00 18.7 126.6871 6.99 2005 -1.11 -1.51329 11.4 561,931.39 1,822,100.00 18.1 143.7826 9.04 2006 -0.55 -2.21439 8.5 595,821.61 1,938,002.50 20.5 148.3301 9.37 2007 -0.57 11.76407 6.566 634,251.14 2,450,896.70 24.8 155.7536 13.04 2008 -0.20 4.190484 15.148 672,202.55 3,240,820.00 33.0 93.64 16.95 2009 -3.27 23.7065 13.935 718,977.33 3,452,990.80 38.0 98.77 23.25 2010 -3.25 -7.23133 11.742 775,525.70 4,194,217.88 32.5 92.31 17.52 2011 -3.10 12.41647 10.333 834,000.83 4,299,155.10 32.4 89.91408 17.40 2012 -2.41 14.87001 11 888,893.00 4,605,319.72 27.10 87.3225 19.88 2013 -2.26 4.304165 9.5 800,925.60 4,194,218.88 27.30 127.87143 18.69 SOURCE: CBN Bulletin of various years, IMF, WDI 2014 DATA FOR THE REGRESSION (GHANA)
P a g e | xxiv
GHANA
YEAR BD RIR INF RGDP GEX MS EXDEP SAV 1970 -7.09 3.75 3.03 676 101.3677 18.90541 0.000102 12.78272
1971 -6.92 3 8.82 193 104.4751 18.9822 0.000103 9.642072
1972 -6.49 0.2 10.81 -292 94.51588 23.71457 0.000133 12.58791
1973 -4.18 -9.6 17.07 285 94.94459 22.66394 0.000116 14.08089
1974 -4.58 -11.3 18.75 1012 103.4763 21.5667 0.000115 9.57061
1975 -9.49 -19.8 29.82 -1186 99.0706 26.23718 0.000115 13.65512
1976 -9.34 -45.4 55.41 -354 100.3386 29.11235 0.000115 8.545556
1977 -5.64 -106.5 116.52 228 101.0561 27.28246 0.000115 10.0077
1978 -1.48 -57.85 73.09 848 101.3285 24.45824 0.000176 4.046489
1979 -1.42 -39.25 54.52 -251 99.92913 21.05489 0.000275 6.608319
1980 -1.39 -34.85 50.15 612.5 100.6884 18.55351 0.000275 4.935592
1981 -7.05 -101.25 117.86 644.2 100.5673 16.5651 0.000275 4.005453
1982 -10.79 -7.85 21.31 611.7 99.64373 17.1649 0.000275 3.733907
1983 -3.93 -107.55 122.97 590.6 100.4331 11.30499 0.000883 3.316707
1984 -4.57 -20.66 40 639.6 102.7266 11.81331 0.003597 4.150391
1985 -2.18 8.08 9.96 677 102.935 13.6185 0.005434 6.635012
1986 -2.36 -6.1 24.8 706.2 103.5596 13.51454 0.008916 5.802418
1987 -2.82 -18.26 39.75 747.6 106.5229 14.20509 0.015365 3.911053
1988 -1.75 -10.36 31.38 817.8 105.8787 14.74606 0.020224 5.417341
1989 -2.02 -2.05 25.26 855.4 107.6006 16.91699 0.026985 5.608394
1990 -1.5 -12.45 35.9 884 108.9724 14.14209 0.032616 5.471648
1991 -3.46 7.91 10.3 928.5 108.5613 15.56278 0.036763 7.317731
1992 -2.36 12.56 13.3 966.4 111.5417 20.52587 0.043685 1.258312
1993 -7.06 2.07 27.7 1011.5 116.1613 19.83546 0.064871 6.048895
1994 -8.81 1.93 34.2 1046.6 111.5039 22.51313 0.095568 12.45386
1995 -8.54 -27.13 70.8 1089 108.4302 21.64299 0.119914 11.5912
1996 -8.44 -9.1 32.7 1138.4 107.9806 20.59611 0.163547 13.21942
1997 -10.17 11.08 20.5 1197.9 120.5813 23.84196 0.204796 4.224961
1998 -9.85 28.43 15.7 1259.1 112.8568 22.85738 0.231166 10.25255
1999 -8.2 17.8 13.8 1318.1 117.5484 24.09283 0.266643 3.45211
2000 -8.54 4.6 40.5 1373.3 118.4439 28.16617 0.544919 5.554686
2001 -6.68 -5.227 21.3 1435 119.5798 31.44533 0.716305 7.0196
2002 -5.18 -9.37 17 1501.8 112.2567 34.10823 0.792417 7.443261
2003 -4.38 -7.01 31.3 1578.6 115.9291 31.04593 0.866764 7.007868
2004 -3.35 -8.443 16.4 1662.5 121.0637 32.72259 0.899495 7.313823
2005 -3.03 -5.387 13.9 1762.6 125.2731 32.11003 0.906279 3.72906
2006 -2.84 -4.707 10.9 1870.5 115.5376 23.26474 0.916452 6.098369
2007 -4.71 -2.84 12.7 1991.4 116.3044 25.71675 0.935248 3.803148
2008 -5.59 -3.026 18.1 2159.2 119.4555 27.46146 1.057858 1.996779
2009 -8.48 -3.349 16 2245.4 113.0106 28.24624 1.4088 7.659583
P a g e | xxv
2010 -5.82 -4.378 8.58 2425.2 116.4243 29.61919 1.421667 9.277813
2011 -8.25 -5.179 8.58 2789.1 105.5796 30.54902 1.659917 24.04829
2012 -12.12 8.08 8.84 3034.3 110.375 31.27799 1.841821 20.51003
2013 -9.993 -1.44 13.5 3250.7 119.7118 28.50346 2.023725 8.13225
SOURCE: Bank of Ghana Statistical Bulletin of various years, IMF, WEO, and WDI 2014 NIGERIA AUGMENTED DICKEY-FULLER (ADF) STATIONARITY TEST RES ULTS AS GENERATED FROM EVIEWS 8. AT LEVEL Null Hypothesis: BD has a unit root Exogenous: Constant, Linear Trend Lag Length: 0 (Automatic - based on SIC, maxlag=9)
t-Statistic Prob.* Augmented Dickey-Fuller test statistic -4.099735 0.0125
Test critical values: 1% level -4.186481 5% level -3.518090 10% level -3.189732 *MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(BD) Method: Least Squares Date: 08/26/15 Time: 23:14 Sample (adjusted): 1971 2013 Included observations: 43 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. BD(-1) -0.579475 0.141345 -4.099735 0.0002
C -1.964088 1.394512 -1.408440 0.1667 @TREND("1970") -0.007341 0.050383 -0.145712 0.8849
R-squared 0.296392 Mean dependent var 0.147849
Adjusted R-squared 0.261211 S.D. dependent var 4.769469 S.E. of regression 4.099494 Akaike info criterion 5.726818 Sum squared resid 672.2339 Schwarz criterion 5.849693 Log likelihood -120.1266 Hannan-Quinn criter. 5.772130 F-statistic 8.424899 Durbin-Watson stat 1.878110 Prob(F-statistic) 0.000884
AT LEVEL Null Hypothesis: EXDEP has a unit root Exogenous: Constant, Linear Trend Lag Length: 0 (Automatic - based on SIC, maxlag=9)
t-Statistic Prob.*
P a g e | xxvi
Augmented Dickey-Fuller test statistic -3.843817 0.0235 Test critical values: 1% level -4.186481
5% level -3.518090 10% level -3.189732 *MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(EXDEP) Method: Least Squares Date: 08/26/15 Time: 23:15 Sample (adjusted): 1971 2013 Included observations: 43 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. EXDEP(-1) -0.536569 0.139593 -3.843817 0.0004
C -12.75599 14.08480 -0.905656 0.3705 @TREND("1970") 1.954681 0.727263 2.687722 0.0104
R-squared 0.270003 Mean dependent var 2.971429
Adjusted R-squared 0.233503 S.D. dependent var 50.05131 S.E. of regression 43.81984 Akaike info criterion 10.46526 Sum squared resid 76807.13 Schwarz criterion 10.58814 Log likelihood -222.0032 Hannan-Quinn criter. 10.51058 F-statistic 7.397359 Durbin-Watson stat 1.894760 Prob(F-statistic) 0.001847
AT 2ND DIFFERNCE Null Hypothesis: D(GEX,2) has a unit root Exogenous: Constant, Linear Trend Lag Length: 9 (Automatic - based on SIC, maxlag=9)
t-Statistic Prob.* Augmented Dickey-Fuller test statistic -4.561458 0.0050
Test critical values: 1% level -4.273277 5% level -3.557759 10% level -3.212361 *MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(GEX,3) Method: Least Squares Date: 08/26/15 Time: 23:19 Sample (adjusted): 1982 2013 Included observations: 32 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. D(GEX(-1),2) -16.83650 3.691036 -4.561458 0.0002
D(GEX(-1),3) 14.47944 3.500726 4.136126 0.0005 D(GEX(-2),3) 13.73803 3.176826 4.324453 0.0003 D(GEX(-3),3) 13.31895 2.960099 4.499494 0.0002
P a g e | xxvii
D(GEX(-4),3) 12.58975 2.759233 4.562772 0.0002 D(GEX(-5),3) 10.88162 2.480687 4.386533 0.0003 D(GEX(-6),3) 8.898648 2.138889 4.160407 0.0005 D(GEX(-7),3) 6.463939 1.740209 3.714462 0.0014 D(GEX(-8),3) 3.241325 1.189286 2.725437 0.0130 D(GEX(-9),3) 0.969074 0.462683 2.094468 0.0492
C -260759.7 102886.0 -2.534452 0.0197 @TREND("1970") 15286.39 4870.854 3.138340 0.0052
R-squared 0.971442 Mean dependent var -22067.15
Adjusted R-squared 0.955735 S.D. dependent var 563222.3 S.E. of regression 118497.3 Akaike info criterion 26.48316 Sum squared resid 2.81E+11 Schwarz criterion 27.03281 Log likelihood -411.7306 Hannan-Quinn criter. 26.66536 F-statistic 61.84846 Durbin-Watson stat 1.883371 Prob(F-statistic) 0.000000
1ST Difference Null Hypothesis: D(INF) has a unit root Exogenous: Constant, Linear Trend Lag Length: 1 (Automatic - based on SIC, maxlag=9)
t-Statistic Prob.* Augmented Dickey-Fuller test statistic -6.587901 0.0000
Test critical values: 1% level -4.198503 5% level -3.523623 10% level -3.192902 *MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(INF,2) Method: Least Squares Date: 08/26/15 Time: 23:21 Sample (adjusted): 1973 2013 Included observations: 41 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. D(INF(-1)) -1.443850 0.219167 -6.587901 0.0000
D(INF(-1),2) 0.371021 0.150925 2.458319 0.0188 C 2.793058 5.088512 0.548895 0.5864
@TREND("1970") -0.116825 0.196835 -0.593518 0.5564 R-squared 0.597405 Mean dependent var 0.268293
Adjusted R-squared 0.564762 S.D. dependent var 22.57055 S.E. of regression 14.89037 Akaike info criterion 8.331775 Sum squared resid 8203.759 Schwarz criterion 8.498952 Log likelihood -166.8014 Hannan-Quinn criter. 8.392652 F-statistic 18.30125 Durbin-Watson stat 2.042006 Prob(F-statistic) 0.000000
1ST DIFFERENCE
Null Hypothesis: D(MS) has a unit root Exogenous: Constant, Linear Trend
P a g e | xxviii
Lag Length: 0 (Automatic - based on SIC, maxlag=9) t-Statistic Prob.* Augmented Dickey-Fuller test statistic -6.331609 0.0000
Test critical values: 1% level -4.192337 5% level -3.520787 10% level -3.191277 *MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(MS,2) Method: Least Squares Date: 08/26/15 Time: 23:22 Sample (adjusted): 1972 2013 Included observations: 42 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. D(MS(-1)) -1.013158 0.160016 -6.331609 0.0000
C 0.672960 1.371470 0.490685 0.6264 @TREND("1970") -0.016317 0.053577 -0.304550 0.7623
R-squared 0.506900 Mean dependent var 0.012697
Adjusted R-squared 0.481613 S.D. dependent var 5.840546 S.E. of regression 4.205141 Akaike info criterion 5.779242 Sum squared resid 689.6453 Schwarz criterion 5.903361 Log likelihood -118.3641 Hannan-Quinn criter. 5.824737 F-statistic 20.04572 Durbin-Watson stat 1.998933 Prob(F-statistic) 0.000001
AT 1ST DIFFERENCE Null Hypothesis: D(RGDP) has a unit root Exogenous: Constant, Linear Trend Lag Length: 0 (Automatic - based on SIC, maxlag=9)
t-Statistic Prob.* Augmented Dickey-Fuller test statistic -5.238360 0.0006
Test critical values: 1% level -4.192337 5% level -3.520787 10% level -3.191277 *MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(RGDP,2) Method: Least Squares Date: 08/26/15 Time: 23:23 Sample (adjusted): 1972 2013 Included observations: 42 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. D(RGDP(-1)) -1.006525 0.192145 -5.238360 0.0000
C 5056.677 11836.99 0.427193 0.6716 @TREND("1970") 623.9178 504.8923 1.235744 0.2239
P a g e | xxix
R-squared 0.420674 Mean dependent var -2106.283
Adjusted R-squared 0.390965 S.D. dependent var 46578.57 S.E. of regression 36350.25 Akaike info criterion 23.90854 Sum squared resid 5.15E+10 Schwarz criterion 24.03266 Log likelihood -499.0793 Hannan-Quinn criter. 23.95403 F-statistic 14.15979 Durbin-Watson stat 1.724490 Prob(F-statistic) 0.000024
AT 1ST DIFFERENCE Null Hypothesis: RIR has a unit root Exogenous: Constant, Linear Trend Lag Length: 0 (Automatic - based on SIC, maxlag=9)
t-Statistic Prob.* Augmented Dickey-Fuller test statistic -7.046315 0.0000
Test critical values: 1% level -4.186481 5% level -3.518090 10% level -3.189732 *MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(RIR) Method: Least Squares Date: 08/26/15 Time: 23:24 Sample (adjusted): 1971 2013 Included observations: 43 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. RIR(-1) -1.080467 0.153338 -7.046315 0.0000
C -9.267837 4.402235 -2.105257 0.0416 @TREND("1970") 0.367696 0.173198 2.122984 0.0400
R-squared 0.554804 Mean dependent var 0.780783
Adjusted R-squared 0.532544 S.D. dependent var 19.23099 S.E. of regression 13.14837 Akaike info criterion 8.057686 Sum squared resid 6915.181 Schwarz criterion 8.180561 Log likelihood -170.2403 Hannan-Quinn criter. 8.102998 F-statistic 24.92406 Durbin-Watson stat 1.936052 Prob(F-statistic) 0.000000
AT 1ST DIFFERENCE Null Hypothesis: D(SAV) has a unit root Exogenous: Constant, Linear Trend Lag Length: 0 (Automatic - based on SIC, maxlag=9)
t-Statistic Prob.* Augmented Dickey-Fuller test statistic -6.001551 0.0001
Test critical values: 1% level -4.192337 5% level -3.520787
P a g e | xxx
10% level -3.191277 *MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(SAV,2) Method: Least Squares Date: 08/26/15 Time: 23:25 Sample (adjusted): 1972 2013 Included observations: 42 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. D(SAV(-1)) -0.964605 0.160726 -6.001551 0.0000
C 0.135832 0.816397 0.166379 0.8687 @TREND("1970") 0.005902 0.032035 0.184245 0.8548
R-squared 0.480650 Mean dependent var -0.008869
Adjusted R-squared 0.454016 S.D. dependent var 3.395744 S.E. of regression 2.509139 Akaike info criterion 4.746506 Sum squared resid 245.5354 Schwarz criterion 4.870625 Log likelihood -96.67662 Hannan-Quinn criter. 4.792000 F-statistic 18.04691 Durbin-Watson stat 1.977379 Prob(F-statistic) 0.000003
GHANA PHILLIP-PERRON (PP) STATIONARITY TEST RESULTS AS GE NERATED FROM EVIEWS 8. AT LEVEL Null Hypothesis: BD has a unit root Exogenous: Constant, Linear Trend Bandwidth: 3 (Newey-West automatic) using Bartlett kernel
Adj. t-Stat Prob.* Phillips-Perron test statistic -4.250001 0.0085
Test critical values: 1% level -4.186481 5% level -3.518090 10% level -3.189732 *MacKinnon (1996) one-sided p-values. Residual variance (no correction) 15.63335
HAC corrected variance (Bartlett kernel) 18.27608
Phillips-Perron Test Equation Dependent Variable: D(BD) Method: Least Squares Date: 08/27/15 Time: 16:24 Sample (adjusted): 1971 2013 Included observations: 43 after adjustments
Variable Coefficient Std. Error t-Statistic Prob.
P a g e | xxxi
BD(-1) -0.579475 0.141345 -4.099735 0.0002 C -1.964088 1.394512 -1.408440 0.1667
@TREND("1970") -0.007341 0.050383 -0.145712 0.8849 R-squared 0.296392 Mean dependent var 0.147849
Adjusted R-squared 0.261211 S.D. dependent var 4.769469 S.E. of regression 4.099494 Akaike info criterion 5.726818 Sum squared resid 672.2339 Schwarz criterion 5.849693 Log likelihood -120.1266 Hannan-Quinn criter. 5.772130 F-statistic 8.424899 Durbin-Watson stat 1.878110 Prob(F-statistic) 0.000884
AT LEVEL Null Hypothesis: EXDEP has a unit root Exogenous: Constant, Linear Trend Bandwidth: 4 (Newey-West automatic) using Bartlett kernel
Adj. t-Stat Prob.* Phillips-Perron test statistic -3.743502 0.0299
Test critical values: 1% level -4.186481 5% level -3.518090 10% level -3.189732 *MacKinnon (1996) one-sided p-values. Residual variance (no correction) 1786.212
HAC corrected variance (Bartlett kernel) 1601.857
Phillips-Perron Test Equation Dependent Variable: D(EXDEP) Method: Least Squares Date: 08/27/15 Time: 16:27 Sample (adjusted): 1971 2013 Included observations: 43 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. EXDEP(-1) -0.536569 0.139593 -3.843817 0.0004
C -12.75599 14.08480 -0.905656 0.3705 @TREND("1970") 1.954681 0.727263 2.687722 0.0104
R-squared 0.270003 Mean dependent var 2.971429
Adjusted R-squared 0.233503 S.D. dependent var 50.05131 S.E. of regression 43.81984 Akaike info criterion 10.46526 Sum squared resid 76807.13 Schwarz criterion 10.58814 Log likelihood -222.0032 Hannan-Quinn criter. 10.51058 F-statistic 7.397359 Durbin-Watson stat 1.894760 Prob(F-statistic) 0.001847
P a g e | xxxii
AT FIRST DIFFERENCE Null Hypothesis: D(GEX) has a unit root Exogenous: Constant, Linear Trend Bandwidth: 2 (Newey-West automatic) using Bartlett kernel
Adj. t-Stat Prob.* Phillips-Perron test statistic -6.472818 0.0000
Test critical values: 1% level -4.192337 5% level -3.520787 10% level -3.191277 *MacKinnon (1996) one-sided p-values. Residual variance (no correction) 3.66E+10
HAC corrected variance (Bartlett kernel) 4.82E+10
Phillips-Perron Test Equation Dependent Variable: D(GEX,2) Method: Least Squares Date: 08/27/15 Time: 16:28 Sample (adjusted): 1972 2013 Included observations: 42 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. D(GEX(-1)) -1.187148 0.191448 -6.200890 0.0000
C -110891.6 68902.43 -1.609400 0.1156 @TREND("1970") 10277.64 3221.966 3.189868 0.0028
R-squared 0.500953 Mean dependent var -9790.337
Adjusted R-squared 0.475361 S.D. dependent var 274125.8 S.E. of regression 198554.7 Akaike info criterion 27.30427 Sum squared resid 1.54E+12 Schwarz criterion 27.42839 Log likelihood -570.3896 Hannan-Quinn criter. 27.34976 F-statistic 19.57448 Durbin-Watson stat 1.562010 Prob(F-statistic) 0.000001
AT FIRST DIFFERENCE Null Hypothesis: D(INF) has a unit root Exogenous: Constant, Linear Trend Bandwidth: 18 (Newey-West automatic) using Bartlett kernel
Adj. t-Stat Prob.* Phillips-Perron test statistic -11.52751 0.0000
Test critical values: 1% level -4.192337 5% level -3.520787 10% level -3.191277 *MacKinnon (1996) one-sided p-values. Residual variance (no correction) 231.7983
HAC corrected variance (Bartlett kernel) 23.78487
P a g e | xxxiii
Phillips-Perron Test Equation Dependent Variable: D(INF,2) Method: Least Squares Date: 08/27/15 Time: 16:30 Sample (adjusted): 1972 2013 Included observations: 42 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. D(INF(-1)) -1.054530 0.159877 -6.595897 0.0000
C 0.912014 5.142990 0.177332 0.8602 @TREND("1970") -0.047574 0.201282 -0.236355 0.8144
R-squared 0.527308 Mean dependent var -0.088095
Adjusted R-squared 0.503067 S.D. dependent var 22.41292 S.E. of regression 15.79965 Akaike info criterion 8.426602 Sum squared resid 9735.527 Schwarz criterion 8.550721 Log likelihood -173.9586 Hannan-Quinn criter. 8.472096 F-statistic 21.75305 Durbin-Watson stat 2.018791 Prob(F-statistic) 0.000000
AT FIRST DIFFERENCE Null Hypothesis: D(MS) has a unit root Exogenous: Constant, Linear Trend Bandwidth: 0 (Newey-West automatic) using Bartlett kernel
Adj. t-Stat Prob.* Phillips-Perron test statistic -6.331609 0.0000
Test critical values: 1% level -4.192337 5% level -3.520787 10% level -3.191277 *MacKinnon (1996) one-sided p-values. Residual variance (no correction) 16.42013
HAC corrected variance (Bartlett kernel) 16.42013
Phillips-Perron Test Equation Dependent Variable: D(MS,2) Method: Least Squares Date: 08/27/15 Time: 16:31 Sample (adjusted): 1972 2013 Included observations: 42 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. D(MS(-1)) -1.013158 0.160016 -6.331609 0.0000
C 0.672960 1.371470 0.490685 0.6264 @TREND("1970") -0.016317 0.053577 -0.304550 0.7623
R-squared 0.506900 Mean dependent var 0.012697
Adjusted R-squared 0.481613 S.D. dependent var 5.840546
P a g e | xxxiv
S.E. of regression 4.205141 Akaike info criterion 5.779242 Sum squared resid 689.6453 Schwarz criterion 5.903361 Log likelihood -118.3641 Hannan-Quinn criter. 5.824737 F-statistic 20.04572 Durbin-Watson stat 1.998933 Prob(F-statistic) 0.000001
AT FIRST DIFFERENCE Null Hypothesis: D(RGDP) has a unit root Exogenous: Constant, Linear Trend Bandwidth: 5 (Newey-West automatic) using Bartlett kernel
Adj. t-Stat Prob.* Phillips-Perron test statistic -4.966708 0.0012
Test critical values: 1% level -4.192337 5% level -3.520787 10% level -3.191277 *MacKinnon (1996) one-sided p-values. Residual variance (no correction) 1.23E+09
HAC corrected variance (Bartlett kernel) 9.61E+08
Phillips-Perron Test Equation Dependent Variable: D(RGDP,2) Method: Least Squares Date: 08/27/15 Time: 16:32 Sample (adjusted): 1972 2013 Included observations: 42 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. D(RGDP(-1)) -1.006525 0.192145 -5.238360 0.0000
C 5056.677 11836.99 0.427193 0.6716 @TREND("1970") 623.9178 504.8923 1.235744 0.2239
R-squared 0.420674 Mean dependent var -2106.283
Adjusted R-squared 0.390965 S.D. dependent var 46578.57 S.E. of regression 36350.25 Akaike info criterion 23.90854 Sum squared resid 5.15E+10 Schwarz criterion 24.03266 Log likelihood -499.0793 Hannan-Quinn criter. 23.95403 F-statistic 14.15979 Durbin-Watson stat 1.724490 Prob(F-statistic) 0.000024
AT LEVEL Null Hypothesis: RIR has a unit root Exogenous: Constant, Linear Trend Bandwidth: 6 (Newey-West automatic) using Bartlett kernel
Adj. t-Stat Prob.*
P a g e | xxxv
Phillips-Perron test statistic -7.447115 0.0000 Test critical values: 1% level -4.186481
5% level -3.518090 10% level -3.189732 *MacKinnon (1996) one-sided p-values. Residual variance (no correction) 160.8182
HAC corrected variance (Bartlett kernel) 97.07979
Phillips-Perron Test Equation Dependent Variable: D(RIR) Method: Least Squares Date: 08/27/15 Time: 16:32 Sample (adjusted): 1971 2013 Included observations: 43 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. RIR(-1) -1.080467 0.153338 -7.046315 0.0000
C -9.267837 4.402235 -2.105257 0.0416 @TREND("1970") 0.367696 0.173198 2.122984 0.0400
R-squared 0.554804 Mean dependent var 0.780783
Adjusted R-squared 0.532544 S.D. dependent var 19.23099 S.E. of regression 13.14837 Akaike info criterion 8.057686 Sum squared resid 6915.181 Schwarz criterion 8.180561 Log likelihood -170.2403 Hannan-Quinn criter. 8.102998 F-statistic 24.92406 Durbin-Watson stat 1.936052 Prob(F-statistic) 0.000000
AT FIRST DIFFERENCE Null Hypothesis: D(SAV) has a unit root Exogenous: Constant, Linear Trend Bandwidth: 1 (Newey-West automatic) using Bartlett kernel
Adj. t-Stat Prob.* Phillips-Perron test statistic -6.002668 0.0001
Test critical values: 1% level -4.192337 5% level -3.520787 10% level -3.191277 *MacKinnon (1996) one-sided p-values. Residual variance (no correction) 5.846080
HAC corrected variance (Bartlett kernel) 5.871761
Phillips-Perron Test Equation Dependent Variable: D(SAV,2) Method: Least Squares Date: 08/27/15 Time: 16:33 Sample (adjusted): 1972 2013
P a g e | xxxvi
Included observations: 42 after adjustments Variable Coefficient Std. Error t-Statistic Prob. D(SAV(-1)) -0.964605 0.160726 -6.001551 0.0000
C 0.135832 0.816397 0.166379 0.8687 @TREND("1970") 0.005902 0.032035 0.184245 0.8548
R-squared 0.480650 Mean dependent var -0.008869
Adjusted R-squared 0.454016 S.D. dependent var 3.395744 S.E. of regression 2.509139 Akaike info criterion 4.746506 Sum squared resid 245.5354 Schwarz criterion 4.870625 Log likelihood -96.67662 Hannan-Quinn criter. 4.792000 F-statistic 18.04691 Durbin-Watson stat 1.977379 Prob(F-statistic) 0.000003
GHANA AUGMENTED DICKEY-FULLER (ADF) STATIONARITY TEST RES ULTS AS GENERATED FROM EVIEWS 8. AT FIRST DIFFERENCE LEVEL Null Hypothesis: D(BD) has a unit root Exogenous: Constant, Linear Trend Lag Length: 0 (Automatic - based on SIC, maxlag=9)
t-Statistic Prob.* Augmented Dickey-Fuller test statistic -6.249993 0.0000
Test critical values: 1% level -4.192337 5% level -3.520787 10% level -3.191277 *MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(BD,2) Method: Least Squares Date: 08/28/15 Time: 06:50 Sample (adjusted): 1972 2013 Included observations: 42 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. D(BD(-1)) -1.015760 0.162522 -6.249993 0.0000
C 0.364548 0.824605 0.442087 0.6609 @TREND("1970") -0.019538 0.032419 -0.602667 0.5502
R-squared 0.500845 Mean dependent var 0.046595
Adjusted R-squared 0.475247 S.D. dependent var 3.481703 S.E. of regression 2.522139 Akaike info criterion 4.756841 Sum squared resid 248.0862 Schwarz criterion 4.880960 Log likelihood -96.89366 Hannan-Quinn criter. 4.802336 F-statistic 19.56604 Durbin-Watson stat 1.979822 Prob(F-statistic) 0.000001
P a g e | xxxvii
AT FIRST DIFFERENCE Null Hypothesis: D(EXDEP) has a unit root Exogenous: Constant, Linear Trend Lag Length: 0 (Automatic - based on SIC, maxlag=9)
t-Statistic Prob.* Augmented Dickey-Fuller test statistic -5.463968 0.0003
Test critical values: 1% level -4.192337 5% level -3.520787 10% level -3.191277 *MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(EXDEP,2) Method: Least Squares Date: 08/28/15 Time: 06:54 Sample (adjusted): 1972 2013 Included observations: 42 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. D(EXDEP(-1)) -0.869719 0.159174 -5.463968 0.0000
C -0.041446 0.023056 -1.797624 0.0800 @TREND("1970") 0.003730 0.001062 3.510195 0.0011
R-squared 0.434635 Mean dependent var 0.004331
Adjusted R-squared 0.405642 S.D. dependent var 0.087428 S.E. of regression 0.067402 Akaike info criterion -2.487524 Sum squared resid 0.177180 Schwarz criterion -2.363404 Log likelihood 55.23799 Hannan-Quinn criter. -2.442029 F-statistic 14.99101 Durbin-Watson stat 2.026269 Prob(F-statistic) 0.000015
AT LEVEL Null Hypothesis: GEX has a unit root Exogenous: Constant, Linear Trend Lag Length: 0 (Automatic - based on SIC, maxlag=9)
t-Statistic Prob.* Augmented Dickey-Fuller test statistic -4.307310 0.0073
Test critical values: 1% level -4.186481 5% level -3.518090 10% level -3.189732 *MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(GEX) Method: Least Squares Date: 08/28/15 Time: 06:55 Sample (adjusted): 1971 2013 Included observations: 43 after adjustments
P a g e | xxxviii
Variable Coefficient Std. Error t-Statistic Prob. GEX(-1) -0.624921 0.145084 -4.307310 0.0001
C 60.92844 14.16708 4.300704 0.0001 @TREND("1970") 0.331206 0.090124 3.675016 0.0007
R-squared 0.317498 Mean dependent var 0.426608
Adjusted R-squared 0.283372 S.D. dependent var 4.930724 S.E. of regression 4.174048 Akaike info criterion 5.762864 Sum squared resid 696.9071 Schwarz criterion 5.885738 Log likelihood -120.9016 Hannan-Quinn criter. 5.808176 F-statistic 9.303924 Durbin-Watson stat 2.000396 Prob(F-statistic) 0.000481
AT LEVEL Null Hypothesis: INF has a unit root Exogenous: Constant, Linear Trend Lag Length: 0 (Automatic - based on SIC, maxlag=9)
t-Statistic Prob.* Augmented Dickey-Fuller test statistic -5.084804 0.0008
Test critical values: 1% level -4.186481 5% level -3.518090 10% level -3.189732 *MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(INF) Method: Least Squares Date: 08/28/15 Time: 06:56 Sample (adjusted): 1971 2013 Included observations: 43 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. INF(-1) -0.747009 0.146910 -5.084804 0.0000
C 40.67183 10.89788 3.732086 0.0006 @TREND("1970") -0.755611 0.342021 -2.209250 0.0329
R-squared 0.394725 Mean dependent var 0.243488
Adjusted R-squared 0.364461 S.D. dependent var 32.75708 S.E. of regression 26.11417 Akaike info criterion 9.430047 Sum squared resid 27277.98 Schwarz criterion 9.552921 Log likelihood -199.7460 Hannan-Quinn criter. 9.475359 F-statistic 13.04284 Durbin-Watson stat 2.209570 Prob(F-statistic) 0.000044
AT FIRST DIFFERENCE Null Hypothesis: D(MS) has a unit root Exogenous: Constant, Linear Trend Lag Length: 0 (Automatic - based on SIC, maxlag=9)
P a g e | xxxix
t-Statistic Prob.* Augmented Dickey-Fuller test statistic -6.053898 0.0001
Test critical values: 1% level -4.192337 5% level -3.520787 10% level -3.191277 *MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(MS,2) Method: Least Squares Date: 08/28/15 Time: 06:57 Sample (adjusted): 1972 2013 Included observations: 42 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. D(MS(-1)) -0.985374 0.162767 -6.053898 0.0000
C 0.090828 0.949272 0.095682 0.9243 @TREND("1970") 0.005847 0.037242 0.157001 0.8761
R-squared 0.485058 Mean dependent var -0.067889
Adjusted R-squared 0.458651 S.D. dependent var 3.965057 S.E. of regression 2.917347 Akaike info criterion 5.047976 Sum squared resid 331.9257 Schwarz criterion 5.172095 Log likelihood -103.0075 Hannan-Quinn criter. 5.093470 F-statistic 18.36835 Durbin-Watson stat 1.902634 Prob(F-statistic) 0.000002
AT FIRST DIFFERENCE Null Hypothesis: D(RGDP) has a unit root Exogenous: Constant, Linear Trend Lag Length: 0 (Automatic - based on SIC, maxlag=9)
t-Statistic Prob.* Augmented Dickey-Fuller test statistic -9.619379 0.0000
Test critical values: 1% level -4.192337 5% level -3.520787 10% level -3.191277 *MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(RGDP,2) Method: Least Squares Date: 08/28/15 Time: 07:06 Sample (adjusted): 1972 2013 Included observations: 42 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. D(RGDP(-1)) -1.396794 0.145206 -9.619379 0.0000
C -53.10327 146.4206 -0.362676 0.7188 @TREND("1970") 6.586028 5.773866 1.140662 0.2610
P a g e | xl
R-squared 0.703620 Mean dependent var 16.65238
Adjusted R-squared 0.688421 S.D. dependent var 804.2075 S.E. of regression 448.9024 Akaike info criterion 15.12024 Sum squared resid 7859023. Schwarz criterion 15.24436 Log likelihood -314.5250 Hannan-Quinn criter. 15.16573 F-statistic 46.29397 Durbin-Watson stat 2.396719 Prob(F-statistic) 0.000000
AT FIRST DIFFERENCE Null Hypothesis: D(RIR) has a unit root Exogenous: Constant, Linear Trend Lag Length: 0 (Automatic - based on SIC, maxlag=9)
t-Statistic Prob.* Augmented Dickey-Fuller test statistic -11.30048 0.0000
Test critical values: 1% level -4.192337 5% level -3.520787 10% level -3.191277 *MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(RIR,2) Method: Least Squares Date: 08/28/15 Time: 07:20 Sample (adjusted): 1972 2013 Included observations: 42 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. D(RIR(-1)) -1.533775 0.135727 -11.30048 0.0000
C -4.024335 9.085124 -0.442959 0.6602 @TREND("1970") 0.176607 0.355619 0.496618 0.6222
R-squared 0.766062 Mean dependent var -0.208810
Adjusted R-squared 0.754065 S.D. dependent var 56.25121 S.E. of regression 27.89603 Akaike info criterion 9.563595 Sum squared resid 30349.34 Schwarz criterion 9.687714 Log likelihood -197.8355 Hannan-Quinn criter. 9.609089 F-statistic 63.85526 Durbin-Watson stat 2.165061 Prob(F-statistic) 0.000000
AT LEVEL Null Hypothesis: SAV has a unit root Exogenous: Constant, Linear Trend Lag Length: 0 (Automatic - based on SIC, maxlag=9)
t-Statistic Prob.*
P a g e | xli
Augmented Dickey-Fuller test statistic -3.704077 0.0327 Test critical values: 1% level -4.186481
5% level -3.518090 10% level -3.189732 *MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(SAV) Method: Least Squares Date: 08/28/15 Time: 07:22 Sample (adjusted): 1971 2013 Included observations: 43 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. SAV(-1) -0.493387 0.133201 -3.704077 0.0006
C 3.199397 1.596396 2.004137 0.0519 @TREND("1970") 0.024792 0.049394 0.501924 0.6185
R-squared 0.257149 Mean dependent var -0.108150
Adjusted R-squared 0.220006 S.D. dependent var 4.547604 S.E. of regression 4.016319 Akaike info criterion 5.685823 Sum squared resid 645.2327 Schwarz criterion 5.808697 Log likelihood -119.2452 Hannan-Quinn criter. 5.731135 F-statistic 6.923296 Durbin-Watson stat 1.962423 Prob(F-statistic) 0.002618
GHANA PHILLIP-PERRON (PP) STATIONARITY TEST RESULTS AS GE NERATED FROM EVIEWS 8. AT FIRST DIFFERENCE Null Hypothesis: D(BD) has a unit root Exogenous: Constant, Linear Trend Bandwidth: 11 (Newey-West automatic) using Bartlett kernel
Adj. t-Stat Prob.* Phillips-Perron test statistic -7.440796 0.0000
Test critical values: 1% level -4.192337 5% level -3.520787 10% level -3.191277 *MacKinnon (1996) one-sided p-values. Residual variance (no correction) 5.906815
HAC corrected variance (Bartlett kernel) 1.568812
Phillips-Perron Test Equation Dependent Variable: D(BD,2) Method: Least Squares Date: 08/28/15 Time: 07:43
P a g e | xlii
Sample (adjusted): 1972 2013 Included observations: 42 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. D(BD(-1)) -1.015760 0.162522 -6.249993 0.0000
C 0.364548 0.824605 0.442087 0.6609 @TREND("1970") -0.019538 0.032419 -0.602667 0.5502
R-squared 0.500845 Mean dependent var 0.046595
Adjusted R-squared 0.475247 S.D. dependent var 3.481703 S.E. of regression 2.522139 Akaike info criterion 4.756841 Sum squared resid 248.0862 Schwarz criterion 4.880960 Log likelihood -96.89366 Hannan-Quinn criter. 4.802336 F-statistic 19.56604 Durbin-Watson stat 1.979822 Prob(F-statistic) 0.000001
AT FIRST DIFFERENCE
Null Hypothesis: D(EXDEP) has a unit root Exogenous: Constant, Linear Trend Bandwidth: 1 (Newey-West automatic) using Bartlett kernel
Adj. t-Stat Prob.* Phillips-Perron test statistic -5.454011 0.0003
Test critical values: 1% level -4.192337 5% level -3.520787 10% level -3.191277 *MacKinnon (1996) one-sided p-values. Residual variance (no correction) 0.004219
HAC corrected variance (Bartlett kernel) 0.004131
Phillips-Perron Test Equation Dependent Variable: D(EXDEP,2) Method: Least Squares Date: 08/28/15 Time: 07:45 Sample (adjusted): 1972 2013 Included observations: 42 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. D(EXDEP(-1)) -0.869719 0.159174 -5.463968 0.0000
C -0.041446 0.023056 -1.797624 0.0800 @TREND("1970") 0.003730 0.001062 3.510195 0.0011
R-squared 0.434635 Mean dependent var 0.004331
Adjusted R-squared 0.405642 S.D. dependent var 0.087428 S.E. of regression 0.067402 Akaike info criterion -2.487524 Sum squared resid 0.177180 Schwarz criterion -2.363404 Log likelihood 55.23799 Hannan-Quinn criter. -2.442029 F-statistic 14.99101 Durbin-Watson stat 2.026269 Prob(F-statistic) 0.000015
P a g e | xliii
AT LEVEL Null Hypothesis: GEX has a unit root Exogenous: Constant, Linear Trend Bandwidth: 2 (Newey-West automatic) using Bartlett kernel
Adj. t-Stat Prob.* Phillips-Perron test statistic -4.261254 0.0082
Test critical values: 1% level -4.186481 5% level -3.518090 10% level -3.189732 *MacKinnon (1996) one-sided p-values. Residual variance (no correction) 16.20714
HAC corrected variance (Bartlett kernel) 15.32707
Phillips-Perron Test Equation Dependent Variable: D(GEX) Method: Least Squares Date: 08/28/15 Time: 07:47 Sample (adjusted): 1971 2013 Included observations: 43 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. GEX(-1) -0.624921 0.145084 -4.307310 0.0001
C 60.92844 14.16708 4.300704 0.0001 @TREND("1970") 0.331206 0.090124 3.675016 0.0007
R-squared 0.317498 Mean dependent var 0.426608
Adjusted R-squared 0.283372 S.D. dependent var 4.930724 S.E. of regression 4.174048 Akaike info criterion 5.762864 Sum squared resid 696.9071 Schwarz criterion 5.885738 Log likelihood -120.9016 Hannan-Quinn criter. 5.808176 F-statistic 9.303924 Durbin-Watson stat 2.000396 Prob(F-statistic) 0.000481
AT LEVEL Null Hypothesis: INF has a unit root Exogenous: Constant, Linear Trend Bandwidth: 1 (Newey-West automatic) using Bartlett kernel
Adj. t-Stat Prob.* Phillips-Perron test statistic -5.029343 0.0010
Test critical values: 1% level -4.186481 5% level -3.518090 10% level -3.189732 *MacKinnon (1996) one-sided p-values. Residual variance (no correction) 634.3717
P a g e | xliv
HAC corrected variance (Bartlett kernel) 555.9835
Phillips-Perron Test Equation Dependent Variable: D(INF) Method: Least Squares Date: 08/28/15 Time: 07:48 Sample (adjusted): 1971 2013 Included observations: 43 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. INF(-1) -0.747009 0.146910 -5.084804 0.0000
C 40.67183 10.89788 3.732086 0.0006 @TREND("1970") -0.755611 0.342021 -2.209250 0.0329
R-squared 0.394725 Mean dependent var 0.243488
Adjusted R-squared 0.364461 S.D. dependent var 32.75708 S.E. of regression 26.11417 Akaike info criterion 9.430047 Sum squared resid 27277.98 Schwarz criterion 9.552921 Log likelihood -199.7460 Hannan-Quinn criter. 9.475359 F-statistic 13.04284 Durbin-Watson stat 2.209570 Prob(F-statistic) 0.000044
AT FIRST DIFFERENCE Null Hypothesis: D(MS) has a unit root Exogenous: Constant, Linear Trend Bandwidth: 3 (Newey-West automatic) using Bartlett kernel
Adj. t-Stat Prob.* Phillips-Perron test statistic -6.057001 0.0001
Test critical values: 1% level -4.192337 5% level -3.520787 10% level -3.191277 *MacKinnon (1996) one-sided p-values. Residual variance (no correction) 7.902992
HAC corrected variance (Bartlett kernel) 7.992544
Phillips-Perron Test Equation Dependent Variable: D(MS,2) Method: Least Squares Date: 08/28/15 Time: 07:49 Sample (adjusted): 1972 2013 Included observations: 42 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. D(MS(-1)) -0.985374 0.162767 -6.053898 0.0000
C 0.090828 0.949272 0.095682 0.9243 @TREND("1970") 0.005847 0.037242 0.157001 0.8761
R-squared 0.485058 Mean dependent var -0.067889
P a g e | xlv
Adjusted R-squared 0.458651 S.D. dependent var 3.965057 S.E. of regression 2.917347 Akaike info criterion 5.047976 Sum squared resid 331.9257 Schwarz criterion 5.172095 Log likelihood -103.0075 Hannan-Quinn criter. 5.093470 F-statistic 18.36835 Durbin-Watson stat 1.902634 Prob(F-statistic) 0.000002
AT LEVEL Null Hypothesis: RGDP has a unit root Exogenous: Constant, Linear Trend Bandwidth: 1 (Newey-West automatic) using Bartlett kernel
Adj. t-Stat Prob.* Phillips-Perron test statistic -5.169553 0.0007
Test critical values: 1% level -4.186481 5% level -3.518090 10% level -3.189732 *MacKinnon (1996) one-sided p-values. Residual variance (no correction) 132800.4
HAC corrected variance (Bartlett kernel) 127526.5
Phillips-Perron Test Equation Dependent Variable: D(RGDP) Method: Least Squares Date: 08/28/15 Time: 07:49 Sample (adjusted): 1971 2013 Included observations: 43 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. RGDP(-1) -0.794349 0.152999 -5.191860 0.0000
C -266.7585 123.3857 -2.161990 0.0367 @TREND("1970") 52.40650 10.10657 5.185392 0.0000
R-squared 0.416193 Mean dependent var 59.87674
Adjusted R-squared 0.387002 S.D. dependent var 482.5857 S.E. of regression 377.8365 Akaike info criterion 14.77401 Sum squared resid 5710417. Schwarz criterion 14.89689 Log likelihood -314.6413 Hannan-Quinn criter. 14.81933 F-statistic 14.25787 Durbin-Watson stat 1.995095 Prob(F-statistic) 0.000021
AT LEVEL Null Hypothesis: RIR has a unit root Exogenous: Constant, Linear Trend Bandwidth: 2 (Newey-West automatic) using Bartlett kernel
Adj. t-Stat Prob.*
P a g e | xlvi
Phillips-Perron test statistic -4.682621 0.0026 Test critical values: 1% level -4.186481
5% level -3.518090 10% level -3.189732 *MacKinnon (1996) one-sided p-values. Residual variance (no correction) 635.7319
HAC corrected variance (Bartlett kernel) 626.2272
Phillips-Perron Test Equation Dependent Variable: D(RIR) Method: Least Squares Date: 08/28/15 Time: 07:50 Sample (adjusted): 1971 2013 Included observations: 43 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. RIR(-1) -0.687027 0.146398 -4.692872 0.0000
C -25.86361 9.544904 -2.709677 0.0099 @TREND("1970") 0.736756 0.348897 2.111671 0.0410
R-squared 0.356045 Mean dependent var -0.120698
Adjusted R-squared 0.323847 S.D. dependent var 31.79206 S.E. of regression 26.14215 Akaike info criterion 9.432189 Sum squared resid 27336.47 Schwarz criterion 9.555063 Log likelihood -199.7921 Hannan-Quinn criter. 9.477501 F-statistic 11.05806 Durbin-Watson stat 2.220150 Prob(F-statistic) 0.000150
AT LEVEL Null Hypothesis: SAV has a unit root Exogenous: Constant, Linear Trend Bandwidth: 3 (Newey-West automatic) using Bartlett kernel
Adj. t-Stat Prob.* Phillips-Perron test statistic -3.682215 0.0345
Test critical values: 1% level -4.186481 5% level -3.518090 10% level -3.189732 *MacKinnon (1996) one-sided p-values. Residual variance (no correction) 15.00541
HAC corrected variance (Bartlett kernel) 14.64485
Phillips-Perron Test Equation Dependent Variable: D(SAV) Method: Least Squares
P a g e | xlvii
Date: 08/28/15 Time: 07:51 Sample (adjusted): 1971 2013 Included observations: 43 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. SAV(-1) -0.493387 0.133201 -3.704077 0.0006
C 3.199397 1.596396 2.004137 0.0519 @TREND("1970") 0.024792 0.049394 0.501924 0.6185
R-squared 0.257149 Mean dependent var -0.108150
Adjusted R-squared 0.220006 S.D. dependent var 4.547604 S.E. of regression 4.016319 Akaike info criterion 5.685823 Sum squared resid 645.2327 Schwarz criterion 5.808697 Log likelihood -119.2452 Hannan-Quinn criter. 5.731135 F-statistic 6.923296 Durbin-Watson stat 1.962423 Prob(F-statistic) 0.002618
NIGERIA ENGEL-GRANGER COINTEGRATION RESULT AS GENERATED FRO M EVIEWS 8. AT LEVEL Null Hypothesis: RESID01 has a unit root Exogenous: Constant Lag Length: 0 (Automatic - based on SIC, maxlag=9)
t-Statistic Prob.* Augmented Dickey-Fuller test statistic -5.936582 0.0000
Test critical values: 1% level -3.592462 5% level -2.931404 10% level -2.603944 *MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(RESID01) Method: Least Squares Date: 08/27/15 Time: 04:54 Sample (adjusted): 1971 2013 Included observations: 43 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. RESID01(-1) -0.867169 0.146072 -5.936582 0.0000
C 0.429302 0.509156 0.843165 0.4040 R-squared 0.462246 Mean dependent var 0.135706
Adjusted R-squared 0.449130 S.D. dependent var 4.477150 S.E. of regression 3.322969 Akaike info criterion 5.284990 Sum squared resid 452.7271 Schwarz criterion 5.366906 Log likelihood -111.6273 Hannan-Quinn criter. 5.315198 F-statistic 35.24301 Durbin-Watson stat 1.702978 Prob(F-statistic) 0.000001
P a g e | xlviii
GHANA ENGEL-GRANGER COINTEGRATION RESULT AS GENERATED FRO M EVIEWS 8. AT LEVEL RESIDUAL Null Hypothesis: RESID01 has a unit root Exogenous: Constant, Linear Trend Lag Length: 0 (Automatic - based on SIC, maxlag=9)
t-Statistic Prob.* Augmented Dickey-Fuller test statistic -3.825501 0.0246
Test critical values: 1% level -4.186481 5% level -3.518090 10% level -3.189732 *MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(RESID01) Method: Least Squares Date: 08/28/15 Time: 11:54 Sample (adjusted): 1971 2013 Included observations: 43 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. RESID01(-1) -0.541540 0.141561 -3.825501 0.0004
C 0.138443 0.645015 0.214636 0.8311 @TREND("1970") -0.004827 0.025614 -0.188458 0.8515
R-squared 0.273687 Mean dependent var 0.002479
Adjusted R-squared 0.237372 S.D. dependent var 2.368815 S.E. of regression 2.068653 Akaike info criterion 4.358886 Sum squared resid 171.1730 Schwarz criterion 4.481761 Log likelihood -90.71606 Hannan-Quinn criter. 4.404199 F-statistic 7.536355 Durbin-Watson stat 1.767630 Prob(F-statistic) 0.001669
Seemingly Unrelated Regression (SUR) Estimation Results Using STATA 13 SUR Estimation Result for Model I . sureg (rirng gexng bdng lnmsng infng) (rirgh gexgh bdgh lnmsgh infgh) Seemingly unrelated regression ---------------------------------------------------------------------- Equation Obs Parms RMSE "R-sq" chi2 P ---------------------------------------------------------------------- rirng 44 4 10.89802 0.7841 34.31 0.0000 rirgh 44 4 8.383868 0.9181 496.03 0.0000 ---------------------------------------------------------------------- ------------------------------------------------------------------------------ | Coef. Std. Err. z P>|z| [95% Conf. Interval]
P a g e | xlix
-------------+---------------------------------------------------------------- rirng | gexng | 1.91e-06 1.52e-06 1.25 0.211 -1.08e-06 4.89e-06 bdng | -.214025 .5106353 -3.42 0.675 -1.214852 .7868018 lnmsng | 12.40231 7.475809 1.66 0.097 -2.250005 27.05463 infng | -.4221472 .1061846 -3.98 0.000 -.6302652 -.2140291 _cons | -34.47242 21.32531 -1.62 0.106 -76.26927 7.324422 -------------+---------------------------------------------------------------- rirgh | gexgh | .7118853 .19213 3.71 0.000 .3353174 1.088453 bdgh | -1.23363 .4557324 -2.71 0.007 -2.126849 -.3404108 lnmsgh | -19.1024 5.312032 -3.60 0.000 -29.51379 -8.69101 infgh | -.9134943 .0472126 -19.35 0.000 -1.006029 -.8209593 _cons | -10.42766 20.52713 -0.51 0.611 -50.66009 29.80478 ------------------------------------------------------------------------------ SUR Estimation Result for Model II . sureg (infng bdng lnrgdpng lnmsng exdepng) (infgh bdgh rgdpgh lnmsgh exdepgh) Seemingly unrelated regression ---------------------------------------------------------------------- Equation Obs Parms RMSE "R-sq" chi2 P ---------------------------------------------------------------------- infng 44 4 13.65895 0.8597 16.85 0.0021 infgh 44 4 26.513 0.7461 5.60 0.2308 ---------------------------------------------------------------------- ------------------------------------------------------------------------------ | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- ifnng | bdng | -1.457205 .5578419 -2.61 0.009 -2.550555 -.3638547 lnrgdpng | -2.236352 2.061049 -1.09 0.278 -6.275934 1.80323 lnmsng | -11.90506 9.045965 -1.32 0.188 -29.63482 5.824706 exdepng | .1373776 .0408192 3.37 0.001 .0573734 .2173818 _cons | 69.77756 24.02346 2.90 0.004 22.69245 116.8627 -------------+---------------------------------------------------------------- infgh | bdgh | .1113837 1.428449 3.08 0.938 -2.688324 2.911091 rgdpgh | -.0064234 .010221 -1.63 0.530 -.0264562 .0136093 lnmsgh | -5.406567 19.38839 -0.28 0.780 -43.4071 32.59397 exdepgh | -5.146021 18.53571 -0.28 0.781 -41.47535 31.18331 _cons | 57.80081 58.54893 0.99 0.324 -56.95298 172.5546 ------------------------------------------------------------------------------ SUR Estimation Result for Model III . sureg (lnrgdpng infng lnsavng rirng bdng) (lnrgdpgh infgh lnsavgh rirgh bdgh)
P a g e | l
Seemingly unrelated regression ---------------------------------------------------------------------- Equation Obs Parms RMSE "R-sq" chi2 P ---------------------------------------------------------------------- lnrgdpng 40 4 1.245208 0.8245 12.90 0.0118 lnrgdpgh 40 4 .5668675 0.7024 15.54 0.0037 ---------------------------------------------------------------------- ------------------------------------------------------------------------------ | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- lnrgdpng | infng | .0145521 .0098601 3.48 0.140 -.0047735 .0338776 lnsavng | .5001784 .3829107 1.31 0.191 -.2503129 1.25067 rirng | .026369 .0111962 2.36 0.019 .0044248 .0483131 bdng | -.0161941 .0344067 -2.47 0.638 -.08363 .0512417 _cons | 10.67361 .9056314 11.79 0.000 8.898608 12.44862 -------------+---------------------------------------------------------------- lnrgdpgh | infgh | -.0140296 .0066474 -2.11 0.035 -.0270582 -.001001 lnsavgh | .0333115 .1161356 0.29 0.774 -.19431 .260933 rirgh | -.0079105 .0065295 -1.21 0.226 -.0207081 .0048871 bdgh | -.0371651 .0236394 -3.57 0.116 -.0834975 .0091673 _cons | 7.013945 .2628574 26.68 0.000 6.498754 7.529136 ------------------------------------------------------------------------------ NIGERIA NORMALITY TEST RESULT AS GENERATED FROM EVIEWS 8.
0
1
2
3
4
5
6
7
8
9
-8 -6 -4 -2 0 2 4 6 8 10
Series: ResidualsSample 1970 2013Observations 44
Mean 0.298047Median 0.575728Maximum 10.51545Minimum -7.279739Std. Dev. 3.479563Skewness 0.350590Kurtosis 3.653747
Jarque-Bera 1.684903Probability 0.430654
P a g e | li
GHANA NORMALITY TEST RESULT AS GENERATED FROM EVIEWS 8.
0
1
2
3
4
5
6
7
8
-15 -10 -5 0 5 10 15 20
Series: ResidualsSample 1970 2013Observations 44
Mean 2.46e-14Median -0.188538Maximum 20.41152Minimum -13.03041Std. Dev. 8.155610Skewness 0.626790Kurtosis 3.054182
Jarque-Bera 2.886401Probability 0.236171
NIGERIA SERIAL CORRELATION RESULT Breusch-Godfrey Serial Correlation LM Test:
F-statistic 6.913589 Prob. F(2,35) 0.0029
Obs*R-squared 12.46014 Prob. Chi-Square(2) 0.0020
Test Equation: Dependent Variable: RESID Method: Least Squares Date: 08/28/15 Time: 22:41 Sample: 1970 2013 Included observations: 44 Presample missing value lagged residuals set to zero.
Variable Coefficient Std. Error t-Statistic Prob. EXDEP 1.540601 1.920988 0.801983 0.4280
GEX -0.001379 0.021096 -0.065388 0.9482 INF 0.041631 0.040618 1.024927 0.3124 MS -0.017322 0.088015 -0.196813 0.8451
RGDP -0.000825 0.001070 -0.771235 0.4457 RIR 0.032013 0.038608 0.829184 0.4126 SAV -0.006863 0.076366 -0.089867 0.9289
RESID(-1) 0.640708 0.172305 3.718447 0.0007 RESID(-2) -0.255295 0.166726 -1.531228 0.1347
R-squared 0.283185 Mean dependent var -0.002694
Adjusted R-squared 0.119342 S.D. dependent var 2.277742 S.E. of regression 2.137510 Akaike info criterion 4.537410 Sum squared resid 159.9132 Schwarz criterion 4.902358
P a g e | lii
Log likelihood -90.82302 Hannan-Quinn criter. 4.672750 Durbin-Watson stat 1.946273
GHANA SERIAL CORRELATION RESULT Breusch-Godfrey Serial Correlation LM Test:
F-statistic 1.136370 Prob. F(2,37) 0.3319
Obs*R-squared 2.546311 Prob. Chi-Square(2) 0.2799
Test Equation: Dependent Variable: RESID Method: Least Squares Date: 08/29/15 Time: 00:39 Sample: 1970 2013 Included observations: 44 Presample missing value lagged residuals set to zero.
Variable Coefficient Std. Error t-Statistic Prob. C 2.601546 7.203218 0.361164 0.7200
GEX 4.41E-07 1.62E-06 0.272622 0.7867 BD -0.131521 0.521669 -0.252116 0.8023 MS -0.140113 0.354515 -0.395224 0.6949 INF -0.012579 0.114862 -0.109512 0.9134
RESID(-1) -0.231770 0.169078 -1.370783 0.1787 RESID(-2) -0.147604 0.165068 -0.894198 0.3770
R-squared 0.057871 Mean dependent var 2.32E-15
Adjusted R-squared -0.094907 S.D. dependent var 11.03810 S.E. of regression 11.55003 Akaike info criterion 7.876163 Sum squared resid 4935.916 Schwarz criterion 8.160011 Log likelihood -166.2756 Hannan-Quinn criter. 7.981427 F-statistic 0.378790 Durbin-Watson stat 1.891342 Prob(F-statistic) 0.887805
NIGERIA White Heteroscedasticity Heteroscedasticity Test: Harvey
F-statistic 0.821468 Prob. F(15,27) 0.6475
Obs*R-squared 13.47456 Prob. Chi-Square(15) 0.5657 Scaled explained SS 15.80805 Prob. Chi-Square(15) 0.3949
Test Equation: Dependent Variable: LRESID2 Method: Least Squares Date: 08/29/15 Time: 00:29 Sample (adjusted): 1971 2013 Included observations: 43 after adjustments
P a g e | liii
Variable Coefficient Std. Error t-Statistic Prob. C -0.729441 8.088995 -0.090177 0.9288
LRESID2(-1) -0.045718 0.187226 -0.244183 0.8089 GEX^2 -3.59E-13 3.55E-13 -1.011540 0.3207
GEX*BD 7.09E-08 2.42E-07 0.293142 0.7717 GEX*MS 1.99E-07 1.12E-07 1.781654 0.0861 GEX*INF 1.36E-07 1.22E-07 1.114483 0.2749
GEX -4.99E-06 2.56E-06 -1.946646 0.0621 BD^2 -0.044081 0.027236 -1.618455 0.1172
BD*MS -0.076072 0.034459 -2.207624 0.0360 BD*INF 0.000674 0.010358 0.065051 0.9486
BD 1.315417 0.606749 2.167976 0.0391 MS^2 -0.031998 0.018107 -1.767170 0.0885
MS*INF 0.009482 0.007192 1.318395 0.1984 MS 0.935376 0.749123 1.248628 0.2225
INF^2 0.002338 0.001502 1.557165 0.1311 INF -0.362856 0.182008 -1.993629 0.0564
R-squared 0.313362 Mean dependent var 2.852148
Adjusted R-squared -0.068104 S.D. dependent var 2.434593 S.E. of regression 2.516131 Akaike info criterion 4.962144 Sum squared resid 170.9347 Schwarz criterion 5.617475 Log likelihood -90.68610 Hannan-Quinn criter. 5.203810 F-statistic 0.821468 Durbin-Watson stat 2.152461 Prob(F-statistic) 0.647511
GHANA White Heteroscedasticity Heteroscedasticity Test: Harvey
F-statistic 1.339767 Prob. F(15,27) 0.2464
Obs*R-squared 18.34849 Prob. Chi-Square(15) 0.2448 Scaled explained SS 22.15187 Prob. Chi-Square(15) 0.1039
Test Equation: Dependent Variable: LRESID2 Method: Least Squares Date: 08/29/15 Time: 00:04 Sample (adjusted): 1971 2013 Included observations: 43 after adjustments
Variable Coefficient Std. Error t-Statistic Prob. C -42.51891 85.83247 -0.495371 0.6243
LRESID2(-1) -0.114461 0.171221 -0.668497 0.5095 GEX^2 -0.011042 0.008434 -1.309319 0.2015
GEX*BD 0.031081 0.027584 1.126760 0.2698 GEX*MS 0.048003 0.020773 2.310870 0.0287 GEX*INF 0.004393 0.004954 0.886867 0.3830
GEX 1.334589 1.620726 0.823451 0.4175 BD^2 -0.059139 0.068240 -0.866635 0.3938
BD*MS -0.013496 0.039491 -0.341760 0.7352 BD*INF 0.004872 0.008761 0.556117 0.5827
BD -3.750597 3.063259 -1.224381 0.2314 MS^2 -0.057792 0.019080 -3.028885 0.0054
P a g e | liv
MS*INF -0.001221 0.002718 -0.449063 0.6570 MS -2.533502 1.828589 -1.385496 0.1772
INF^2 0.000513 0.000558 0.919896 0.3658 INF -0.439677 0.541411 -0.812094 0.4238
R-squared 0.426709 Mean dependent var 2.671298
Adjusted R-squared 0.108214 S.D. dependent var 2.469730 S.E. of regression 2.332275 Akaike info criterion 4.810388 Sum squared resid 146.8667 Schwarz criterion 5.465719 Log likelihood -87.42335 Hannan-Quinn criter. 5.052054 F-statistic 1.339767 Durbin-Watson stat 1.820869 Prob(F-statistic) 0.246393
Ramsey RESET Test Model I Ramsey RESET Test Equation: UNTITLED Specification: RIR C GEX BD LOG(MS) INF Omitted Variables: Squares of fitted values
Value df Probability
t-statistic 1.323514 38 0.1936 F-statistic 1.751689 (1, 38) 0.1936 Likelihood ratio 1.982911 1 0.1591
F-test summary:
Sum of Sq. df Mean
Squares Test SSR 228.6885 1 228.6885 Restricted SSR 5189.708 39 133.0694 Unrestricted SSR 4961.020 38 130.5531 Unrestricted SSR 4961.020 38 130.5531
LR test summary: Value df
Restricted LogL -167.3786 39 Unrestricted LogL -166.3872 38
Unrestricted Test Equation: Dependent Variable: RIR Method: Least Squares Date: 09/02/15 Time: 17:18 Sample: 1970 2013 Included observations: 44
Variable Coefficient Std. Error t-Statistic Prob. C -24.94845 23.02929 -1.083336 0.2855
GEX 2.41E-06 1.70E-06 1.413653 0.1656 BD -0.169056 0.547982 -0.308507 0.7594
LOG(MS) 8.541646 8.140194 1.049317 0.3007 INF -0.195781 0.187274 -1.045423 0.3024
FITTED^2 -0.027971 0.021134 -1.323514 0.1936 R-squared 0.415318 Mean dependent var -1.674357
Adjusted R-squared 0.338386 S.D. dependent var 14.04725 S.E. of regression 11.42599 Akaike info criterion 7.835781
P a g e | lv
Sum squared resid 4961.020 Schwarz criterion 8.079080 Log likelihood -166.3872 Hannan-Quinn criter. 7.926008 F-statistic 5.398512 Durbin-Watson stat 2.279542 Prob(F-statistic) 0.000758
Model II Ramsey RESET Test Equation: UNTITLED Specification: INF C BD LOG(RGDP) LOG(MS) EXDEP Omitted Variables: Squares of fitted values
Value df Probability
t-statistic 4.195405 38 0.0002 F-statistic 17.60142 (1, 38) 0.0002 Likelihood ratio 16.74740 1 0.0000
F-test summary:
Sum of Sq. df Mean
Squares Test SSR 2591.293 1 2591.293 Restricted SSR 8185.677 39 209.8892 Unrestricted SSR 5594.384 38 147.2206 Unrestricted SSR 5594.384 38 147.2206
LR test summary: Value df
Restricted LogL -177.4042 39 Unrestricted LogL -169.0305 38
Unrestricted Test Equation: Dependent Variable: INF Method: Least Squares Date: 09/02/15 Time: 17:16 Sample: 1970 2013 Included observations: 44
Variable Coefficient Std. Error t-Statistic Prob. C -131.1249 51.14231 -2.563923 0.0144
BD 2.828763 1.127489 2.508904 0.0165 LOG(RGDP) 7.785734 3.062567 2.542225 0.0152
LOG(MS) 18.49161 10.49336 1.762221 0.0861 EXDEP -0.396728 0.132946 -2.984126 0.0049
FITTED^2 0.073607 0.017545 4.195405 0.0002 R-squared 0.495512 Mean dependent var 19.07782
Adjusted R-squared 0.429132 S.D. dependent var 16.05893 S.E. of regression 12.13345 Akaike info criterion 7.955933 Sum squared resid 5594.384 Schwarz criterion 8.199232 Log likelihood -169.0305 Hannan-Quinn criter. 8.046160 F-statistic 7.464784 Durbin-Watson stat 1.665358 Prob(F-statistic) 0.000058
P a g e | lvi
Model III Ramsey RESET Test Equation: UNTITLED Specification: LOG(RGDP) C INF LOG(SAV) RIR BD Omitted Variables: Squares of fitted values
Value df Probability
t-statistic 0.820831 38 0.4169 F-statistic 0.673764 (1, 38) 0.4169 Likelihood ratio 0.773312 1 0.3792
F-test summary:
Sum of Sq. df Mean
Squares Test SSR 1.298113 1 1.298113 Restricted SSR 74.51111 39 1.910541 Unrestricted SSR 73.21299 38 1.926658 Unrestricted SSR 73.21299 38 1.926658
LR test summary: Value df
Restricted LogL -74.02198 39 Unrestricted LogL -73.63533 38
Unrestricted Test Equation: Dependent Variable: LOG(RGDP) Method: Least Squares Date: 09/02/15 Time: 17:14 Sample: 1970 2013 Included observations: 44
Variable Coefficient Std. Error t-Statistic Prob. C 21.86404 16.52884 1.322781 0.1938
INF 0.180836 0.188766 0.957990 0.3441 LOG(SAV) 9.133090 9.522567 0.959100 0.3436
RIR 0.236568 0.246135 0.961131 0.3426 BD 0.022446 0.060217 0.372760 0.7114
FITTED^2 -0.240853 0.293425 -0.820831 0.4169 R-squared 0.273250 Mean dependent var 11.97732
Adjusted R-squared 0.177625 S.D. dependent var 1.530620 S.E. of regression 1.388041 Akaike info criterion 3.619788 Sum squared resid 73.21299 Schwarz criterion 3.863086 Log likelihood -73.63533 Hannan-Quinn criter. 3.710014 F-statistic 2.857512 Durbin-Watson stat 0.464350 Prob(F-statistic) 0.027567