ECONOMIC ANALYSIS OF PUBLIC EXPENDITURE

GROWTH AND OPTIMAL SIZE OF PUBLIC SECTOR

IN NIGERIA

BY

NWAFEE, FRANCIS IKECHUKWU

PG/MSC/13/65868

DEPARTMENT OF ECONOMICS

FACULTY OF SOCIAL SCIENCES

UNIVERSITY OF NIGERIA, NSUKKA

DECEMBER, 2015

0
 TITLE PAGE

ECONOMIC ANALYSIS OF PUBLIC EXPENDITURE

GROWTH AND OPTIMAL SIZE OF PUBLIC SECTOR IN
NIGERIA

Being a research project submitted to the Department of Economics, Faculty of Social

Sciences, University of Nigeria Nsukka in partial fulfilment of the requirement for award of
Masters of Science Degree in Economics (M.Sc Economics)

BY

NWAFEE, FRANCIS IKECHUKWU

PG/M.SC/13/65868

DECEMBER, 2015

i
 APPROVAL PAGE

This thesis report has been approved for the Department of Economics, University of Nigeria,
Nsukka.

By

Prof. (Mrs) S. I. Madueme Prof. (Mrs) S. I. Madueme

Project Supervisor Head of Department

Prof. I. A. Madu
Dean, Faculty of Social Sciences External Examiner

ii
 CERTIFICATION

Nwafee, Francis Ikechukwu, a postgraduate student of the Department of Economics, with

registration number PG/MSC/13/65868 has satisfactorily completed the requirements for the
degree of Master of Science in Economic. The work embodied is original and has not been
submitted in part or full for any other Diploma or Degree of this or any other university.

Prof. (Mrs) S. I. Madueme Prof. (Mrs) S. I. Madueme

Project Supervisor Head of Department

iii
 DEDICATION

To my mother, Mrs Okwuchukwu Caroline Igboanugo-Nwafee.

iv
 ACKNOWLEDGEMENT

I appreciate the wonderful experience with staff and students of Department of Economics,
University of Nigeria Nsukka during the course of this M.Sc programme. In a very special
way, I salute the effort of my project supervisor and Head of Department, Professor (Mrs) S.
I. Madueme, especially for her dedication to duty and her friendly approach to issues of
concern. I also commend her academic impact alongside fellow staff such as Rev. Fr. (Prof.)
H. E. Ichoku, Professors Emeritus F. E. Onah and N. I. Ikpeze, Prof. C. C Agu, Dr. N. E.
Urama, Dr. M. O. Oduh, Dr. E. Ukwueze, Dr. E. O. Nwosu and Dr. A. I. Ifelunini.

The contribution of my mother, Mrs O. I. Igboanugo and my brother, Mr Anthony O.

Igboanugo is also memorable. Others include Sir Micheal Ugwu, Uche Chimobi J., Amakom
Tochukwu Stanley, Ebeke Nnenna Beatrice, Mrs Urama Edith and Mr Ojonta Obed for their
hands of fellowship during this study programme. Lastly, I warmly acknowledge the
opportunity, friendship, support and cooperation of my classmates. I thank you all.

F. I. Nwafee
December, 2015.

v
 TABLE OF CONTENTS

Title Page ... ... ... ... ... ... ... ... ... ... i
Approval page ... ... ... ... ... ... ... ... ... ii
Certification ... ... ... ... ... ... ... ... ... ... iii
Dedication ... ... ... ... ... ... ... ... ... ... iv
Acknowledgements ... ... ... ... ... ... ... ... ... v
Table of contents ... ... ... ... ... ... ... ... ... vi
List of tables ... ... ... ... ... ... ... ... ... ... ix
List of figures ... ... ... ... ... ... ... ... ... ... x
Abstract ... ... ... ... ... ... ... ... ... ... xi
Chapter One: Introduction ... ... ... ... ... ... ... ... 1
1.1 Background of the study ... ... ... ... ... ... ... 1
1.2 Statement of the problem ... ... ... ... ... ... ... 4
1.3 Research questions ... ... ... ... ... ... ... ... 8
1.4 Research objectives ... ... ... ... ... ... ... ... 8
1.5 Research hypotheses ... ... ... ... ... ... ... ... 9
1.6 Scope of the study ... ... ... ... ... ... ... ... 9
1.7 Policy relevance of the study ... ... ... ... ... ... ... 9
Chapter Two: Fiscal Policy and Government Expenditure Growth in Nigeria ... 11
2.1 Pre-independence era ... ... ... ... ... ... ... ... 11
2.2 Post-independence and military era ... ... ... ... ... ... 12
2.3 Return to democratic era ... ... ... ... ... ... ... 16
2.4 The search for optimal size of public sector in Nigeria ... ... ... 17
Chapter Three: Literature Review ... ... ... ... ... ... ... 19
3.1 Conceptual framework ... ... ... ... ... ... ... 19
3.2 Theoretical literature ... ... ... ... ... ... ... ... 23
3.2.1 Rostow and Musgrave stages of development growth theory ... 24
3.2.2 Wagner's Law of expanding state activity ... ... ... ... 25
3.2.3 Peacock and Wiseman displacement effect model ... ... ... 26
3.2.4 Niskanen model of budget maximizing bureaucrats ... ... ... 27
3.2.5 Theoretical background on optimal size of government and the BARS
curve ... ... ... ... ... ... ... ... ... 28
3.2.6 Classification of public expenditure ... ... ... ... ... 30

vi
 3.2.7 Classical reasons for increased government expenditure ... ... 32
3.3 Empirical literature ... ... ... ... ... ... ... ... 35
3.3.1 Studies on determinants of public expenditure and economic growth
from abroad ... ... ... ... ... ... ... ... 35
3.3.2 Studies on determinants of public expenditure and economic growth
from Nigeria ... ... ... ... ... ... ... ... 38
3.3.3 Studies on the size of public sector from abroad ... ... ... 40
3.3.4 Studies on the size of public sector from Nigeria ... ... ... 42
3.4 Summary of literature review ... ... ... ... ... ... ... 43
3.5 Limitations of previous studies and motivation ... ... ... ... 44
Chapter Four: Methodology ... ... ... ... ... ... ... 47
4.1 Theoretical framework ... ... ... ... ... ... ... 47
4.2 Model specification ... ... ... ... ... ... ... ... 49
4.3 Estimation technique and model justification ... ... ... ... 53
4.4 Data sources and software for analysis ... ... ... ... ... 54
Chapter Five: Data Presentation and Analysis ... ... ... ... ... 55
5.1 Introduction to data analysis ... ... ... ... ... ... 55
5.2 Definition of variables and descriptive analysis ... ... ... ... 55
5.3 Stationarity and cointegration test ... ... ... ... ... ... 57
5.4 Regression result for average growth rate of public expenditure ... ... 58
5.4.1 Economic and statistical criteria ... ... ... ... ... 58
5.4.2 Post-estimation diagnostic check for average growth rate of public
expenditure ... ... ... ... ... ... ... ... 60
5.5 Regression result for determinants of public expenditure growth ... ... 62
5.5.1 Post-estimation diagnostic check for determinants of public expenditure
growth … … … … … … … … 66
5.6 Determinants of real gross domestic product in Nigeria ... ... ... 68
5.6.1 Post-estimation and diagnostic check for determinants of real gross
domestic product ... ... ... ... ... ... ... 71
5.7 Optimal size of the public sector in Nigeria ... ... ... ... ... 73
5.8 Evaluation of research hypotheses ... ... ... ... ... ... 75
Chapter Six: Summary, Conclusion and Policy Recommendation ... ... 78
6.1 Summary of major findings ... ... ... ... ... ... ... 78

vii
6.2 Conclusion ... ... ... ... ... ... ... ... ... 79
6.3 Policy recommendation ... ... ... ... ... ... ... 80
6.4 Limitation of the study and recommendation for further research ... ... 81
References ... ... ... ... ... ... ... ... ... ... 83
Appendix ... ... ... ... ... ... ... ... ... ... 91

viii
 LIST OF TABLES

Table 4.1: Summary of key variables ... ... ... ... ... ... ... 54

Table 5.1: Definition of variables ... ... ... ... ... ... ... 55

Table 5.2: Descriptive statistics ... ... ... ... ... ... ... 56

Table 5.3: Augmented Dickey-Fuller test statistic ... ... ... ... ... 57

Table 5.4: Average growth rate of public expenditure ... ... ... ... 58

Table 5.5: Model fit and diagnostic check for average growth rate of public

Expenditure ... ... ... ... ... ... ... ... ... 60

Table 5.6: Determinants of public expenditure growth ... ... ... ... 62

Table 5.7: Model fit and diagnostic check for determinants of public expenditure

growth ... ... ... ... ... ... ... ... ... ... 67

Table 5.8: Regression result for determinants of real gross domestic product ... 68

Table 5.9: Model fit and diagnostic check for real gross domestic product ... ... 72

ix
 LIST OF FIGURES

Figure 1.1: Federal government annual expenditure 1961-2013 ... ... ... 5

Figure 1.2: Relative size of government sector ... ... ... ... ... 7

Figure 3.1: Government expenditure, economic growth and optimal size of public

Sector ... ... ... ... ... ... ... ... ... ... 22

Figure 3.2: BARS Curve ... ... ... ... ... ... ... ... 29

Figure 3.3: The three sides of public expenditure growth ... ... ... ... 35

Figure 5.1: Optimal level of federal government spending in Nigeria ... ... 74

x
 ABSTRACT

Public expenditure1 is not only used for stimulation of aggregate demand, it is also a useful
tool for redistribution of wealth and provision of public goods. Consequently, the size and
object of government expenditure becomes a leading policy issue of general concern in most
national and international gathering. To proffer the much needed professional guidance on
such occasion at the local level, this study therefore carried out economic analysis of public
expenditure growth and optimal size of public sector in Nigeria, with federal government as
the case study. The Ordinary Least Square econometric technique was adopted to guide the
study using Eviews 8 econometric software for data analysis. Annual time series data (1961-
2014) from Central Bank of Nigeria Statistical Bulletin 2009 and 2014 was used alongside
2014 World Development Indicators from the World Bank. Having performed necessary pre-
and post diagnostic tests, the results show that average annual government expenditure
growth was 1.91%. Again, tax revenue from oil sector was found to be the most influential
determinant of public expenditure with 32.66% contribution to its growth. Other influential
determinants were tax revenue from non-oil sector and population growth rate which
accounted for 24.29% and 22.49% respectively. It was also found that one of the core
objectives of minimizing the widely decried government wastefulness has not been
materialized as dummy for Fiscal Responsibility Act of 2007 could not influence growth rate
of public expenditure in Nigeria. Lastly, evidence of non-linear relationship between
government expenditure and economic growth was found in Nigeria as predicted by Armey
Curve. As a result, approximately 26.5% was found to be the optimal size of public sector in
Nigeria. It was concluded that above annual growth rate of government expenditure is not
harmful for economic growth; Nigeria remains oil dependent state in terms of government
spending; and that government expenditure still operates along growth part of the non-linear
curve. Among its recommendations was possible increase in government spending on sectors
with high multiplier effect, further diversification of revenue base away from oil and
promotion of transparency and accountability in government business.

1
Public expenditure (spending) and government expenditure (spending) were used interchangeably
to mean the same thing in this study.

xi
 CHAPTER ONE

INTRODUCTION

1.1 Background of the study

Economic analysis of public expenditure growth is all about how growth of public
expenditure over time influences overall welfare of the economy. It involves a critical
examination of trends over time and across components of public sector spending within the
economy. According to Brown and Jackson (1990), economic analysis is necessary for
effective evaluation of government performance in the course of carrying out its obligation in
modern economies. This argument anchors on the premise that public expenditure reflects the
policy choices of government, which directly or indirectly shapes the fortune of entire
economy. For Edame (2014), public expenditure refers to the expenses government incurs for
its own maintenance, society and the overall economy. It varies and ranges from defence,
general administration, health, education, electricity generation and supply, roads,
telecommunications and water supply among others.

Before 1936, the classical economic thinking that characterized most nation states was
heavily built around laissez faire ideology. This ideology placed minimum government
interference on economic affairs of individuals and state. It restricted the growth of
government expenditure to the need for maintenance of law and order, and defence against
external aggression. Public expenditure however, witnessed unprecedented growth
immediately after the emergence of Keynesian macroeconomic revolution in late 1930s. For
instance, Brown and Jackson (1990) quoted Peacock and Wiseman (1961) as revealing that
government expenditure as percentage of gross domestic product (GDP) in the United
Kingdom hovered around 9% and 15% between 1890 and 1914. It subsequently reached
about 48% during World War I (1914-1918) and declined to average of about 27%, until
1937 when it started rising astronomically. Following the outbreak of World War II (1939-
1945), government expenditure rose to about 60% of GDP in the U.K but declined to average
of about 44.9% from 1948 to 1987. The trend was similar across major economies, including
the U.S.A (Musgrave & Musgrave, 2004).

Existence of rapid growth of public expenditure and the need for its regulation led to some
theoretical postulations. Some of them include political instability inducing government
expenditure, also known as Peacock and Wiseman Displacement theory (Peacock and

1
Wiseman, 1961); stages of development growth theory of public expenditure (Musgrave,
1969; 1974 and Rostow, 1971); economic growth inducing government expenditure or
Wagner’s law (Bird, 1971); government decentralisation inducing government expenditure
which is also known as the Leviathan theory (Rodden, 2003); and revenue-spend theory
which suggests that the spending level in an economy should adjust equivalently to the
volume of revenue generated in the economy. All these were rooted in traditional economic
thinking of the classical, Keynesian, and Wagnerian schools of thought.

Economists recently however, observed potential consequences of unregulated rise in

government expenditure on economic growth. So far, a number of seasoned theoretical
studies support the claim that rapid increase in democratic institution and rising demand for
welfare state positively correlate with high profile growth in relative size of public sector
spending. In their inquiries, Barro (1989), Scully (1994 & 2003), Armey and Armey (1995)
and Rahn and Fox (1996) came up with similar argument that long-run relationship between
size of government sector and economic growth has a concave shape. It implies that when
government sector is very small, long-term economic growth can be accelerated through
increase in provision of public goods. The marginal economic growth will be positive but
decreasing as the size of government sector increases. It will become negative when
additional charges from increment of government sector became harmful to the benefits
resulting from increasing the productivity. The turning point at which increase in government
expenditure leads to maximum increase in economic growth is therefore known as the
optimal size (level) of public sector. According to Vedder and Gallaway (1998) and Facchini
and Melki (2011), it is the point at which increase in government spending (as proportion of
GDP) results to maximum increase in economic growth. However, exact position of this
turning point remains unknown and contentious among researchers. Mutascu and Milos
(2009) argue that whatever becomes an economy’s optimal size of public sector depends
largely on structural factors such as economic cycle, structure of public expenditure and fiscal
pressure. This fact notwithstanding, the importance of knowledge of optimal size of public
sector cannot be overstated in the face of increasing need for effective macroeconomic
stabilization policy in contemporary time. At the peak of its benefit remains provision of
insight to the government on how to strike balance between supply of public goods and
achievement of sustainable economic growth (Chobanov & Maldenova, 2009).

2
As a case study, Nigeria is a middle income country and mixed economy. She is an emerging
market with expanding financial, service, communications, technological and entertainment
sectors. According to Worldometers (2015), Nigeria worth about $568.51 billion in 2014
GDP and represents about 0.92% of world economy. She ranks 26th in the world and 1st in
Africa. In similar way, her population is estimated at about 177.8 million in 2014 as she ranks
7th in the world (Worldometers, 2015). The country has however witnessed unusual high
growth in government expenditure in the last two and half decades. From 1960 to 2013 for
example, government expenditure has grown from $291.48 million to $57.49 billion
respectively. This growth however, is yet to be translated into citizens’ welfare as more
Nigerians are poor today as was never before (Nasiru, 2012). Interested analysts such as
Omotoso (2010) and Ikeji (2011) blame this unwanted outcome on regular change in revenue
allocation and expenditure policy, which was widely accused of benefiting the federal
government at the expense of her federating units. Hence, justification for high growth in
federal government spending becomes a source of unending socio-political and economic
development debate in Nigeria.

Professional analysis of public expenditure growth and optimal size of the public sector now
becomes more relevant at a time when development challenges such as poor infrastructure,
high level of unemployment, insecurity of life and property are very common. These
challenges persist despite huge government expenditure budgeted annually to solve them.
Among several others, Okafor and Eiya (2011) argue that urbanization, degree of trade
openness, government revenue, inflation, external reserves, population density, type of
government regime, policy instability and public debt remain key growth drivers of public
expenditure in Nigeria. Sequel to this argument, diverse fiscal policy measures were adopted
by Nigerian government in the course of managing public expenditure. Some of these
policies were channelled towards reduction in total expenditure and diversification of revenue
mobilization strategy. Others include promotion of transparency cum accountability in
management of national resources. The quest for effective fiscal framework culminated in
enactment of Nigeria’s Fiscal Responsibility Act of 2007. This Act was aimed at minimizing
wasteful government spending, encouragement of transparency, consolidation of
accountability and promotion of overall fiscal discipline.

The problem of increasing government spending however remains with us despite above
control measures. For example, within the periods 1967-1970, 1973-1975, and 1980-1981,

3
public expenditure growth rate varied from 1.12% to 58.96%, 4.48% to 116.84%, and
102.09% to -23.75%; while economic growth rate fluctuated between -4.25% and -1.25%,
5.8% and 6.2%, and -3.2% and -2.9% respectively (Central Bank of Nigeria, 2009; 2012). In
recent time, growth in government expenditure declined from 21.48% in 2010 to -2.27% in
2012 and rose to 14.23% in 2013, while the economy grew at the rate of 7.84%, 6.75% and
7.31% at the same time. Correspondingly, if the claim by Vedder and Gallaway (1998) holds
that “no government is too little, but all-encompassing government is too much,” the question
that comes to mind becomes ‘what is the right size of government from the standpoint of
maximizing economic welfare in Nigeria?’ ‘Has the growth of public sector in Nigeria
proceeded too little, too much, or about right from the standpoint of increasing the output of
goods and services?’ ‘Should Nigeria expand government financial activities faster or slower
than growth rate of the economy so as to expand output of goods and services?’

These questions are yet to receive desired response from researchers. Apart from few studies
like Okafor and Eiya (2011) which estimated the determinants of growth in government
expenditure; Ekeoch and Oduh (2012), Oriakhi and Arodoye (2013), Olaleye, Edun, Bello
and Taiwo (2014) and Alimi (2014) which examined the optimal size of the public sector in
Nigeria, dozens of other studies were more interested and focused on impact of government
expenditure on economic growth. Interestingly enough, none of the above few studies care to
estimate the rate at which public expenditure grows in Nigeria. This knowledge gap makes it
difficult to establish dependable link between actual and expected rate of government
expenditure growth in the economy. Again, little or no attention was paid to the issue of
possible ways of checkmating growth of public expenditure. For this reason, the marginal
efficiency of rising components of government spending cannot be equally predicted. In
addition, detailed study on optimal size of public sector in Nigeria is scarce.

1.2 Statement of the problem

Nigeria has witnessed continuous rise in government expenditure due to growing receipts
from oil revenue, increased demand for public utilities and general effect of inflation on the
economy (Nurudeen & Usman, 2010). Again, the increasing need for security in the face of
rising civil unrest from militancy to terrorism also increased government expenditure. For
example, Central Bank of Nigeria (2012) and Budget Office of the Federation (2014) reveal
that government expenditure on defense were N444.6 billion, N233 billion, N264 billion,

4
N348 billion, N921.91 billion, N1,055 billion and N968.127 billion from 2008 to 2014
respectively. In addition, public sector analysts believed that steady rise in government
expenditure in Nigeria was underpinned by persistent rise in cost of general administration,
emolument of public office holders, misallocation of resources, government inefficiencies,
personal interest of politicians and bureaucrats, debt services and debt repayment. Available
statistics from Central Bank of Nigeria (2009; 2012) shows that total government expenditure
rose significantly in the last five decades. For instance, total recurrent expenditure rose from
N716.10 million, N4,085.20 million, N36,219.60 million, N416,600 million, N3,109,378.51
million to N3,365,760.00 million. While total capital expenditure was N187.80 million,
N10,163.30 million, N24,048.60 million, N239,450.90 million, N883,874.50 million and
N1,621,480.00 in 1970, 1980, 1990, 2000, 2010 and 2013 respectively. The image was
clearly painted in Figure 1.1 below.

Figure 1.1: Federal government annual expenditure 1961-2013

Source: Plotted by the researcher from CBN Statistical Bulletin, 2009 and 2012

Despite continuous rise in annual government spending as shown in Figure 1.1 above,
Nigeria still faces rising incidence of poverty and ranks among the poorest countries in the
world. Many Nigerians continued to wallow in abject poverty, while more than 50% live on
less than US$2 per day (Nurudeen & Usman, 2010). Apart from inadequate infrastructure

55
(especially roads and power supply) which impedes industrialization and promotes high level
of unemployment, macroeconomic indicators like balance of payments, inflation rate,
exchange rate, and national savings reveal that Nigeria has not fared well in the last couple of
years (Central Bank of Nigeria, 2009).

In attempt to finance rising expenditure given above situation, government may increase
taxes (borrowing) for instance. Higher income tax tends to discourage individual from
working for long hours or even searching for jobs. This, in turn, reduces income and
aggregate demand (Nurudeen & Usman, 2010). Similarly, higher profit tax may lead to
increased production costs and decreased investment expenditure as well as profitability of
firms. When firms faced downward trend in their profitability, they usually shut down
beyond their breakeven point. This action may result to reduction in national output, followed
by rise in unemployment rate. Risen unemployment rate will on the other hand, translate into
reduction in household income and leads to deterioration in standard of living among
citizenry. Furthermore, if growth in public sector spending is finance through borrowings
from the banks, such borrowings culminates in rising interest rate which will consequently
lead to rise in the cost of capital for the private sector. This, however, will crowd out
(compete away) private investment and deteriorate economic growth (Nwosa, Adebiyi &
Adedeji, 2013). Despite above chains of consequences of high government taxes and
borrowings, information from Debt Management Office as cited in Federal Ministry of
Finance (2014) suggest that total public debt maintained steady rise immediately after the
2005 debt forgiveness. It rose from US$21,398.91 million, US$25,817.42 million,
US$35,097.43 million, US$42,225.6 million to US$48,496.23 million during 2008-2012
fiscal year. It however, slightly decline to US$48,361.37 in 2013. Similarly, annual debt
services increased by 20.32% in 2014, from N591.76 billion in 2013 to N712 billion in 2014.
This increment represented a rise in outflow of resources which could have been invested to
generate employment for the populace.

In addition to the situation in which Nigeria finds itself, there exists another dimension to the
problem. It is the fact that expanding size of public sector due to increasing government
spending, at some point, leads to decrease in economic growth. Though Ighodaro and Oriakhi
(2010) and Egbetunde and Fasanya (2013) found increase in government expenditure on
socio-economic and physical infrastructures to encourage economic growth in Nigeria,
Glomm and Ravikumar (1997) and Ekeocha and Oduh (2012) contended that higher

6
government expenditure, if not regulated, may slowdown overall performance of the
economy. Similarly, Vedder and Gallaway (1998) argued that as government expenditures
grows incessantly, the law of diminishing returns begins to operate and beyond some point,
further increase in government expenditures contributes to economic stagnation and decline.
Vaish (2002) also supported this claim for two reasons. First, operations are often conducted
inefficiently; hence they reduce overall productivity of the economic system. Second,
excessive government spending distorts economic incentives and results in sub-optimal
economic decisions. This claim backs up Nurudeen and Usman (2010) who opined that
politicians’ and bureaucrats’ desire to remain in power in Nigeria and other developing
economies sometimes spurs them to increase government expenditure in unproductive
projects, or in goods that the private sector can produce more efficiently.

The debate on the relationship between government size and economic growth today has
shifted from ‘which sign’ (positive or negative) to ‘what is the optimal size of public sector’
(Mueller, 2003). The above shift draws attention to Figure 1.2 below as the question now
becomes, ‘at what percentage point would increase in government expenditure guarantee
maximum economic growth in Nigeria?’

Figure 1.2: Relative size of government sector

Source: Plotted by the researcher from CBN Statistical Bulletin, 2009 and 2012

77
Since it is desirable at all times that government expenditure stimulates growth and enhances
economic welfare, this study therefore considers it imperative to investigate the growth rate
and growth drivers of government expenditure in Nigeria. Moreover, there is growing need to
investigate credibility (effectiveness) of the Fiscal Responsibility Act of 2007. At the heart of
these needs stands out the need to estimate optimal size of government sector. Achievement
of this feat will give a new lease to effective policy-making in the country. It will help to
establish the needed link between current economic situation and expected government fiscal
disposition. This will help the economy establish a strong fiscal regime in order to maximize
economic welfare through effective government spending.

1.3 Research questions

In view of above stated problems, the following research questions will guide the study;

1) What is the average growth rate of public expenditure in Nigeria?

2) What are the determinants of public expenditure growth in Nigeria?
3) How has Fiscal Responsibility Act affected the rate of public expenditure growth in
Nigeria?
4) What is the optimal size of public sector in Nigeria?

1.4 Research objectives

The broad objective of this study is to analyze public expenditure growth and optimal size of
public sector in Nigeria. This can therefore be achieved with the following specific
objectives.

1) Ascertain average growth rate of public expenditure in Nigeria.

2) Investigate determinants of public expenditure growth in Nigeria.
3) Examine effect of Fiscal Responsibility Act on growth rate of public expenditure in
Nigeria.
4) Estimate the optimal size of public sector in Nigeria.

8
1.5 Research hypotheses

The aforementioned research questions prompted the following null hypotheses;

Ho1: Average growth rate of public expenditure in Nigeria is not significant.

Ho2: The effect of determinants of public expenditure growth in Nigeria is not significant.
Ho3: There is no significant effect of Fiscal Responsibility Act on rate of public
expenditure growth in Nigeria.
Ho4: Optimal size of public sector in Nigeria is less than 15%.

1.6 Scope of the study

The scope of this study is limited to carrying out an economic analysis of public expenditure
growth and estimate optimal size of government sector in Nigeria. The study will cover the
periods 1961-2014 using data on federal government public spending. The researcher intends
to obtain information on growth of public spending across different regimes in the country.
Although there are many factors that contribute to growth of government spending as were
found in empirical studies, this study shall utilize population growth rate, tax revenue (from
oil and non-oil sector), inflation rate, official external reserve, and public debt growth rate as
proximate variables for determinant of total government expenditure growth rate. Again,
public expenditure as percentage of GDP will be used alongside real gross domestic product
(RGDP) to estimate optimal size of public sector. Finally, a dummy will be generated to
investigate the impact of Fiscal Responsibility Act of 2007 on the rate of growth of
government expenditure in Nigeria.

1.7 Policy relevance of the study

The importance of this research draws its justification from the fact that it will serve as a
veritable instrument for policy articulation. It will also be of great relevance to institutions
and individuals. Specifically, its significance is expected to benefit the following:

The federal government will find this study very useful as it will provide valuable
information on determinants of growth in public sector spending as well as what the optimal
size of public sector should be in Nigeria. By so doing, unnecessary public sector spending
will be significantly curtailed, especially in the face of declining oil revenue. The study will

9
also be relevant in reversing the rising public debt and its servicing cost and thereby help to
free more resources of national development.

NGOs and International Development Agencies will also find this study important. It will
feed them with reliable information on key issues on government expenditure in Nigeria. In
addition, this study will add to existing knowledge on the scope of government spending in
Nigeria. It will provide the needed insight for researchers and students of public sector
economics as it tends to estimate optimal size of the public sector in Nigeria. The study will
examine determinants of government expenditure growth in the past 54 years.

10
 CHAPTER TWO

FISCAL POLICY AND GOVERNMENT EXPENDITURE GROWTH IN NIGERIA

This chapter intends to briefly review key government policies that contributed to growth in
federal government expenditure in Nigeria. A preamble of the structure of Nigeria’s fiscal
policy before independence will be presented and the rest will be discussed under the title
‘Post-independence and military era’, ‘Return to democratic era’ and ‘The search for optimal
size of public sector in Nigeria.’

2.1 Pre-independence era

Various principles have been recommended by the commissions/committees of revenue

allocation in Nigeria. These include basic needs, minimum material standards, balanced
development, derivation, equality of access to development opportunities, independent
revenue/tax effort, absorptive capacity, fiscal efficiency, minimum responsibility of
government, population, social development factor, equality of states, landmass and terrain
and internal revenue generation effort (Lukpata, 2013). There are two components of
revenue allocation formula used for the disbursement of Federation Account known as
Vertical Allocation Formula (VAF) and Horizontal Allocation Formula (HAF). The VAF
shows the percentage allocated to the three tiers of government i.e. federal, states and local
governments. This formula is applied vertically to the total volume of disbursable revenue in
the Federation Account at a particular point in time. For Bashir (2008), the VAF allows every
tier of government to know what is due to it; the federal government on one hand and the 36
states and 774 local governments on the other. Horizontal allocation formula is applicable to
states and local governments only. More attention was paid to VAF component in this study.

Before the amalgamation of Nigeria in 1914, the component units, the protectorate of
Northern and Southern Nigeria and the colony of Lagos - each enjoyed complete fiscal
autonomy. Again according to Ejeh and Orokpo (2014), before the political unification, a
unified fiscal system had already been operational until a centralized budgeting system was
introduced in 1926. However, with adoption of regionalism in 1946, a decentralized fiscal
structure was evolved. For this reason, fiscal arrangements were influenced by political and
constitutional factors before introduction of republican constitution in 1963.

11
Fiscal laws in Nigeria tend to give more powers to the federal government than other sub-
federal units combined. This is the key argument upon which continuous rise in federal
government expenditure was anchored in Nigeria. In fact, according to Omotoso (2010) and
Ikeji (2011) there is an increased dependence of sub-federal units on the federal government
particularly for their finances. State and local governments were neither given any strong
fiscal incentive nor encouraged to generate revenue internally. Hence, they are financially
weak and cannot effectively meet basic developmental needs of the local people. Regarding
revenue allocation and its importance for stable economic and political development in
Nigeria, Ikeji (2011) quoted the Report of the Political Bureau of 1987 thus;

Revenue allocation or the statutory distribution of revenue from the

Federation Account among the different levels of government has been one
of the most contentious and controversial issues in the nation’s political life.
So contentious has the matter been that none of the formulae evolved at
various times by a commission or by decree under different regimes since
1964 has gained general acceptability among the component units of the
country. Indeed, the issue, like a recurring decimal, has painfully remained
the first problem that nearly every incoming regime has had to grapple with
since independence. In the process, as many as thirteen different attempts
have been made in devising an acceptable revenue allocation formula, each
of which is more remembered for the controversies it generated than issues
settled (p. 121-122).

Given this situation, Ikeji contends that true federalism cannot be strengthen or guarantee. As
a result, there are discontentment, conflicts and agitation by the two other tiers against
wasteful and skyrocketing federal government expenditure that sap them financial strength to
perform their constitutional duty. Thus, for any federation to be sustained there must be fiscal
decentralization and financial autonomy but not fiscal centralization as the case in Nigeria.
Among factors that account for this include the growing importance of crude oil, the civil
war, military incursion into politics, centralizing tendencies of the military and state creation
exercises. These resulted to significant and steady rise in federal government expenditure
profile due to her large revenue mobilization and retention power in the nation’s fiscal policy
framework.

2.2 Post-independence and military era

Foundation of high profile growth of federal government expenditure in Nigeria was

reinforced between 1946 in which the nation adopted a federal constitution and 1975 which
recorded issuance of Decree No. 6, 1975 of the Murtala/Obasanjo regime. As noted above,

12
several factors played into the evolution of a more centralized revenue and expenditure
profile. They gave rise to the following recommendation of various commission, committee
and decree which was discussed by Ikeji (2011), Arowolo (2011) and Lukpata (2013). They
include;

§ Phillipson Commission, 1946: Derivation and even development.

§ Hick-Phillipson Commission, 1951: It recommended derivation, need, national
interest, fiscal autonomy of regions, special grants to regions for education, the policy
and capitation.
§ Chick Commission, 1953: Derivation and fiscal autonomy. Under this commission
most of the country’s revenue derived from exports and imports went to the regions
on the basis of derivation and consumption.
§ Raisman Commission, 1957: This suggested need, even or balanced development,
derivation and fiscal autonomy. It established a distributable pool account (DPA) and
allocated the following percentages of the DPA to the regions – North 40%, West
31%, East 24% and Southern Cameroon 5%. Raisman Commission was first to
introduce an objective formula for horizontal revenue allocation model among the
regions.
§ Binns Commission, 1964: Binns recommended the same principles as in Raisman,
but only changes the percentage as follows; North 42%, East 30%, West 20%,
Midwest 8%.

Shortly after the implementation of Binns Commission’s recommendation was the eruption
of political crisis that ushered in January 15, 1966 which mark the first military coup in
Nigeria. Military take over represented the end of what Omotoso (2010) called the “Old
federalism” of the first republic and establishment of a new federalism marked by increased
centralisation of political authority, ascendancy of federating forces, greater structural
differentiation of the constituent states, increases in federal government largesse, and
expansion of policy-making and execution functions of the federal civil service. The
centralisation tendency went into formulation of national development planning strategies in
which federal government desired to occupy the ‘commanding height’ and provides
leadership and administration necessary to achieve national objectives. Similarly, as part of
its centralisation policy, the military regime ensured a progressive concentration of fiscal
power in the federal government. This was as a result of the windfall in oil-generated wealth.

13
The phenomenon then conditioned entire political economy of the country as every political
component of the state depends on oil-resource federation account. According to Donald
Williams (1992), as quoted by Olaopa (2013),

Over the period 1969-1979, the federal government expenditure rose about
twenty times from ₦548.2 million to ₦18.5 billion. The extent to which the
federal government assumed superiority over fiscal policy in relation to the
state government is illustrated by the fact that federal spending rose from 60%
of all outlays in 1968/69 to approximately 87% by 1980/81 (p. 37).

Hence, intervention of the military can be define as one of the key consolidating factor
behind fiscal centralization and the consequential fast growing federal government
expenditure in Nigeria. Below are recommendations of various decree, commission and Act
that help in shaping the nation’s fiscal environment during military regime and civilian
regime before the return of democratic rule in 1999.

§ Decree No. 15 of 1967: In this place population was added to Raisman formula. Each
of the six northern states was to receive 7%. The remaining 58% was to be shared
among the six states in the south according to population.
§ Dina Commission, 1968: Dina upheld principles of derivation, need, even
development, special grants and added minimum responsibility of government or
equality of states. It also introduced a special grants account and recommended
establishment of a permanent revenue planning and fiscal commission.
§ Decrees No. 13 of 1970; No. 9 of 1971; and No. 6 of 1975: Under these decrees,
population and equality of states received 50% of share of the DPA each, that is, 50%
(population), 50% (equality). Again, following post war reconstruction, rehabilitation
and reconciliation policy, it increased share of the federal government because of
increased responsibility. Decree No. 9 of 1971 & No. 6 of 1975 reduced mining rents
and royalties from 45% to 20%, gave more functions to federal government,
particularly in education and all erstwhile regional universities became federal and
federal also undertook universal free primary education.
§ Aboyade Commission, 1977: Having rejected the five point criteria, the commission
recommended federation account, into which all federally collected revenue
(excluding personal income tax of the personnel of the armed forces, the police,
external affairs ministry, residents and non residents of the Federal Capital Territory)
would be paid and shared among the three levels of government. Vertical allocation

14
 formula recommended was 57% federal, 30% state, 10% local governments and 3%
special grants account. Horizontal formula suggest equality of and access to
development opportunities 25%, minimum standard for national integration 22%,
absorptive capacity 20%, independent revenue and minimum tax effort 18% and fiscal
efficiency 15%. However, vertical formula recommendation was accepted, but
horizontal formula was dubbed ‘too technical’ and rejected.
§ Okigbo Commission, 1980: The commission revised revenue allocation formula as
follows – 55% federal, 30% states, 8% local governments, and 7% special funds. On
horizontal revenue allocation formula, it was as follows; population (40%), national
minimum standard for national integration (40%), social development factor (15%),
and internal revenue effort (5%). But it should be noted that Supreme Court declared
it null and void which paved the way for Revenue Act of 1981.
§ Revenue Act, 1981: this Act revised the revenue allocation formula of Okigbo
commission as follows – 55% federal, 30.5% for states, 10% local government, and
2.5% special funds.
§ Decree No. 36 of 1984: Muhammadu Buhari led military regime revised the revenue
allocation formula as follows – 55% federal, 30% states, 15% local government, and
2.5% special funds.
§ National Revenue Mobilization, Allocation and Fiscal Commission (NRMAFC),
1989: The vertical allocation formula was revised as follows – 47% federal, 30%
states, 15% local government, and 8% special funds. However, the military
government approved 50% federal, 30% states, 15% local government, and 5%
special funds. On horizontal formula – equality of states (40%), population (30%),
internal revenue effort (20%), social development factor (10%), government approved
this but added 10% for landmass and terrain.
§ Armed Forces Ruling Council (AFRC), 1993: This decision-making body under
General Abacha revised the revenue allocation formula as follows – 48% federal,
24% states, 20% local government, and 7.5% special funds; equality of states (40%),
population (30%), internal revenue effort (10%), social development factor (10%),
landmass and terrain (10%).

15
2.3 Return to democratic era

With enthronement of democracy in 1999, controversies regarding the country’s fiscal

operations took a new turn with federal government being accused by oil producing states for
not honouring the derivation principles as stated in 1999 federal constitution (Ejeh & Orokpo,
2014). This made the federal government to introduce the on-off shore dichotomy implying
that oil found in the sea cannot be ascribed to the adjourning state. With this on ground, states
in Niger Delta led the struggle for resource control and sued the federal government with the
matter ended up in the Supreme Court. In defence of the federating units, NRMAFC rejected
on several occasions the interference of the president and Federal Ministry of Finance on the
formula for revenue sharing and insists on proper interpretation of the constitution. For
example in January 2004, the Federal Ministry of Finance in a letter to the commission gave
federal government a share of 54.68% and a grant of 2% of the states. The NRMAFC
disagreed with the ministry of its non-compliance with provisions of section 164(1) of the
1999 Constitution (Ekpo, 2004).

In 2001, the fiscal body drafted proposal with the sharing formula; the federal 41.3%, state
31%, local governments 16% and special fund 11.7%. However, this particular proposal was
crippled following the Supreme Court pronouncement on resource control in April 2002 and
at the same time rendered allocation to special fund unconstitutional. Following this
development is similar revenue allocation proposal and recommendation as follows;

§ Executive Order, May 2002: Federal 56%, state 24% and local government 20%.
§ Executive Order, July 2002: Federal 54.68%, state 24.72% and local government
20.60%.
§ RFMAC Proposal, January 2003: Federal 46.63%, state 33% and local government
20.37%.
§ RMFAC Proposal, September 2004: Federal 47.19%, state 31.1% and local
government 15.21%. Special funds was renamed ‘National priority services funds’
with General Ecological Fund (1.50%); Solid Minerals Development Fund (1.75%);
National Agricultural Development Fund (1.75%) and National Reserve Fund
(1.50%). These comprise a total of 6.50%.
§ Presidential Proposal, January 2005: This is the same as RMFAC’s proposal of
2004. It only added to horizontal formula and empowers state derivation funds board

16
 to manage 13% derivation (Omotoso, 2010; Arowolo, 2011; and Ejeh & Orokpo,
2014).

By the year 2008, the fiscal body had a new proposal for revenue sharing table before the
National Assembly. It had proposed 53.69% for the federal government. Hence, it is obvious
that Nigerian fiscal federalism has unduly benefited the centre in revenue allocation and
empowered it to continually increase the volume of its expenditure. To buttress this point,
some researchers contend that various criteria used for allocating resources among the three
ties of government in Nigeria were not based on rational consideration but on the basis of
other primordial considerations. Based on this line of thought, Sagay (2008) argued that ‘the
federal government has no business having more than 35% of the revenue since it has little
work to do. This unmerited favour to the federal government according to Sagay (2008),
‘accounted for why it can afford to spend money anyhow’. The federal government, on the
other hand cannot be dissociated from this revenue allocation ‘pathology’ since it
appropriates and concentrates too much money at the centre leading to waste and corruption.

2.4 The search for optimal size of public sector in Nigeria

The fact that governance represents more than a means of providing common good cannot be
understated. Governance can be related to government capacity to help the citizens achieve
individual satisfaction and material prosperity. This responsibility suggests that governance
also requires financial cost. Factors responsible for the rising cost of governance in Africa
and Nigeria in particular include but not limited to undue inflation of project cost by
government agents, undue equality in representation of multiple ethnic groups, embezzlement
and mismanagement of public funds, high profile population growth and extra large civil
service sector (Fluvian, 2006; Adewole & Osabuohien, 2007; Warimen, 2007; and
Ejuvbekpokpo, 2012).

Since resources are scarce and should be optimally utilized, it entails that government
expenditure should not be excluded from factors that calls for optimal usage. Again, the role
of government expenditure in simulation of economic activities in modern states and
associated consequences of its abuse on peoples’ welfare now compels people to demand the
search for optimal size of the public sector with respect to its expenditure. The argument
behind above demand was the need to regulate the wave of rising government wastages and
dominance of political class interest over common welfare of the masses. A good example for

17
justification of the search in Nigeria has to do with above illustration on the ever increasing
revenue and expenditure profile of the federal government and countless cases of money
laundering allegation on government agents. The question now becomes what should the
optimal size of public sector be when compared to private sector. Hence, ascertaining the
optimal size of government is necessary according to Barro (1990). This is because of the
strong theoretical assertion which was backed by empirical and statistical evidence that as
government expenditure grows, the law of diminishing return sets in. Until at a point,
government expenditure will become negatively related to economic growth.

18
 CHAPTER THREE

LITERATURE REVIEW

3.1 Conceptual framework

Public expenditure refers to government expenditure or government spending. It is incurred

by central, state and local governments of a country. Conventionally, public expenditure has
to do with the expenditure incurred by public authorities like central, state and local
governments to satisfy the collective social wants of the people. For Leland (2005), public
expenditure can be defined as money allocated by governments for the provision of public
goods and services. It is the value of goods and services bought by the state and its
articulations (Piana, 2001). In democracy, public expenditure is an expression of people's
will, managed through political parties and institutions. It can be financed through taxes,
public debt, money emission and international aids. According to Brown and Jackson (1990)
total public expenditure is the sum of expenditure on current and capital account of the
public sector and is by definition equal to the sum of consolidated public sector receipts. In
their classification, total public expenditure is made up of public sector consumption, public
sector investment, subsidies, current grants, capital transfers, debt interests and net lending to
the private sector and overseas. Similarly, Rutherford (2002) conceived public expenditure as
something that has to do with the expenditure of central, regional and local governmental
organizations on intermediate and final goods and services. Such expenditure is undertaken to
achieve a variety of goals including the redistribution of benefits in kind, the provision of
public goods, correction of disequilibria in markets and the regulation of industry.

This study however adapts above definition of Piana (2001) but limits such expenditure to
that incurred by the federal government in Nigeria. Thus public expenditure is define as the
total monetary equivalent of commodities bought and transfer payments by federal
government and its agencies for the provision of public goods and services. It is directed
mainly towards satisfaction of collective wants of the society.

Hence, the growth in public expenditure has to do with rising trend in above define
government spending over time. It can also be seen as growth on the amount spent by
government in the course of its duty and responsibility to the citizens and aliens. Concept of
public expenditure growth can best be illustrated with the emergence of Keynesian economic
philosophy of the 1930s. Prior to this development, the size of public expenditure was very

19
small throughout the 19th Century when most governments followed laissez faire economic
policies. Their functions were restricted to defending aggression and maintaining laws and
order. However, the early 20th Century witnessed sharp rise in government expenditure
everywhere due to sudden breakthrough of Keynes’ ideology of government expenditure as
the ‘missing wheel.’ In several countries however, the increasing ratio of public expenditure
to the national income has excited political debate. The ratio increases whenever a collectivist
ideology influenced policy-makers or special interest groups succeed in obtaining public
subsidization. But in many economies, the increase can be attributed to an expanding or
ageing population, to public sector output being relatively more costly than production in the
private sector, or to a greater use of bureaucratic methods (Rutherford, 2002). Thus, public
expenditure growth is a function of increasing trend in the size of public expenditure resulting
from initial positive correlation between public expenditure and economic welfare in modern
states.

In attempt to define the public sector, System of National Accounts (1993) in Chobanov &
Mladenova (2009) maintains that “The general government sector consists of the totality of
institutional units which, in addition to fulfilling their political responsibilities and their role
of economic regulation, produce principally nonmarket services (possibly goods) for
individual or collective consumption and redistribute income and wealth.” Similarly, optimal
size of public sector was defined to be the total amount of public expenditures as percentage
of GDP which ensures maximum economic welfare at any given point within the economy
(Ekinci, 2011). Theoretically, the relationship between optimal government expenditure and
economic growth has been associated with Armey curve.

According to Altunc and Aydin (2013) and Olaleye, et al. (2014) the concept was built on the
Laffer curve by Armey and Armey (1995) to theorize the effect of government interference in
relation to economic growth. The Armey curve demonstrates the relation between
government expenditure and economic development and hypothesizes that an optimal size of
government expenditure exists at the threshold of government expenditure which guarantee
optimal rate of economic growth. And any further increment in relative size of government
expenditure will have negative effect or slowdown economic growth (Thanh, 2009). Thus for
the purpose of this study, optimal size of the public sector was define as the point at which
relative size of government expenditure as percentage of GDP guarantees maximum
economic growth measured as relative change in RGDP overtime. Most empirical studies of

20
developing countries suggest that public expenditure policy not only accelerates economic
growth and promotes employment opportunities. It also plays useful role in reducing poverty
and inequalities in income distribution. Going by claims of Piana (2001), public expenditure
plays four main roles such as; (a) contribution to current effective demand; (b) expression of
coordinated impulse on the economy which can be used for stabilization, business cycle
inversion, and growth purposes; (c) increment of the public endowment of goods for
everybody; and (d) leads to a rise in positive externalities to economy and society as a whole
(or in specific sectors and geographical areas), especially its capital component.

Given the above roles, it becomes necessary that government expenditure should be
encouraged to grow. These facts notwithstanding, theoretical literature concerned with impact
of public sector spending on economic growth suggest that when public expenditure is zero,
all the goods and services to be produced by the state are offered by the private sector instead.
The rate of growth would be very low in this scenario due to existence of market failure.
When the amount of public expenditure rises, it triggers economic growth but only up to an
optimal level. Any further increase has a negative impact which comes as a result of
government failure. This is true when the size of public sector becomes too large in such a
way that government begins production of those commodities which it has less comparative
advantage than the private sector. Put differently, when government expenditure is below
4.55% of GDP according to Ekinci (2011), the economy may likely fall into the trap of
possible recession as was noted by Keynes (1936). When government expenditure
approaches 100% however, the economy may slip into the problems of command economy
such as lack of enterprises, lack of freedom of choice among consumers and poor economic
performance as was seen in the old Soviet Union and similar economies.

The consequences associated with either of the extreme positions become the harbour upon
which the argument for estimation of optimal size of public sector anchored in this study. It
becomes necessary therefore, to identify a point at which the combination of public sector
and private sector expenditure guarantees highest possible level of economic growth and
welfare in the society. The unknown ideal point is what was defined in this study as optimal
size of public sector in Nigeria. Figure 3.1 below illustrates the link between public
expenditure, economic growth and optimal size of the public sector.

21
Figure 3.1: Government expenditure, economic growth and optimal size of public sector

Economic growth
Panel A
A
1

B
1
F
E
G
*
U4
U3
U2
C U1
U0
D
O
Size of private sector
Economic growth

G •
*

Panel B

Government expenditure

O
Size of public sector
Economic growth

F Growth path
E
•
•
G •

Panel C
B •

Public Private
A •
sector sector

O G
*
Size of the economy

Source: Researcher’s conceptualization

Panel A, B and C represented activities of the private sector, public sector and the entire
economy respectively. From microeconomic assumptions of profit maximization, the private
sector can only produce at points G, E and F in Panel A. This is because the points ensure
profit maximization. For this same reason, firms can afford to invest more resource at any
point along Indifference curve U2 and beyond. On the other hand, firms cannot produce at

22
any point along Indifference curve U0 and U1 because points A, B, C and D are below the
firm’s profit maximizing points. Thus the distance |OG| along the horizontal axis was left for
public sector investment since it represented the provision of pure public goods which the
private sector cannot effectively produce in the economy.

The existence of line |OG| in Panel B suggests that government expenditure is necessary for
economic growth. As size of the public sector rises form point O to B through A, the
economy grows. This positive relationship continues until at point G where the marginal
increase in economic growth resulting from a unit increase in size of the public sector
becomes zero. Beyond this point, additional increase in public sector leads to a decrease in
economic growth. Thus, the shaded region represents proportion of the economy which
permits expansion of only the relative size of the private sector to ensure steady growth of the
economy.

Panel C combines output of Panel A and B. It summarizes the link between government
expenditure, economic growth and optimal size of public sector. Government expenditure is
necessary for economic growth. Line |OG| along the horizontal axis represents the provision
of pure public goods which can only be optimally produced by government. Furthermore,
point G that separated the economy into public and private sector represents the provision of
mixed goods in the economy. Both government and private sector can jointly produce at that
point. Therefore, point G represents the optimal size of the public sector above which
government spending is not expected to exceed. Beyond point G stands the provision of pure
private goods. Only the private sector is expected to expand along this path. The combination
of public and private sector contribution will ensure economic growth along the growth path
in Panel C. In summary, this study intends to investigate optimal size of federal government
sector ably represented by point G in Panel B and C of above Figure 3.1

3.2 Theoretical literature

The macromodels that will be considered in this section differs from short-run
macroeconomic forecasting models in their treatment of public expenditure, in so far as the
latter take government expenditure as exogenously determined. The macromodels that
interest us seek to explain how government expenditure has behaved over the long term by
analysing the time pattern of public expenditure.

23
3.2.1 Rostow and Musgrave stages of development growth theory

This approach to explanation of public expenditure growth is best represented by the works
of Musgrave (1969; 1974) and Rostow (1971). This theory defined three phases that every
economy will undergo in their economic development cycle. They include early stage, middle
stage and maturity stage. According to the theory, in the early stages of economic growth and
development, public sector investment as a proportion of the total investment of the economy
is found to be high. The public sector is, therefore, seen to provide social infrastructure
overheads such as roads, transportation systems, health, law & order, education and other
investment in human capital. This public sector investment, it is argued, is necessary to gear
up the economy for take-off into the middle stages of economic and social development.

At the middle stages of growth, the government continues to supply investment goods but
this time public investment is complementary to the growth in private investment. After the
second stage the increasing need in high income societies for skilled labour leads education to
become increasingly an investment good for society as a whole. Increased population
movement leads to the development of urban slums. Such factors and others lead once again
to an increase in public expenditure in relation to total output. According to Rostow, once the
economy reaches the maturity stage the mix of public expenditures will shift from
expenditures on infrastructure to increasing expenditure on education, health and welfare
services. During all the stages of development, market failures exist which can frustrated the
push towards maturity; hence the increase in government involvement in order to deal with
these market failures.

In the context of above theory, Nigerian economy can be placed within middle stage of
economic and social development where public expenditure increase is being augmented by
expanding private sector in the provision of infrastructure and corporate social responsibility.
This is evidently true in Nigeria where decomposition of annual federal government budget
suggest that about three-quarter of the expenditure goes into maintenance of her workforce
and provision of infrastructure.

24
3.2.2 Wagner's Law of expanding state activity

The 19th Century German economist, Adolf Wagner, was concentrated to explain the share of
gross national product (GNP) taken up by the public sector (Brown & Jackson, 1990).
Though he never intends to state his idea as a law, it was formalize by Bird (1971) in his
Wagner's law of expanding state activity. In its simple form, Wagner’s law states that as per
caput income in an economy grows, the relative size of the public sector will grow also. The
theory perceives the state as existing independently of the individuals in society and has a
general responsibility for society as a whole. This contrasts strongly with the standard
neoclassical view that the state should merely reflect the views of individual citizens.

It went further to list three functions of the state as provision of administration and
protection; ensuring stability; and provision for the economic and social welfare of society as
a whole. According to Wagner in Bird (1971), public expenditure on the first function of the
state (which included law, order and defence) would grow because the increasing division of
labour would lead to the breakdown of communal relationships and requires the state to take
over functions previously done by families and local communities. In the process,
administration would become more centralised and administrative units larger. The increased
division of labour would be accompanied by the development of new technological processes
which would lead to the growth of monopolies in the private sector.

In his view, private sector monopolies would not adequately accommodate the social needs of
society as a whole and would therefore need to be replaced by public corporations. Again, if
private sector firms became too large, the economy would become unstable because problems
for individual firms would become problems for the whole society. Finally, government
would need to expand to provide social benefits and services which Wagner saw as not open
to economic evaluation. The theory recognized that as an economy became industrialized the
nature of the relationships between the expanding markets and the agents in these markets
would become more complex. This complexity of market interaction would result in a need
for commercial laws and contracts, which would require the establishment of a system of
justice to administer such laws and regulate the system. Urbanization and high-density living
would also result in externalities and congestion requiring public sector intervention and
regulation, which will increase government expenditure and therefore size of the public
sector.

25
Comprehensive review of trends in Nigerian government expenditure growth reveal that the
economy has witnessed consistent breakdown in communal relationships. Subsequently,
there has been a more centralised and larger administrative unit as was predicted by Waganer.
This example was illustrated in empirical literature discussed in section 2.4.

3.2.3 Peacock and Wiseman displacement effect model

This theory was championed by Peacock and Wiseman (1961) with the central message that
political instability induces government expenditure and left incidences of ratchet effect on
the entire economy. The authors perceive taxation as setting constraint on government
expenditures. According to the theory, as the economy and thus income grows, tax revenue at
constant rate would rise and thereby enable public expenditure to grow in line with GNP. In
normal times, therefore, public expenditure would show a gradual upward trend, even though
within the economy there might be a divergence between what people regarded as being a
desirable level of taxation. During periods of social upheaval, however, this gradual upward
trend in public expenditure would be disturbed. These periods would coincide with war,
famine or some large-scale social disaster which would require rapid increase in public
expenditures.

In order to finance the increase in public expenditures the government would be forced to
raise taxation levels. This raising taxation levels would therefore, be regarded as acceptable
to the electorate during periods of crisis. This phenomenon was referred to as the
displacement effect. Public expenditure is displaced upwards and for the period of the crisis
displaces private expenditure. The process represents an upward shift in the trend line of
public expenditure. Following the period of crisis however, public expenditure does not fall
to its original level (the ratchet effect incidence). This is because of inspection effect which
they suggest arises from voters’ keener awareness of social problems and more needs for
government intervention during the period of upheaval. The government therefore expands its
scope of services to improve these social conditions. Again, the fact that electorates’
perception of tolerable levels of taxation does not return to its former level encourages the
government to finance these higher levels of expenditure originating in the expanded scope of
government and debt charges. Finally, the crisis also leads to an increase in the concentration
of power in the hands of central government and this is also not reversed after the crisis. In
the views of the proponents, such crisis like war and depression is not fully paid for from

26
taxation. Countries have to borrow and debt charges have to be met after the event. All these
result to increase in public expenditure and relative size of the public sector.

Several of the ideas above and some others were grouped by Peacock and Wiseman under the
heading ‘permanent influences on government expenditure from the changing nature of
society.’ They were listed to include the level of development, population size and the age
composition of the population, social insurance, increased mobility of society, and effects of
war, war-related and defence expenditure. They further contend that increased output per
capita (affluence) results to the breakdown of the extended family and increasing population
lead to growth of conurbations and urbanisation which is associated with increasing
interdependence and externalities as was noted in Wagner’s law above.

Once again, Nigeria economy partly reflects predictions of this theory. Trends in federal
government spending respond to sudden changes in socio-political and economic atmosphere
in the country such as periods of crisis (religious, political and communal), outbreak of
epidemics and economic downturns.

3.2.4 Niskanen model of budget maximizing bureaucrats

The economic analysis of budget maximizing bureaucrats and how they contribute to growth
of government spending was formally established by Niskanen (1968) and was followed up
in his book Bureaucracy and Representative Government in 1971. The bureaucrat is assumed
to be a monopoly supplier and the legislature to be a monopoly buyer (monopsonist) in the
production process of public goods through the state budget. Thus the relationship between
the two groups is one of bilateral monopoly. The bureaucrat attempts to obtain as large a
budget as possible whilst the voters intend to maximize benefit from such budget. In its
simple form, the model suggests that public sector bureaucracy has the tendency to expand
output of the public sector to twice the level of output that would be regarded as socially
optimal. If we assume a relationship between public output and budget size (public
spending), then it follows that the size of public sector is twice its social optimum. Niskanen
used simple microeconomic argument to support this view. In his analysis, ‘a private firm
selling products and attempting to maximize profits will increase its output only to the point
where marginal revenue equals marginal cost, but a public sector bureau providing services
free of charge will go on providing the product until the marginal benefit is equal to zero.

27
Hence it will over-provide the product and its unit costs will be higher than those of the
private firm’.

How does this come about? The suppliers of public services (bureaucrats) have greater
knowledge about the details of production and cost functions than their sponsors (their
political pay-masters), i.e., those who provide the budget. Knowledge is asymmetric or
impacted in this case. Thus when negotiating a budget, the bureaucrat is able to extract
consumer surplus (the difference between the total benefit and total cost curves) and divert it
into more output and hence a bigger budget. Since the bureaucrat’s utility is assumed to be an
increasing function of budget size, there is an incentive to behave in this way (Brown &
Jackson, 1990). Since the utility of the bureaucrat is an increasing function of budget size, it
therefore implies that government spending will rise relative to the budget size.

This model explain what happens to annual federal budget in Nigeria given evidence of
highly inflated costs, over supply and duplication of certain items, allocation for unnecessary
and sometimes ambiguous items. It also include rising allocation for overheads and
emoluments of public servants and their wards. All these translate to rising government
expenditure, especially as it relates to recurrent expenditure.

3.2.5 Theoretical background on optimal size of government and the BARS curve

The size of the public sector varies greatly around the world, and so do its responsibilities.
What in some countries are considered a necessity of supply of public goods may in other
countries be seen as an act of interference in private sector’s responsibility. Empirical
evidence has shown that up to a point, government expenditure will boost economic growth,
but past some point, the extra spending is mostly wasted as it becomes negatively related to
economic growth. It suggests that the relationship between government spending and
economic growth is non-linear, though many researchers established both positive and
negative linear relationship in the past. Behind contribution to this theoretical knowledge are
Barro (1989) Armey and Armey (1995) Rahn and Fox (1996) and Scully (1994, 2003) whose
theoretical and empirical studies popularized the existence of an optimal size of government
as depicted by a concave shape or an inverted “U” curve. It was referred to as ‘BARS Curve’
after Barro, Armey, Rahn, and Scully (Chobanov & Mladenova, 2009; Alimi, 2014).

28
BARS argue that non-existence of government causes a state of anarchy. Hence, it suggests
that absence of rule of law, protection of property rights and lack of collective infrastructure
would lead to poor productivity. The implication is that there will be little incentive to save
and invest, which will result to low levels of wealth creation. In likewise manner where all
economic decisions are made by government, output per capita is also low (Pevcin, 2004). It
however suggests that no economy can achieve significant level of economic growth without
government intervention and contribution. The key message conveyed by BARS curve
remains that excessively large governments have tendency to become increasingly less
productive, and thereby reducing output growth (Alimi, 2014). Therefore, output would be
high when there is optimal mix of private and government initiative regarding the allocation
of resources, which according to Facchini and Melki (2011) will enhance economic growth.
Within this framework, government involvement in the economic process is a necessary but
not sufficient condition for growth. Government can have a positive influence on economic
growth until a certain scope and beyond it, it can become harmful.

Figure 3.2: BARS Curve

Economic growth

O G Size of public sector (as % of GDP)

*

Source: Chobanov & Mladenova (2009)

As relative size of public sector in Figure 3.2 expands from zero (complete anarchy) along
the horizontal axis, the economy will grow along the vertical axis. This implies that as
relative size of government grows, expenditures are channelled into less productive (and later
counterproductive) activities, causing the rate of economic growth to diminish and eventually
decline (Chobanov & Mladenova, 2009). The curve has a concave shape due to decreasing

29
marginal returns. This suggests that proportional increase in government spending yields less
than proportional increase in economic growth. Alimi (2014) pointed out that with accrued
positive externalities, additional percentage rise in government contribution to economic
activities will still create higher economic efficiency in form of positive slope. However, at
point G in above figure the marginal benefit from increased government spending becomes
zero.

Several attempts were made to determine the level of government spending at which growth
is optimized. While an expansion in size of government usually results in greater provision of
services according to Ekeocha and Oduh (2012), it can also lead to slower rates of growth
because of greater bureaucracy and the crowding-out of private sector driven initiatives.
Growth-enhancing features of government begin to diminish when adverse effect of
government expansions result in reduction of output growth (Herath, 2012). This growth-
enhancing region according to Friedman (1997) was assured at any point in between 0.1% to
14.9%. He contends that average contribution of the public sector in the economy is positive.
As public share of national income increases from 15% to 50% the marginal contribution of
public sector will be negative. Thus based on development level of countries, the optimal
level of public spending should be between 15% and 50%.

Pevcin (2004) argue that with the increasing trend on promotion of concept of welfare state,
the size and role of government in the economy have increased in most countries leading to a
significant proportion of the national output. Existing literature considered the advantages
and disadvantages of having a government that is much larger or much smaller than what is
required. It stressed the need for determination of optimal size and structure of government
for maximising economic benefits (Ekeocha & Oduh, 2012).

3.2.6 Classification of public expenditure

Classification of public expenditure refers to the systematic arrangement of different items on

which the government incurs expenditure. Different economists looked at public expenditure
from different points of view which include;

§ Functional classification: This has to do with classification of public expenditure on

the basis of functions for which they are incurred (Connolly & Munro, 1999). The

30
 government performs various functions like defence, social welfare, agriculture,
infrastructure and industrial development.
§ Revenue and capital expenditure: Revenue expenditures are current expenditures or
consumption expenditures incurred on civil administration, defence forces, public
health, education and maintenance of government machinery. This type of
expenditure is of recurring type which is incurred yearly. On the other hand, capital
expenditures are incurred on building durable assets like highways, irrigation projects,
buying machinery and equipment. They are non recurring type of expenditures in the
form of capital investments. Such expenditures are expected to improve the
productive capacity of the economy (Musgrave, R & Musgrave, P, 2004).
§ Transfer and non-transfer expenditure: According to Pigou (1928) and Egbetunde
and Fasanya (2013) transfer expenditure relates to the expenditure against which there
is no corresponding returns. Such expenditure includes public expenditure on national
old age pension schemes, interest payments, subsidies, unemployment allowances,
welfare benefits to weaker sections, etc. By incurring such expenditure, the
government does not get anything in return. It adds to the welfare of the people,
especially those that belong to non-labour force and results in redistribution of money
incomes within the society. The non-transfer expenditure relates to expenditure which
results in creation of income or output. They include development (expenditures that
promote economic growth and development) as well as non-development expenditure
that results in creation of output directly or indirectly. Economic infrastructure such as
power, transport, irrigation; social infrastructure such as education, health and family
welfare; internal law and order and defence; public administration etc were all
included. By incurring such expenditure, the government creates a healthy
environment for economic activities.
§ Productive and unproductive expenditure: This was classified by the classical
economists on the basis of creation of productive capacity. Productive expenditure is
all about expenditure on infrastructure development, public enterprises or
development of agriculture. These lead to increase in productive capacity of the
economy and yield income to the government. Unproductive expenditure accounts for
expenditures in the nature of consumption such as defence, interest payments,
expenditure on law and order, and public administration. It does not create any
productive asset which can bring income or returns to the government directly.

31
 § Benefit incidence classification: Public expenditure can be classify on the basis of
benefits they confer on different groups of people such as (a) Common benefits to all-
Expenditures that confer common benefits on all the people like expenditure on
education, public health, transport, defence, law and order, general administration; (b)
Special benefits to all-Expenditures that confer special benefits on all. For example,
administration of justice, social security measures, community welfare and; (c)
Special benefits to some-Expenditures that confer direct special benefits on certain
people and also add to general welfare such as old age pension, subsidies to lower
class of the society, unemployment benefits (Brown & Jackson, 1990).
§ Hugh Dalton's classification: Dalton (1920) classified public expenditure into
expenditures on political executives-to maintain ceremonial heads of state, like the
president; administrative expenditure-to maintain the general administration of the
country, like government departments and offices; security expenditure-to maintain
armed forces and the police forces; expenditure on administration of justice-include
maintenance of courts, judges, public prosecutors; developmental expenditures-to
promote growth and development of the economy, like expenditure on infrastructure,
irrigation; social expenditures on public health, community welfare, social security,
etc; and public debt charges-payment of interest and repayment of principle amount.

3.2.7 Classical reasons for increased government expenditure

Many factors contributed to the rapid and continual growth in public expenditures. States are
not only undertaking new activities, but also entering into former activities more extensively.
The mere fact that population is becoming more dense has caused a number of expenditures
to increase. On the whole, governments are institutions of increasing cost. A large part of its
services cost more per capita as population increases. Key factors that contribute to increased
government expenditure in Nigeria and other countries were comprehensively summarized by
UNESCO-Nigeria Project (2010) as follows:

§ Evolution of welfare state: This changed the role of public expenditure radically. The
state now provides social security scheme such as highly subsided or free education,
medical aid, old age provisions, provident funds, accident benefits etc.

32
 § Expansion of public sector: Most governments expanded the scope of public sector
undertakings. They now participate in various economic sectors like manufacturing,
trade and commerce, banking and agriculture.
§ Increase in population: Most countries especially the developing ones have
witnessed phenomenal increase in population. This necessitated the provision of
increased educational and health services as well as provision of employment
opportunities, housing and public utilities.
§ Economic development: This function has become important in developing countries
where there is acute shortage of entrepreneurial skill and private financial capital.
Thus, the government becomes important player in the agricultural and industrial
sectors.
§ Increase in the number of democratic institutions: There has been tremendous
growth in democratic institutions at various levels of government. Such democratic
institution include the upper and lower houses of parliament in the parliamentary
system of government or senate and congress in the presidential system as well as the
state houses of assembly and the local government councils.
§ Increase in defence expenditure: Governments are also spending enormous sums of
money on the prosecution of wars or on preventing them. For example, Nigeria spent
huge sum of money on the prosecution of civil war and still spends in fighting
militancy and terrorism.
§ Mismanagement: Large sums of money are wasted due to mismanagement. This is
due to defective and inefficient financial and administrative machinery. Also there has
been duplication of government and administrative agencies, mismatching of
resources and extravagance.

In addition to above factors, Chand (2014) included income elasticity and increase in per
capita income and inflation among important contributing factors of growth in government
expenditure. Apart from the popular view that public expenditure grows because there is a
growing demand for public goods, Tarschys (1975) raised an argument for production
(supply) and finance induced growth of public expenditure. In his own words, “In politics, as
on the market, we have supply and demand, and there is no reason to believe that the
suppliers of public goods are idly awaiting the decisions of the consumers. Their interests are

33
also at stake in the determination of government expenditures, and it seems likely that they
have their part in the decision-making process.”

Who, then, are the suppliers of public goods? In one sense, it is the political actors who offer
the electorate a choice between several 'packages' of policy outputs. Yet a more solid supplier
interest is represented by the actual producers of the public goods: the civil servants
(bureaucrats). A second category consists of those in private enterprises who sell most of
their production to the government. When the growth of public expenditures is viewed from
the 'producer perspective', emphasis is placed on the efforts by various supplier collectives to
influence the scope and substance of government activities. Such efforts may be made either
through regular bureaucratic channels or through producer politics in the electoral or
parliamentary arena. As the public sector grows, it seems likely that the producers’ interest
will carry greater weight in the political process (Tarschys, 1975).

A third angle from which the expansion of the public sector may also be viewed could be
called the 'financial perspective'. This presupposes the early argument by Peacock and
Wiseman (1961) that citizens’ notion of tolerable tax rate often create in them the
unwillingness to contribute to the financing of public goods in general. Thus, a rise in
government finance brings about rise in production of public goods, which ordinarily are not
consumption (demand) induced by nature. According to Tarschys (1975), whereas
transactions in the private market are a function of two variables, i.e., demand and supply;
transactions on the public goods market are a function of three variables, i.e., demand, supply
and finance. Graphically the relationship can be illustrated by an equilateral triangle in Figure
3.3. A given volume of government activity presupposes certain quantum of each variable.
An increase in size of public sector would require approximate increase among supply
(production), demand (consumption), and finance. All three sides of the triangle must grow.
Advocates of the first perspective obviously presume consumer demand to be the
independent variable. That is, when more or new public goods are required by the citizens,
the system will respond by arranging for more finance and more production. Yet one might
also argue that the causal pattern is different. Tarschys (1975) contends that the opposite can
also be the case. This means that increase in public consumption could be a consequence
either of new resources (on the finance side) or of expanded production.

34
Figure 3.3: The three sides of public expenditure growth

Production Consumption
(Supply) (Demand)

Finanzbranche

Source: Tarschys (1975).

Three levels of explanation of growth in government expenditure were offered to include

socio-economic, ideological-cognitive, and political-institutional. Explanations on the socio-
economic level refer to changes in technological, economic, social, or demographic structure
while explanations on the ideological-cognitive level focus on changes in knowledge, beliefs
and desires (Tarschys, 1975). Lastly, Tarschys argues that mere existence of a belief or a
demand does not make it politically efficient. To make impact on the decision-makers it must
be articulated, aggregated, and channelled so as to affect the political calculus. Shifts on the
political-institutional level may therefore influence the effective policy output and thereby
increase government expenditure.

3.3 Empirical literature

A good number of empirical studies on government expenditure and economic growth were
done at both local and international level. Researchers are now increasingly interested on the
issues of optimal size of public sector. This is due to increasing trend in financial and finance
related recession and its spill-over effect on global economy. However, while some of the
review literature presented both positive and negative relationship between government
expenditure and economic growth, others found different rates of optimal size of government
sector within and across countries.

3.3.1 Studies on determinants of public expenditure and economic growth from

abroad

Ram (1986), Liu, Hsu and Younis (2008), and Ranjan and Sharma (2008) examine
government size and economic growth from 1974 to 1984, government expenditure and
economic growth from 1947 to 2002, and impact of government developmental expenditure

35
on India’s economic growth from 1950 to 2007 respectively. While Ram (1986) and Liu, Hsu
and Younis (2008) adopted cross sectional approach of fixed and random effect, and OLS
technique and found a significant positive relationship between government expenditure and
growth in India respectively. Ranjan and Sharma (2008) adopted standard time series
technique of unit root test and cointegration analysis and found that government expenditure,
investment and trade have significant positive impact on economic growth in India.

In a related research in Saudi Arabia by Abdullah (2000), an OLS estimation technique

alongside basic time series diagnostic test was employed to analyse the relationship between
government expenditure and economic growth from 1964 to 1995. The researcher reported
that the size of government is very important in the performance of economy due to its
significant positive influence.

While treading similar path, Yasin (2003) contributed to knowledge when he re-examined the
effect of government spending on economic growth. The study used panel data set from 26
Sub-Saharan Africa countries between 1987 and 1997. Its analytical model was derive from
an aggregate production function using OLS estimation technique in which government
spending, foreign assistance for development and trade-openness were explicitly specified as
input factors. Fixed-effects and random-effects estimation techniques were applied to the
model. The results from both estimation techniques indicate that government spending, trade-
openness, and private investment spending all have positive and significant effect on
economic growth. Foreign development assistance and the growth rate in population are
statistically insignificant. A test of a restricted version of the model indicates that the
contributions of foreign development assistance and the growth rate in population on
economic growth are statistically zero.

Again, Jiranyakul (2007) examined using the Granger causality test, the causation between
government expenditures and economic growth in Thailand. Time series data from 1993 to
2004 were retrieved from the International Monetary Fund’s International Financial Statistics
and Thailand National Economic and Social Development Board. The study found that there
is no cointegration between government expenditures and economic growth. However,
unidirectional causality from government expenditures to economic growth exists. In
addition, estimates from the ordinary least square (OLS) confirms the strong positive impact
of government spending on economic growth during the period of investigation.

36
A study on the impact of public spending on growth was carried out by Moreno-Dodson
(2008). It employed data from seven fast-growing countries drawn across Asia and Africa
between 1970 and 2005. The study adopted a panel analysis and obtained similar data from
few countries in South America which it used as a comparison group. The overall results
suggest that public spending, especially its productive components, has indeed a statistically
significant and positive impact on the growth rate of real GDP per capita in the first group. A
similar link could not be established robustly for the comparison group. In addition, the joint
net effect of fiscal policy (calculated as sum of the coefficients of expenditures, revenues, and
the fiscal balance) is also positive and statistically significant only for the first group.
Furthermore, inflation is negatively correlated with growth mainly for the fast-growing
group. This indicated that reducing inflation leads to faster growth for these countries and
therefore growth is more responsive to improvements in macroeconomic stability. When the
two groups were combined in a panel data analysis, it was however found that both the
economic and statistical significance as well as magnitude of total public expenditure droped
substantially. Similarly, when the different components of public spending were
disaggregated, productive and core expenditures become insignificant in explaining growth in
the pooled sample. These results indicate that when a more heterogeneous group of
developing countries (in terms of growth performance) is included in the study, the
significance of public spending and other budget components drops. In Moreno-Dodson’s
view, this may partially explain why some previous empirical studies that mixed countries
with very different growth patterns could not find a statistically significant link between
government spending and economic growth.

Also in this category, Alexiou (2009) investigated the impact of government spending on
economic growth in South Eastern Europe from 1995 to 2005. The study applied two
different panel data methodologies to seven transition economies in the South Eastern Europe
and generated significant results which, if considered, may enhance the economic
performance of the countries in the region. More specifically, the evidence generated indicate
that four out of the five variables used in the estimation i.e. government spending on capital
formation, development assistance, private investment and trade-openness all have positive
and significant effect on economic growth. Population growth in contrast, is found to be
positive but statistically insignificant.

37
3.3.2 Studies on determinants of public expenditure and economic growth from
Nigeria

Nurudeen and Usman (2010) observe that rising government expenditure has not translated to
meaningful development in Nigeria after investigating the effect of government expenditure
on economic growth. It was shown that total capital expenditure (TCAP), total recurrent
expenditures (TREC), and expenditure on education (EDU) have negative effect on economic
growth while that of transport and communication (TRACO) and health (HEA) result to
increase in economic growth. The study employs cointegration and error correction methods
to analyse the time series data from 1960 to 2007.

Although his study does not support the existence of Wagner’s law at both aggregate and
disaggregate levels in Nigeria, Louis (2012) found that all the variables except total recurrent
expenditure cause economic growth after carrying out cointegration and the Toda-Yamamoto
Granger Causality test on annual data from 1961-2009. In contrast, Nasiru (2012) investigates
the relationship between government expenditure (disaggregated into capital and recurrent)
and economic growth in Nigeria over the period 1961-2010. He employs the Bounds Test
approach to cointegration based on unrestricted Error Correction Model and Pairwise
Granger Causality tests. The results indicate that there exists no long-run relationship
between government expenditure and economic growth in Nigeria only when real GDP is
taken as dependent variable. In addition, the causality result reveals that government capital
expenditure granger causes economic growth. There was no causal relationship between
government recurrent expenditure and economic growth.

Again, a study aimed at investigating impact of cost of governance on economic development

in Nigeria was conducted by Ejuvbekpokpo (2012). Cost of governance was captured by
recurrent and capital administrative expenditures, while gross domestic product is used as a
proxy for economic growth. Using data from 1970 to 2010 and ordinary least squares (OLS)
econometric technique, the study reveals that cost of governance hampers economic
development in Nigeria.

Nwosa, Adebiyi and Adedeji (2013) examined the relationship between components of public
spending and private investments in Nigeria for the period 1981 to 2010. It adopted the
regression approach of error correction model (ECM). The study shows that components of
public spending have different impact on private investment both in the long-run and the

38
short-run. Specifically, recurrent and government final consumption expenditure had positive
(crowd-in) effect on private investment while capital expenditure had negative (crowd-out)
effect on private investment. Similarly, Chude, N and Chude, D (2013) investigate effects of
public expenditure in education on economic growth in Nigeria over the period 1977 to 2012,
with particular focus on disaggregated and sectoral expenditures analysis. The error
correction model indicates that total expenditure on education is positively related to
economic growth in Nigeria.

In likewise manner, Appah and Ateboh-Briggs (2013), Oyinlola and Akinnibosun (2013) and
Agbonkhese and Asekome (2014) employ OLS, vector error correction model (VECM) and
cointegration technique to analyse time series data between 1981-2011, 1961-2010 and 1970-
2009 respectively. They found positive relationship between disaggregated government
expenditure and economic growth. In particular, the pattern of public expenditure of
administration, social and community services, economic services, and transfers positively
affects the economic growth of Nigeria. Above studies came almost the same time with
Egbetunde and Fasanya (2013) who analysed impact of public expenditure on economic
growth in Nigeria from 1970 to 2010. It adopted the bounds testing (ARDL) approach and
found impact of total public spending on growth to be negative. Recurrent expenditure was
however found to have little significant positive impact on growth.

In attempt to empirically investigate the relationship between government expenditure and

economic growth in Nigeria, Olulu, Erhieyovwe and Ukavwe (2014) employ OLS estimation
on time series data from 1980 to 2010. The study found inverse relationship between
government expenditure on health and economic growth. Government expenditure on
education was however found to be insufficient to cater for the expending sector in Nigeria. It
was also discovered that government expenditure in Nigeria could increase foreign and local
investments. This result does not tally with Edame (2014) who analyze the trends in public
expenditure on infrastructure and economic growth in Nigeria from 1970 to 2010. Edame
however found a strong evidence that administration, external reserves, government revenue,
population density and rate of urbanization could collectively or individually influence
infrastructural growth vis-à-vis long-run economic growth. It was based on OLS regression
and subsequent Granger Causality test carried out on samples of the study.

39
3.3.3 Studies on the size of public sector from abroad

While considering general government spending, Peden (1991) estimated that the optimal
size of the U.S. government from 1929 to 1986. After basic time series diagnostic checks,
Barro (1989) methodology was adopted by the study and the result shows that optimal size of
government sector is about 20% of GDP. This study was closely followed by that of Scully
(1994). He employed time series data from 1929 to 1989 with which the optimal size of
government in the U.S was investigated using OLS technique. The result was not far different
from that of Peden (1991) as it shows that optimal growth-maximizing average rate for
federal and state on one hand and local taxes on the other hand is between 21.5% and 22.9%
of GNP respectively.

Surprise enough, Vedder and Gallaway (1998) estimated optimal size of federal government
spending based on the Armey Curve methodology in the United States within the period
1947–1997. It was shown that optimal size of government spending was about 17.45% of
GDP. This means that federal spending of about 22% at the beginning of 1990s (in Scully,
1994) was roughly 20% too large from the standpoint of growth optimization. Result of
above study was in line with Leland (2005) who carried out a study on fiscal characteristics
of public expenditures in the U.S from 1920 to 2003. The study employed OLS approach and
found that public expenditures in general increased 60-fold during the twentieth century,
while per capita spending increased nearly 20 times during the same period. Furthermore, it
was observe that government expenditures were only 13% of personal income in 1927, but
then grew rapidly during WWII, reaching 45% of personal income in 1946.

Karras (1996) estimates the optimal government size for several sets of economies by
investigating the role of public services in the production process. The study spans from 1984
to 1995 and adopted panel data analysis which was anchored on production function
methodology to capture various country group effects. He assumes government services are
optimally provided when their marginal product equals unity (the "Barro rule"). The
empirical results suggest that government services are significantly productive; they are
overprovided in Africa, underprovided in Asia, and optimally provided everywhere else; the
optimal government size is 23% for the average country but ranges from 14% for the average
OECD country to 33% in South America; and the marginal productivity of government
services is negatively related to government size.

40
After an assessment of growth-enhancing size of government in the Caribbean, Lewis-Bynoe,
Winston, and Kerry-Ann (2004) used a simple production function on annual observation for
the period 1975 to 2002. They posit that between 1990 and 1994, the average size of the
public sector in the Caribbean was just 16% of GDP. In five years thereafter, the ratio
climbed and stood at 22% of GDP. While an expansion in size of government usually results
in the greater provision of services, it can also lead to slower rates of growth because of
greater bureaucracy and the crowding-out of private sector driven initiatives.

Gunalp and Dincer (2005) estimated productivity of government services and the optimal
government size for a set of transition countries and employed the empirical methodology
based on Barro (1989, 1990). The study used annual data from 1990 to 2001 for 20 transition
countries and found that government services were productive, and on average it was
optimally provided. The marginal productivity of government services was negatively related
to government size. At the peak of its findings lies that optimal government size is about
17.3% for the average transition country.

In a study of five Gulf countries of the Middle East, Aly and Strazicich (2006) examined if
the size of government consumption relative to national output is optimal. The authors
adopted the methodology suggested in Barro (1990) and examined marginal productivity of
government consumption from 1970 to 1992. Results reveal that government consumption is
productive, but the size of government is too large to be optimal. Similarly, Mavrov (2007)
studied the relationship between government spending and economic growth rate in the
period 1990-2004 in Bulgaria. The Armey curve was adopted in the study. The optimal rate
of the spending (output ratio) was calculated on the basis of that curve. He concluded that
current ratio of government spending to total output which is 40% was above the optimal rate
of 28% found in the study. Among its recommendation were gradual decrease in government
spending (output ratio) and an increase in efficiency of government spending programs.

In the same vein, Herath (2010) studied the size of government and economic growth in Sri
Lanka form 1959 to 2003. Armey curve methodology was adopted by the study. The
researcher observed that optimal size of government in Sri Lanka was about 27% which is
less than 29% of actual government size in the country. He went ahead to suggest downward
review of government spending in the country so as to avert possible downward trend in
overall economic growth.

41
3.3.4 Studies on the size of public sector from Nigeria

Udah (2010) used the cointegration and error correction frameworks of analysis to investigate
extent to which government size and other factors succeeded in improving the conditions
needed to stimulate private investment in Nigeria. The study employed time series data from
1980 to 2008. The results show, surprisingly, that government size did not complement
private investment initiative. This unusual observation was attributed to, among others,
inefficiency in government expenditure and poor service delivery. While using a different
theoretical framework, Ekeocha and Oduh (2012) carried out an empirical investigation into
the optimal size of government in Nigeria. The study empolyed annual observation between
1970 to 2006 and adopted OLS estimation techique in its analysis based on Armey and
Armey (1995). The result shows that federal government expenditure as a percentage of GDP
has a positive coefficient. It equally estimated the optimal size of Nigerian federal
government expenditure to be about 22.47% of GDP.

Contrary to above findings of Udah (2010), Oriakhi and Arodoye (2013) examined the
relationship between government size and economic growth in Nigeria using annual time
series data from 1970 to 2010. In order to fully account for feedbacks, a vector autoregressive
model is utilized. The results show that there is long-run relationship between government
size and economic growth. The Forecast Error Variance Decomposition (FEVD) results show
that the main sources of Nigeria economic growth variation are due largely to “own shocks”,
government size and real gross domestic product per head innovations.

A recent discussion on the theoretical and empirical basis for the existence of optimal size of
government was carried out by Olaleye, et al. (2014). It adopted OLS technique and use time
series data that covers between 1983 and 2012. The study found that the optimal size of
government in Nigeria (the share of overall government spending that maximizes economic
growth) is 11% of GDP. Similarly, Alimi (2014) used annual Nigeria country-level data for
the period 1970 to 2012 to investigate existence of optimal government size in developing
economies. The study adopted the methodology put forward by Barro (1990) and Armey
(1995). It measured size of government in two ways; share of total expenditures to GDP and
share of recurrent expenditures to GDP. The researcher found that optimal size is about
19.81% and 10.98% respectively.

42
3.4 Summary of literature review

Evidence from reviewed literature in this study was broadly classified into studies on public
expenditure and economic growth on one hand, and optimal size of the public sector on the
other hand. This was done to enhance summary of thematic issues and findings by
researchers.

Public expenditure was popularly acknowledged among researchers as veritable tool for
production of public goods as well as for economic growth in both developed and developing
economies. For instance Brown and Jackson (1990), Piana (2001), Rutherford (2002) and
Leland (2005) saw it as money allocated by governments for provision of public goods and
services. It also serves as a tool for income redistribution, poverty and inequality reduction,
and stimulation of economic growth. Notwithstanding above mention key functions, the
classical economists however relegated public expenditure to mere maintenance of law and
order. Again, no meaningful growth was expected in government expenditure except during
the time of social upheaval other national emergencies. Most recent studies however, greatly
acknowledge the unavoidable circumstance behind public expenditure growth in modern
states. In addition to Keynesian revolution in public expenditure growth, several factors were
identified as contributors to growth of public expenditure across economies. These include
the quest for economic development (Musgrave, 1969, 1974; Rostow, 1971; Bird, 1971;
Brown & Jackson, 1990; Leland, 2005), social upheaval (Peacock & Wiseman, 1961; Leland,
2005), population growth, mismanagement and welfare state ideology (Tarschys, 1975; Ram,
1986; Yasin, 2003; UNESCO-Nigeria Project, 2010; Chand, 2014), and growth of the civil
service sector (Niskanen, 1971; Fluvian, 2006; Adewole & Osabuohien, 2007; Warimen,
2007; Ejuvbekpokpo, 2012).

In addition to its growth diamension, public expenditure was classified into functional
(Connolly & Munro, 1999), revenue and capital (Musgrave & Musgrave, 2004), transfer and
non-transfer (Pigou, 1928; Fasanya, 2013), benefit incidence (Brown & Jackson, 1990),
productive and unproductive, and Hugh Dalton’s category (Dalton, 1920).

In a bid to investigate the link between public expenditure and economic growth, except for
Nwosa, Adebiyi and Adedeji (2013), Egbetunde and Fasanya (2013) and Olulu, Erhieyovwe
and Ukavwe (2014) who found capital expenditure component, total public spending and
government expenditure on health to be negatively related to economic growth respectively,

43
others such as Jiranyakul (2007), Liu, Hsu and Younis (2008), Ranjan and Sharma (2008),
Moreno-Dodson (2008), Alexiou (2009), Appah and Ateboh-Briggs (2013), Oyinlola and
Akinnibosun (2013) and Agbonkhese and Asekome (2014) found positive relationship
between government spending and economic growth.

The argument on the exact direction of relationship between public expenditure and
economic growth however evolved into a consensus that non-linear relationship exist
between the two. For instance, Mueller (2003), Chobanov and Mladenova (2009), Thanh
(2009), Ekinci (2011), Facchini and Melki (2011), Ekeocha and Oduh (2012) and Altunc and
Aydin (2013) acknowledged that there is no single threshold for optimal size of the public
sector for all economies. They all affirm that optimal size of government sector varies across
countries depending on the country’s level of development and other non-economic factors.
The common view held among researchers was that on average, optimal size of the public
sector lies approximately between 15% and 50% of the GDP across developed and
developing economies (Barro, 1989; Armey & Armey, 1995; Rahn & Fox, 1996; Friedman,
1997; Scully, 1994, 2003). Hence, no economy can firmly lay claim to its own optimal size of
public sector over a long term since such optimum level is subject to change over time.

3.5 Limitations of previous studies and motivation

In a study on determinants of government expenditure growth in Nigeria, Okafor and Eiya

(2011) made a right choice of variables such as inflation rate, population growth rate, public
debt growth rate and tax revenue growth rate. These variables were known for the role they
play as key public expenditure growth derivers in Nigeria. However, the study has a very
limited number of observations (1999-2008). Consequently, inferences drawn from such
study with only five degrees of freedom cannot be reasonably reliable for policy purposes.

Much studies which investigated impact of government expenditure on economic growth in

Nigeria such as Ejuvbekpokpo (2012), Egbetunde and Fasanya (2013), Chude, N. and Chude,
D. (2013), Oyinlola and Akinnibosun (2013), Agbonkhese and Asekome (2014), Edame
(2014), and Olulu, Erhieyovwe and Ukavwe (2014) used observation from 1970 upwards.
The use of few numbers of observations affected the results of above studies. For example,
while some observed positive relationship, others found negative relationship with respect to
the same variable. This trend was also seen in the level of significance of such variables as
they vary heavily across studies. Other studies which have relatively large number of

44
observations like Nurudeen and Usman (2010), Nasiru (2012), Louis (2012), and Appah and
Ateboh-Briggs (2013) were interested in short-run relationship between government
expenditure and economic growth. As such, they adopted approaches such as ECM, ARDL
and VEC estimation techniques. Among above studies, none has estimated the growth rate of
public expenditure over time or investigated the existence of structural break in the growth
rate of public expenditure

Moreover, the theoretical argument for existence of non-linear relationship between

government expenditure and economic growth made the linear approach in above studies
relatively less important. To confirm this claim, studies within Nigeria contradicted
themselves as both positive and negative relationship between government expenditure and
economic growth were found using the same variables. The studies by Agbonkhese and
Asekome (2014) and Olulu, Erhieyovwe and Ukavwe (2014) are good examples respectively.

Only very few recent studies on optimal size of public sector in Nigeria were done. Among
such studies are Udah (2010), Ekeocha and Oduh (2012), Oriakhi and Arodoye (2013),
Olaleye, et al. (2014) and Alimi (2014). None of the above made use of observations beyond
1970. Again, none of the studies used observation beyond 2012. This point was emphasized
by Olaleye, et al. (2014) who conceded that limitations in model specification and data might
have led to underestimation of their result which suggest that optimal size of the public sector
in Nigeria is 11% of the GDP. Apart from Ekeocha and Oduh (2012) who carried out basic
time series diagonistic analysis before estimation of their model, others estimated their
models without paying due attention to the data generating process of the series they used.
This approach was known for its vulnerability in producing spurious relationship between
regressor(s) and regressand. This is because the true data generating process of the series
might be unknown to the researcher.

As a result of above identified limitations of previous studies, the researcher was motivated to
carry out economic analysis of public expenditure growth and optimal size of the public
sector in Nigeria from 1961 to 2014. The prime motive is to assess behaviour of the key
variables throughout the span of policy changes within the country. To achieve this, the study
intends to estimate average growth rate of public expenditure in Nigeria which hitherto, has
not been attempted. It will also examine influence of key growth drivers of public
expenditure. At the same time, the study will pay attention to impact of Fiscal Responsibility

45
Act of 2007 on the rate of public expenditure growth. This is important as it will validate the
effectiveness or otherwise of the above law in Nigeria.

In addition to above motivation, the study intends to estimate optimal size of public sector in
Nigeria using 54 observations (1961-2014). This remains the longest time horizon with
respect to this research topic in Nigeria. The intension of this researcher aligns with
observations of Vish (2002), Mutascu and Milos (2009), and Facchini and Melki (2011) that
studies which involve larger observations tends to produce more accurate data on economy’s
optimal size of public sector than studies with smaller number of observations. Therefore,
with its 54 time series observations, this study remain best fitted to achieve a better estimate.

46
 CHAPTER FOUR

METHODOLOGY
4.1 Theoretical framework

Based on above reviewed theoretical literature, the study adopted the theoretical approach of
Institutional Model of public expenditure growth and Armey framework respectively. While
the former incorporated key features of stages of development growth model, Wagner’s law,
displacement effect model and Niskanen model, it also provided the foundation for
explanation of interaction between public expenditure growth and its key explanatory
variables. The latter on the other hand, emphasized approach to estimation of non-linear
relationship between economic growth and relative size of the public sector. On the part of
institutional model proposition, public expenditure was seen as being primarily determined
by institutional related and country-specific factors which differs among economies (Downs,
1957; Buchanan & Tulloch, 1962; and Black, 1969). One of the key assumptions of the
model is that outcomes of public sector expenditure growth is determined, in part, by
institutions, their operational mechanisms and the people working in those institutions
(Doessel & Valadkhani, 2003). In this study therefore, the model is specified thus;

G = f(T, B, FRA, BI) ... ... ... ... ... ... ... (4.1)

Equation (4.1) indicates that public expenditure (G) is determined by institutionally induced
factors such as tax revenue (T), government borrowings (B), foreign reserve assets (FRA)
and bureaucratic inertia (BI). The framework suggests that fiscal institutions in form of
arrangements or rules do affect outcomes of public sector spending. While explaining the
dynamics of institutional framework, Mankiw (2009) and Blanchard, Amighini and Giavazzi
(2010) contend that fundamental changes in public sector spending come from both tax
revenue and government borrowing. Therefore, given government expenditure G, tax revenue
T and government borrowing B, the relationship can be stated as follows:

G = f(T, B) ... ... ... ... ... ... ... ... (4.2)

Based on equation (4.2), Peacock and Wiseman (1961) argue that rise in either T or B, or
both, leads to rise in government ability to finance its expenditure on public goods.
Furthermore, Zhou (2007) and Pugel (2007) explained that when adverse shocks occur,

47
international reserve held by the central bank provides alternative source of international
liquidity by cushioning some of the losses in tax revenue. Hence, government will find itself
on a position to conduct counter-cyclical fiscal policies in form of higher spending and lower
tax rate during economic downturns. Lastly, above reviewed Niskanen model of budget
maximizing bureaucrats after Niskanen (1968) supports the view that large government
expenditure is often a rising function of bureaucratic institution.

A key advantage of this framework lies in its capacity to incorporate key determinants of
government spending. However, its prime limitation was found in its inability to link up how
changes in government transfer influences change in government expenditure.

The Armey theoretical framework was anchored on Armey and Armey (1995) who explain
the notion of optimal size of government. It argues that non-existence of government causes a
state of anarchy and low levels of output per capita. Hence, in the absence of law and order,
there will be little incentive to save and invest among economic agents. Vedder and Gallaway
(1998) and Alimi (2014) summarized the framework by stating that its key message remains
that excessively large governments have tendency to become increasingly less productive,
and thereby reduces output growth. However, where there is a mix of private and public
decisions on the allocation of resources, output should be larger (Pevcin, 2004). The model is
represented thus:

Q = f(G, Ω) ... ... ... ... ... ... ... ... (4.3)

While Q measures the output of the economy, G measures components of state intervention
in the economy and Ω is a proxy for core exogenous factors that also influence output.
Accordingly, equation (4.3) can be specified in the following non-linear regression equation:

Qt = + Gt - Gt2 + ut ... ... ... ... ... ... (4.4)

The idea behind apriori signs in the parameters of G in equation (4.4) was based on the
framework’s argument that while G capture beneficial incidence of public spending, G2 will
indicate the effect of decreasing marginal productivity of government spending on output
(Scully, 1994; Mutascu & Milos, 2009). Finally, the application of differential calculus on
equation (3.4) will yield the optimal value for relative size of public sector to GDP.

48
4.2 Model specification

To realize objectives of the study, three models were specified as follows;

Model 1: Average growth rate of public expenditure

To estimate average growth rate of public expenditure in Nigeria, let GEt denote real total
federal government expenditure at time t and GE0 the initial value of total federal government
expenditure (i.e., the value at the end of 1960). The relationship may be defined thus;

GEt = GE0(1 + r )t ... ... ... ... ... ... (4.5)

The letter r is the compound (over time) rate of growth of GE.

Taking logarithm of equation (4.5) will yield

Log(GEt) = Log(GE0) + t Log((1 + r)) ... ... ... ... (4.6)

Let β0 = Log(GE0) and β1 = Log((1 + r))

Thus equation (4.6) becomes

Log(GEt) = β0 + β1t ... ... ... ... ... ... ... (4.7)

Adding a vector of control variables and error term to equation (4.7) yields an econometric
relationship in equation 4.8.

Log(GEt) = β0 + β1t + β2Πt + µt ... ... ... ... ... (4.8)

Where:

Log(GE) = log of government expenditure

t = time (1961 - 2014)

Π = vector of control variables (tax revenue, public debt and population growth rate)

βi = parameters (i = 0, 1, 2, ...)

µ = error term

Equation (4.8) represents linear regression growth model.

49
Model 2: Determinants of public expenditure growth

In order to investigate determinants of public expenditure growth in Nigeria, key explanatory

variables were chosen. The channels through which the explanatory variables influence the
dependent variable were specified in the following equation.

GE = f(OLR, NOLR, PD, PGR, OER, INF, GEt-1) ... ... ... (4.9)

Where:

GE = government expenditure (N billion)

OLR = tax revenue from oil (N billion)

NOLR = tax revenue from non-oil (N billion)

PD = public debt (N billion)

PGR = population growth rate (%)

OER = official external reserve ($ million)

INF = inflation rate (%)

GEt-1 = lagged value of government expenditure (N billion).

The choice of above listed indicators was in line with estimate of important determinants of
public expenditure growth across the globe. In addition to the support they gained from
theoretical literature, they were also employed as core variables by Okafor and Eiya (2011)
and Edame (2014) in their respective studies on determinants of public expenditure in
Nigeria. It is also important at this juncture to note that inflation rate can positively or
negatively correlate with with government ependiture. According to Han and Mulligan
(2008) and Nguyen (2014), inflation would likely exhibit both positive and negative link with
government spending. This projection suggests that government spending is positively related
to inflation when the former crowds-out private investment, and vis-à-vis. This concurs with
theoretical evidence that government spending stimulates aggregate demand (Keynesian
school) but also crowds-out private investment (Neoclassical school).

Equation (4.9) above can be express as follows;

50
 GEt = δ0 + δ1OLRt + δ2OLRt + δ3PDt + δ4PGRt + δ5OERt + δ6INFt + δ7GEt-1 + εt

... ... ... ... ... ... ... ... .. ... (4.10)

Again, FRA was added to equation (4.10) to enable the researcher investigate impact of
Fiscal Responsibility Act of 2007 (enacted to promote transparency and accountability in
government spending) on rate of growth of public expenditure in Nigeria. Thus equation
(4.10) now becomes

GEt = δ0 + δ1OLRt + δ2NOLRt + δ3PDt + δ4PGRt + δ5OERt + δ6INF + δ7GEt-1 +

λ1FRAt + εt .... ... ... ... ... ... ... (4.11)

In order to tally with demands of objective two of this study, equation (4.11) can be transform
as follows

Log(GEt) = δ0 + δ1Log(OLRt) + δ2Log(NOLRt) + δ3Log(PDt) + δ4PGRt-1 + δ5OERt-1

+ δ6INFt-1 + δ7Log(GEt-1) + λ1FRAt + εt ... ... ... ... (4.12)
Where:

FRAt = dummy for Fiscal Responsibility Act =1 for (2007-2014); 0 otherwise.

δ and λ = parameters

ε = error term

t = time (1961-2014)

In line with postulations of above theoretical framework in section 4.1, parameters of all the
explanatory variables in equation (4.12) are expected to be greater than zero except FRA
which is expected to be greater than zero. INF can be either greater or less than zero. That is
to say that δ1>0, δ2>0, δ3>0, δ4>0, δ5>0, δ7>0, but λ1<0 and δ6>0 or <0. Again, to avert
possible violation of OLS assumption of ‘no serial correlation’, first difference of PGR and
OER may also be employ (i.e., Δx = xt – xt-1).

Model 3: Determinants of real gross domestic product and estimation of the optimal size
of the public sector

The model for optimal size of public sector will examine the relationship between real GDP
and federal government expenditure following the analyses carried out by Vadder and

51
Gallaway (1998), Mutascu and Milos (2009), Ekeocha and Oduh (2012) and Olaleye, et al.
(2014) which were founded on Armey and Armey (1995) theoretical framework as follows;

RGDPt = + SPSt - SPSt2 + Ωt + u t ... ... (4.13)

Where:

RGDP = real gross domestic product (N billion)

SPS = size of public sector (government expenditure ratio to GDP)

SPS2 = squared size of public sector

Ω = vector of relevant explanatory variables such as broad money supply (M2), inflation rate,
trade openness, tax revenue and gross fixed capital formation.
αj = parameters (j = 0, 1, 2, ...)

u = error term

t = time (1961- 2014)

It is expected that parameter of SPS will have positive sign ( > 0) in order to show
beneficial effects of government spending on output growth (GDP). On the contrary, squared
term (SPS2) is expected to have negative sign ( < 0) to indicate the decreasing marginal
productivity of government spending.

Hence to arrive at an optimal value for the size of government sector, equation (4.13) will be
differentiated with respect to SPS as follows;

( )
= – 2( )SPS = 0 ... ... ... ... ... (4.14)
( )

Equation (4.14) satisfies the first-order condition for optimization which requires that first
derivative of RGDP with respect to SPS, i.e., f’(.) be equal to zero. Making SPS subject of the
formula in equation (3.14) gives us

SPS* = ... ... ... ... ... ... ... (4.15)

( )

52
4.3 Estimation technique and model justification

The econometric method of research analysis is employed in this study specifically using the
ordinary least square (OLS) method. The choice of this method was based on its “Best Linear
Unbiased Estimator” (BLUE) properties (Gujarati & Porter, 2009; Wooldrige, 2013).
Similarly, the model readily helps to ascertain quantitative impact of certain factors on a
given phenomenon under study when compare to competing models (Greene, 2012).
However, appropriate statistical properties of the time series data will be initially examine
such as stationarity and cointegration test so as to forestall possible spurious result. In
addition, post estimation diagnostic checks such as test for autocorrelation, regression
specification error test and normality test will also be carried out. These measures shall be
taken to ensure conformity of available data with assumptions of OLS estimation technique
as well as guarantee reliability of the regression results.

The choice of three separate models by the researcher was informed by the need to
appropriately capture objectives of the study. This approach conforms to prescription of
Tinbergen (1978) on the need to use specific and most suitable policy instrument to achieve a
given policy objective (target). While equation (4.8) is targeted to realize objective 1 of the
study, equation (4.12) will be used to achieve objectives 2 and 3. As stated above, the
realization of optimal size of government sector will begin by employing differentiation
calculus on quadratic function of government size in equation (4.13). Finally, output of
equation (4.15) will be used to realize objective 4 of the study.

Key variables of interest in this study were summarized by the researcher in Table 4.1 below.

53
Table 4.1: Summary of key variables

Variables Definition Proxy Source

Federal government total Growth of federal government CBN Statistical Bulletin 2009
GE
expenditure. expenditure. and 2014
Bureaucratic inertia or
Lagged value of government CBN Statistical Bulletin 2009
GEt-1 incrementalism in government
expenditure. and 2014
spending
A dummy for Fiscal
Measure for transparency, To be generated by the
Responsibility Act which takes
FRA accountability and fiscal researcher using econometric
1 for 2007-2014 and 0 for
discipline. software
1961-2006.
CBN Statistical Bulletin 2009,
TAX Federal tax revenue Government revenue 2014 and
http://data.worldbank.org
CBN Statistical Bulletin 2009
INF Inflation rate Price stability
and 2014
CBN Statistical Bulletin 2009,
OER Official external reserve International liquidity position
and http://data.worldbank.org
CBN Statistical Bulletin 2009
PD Public debt Government revenue
and 2014
Federal government CBN Statistical Bulletin 2009
SPS Relative size of public sector
expenditure ratio to GDP. and 2014
CBN Statistical Bulletin 2009,
RGDP Real gross domestic product. Economic growth 2014 and
http://data.worldbank.org
PGR Relative change in population. Population growth rate http://data.worldbank.org

4.4 Data sources and software for analysis

As a time series analysis, the study shall utilize secondary data sourced from Central Bank of
Nigeria Statistical Bulletin (2009 & 2014). In addition, data will also be sourced from World
Bank’s World Development Indicators 2015 online database through its website address,
http://data.worldbank.org. The study shall employ Eviews 8.0 econometric software for its
analysis.

54
 CHAPTER FIVE

DATA PRESENTATION AND ANALYSIS

5.1 Introduction to data analysis

This chapter summarized and presented brief descriptive analysis of sample data in this study.
It further presented and interpreted the regression results alongside results of basic diagnostic
test carried out in the research. The analysis was presented in three sections to replicate the
three regression models built in chapter three, which conform to objectives of the study in
chapter one. The first result presents estimate of average growth rate of public expenditure in
Nigeria. In the result, time trend and other control variables were regressed on annual
government expenditure. The second result estimated determinants of government
expenditure where impact of government revenue, public debt and other control variables on
government expenditure was examined. Finally, the result of non-linear relationship between
size of public sector (measured as ratio of government expenditure to GDP) and economic
growth was presented and discussed. The last presented result enabled the research to
estimate optimal size of public sector in Nigeria. However, the descriptive statistics and
Augmented Dickey-Fuller Unit Root test on the variables were first presented below.

5.2 Definition of variables and descriptive analysis

Table 5.1: Definition of variables

Variable Name Variable Label

FRA Fiscal Responsibility Act (1= 2007-2014, 0= 1961-2006)
GE government expenditure (₦'billion)
GFCF gross fixed capital formation (₦’million)
INF inflation rate (%)
M2 broad money supply (₦'billion)
NOLR tax revenue from non-oil sector (non oil revenue, ₦'billion)
OER official external reserve (foreign reserve) ($'million)
OLR tax revenue from oil sector (oil revenue, ₦'billion)
PD public debt (₦'billion)
PGR annual population growth rate (%)
RGDP real gross domestic product (₦'billion)
SPS size of public sector (government expenditure ratio to GDP)
SPS2 square of size of public sector
T time (1961-2014)
TAX tax revenue (₦'billion)
TO trade openness (net export ratio to GDP)

55
Table 5.2: Descriptive statistics

Variable Obs Mean Std. Dev. Min Max

FRA 54 0.148148 0.358583 0 1
GE 54 832.1180 1475.952 0.16 5260.77
GFCF 54 1241957 3092069 258 11723098
INF 54 16.26944 16.13493 -3.73 72.84
M2 54 2119.659 4498.119 0.3 17680.52
NOLR 54 409.3713 805.9900 0.22 3275.12
OER 54 10716.14 16038.68 112.36 53599.28
OLR 54 1353.982 2430.079 0 8879
PD 54 154.4128 324.0891 0.06 1158.52
PGR 54 2.544074 0.235849 2.04 3.05
RGDP 54 24558.19 15257.34 8758.73 67977.46
SPS 54 16.80204 6.626924 6.45 34.19
SPS2 54 325.4113 253.6121 41.6025 1168.956
TAX 54 1748.776 3177.187 0.22 11116.90
TO 54 44.42130 17.09303 19.62 81.81

Out of 54 years of observation in the study, the dummy for Fiscal Responsibility Act (FRA),
government expenditure (GE) and gross fixed capital formation (GFCF) recorded annual
mean values of 0.1481481, N832.118 billion and N1,241,957 million with standard deviation
of 0.3585825, 1475.952 and 3092069 respectively. While FRA and GE sequentially has
minimum and maximum values of 0 & 1 and N0.16 billion & N5,260.77 billion, GFCF has
N258 million & N11,723,098 million. Similarly, inflation rate (INF), broad money supply
(M2) and non-oil revenue (NOLR) have annual average values of 16.27%, N2,119.659
billion and N409.3713 billion each. At the same time their standard deviation, minimum and
maximum values were respectively; 16.13493, -3.73% & 72.84%; 4498.119, N0.3 billion &
N17,680.52 billion; and 805.99, N0.22 billion & N3,275.12 billion.

For official external reserve (OER), oil revenue (OIL), public debt (PD) and population
growth rate (PGR), the table sequentially shows their mean, standard deviation, minimum
and maximum values to be $10,716.14 million, 16038.68, $112.36 million & $53,599.28
million; N1,353.982 billion, 2430.079, N0 billion & N8,879 billion; N154.4128 billion,
324.0891, N0.06 billion & N1,158.52 billion; and 2.54%, 0.235849, 2.04% & 3.05%. The
mean value of real gross domestic product (RGDP) is N24558.19 billion. It has standard
deviation value of 15257.34, while it recorded minimum and maximum values of N8,758.73
billion and N67,977.46 billion during the period. Lastly, mean values of size of public sector

56
(SPS), tax revenue (TAX) and trade openness (TO) are 16.8%, N1748.776 billion and
44.42%. In a similar manner, their standard deviation, minimum and maximum values are
6.626924, 6.45% & 34.19%; 3177.187, N0.22 billion & N11,116.9 billion; and 17.09303,
19.62% & 81.81% respectively.

5.3 Stationarity and cointegration test

The table below shows result obtain form stationarity test (unit root) which follows the test
statistic developed by Dickey and Fuller (1979). Variable of interest, estimated ADF test
statistic value, probability value and order of integration of those variables were reported. It
also reported the result of cointegration test carried out in the study.

Table 5.3: Augmented Dickey-Fuller test statistic

Variable ADF test statistic p-value Order of integration

GE -3.002409 0.0036** I(0)
GFCF -5.395870 0.0000** I(1)
INF -2.251870 0.0248* I(0)
M2 -4.146951 0.0001** I(2)
NOLR -9.437909 0.0000** I(2)
OER -5.695329 0.0000** I(1)
OLR -8.491403 0.0000** I(1)
PD -6.468044 0.0000** I(2)
PGR -8.080581 0.0000** I(1)
RGDP -8.747016 0.0000** I(2)
SPS -3.481481 0.0123* I(0)
SPS2 -4.302924 0.0012** I(0)
TAX -8.052821 0.0000** I(1)
TO -9.997272 0.0000** I(1)
Cointegration -1.864065 0.0599 I(1)
Note: ** p-value < 0.01; * p-value < 0.05

In line with Dickey and Fuller (1979), a variable is stationary if its ADF test statistic value is
greater than its critical value at given percentage level of significance, say at 5% or 1% level.
Table 4.3 above shows that government expenditure (GE), inflation rate (INF), size of
government sector (SPS) and square of government expenditure (SPS2) are stationary at
level, that is, they are integrated of order zero (I(0)). Gross fixed capital formation (GFCF),
official external reserve (OER), oil revenue (OIL), population growth rate (PGR), tax revenue
(TAX) and trade openness (TO) were stationary after first difference. This imply that the

57
variables are integrated of order one (I(1)). Lastly, money supply (M2), non-oil revenue
(NOLR), public debt (PD) and real gross domestic product (RGDP) were stationary after
second difference. That is to say the variables are integrated of order two (I(2)) (See
Appendix 1 for stationarity result).

Since one of the dependent variables (RGDP) is of the same order of integration with M2 as
was contain in equation 4.13, the researcher estimated the linear combination of the duo at
their stationary form and obtained their residuals which was subjected to Augmented Dickey-
Fuller unit root test. The result in Table 5.3 and Appendix 2 show that ADF test t-statistic of -
1.864065 is not significant at 0.05 given its estimated p-value of 0.0599. The result suggests
that there is no long-run relationship between RGDP and M2 in this study. Hence, there is no
need to carry out error correction mechanism since the identical order of integration between
the two variables is mere coincidence, but not as a result of unobserved long-run relationship.

5.4 Regression result for average growth rate of public expenditure

Below is the OLS estimate of average growth rate of public expenditure in Nigeria from 1961
to 2014. Time trend alongside other control variables were regressed on the log of
government expenditure. The research objective is to ascertain average growth rate of
government expenditure using estimated coefficient of Time (T).

Table 5.4: Average growth rate of public expenditure

Variable Coefficient HAC Std Error t-statistic P-value

T 0.019059 0.018435 1.033831 0.3085
LOG(TAX) 0.725239 0.074296 9.761447 0.0000**
LOG(PD) 0.093175 0.037503 2.484453 0.0181*
PGR 0.04528 0.126858 0.356936 0.7233
Constant -37.306 36.33886 -1.02662 0.3119
R-squared 0.993489
DW-statistic 1.304686
Note: ** p-value < 0.01; * p-value < 0.05

5.4.1 Economic and statistical criteria (apriori expectation and first order condition)

Anticipated economic relationship between dependent variable and independent variables

was examined in this section. To validate the effectiveness of such relationship, the

58
researcher employed exact (true) level of significance (p-value) approach in testing the
research hypotheses. This implies that any estimated coefficient with corresponding p-value
less than or equal to (<=) 0.05 is considered statistically significantly different from zero. As
such, the conclusion shall be that effect of such variable on the dependent variable cannot be
underestimated as it significantly influences the dependent variable.

Time trend (T)

The coefficient of T is 0.019059 which shows the percentage average response of

government expenditure (GE) per unit change in time (T). This implies that when effects of
other variables are held constant, a year increase in time leads to about 1.91% increase in
government expenditure on the average. The positive coefficient of time parameter in this
study conforms to apriori expectation because it is rational that government spending should
be positively correlated with increasing demand for government responsibility in the society
over time. However, effect of time changes on government expenditure is considered
statistically insignificantly different from zero as estimated p-value (0.3085) of its coefficient
is greater than 0.05. It suggests that though government expenditure increases on average by
about 1.91%, this magnitude of change is not statistically significant.

Tax revenue (LOG(TAX))

Tax revenue has positive coefficient value of 0.725239 which suggest a positive correlation
with government expenditure. This is in line with apriori expectation that increasing tax
revenue will improve government capacity to finance its expenditure. Hence, when effects of
other variables are assumed to be zero, a percentage rise in tax revenue will bring about
average yearly increase of 72.52% in government expenditure. This effect is statistically
significantly different from zero given estimated p-value (0.0000) of its coefficient which is
less than 0.05. The result suggests that tax revenue is significant determinant of government
expenditure in Nigeria.

Public debt (LOG(PD))

The coefficient of public debt is 0.093175. It suggests a positive relationship between public
debt and public expenditure in this study. All other thing being equal, a percentage increase
in public debt will increase public expenditure on average by approximately 9.32% each year.
Though the effect of public debt on public spending is statistically significantly different

59
from zero at 5% due to p-value of its coefficient (0.0181), the magnitude of its coefficient
does not prove it a key growth driver of public spending in Nigeria as was observed by World
Economic Forum (2014) which rated Nigeria 18th among 143 countries in terms of ratio of
public debt to GDP.

Population growth rate (PGR)

The coefficient of population growth rate (0.04528) is well behaved as it displayed positive
relationship with government expenditure. It suggests that as population of the country
grows, average government spending on the population also grows. Therefore, holding all
other variables constant, a percentage increase in population growth rate will cause
government expenditure to increase by about 4.53% on average per annum. The result further
suggested that population growth rate is not a significant determinant of government
expenditure as its p-value is 0.7233. This value is greater than expected 0.05 which is the
bases for decision taking on the study’s hypotheses.

5.4.2 Post-estimation diagnostic check for average growth rate of public expenditure

This section presented the econometric criteria and second order condition for evaluation of
above regression result. Goodness of fit of the model, test for serial autocorrelation, model
specification error test and normality test were carried out here.

Table 5.5: Model fit and diagnostic check for average growth rate of public expenditure

Test Estimated statistic p-value

R-squared 0.993489
F-statistic 1296.905 0.000000**
DW-statistic 1.304686
RESET F-stat (0.010906) 0.9175
Normality Test JB-stat (2.343488) 0.309826
Note: ** p-value < 0.01; * p-value < 0.05

R-squared (Goodness of fit)

The R-squared measures proportion of variation in the dependent variable (GE) which was
explained by independent variables in Table 5.4 above. From Table 5.5, estimated R-squared
value of 0.993489 was obtained. It shows that about 99% of total variation in government
expenditure was explained by time trend, tax revenue, public debt and population growth

60
rate. The 99% goodness of fit in this study shows that the result is very reliable for policy
prescription.

Test for autocorrelation

To avoid associated effects of serial correlation, the researcher tested the model for
autocorrelation using estimated d-statistic after Durbin and Watson (1951). From Table 5.5,
the null hypothesis of ‘no autocorrelation’ was rejected at 5% level of significance since
dU(1.789) ≮ d(1.304686) < 4-dU(2.211). While d is estimated d-statistic in Table 5.5, dU is
upper boundary of d-critical value for 39 observations and 5 parameters from Durbin-Watson
statistic table. This implies that assumption of no auto-, or serial, correlation in the error
terms that underlies the classical linear regression model (CLRM) will be violated. In a bid to
resolve this problem the researcher instead employed Newey and West (1987)
heteroscedasticity and autocorrelation-consistent (HAC) standard errors technique as reported
in Table 5.4. This approach was identified by Gujarati and Porter (2009) as the best option
when OLS estimate suffers from pure autocorrelation becaues of its ability to correct
estimated standard errors and as well solve twin problem of heteroscedasticity whenever it
existed.

Regression specification error test

To further investigate whether observed autocorrelation was a result of wrong model

specification, the researcher adapted the regression specification error test (RESET)
developed by Ramsey (1969). This approach was aimed at verifying whether regression
model was well specified or not. The result in Table 4.5 shows that observed F-statistic is
0.010906 with (1, 33) degree of freedom and estimated p-value of 0.9175. Hence, the null
hypothesis that ‘the model was not mis-specified’ was not rejected given the p-value of
0.9175 which is far greater than 0.05. The researcher therefore concludes that the model was
correctly specified. Hence, above corrected autocorrelation was independent of wrong model
specification. It is rather a case of pure autocorrelation.

Normality test

To maintain one of the key assumptions of OLS, the Jarque-Bera (JB) normality test was
employed in line with Jarque and Bera (1987). The statistic is an asymptotic test which
follows Chi-square distribution with 2 df. From Table 5.5 the researcher do not reject the null

61
hypothesis that ‘the residuals are normally distributed’ since estimated p-value of JB-statistic
(0.309826) is greater than 0.05. The result shows that the probability of rejecting a true null
hypothesis (Type I error) is reasonably high. As such the null hypothesis was not rejceted
(See Appendix 3A, 3B and 3C for diagnostic check for average growth rate of public
expenditure).

5.5 Regression result for determinants of public expenditure growth

The result below was obtained from OLS regression of oil revenue, non-oil revenue, public
debt, population growth rate, official external reserve, inflation rate, lag of government
expenditure and dummy for Fiscal Responsibility Act on government expenditure. The
objective is to investigate determinants of public expenditure growth in Nigeria. Statistical
inference and interpretation of the result is based on already established rules in section 5.4.1
above.

Table 5.6: Determinants of public expenditure growth

Variable Coefficient HAC Std Error t-statistic p-value

LOG(OLR) 0.326753 0.105287 3.103467 0.0041**
LOG(NOLR) 0.242974 0.086348 2.813887 0.0086**
LOG(PD) 0.062094 0.042475 1.461897 0.1542
D(PGR(-1)) 0.224977 0.090815 2.477318 0.0191*
D(OER(-1)) 2.63E-06 4.31E-06 0.611498 0.5455
INF(-1) 0.001315 0.002543 0.517110 0.6089
LOG(GE(-1)) 0.288735 0.156794 1.841485 0.0755
FRA 0.055295 0.137813 0.401235 0.6911
Constant 0.759440 0.084975 8.937235 0.0000**
R-squared 0.995684
DW-statistic 1.868273
Note: ** p-value < 0.01; * p-value < 0.05

Oil revenue (LOG(OLR))

The result presented above shows that tax revenue for oil sector has a positive coefficient
(0.326753). This implies that oil revenue contributes positively to public expenditure in
Nigeria. Thus, holding the effects of other variables constant, a percentage increase in oil
revenue leads to mean annual increase of about 32.68% in public expenditure. This is
explained by the fact that Nigeria government depends heavily on oil revenue as it accounts

62
for about 90% of export revenue and over 80% of the government budget (Anyaehie & Areji,
2015). Moreover, oil revenue in this study is statistically significant using estimated p-value
of its coefficient. The p-value which is 0.0041 is less than 0.05 and suggests that oil revenue
has significant impact on government expenditure in Nigeria. This study agrees with Okafor
and Eiya (2011) who found tax revenue from oil and non-oil sector as significant determinant
of government expenditure in Nigeria.

Non-oil revenue (LOG(NOLR))

The coefficient of non-oil revenue (0.242974) is positive and significant given estimated p-
value of 0.0086. The result suggests that when effects of all other variables are assumed to be
fixed, a percentage increase in non-oil revenue will lead to annual mean increase of about
24.3% in government expenditure. One of the most interesting features of this result is the
improvement in magnitude of non-oil revenue coefficient which tends to be insignificant in
some past studies including Ekeocha and Oduh (2012). This is an indication that Nigeria
economy is gradually becoming diversified with non-oil sector topping the chart as was
reported in National Bureau of Statistics rebased GDP presented by Kale (2014). In addition
the incidence of oil doom which befall Nigeria and other oil-dependent economy in recent
time aslo explain the sudden rise of non-oil sector contribution to government expenditure in
this study. This result is in line with apriori expectation as was predicted in section 4.2, page
51. It suggests that as non-oil revenue increase, government expenditure will also increase
because such tax income will provide the government with revenue to settle its expenditure.

Public debt (LOG(PD))

Unlike the result in Table 5.4, public debt has a positive but insignificant relationship with
government expenditure in Table 5.6. Given its estimated coefficient value of 0.062094 and
p-value of 0.1542, the result implies that ceteris paribus, a percentage rise in public debt will
translate to annual average increase of about 6.21.% in government expenditure. This effect is
however not statistically significantly different from zero at 5% level of significance because
estimated p-value of its coefficient of 0.1542 is greater than 0.05. This result best validated
above ranking of Nigeria by World Economic Forum’s Global Competitiveness Report 2014-
2015 as one of the countries with least incidence of public debt ratio to GDP. It simply
demonstrates that though Nigeria government borrows to finance its expenditure, it does not
rely on such borrowing. Rather, the government mainly finance its expenditure through tax

63
revenue from oil and non-oil sector as was shown by magnitude of their respective coefficient
values of 32.67% and 24.29%.

Population growth rate (D(PGR(-1)))

The coefficient of annual population growth rate displayed positive correlation with
government expenditure. From the result, a percentage increase in one year lag of population
growth rate leads to predicted annual average increase of about 22.5% in government
expenditure, all other variables held constant. Surprise enough, impact of population growth
rate significantly impacts on government spending in Nigeria as its p-value (0.0191) is less
than 0.05. The result supported the claim by Central Intelligence Agency (2014) which rated
Nigeria 33 rd out of 232 countries in terms of population growth rate. Given Nigeria’s position
as the 7th largest populous country with estimated population of about 177.8 million people in
2014 (Worldometers, 2015), its annual mean growth rate of 2.54% (Table 5.2) is by every
means a key factor which can influence government spending. This result is in line with
apriori expectation. The fact that population growth rate significantly influences government
expenditure was further explained by the report which placed Nigeria in 13th position out of
223 countries given its mean fertility rate of 5.25 children born per woman (Central
Intelligence Agency, 2014).

Official external reserve or foreign reserve (D(OER(-1)))

Although coefficient of Nigeria’s foreign reserve indicated that it positively correlate with
government expenditure (see appendix 6), it is however not significant. From Table 4.6 the
result shows that when effects of other variables are held constant, a million dollar increase in
a year lag of external reserve will on average increase government expenditure by
0.00000263 (0.000263%) per annum. The result conforms to apriori expectation. Its
insignificant impact on government expenditure can be readily explained by the fact that
government hardly choose the option of financing its expenditure through depletion of
foreign reserve. This is because the option can best be chosen during adverse severe
economic shocks (see section 4.1) which Nigeria rarely found itself during the span of this
study.

64
Inflation rate (INF(-1))

The coefficient of inflation rate is 0.001315 and its p-value is 0.6089. It demonstrates a
positive but insignificant relationship with government expenditure. Thus, when impact of all
other variables are assumed to be zero, a percentage increase in one year lag of inflation rate
will lead to 0.001315% mean annual increase in government expenditure. The result concurs
with economic apriori expectation. Its positive coefficient suggests that government spending
likely crowds out private investment (Han & Mulligan, 2008). As such, rise in inflation rate
leads to rise in government spending due to depreciating value of local currency in such
situation. While findings of this study contradicted that of Okafor and Eiya (2011) in sign, it
agrees with same in relative significance such that both found the impact of inflation rate to
be statistically insignificant.

Lag of government expenditure (LOG(GE(-1)))

Lagged value of government expenditure also has positive coefficient. It is in line with
economic apriori expectation since increase in past value of government expenditure will
likely result to increase in current government expenditure due to bureaucratic inertia,
especially as it concerns public sector spending. The result shows that ceteris paribus, a
percentage increase in past value of government expenditure leads to annual mean increase of
about 28.87% in current government expenditure. Although this result is statistically
insignificantly different from zero at 5% level due to its estimated p-value of 0.0755, it
implies that government spending in Nigeria is fairly underscore by incrementalism principle
especially in its annual budgeting system. The result does not however agree with Edame
(2014) who found lagged value of government expenditure to be negatively correlated with
its current value. The reason for positive correlation of past value of government expenditure
with its current value is obvious. Nigeria currently belongs to countries that practice line item
budgeting system where expenditure are made according to ‘line item’, which encourages
historical and incremental approach to budgeting without necessarily examining the result of
existing programmes as explained by Department for International Development (2005).

Fiscal Responsibility Act (FRA)

The coefficient of FRA in Table 5.6 shows that Fiscal Responsibility Act dummy is
positively related to government expenditure in this study. Its estimated coefficient
(0.055295) means that post-FRA period (2007-2014) contributes more to government

65
expenditure than pre-FRA period (1961-2006) by about 5.53%. Although impact of the
variable is not statistically significant due to its large p-value (0.6911), the relationship does
not tally with apriori expectation since the Act was expected to initiate transparency and
accountability and at the same time scale down wastefulness in government spending. The
result simply implies that when effects of all other variables are held constant, post-FRA
period contributes more to growth of government expenditure on average than pre-FRA
period by 5.53%. Hence, it can be said that while government expenditure grows on average
by 0.75944 from 1961 to 2006, it subsequently recorded average annual growth rate of about
0.814735 from 2007 to 2014. The argument form the result is that Fiscal Responsibility Act
has not been effective in actualizing its objectives, especially that which reads, “To secure
greater accountability and transparency in fiscal operations within a medium term fiscal
policy framework” (Federal Republic of Nigeria, 2007).

Government expenditure (GE)

The intercept coefficient is positive and suggests by how much government expenditure
grows in the absence of all the explanatory variables in Table 5.6. It is the autonomous rate of
growth captured by δ0 in equation 4.11. The result implies that when effects of other variables
are held constant, government expenditure will grow on average by 0.75944 per annum. This
coefficient is statistically significantly different from zero at 5% level of significance due to
its p-value which is 0.0000. The implication is that explanatory variables in the regression
analysis are not the only factors that explain the growth of government expenditure in
Nigeria.

5.5.1 Post-estimation diagnostic check for determinants of public expenditure growth

Similar to section 5.4.1 above, this presented the criteria for evaluation of regression result of
the determinants of public expenditure growth. As such estimates of R-squared test,
autocorrelation test, Ramsey’s regression specification error test and normality test were
interpreted as appeared in Table 5.7.

66
Table 5.7: Model fit and diagnostic check for determinants of public expenditure
growth

Test Estimated statistic p-value

R-squared 0.995684
F-statistic 865.1956 0.000000**
DW-statistic 1.868273
RESET 0.221625 0.6413
Normality Test 3.932887 0.490032
Note: ** p-value < 0.01; * p-value < 0.05

R-squared (Goodness of fit)

The regression model’s ‘goodness of fit’ captured by R-squared is very high to the tune of
0.995684. It suggests that explanatory variables in the model explained about 99.56% of
changes in government expenditure growth in Nigeria during the study period. This is an
indication of a good estimate since the proportion of unexplained variation in the dependent
variable (GE) is very insignificant (about 0.44%).

Test for autocorrelation

Given 39 observations and 9 parameters in the model, the null hypothesis of ‘no
autocorrelation’ was rejected at 5% level of significance since d U(2.085) ≮ d(1.868273) < 4-
dU(1.915). Hence, the researcher employed Newey-West heteroscedasticity and
autocorrelation-consistent (HAC) standard errors technique to correct for existing
autocorrelation in the estimated model.

Regression specification error test

After carrying out the regression specification error test (RESET) as was specified by
Ramsey (1969), the null hypothesis of ‘no misspecification’ was not rejected at 5% level of
significance given observed F-statistic (0.221625) with (1, 29) degree of freedom and p-value
(0.6413). The result implies that the model is correctly specified.

Normality test

The result form Jarque-Bera (JB) normality test employed to ascertain the normal distribution
of random errors (residuals) after OLS shows estimated JB-statistic (3.932887) and p-value
(0.490032). Thus, the null hypothesis that ‘residuals are normally distributed’ was not

67
rejected at 5% level of significance. The researcher concludes that residuals in the model
were normally distributed. See Appendix 4A, 4B and 4C for Eviews output of diagnostic
check for determinants of public expenditure growth.

5.6 Determinants of real gross domestic product in Nigeria

OLS estimate of effects of SPS, SPS2, GFCF, TO, TAX, M2 and INF on RGDP was carried
out in this section. The objective is to calculate the optimal size of the public sector using
estimated coefficient of SPS and SPS2. However, the result of impact of above explanatory
variables on RGDP will be interpreted first before the calculation of optimal size of the
public sector from output of Table 5.8 below.

Table 5.8: Regression result for determinants of real gross domestic product

Variable Coefficient HAC-Std Error t-statistic P-value

SPS 874.6039 191.4117 4.569228 0.0000**
2
SPS -16.4957 4.699807 -3.50987 0.0010**
GFCF -0.00083 0.000566 -1.46185 0.1506
TO 83.88714 38.30862 2.189772 0.0337*
TAX 2.607451 1.054108 2.473609 0.0171*
M2 2.097880 0.588602 3.564172 0.0009**
INF 20.35633 20.63339 0.986572 0.3290
Constant 3194.016 1962.226 1.627751 0.1104
R-squared 0.965116
DW-statistic 1.510105
Note: ** p-value < 0.01; * p-value < 0.05

Size of public sector or government expenditure ratio to GDP (SPS)

The result in Table 5.8 shows that coefficient of size of government sector (874.6039) is
positively related to real gross domestic product (RGDP). The result concurs with economic
apriori expectation as was prescribed by theoretical literature such as Barro (1989) among
others. On the other hand, it represented beneficial incidence of public sector spending in the
economy. Therefore, holding effects of other variables constant, a percentage increase in size
of government sector will lead to annual mean increase of about N874.60 billion in RGDP. In
order to validate this claim, the result also shows that estimated p-value of SPS coefficient
(0.0000) is statistically significantly different from zero even at 1% level. It is evidence
enough to confirm the usefulness of government expenditure in promoting economic growth

68
in Nigeria as was found in the studies by Ekeocha and Oduh (2012), Appah and Ateboh-
Briggs (2013) and Oyinlola and Akinnibosun (2013) among others. Hence, the primary
function of government expenditure which is provision of public goods is well justified in
this study.

Square of size of public sector (SPS2)

The coefficient of square of size of government sector is -16.4957. It shows a negative

relationship between square of government sector and real gross domestic product in Nigeria.
Above coefficient of SPS2 implies that when effect of other variables are assumed to be
insignificant, a percentage increase in square of size of government sector will on average
decrease real gross domestic product by approximately N16.5 billion per annum. This result
is statistically significant given estimated p-value of its coefficient (0.0010) which is less than
0.05. More so, the negative relationship between the two variables meets economic apriori
expectation as was contended by Leland (2005) and Facchini and Melki (2011) that
expansion of government expenditure beyond certain proportion of the GDP will result to
diminishing marginal returns on economic growth. This findings also validates the argument
put up earlier by the researcher that government expenditure should not be allow to grow
incessantly because of possible incidence of government failure in the system.

Gross fixed capital formation (GFCF)

The coefficient of gross fixed capital formation is -0.00083 which implies a negative
relationship between GFCF and RGDP. This relationship can be interpreted to mean that
when effects of other explanatory variables are held constant, a billion naira increase in gross
capital formation will lead to annual mean decrease of about N0.00083 million in real gross
domestic product. Ideally, it is expected that a rise in gross fixed capital formation induce a
rise in real gross domestic product since the former creates enabling environment (in terms of
opportunities) for growth of the latter. Thus this result does not conform to economic apriori
expectation. This notwithstanding, it is also not statistically significant in this study based on
its p-value (0.1506) which is greater than 0.05. This deviation could be the result of
consistent volatile and low investment attitude in Nigeria when it comes to fixed capital
formation and infrastructural development. A good example is consistent low rating of the
country based on quality of overall infrastructure in the past three editions of Global

69
Competitiveness Report of World Economic Forum where it ranks 117, 129 and 133 for
2012-2013, 2013-2014 and 2014-2015 respectively.

Trade openness or net export ratio to GDP (TO)

Table 5.8 shows that trade openness is positively related to real gross domestic product in this
study. With estimated coefficient of 83.88714 the result implies that ceteris paribus, a
percentage increase in trade openness results to average annual increase of about N83.89
billion in real gross domestic product. This output is in agreement with economic apriori
expectation because when net export ratio to GDP rises, domestic economy will benefit from
international trade and hence maximize its real gross domestic product. Furthermore, the
estimated relationship between TO and RGDP in this study proves to be statistically
significantly different from zero at 5% level given its p-value (0.0337) which is less 0.05.

Tax revenue (TAX)

Tax revenue displayed a positive relationship with real gross domestic product base on its
coefficient (2.607451). The positive coefficient displayed in the result conforms to economic
apriori expectation since it is rational that increased government revenue can induce
government capacity to afford more expenditure. Therefore, holding other variables constant,
a billion naira increase in tax revenue leads to annual mean increase of about N2.607 billion
in real gross domestic product. This result validated the theoretical linkage between income
and expenditure as speculated by Keynes in 1930s (Mankiw, 2010) and also agree with
empirical literature afterwards. To augument this assertion, the result revealed that impact of
tax revenue on real gross domestic product is statistically significantly different form zero
when judged with estimated p-value of its coefficient (0.0171).

Broad money supply (M2)

Similar to above relationship between TAX and RGDP, broad money supply also displayed
positive relationship with real gross domestic product as it generated coefficient value of
2.097880. It implies that holding effect of other variables constant, a billion naira rise in
broad money supply will lead to mean increase of approximately N2.1 billion in real gross
domestic product each year. This outcome is in conformity with economic apriori
expectation for the fact that when monetary authority increased the amount of money supply,
economic agents will take advantage of expanding investment activities as more money is

70
available to spend. In that way, more returns on investment will add up to the economy’s
level of real gross domestic product. From above result, estimated coefficient of M2 has
corresponding p-value of 0.0009 which shows that impact of broad money supply on real
gross domestic product is statistically significantly different from zero in this study. With this
information, broad money supply proved itself a core determinant of real gross domestic
product in Nigeria.

Inflation rate (INF)

The coefficient of inflation rate is positive (20.35633). It implies that a percentage increase in
inflation rate will lead to annual mean increase of about N20.36 billion in real gross domestic
product, all other variables held constant. Inflationary trend by its very nature has the
tendency to increase nominal gross domestic product but decrease real gross domestic
product due to its negative correlation with capital investment expenditure over time. Hence,
the result does not conform to economic apriori expectation. On the other hand, estimated
coefficient of inflation rate proves to have no significant impact on real gross domestic
product based on evidence of its p-value (0.3290) in Table 5.8 above.

Real gross domestic product (RGDP)

Coefficient of the intercept term (constant) in Table 5.8 is 3194.016. It measures the rate by
which real gross domestic product changes in the absence of any explanatory variables in the
model. Its positive coefficient implies that when effects of other variables in the model are
held constant, real gross domestic product will record mean annual increase of about N3194
billion. However, this impact is not statistically significantly different from zero at 5% level
since the p-value of its coefficient is 0.1104.

5.6.1 Post-estimation and diagnostic check for determinants of real gross domestic
product

In line with the approach in section 5.4.1 and 5.5.1 above, estimated R-squared,
autocorrelation test, Ramsey’s regression specification error test and normality test for output
of Table 5.8 were interpreted in this section.

71
Table 5.9: Model fit and diagnostic check for determinants of real gross domestic product

Test Estimated statistic p-value

R-squared 0.965116
F-statistic 181.8072 0.000000**
DW-statistic 1.510105
RESET 1.288544 0.1798
Normality Test 0.855929 0.651835
Note: ** p-value < 0.01; * p-value < 0.05

R-squared (Goodness of fit)

The R-squared value is 0.965116. It shows that about 96.5% of temporary variation in real
gross domestic product was explained by the independent variables in the model. It indicated
a good measure of goodness of fit due to its high value.

Test for autocorrelation

With 54 observations and 8 parameters in the model, the null hypothesis of ‘no
autocorrelation’ was rejected at 5% level of significance since d U(1.930) ≮ d(1.510105) < 4-
dU(2.07). The researcher therefore employed Newey-West heteroscedasticity and
autocorrelation-consistent (HAC) standard errors technique to correct for existing
autocorrelation and possible heteroscedasticity in the model.

Regression specification error test

The result of regression specification error test (RESET) in this model shows observed F-
statistic (1.288544) with (1, 45) degrees of freedom and p-value (0.1798). Based on the p-
value, the null hypothesis of ‘no model misspecification’ was not rejected at 5% level of
significance and conclusion was reached that the model is well specified at 5% level of
significance.

Normality test

Jarque-Bera (JB) normality test shows that JB-statistic is 0.855929 while its p-value is
0.651835. The researcher therefore do not reject the null hypothesis that ‘residuals are
normally distributed’ at 5% level of significance. Hence, conclusion was reached that
residuals in the model were normally distributed. See Appendix 5A, 5B and 5C for Eviews
output of diagnostic check for determinants of real gross domestic product.

72
5.7 Optimal size of the public sector in Nigeria

Following equations 4.14 and 4.15 in chapter three, optimal size of public sector was
computed by taking first derivative of equation 5.2 below with respect to SPS. From the
output, SPS was made subject of formula. To further complement the result from statistical
estimation, Armey curve was plotted in Excel worksheet using same values in equation 5.2.

From the regression output in Table 5.8, the following equation was generated;

RGDP = 3194.016 + 874.6039SPS - 16.4957SPS2 - 0.00083GFCF + 83.88714TO +

2.607451TAX + 2.09788M2 + 20.35633INF ... ... ... ... (5.1)

From equation 5.1, quadratic function for optimal size of public sector was presented below.

RGDP = 3194.016 + 874.6039SPS - 16.4957SPS2 ... ... ... (5.2)

First derivative of equation 5.2 yields

( )
= 874.6039 – 2(16.4957)SPS ... ... ... ... (5.3)
( )

Equating equation 5.3 to zero and calculating the optimal level of SPS gives;

0 = 874.6039 – 2(16.4957)SPS ... ... ... ... ... (5.4)

Making SPS subject of the formula gives;

.
SPS* = ⇒ 26.51 ... ... ... ... ... (5.5)
( . )

Graphically, optimal level of public sector spending in Nigeria was represented below using
estimate of equation 5.2.

73
Figure 5.1: Optimal level of federal government spending in Nigeria

16000 26.5%

14000

12000
Economic growth

10000

8000
Squared federal
6000 government expenditure
ratio to GDP
4000

2000

0
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43
Public sector spending ratio to GDP

Source: Plotted by the researcher from estimate of equation 5.2

Figure 5.1 show that government expenditure is an increasing function of GDP. The curve of
government expenditure ratio to GDP however attained its maximum at 26.51 along the
horizontal axis. It implies that as relative size of public sector grows, government expenditure
is channelled into less productive activities and thereby causing economic growth to
diminish. Marginal increase in economic growth however becomes zero at the point where
size of public sector is 26.51%. Further increase in relative size of public sector becomes
counterproductive, hence leading to decline in economic growth. From the perspective of
maximizing economic growth, Figure 5.1 show that any federal government spending as ratio
of GDP beyond 26.51% becomes harmful. The result is in line with theoretical postulation of
non-linear relationship between government expenditure and economic growth as was found
in Armey and Armey (1995), Rahn and Fox (1996) and Scully (1994, 2003). It also conform
to findings of empirical studies in Nigeria such as Ekeocha and Oduh (2012), Alimi (2014)
and Olaleye, et al. (2014) whose estimate predicted optimal level of government spending
value of 22.47%, 19.81% and 11% respectively.

Based on actual federal government spending in Nigeria, it can be seen that the threshold of
26.51% was exceeded between 1975 and 1977 when it spent 27.67%, 29.47% and 27.99%
respectively. This was also repeated in 1993 and 1999 with corresponding total relative

74
74
spending of about 34.19% and 29.67%. It is not surprising that Nigeria economy witnessed
decline in growth of real gross domestic product (RGDP) form 11.16% in 1974 to -5.23% in
1975. Although the economy recorded growth rate of 9.04% and 6.02% in 1976 and 1977, it
did not catch up with its previous rate in 1974. This period could be defined as an era of
reckless federal government spending of proceeds from oil boom of the 1970s. For instance,
Nigeria hosted the World Black and African Festival of Arts and Culture, popularly known as
FESTAC ’77 during the period. Similarly, the economy managed to grow by 2.09% in 1993
when the country was thrown in political tension by Babangida’s military government
annulment of popular June 12 presidential democratic election. Lastly, in 1999 when the
threshold was also exceeded, the economy boosted of only 0.47% growth when compared to
2.72% it recorded in 1998. The year was also remarkable for high government spending to
usher in new democratic government after 16 years of uninterrupted military reign.

5.8 Evaluation of research hypotheses

Working hypotheses of the study were evaluated based on regression results obtain from
estimated models in this work. It was based on statistical criteria (Significance of F- and t-
statistic) and econometric criteria (result of diagnostic checks).

Research hypothesis 1: Average growth rate of public expenditure in Nigeria is not

significant.

Average rate of public expenditure growth was captured by coefficient of time trend (T).
From regression result in Table 5.4, T recorded estimated coefficient, t-statistic and p-value
of 0.019059, 1.033831 and 0.3085 respectively. The decision rule is to reject null hypothesis
if p-value of t-statistic is less than 0.05, but do not reject if otherwise. Since 0.3085 is not less
than 0.05, the researcher does not reject the null hypothesis. The study therefore concludes
that annual average growth rate of public expenditure is not statistically significantly different
from zero at 5% level.

To confirm reliability of above decision, R-squared (0.993489) and p-value of F-statistic

(0.000000) in Table 5.5 suggest that goodness of fit of the model is very high and collective
impact of all the regressors on government expenditure is also highly significant. Although
DW-statistic (1.304686) suggest presence of first order serial correlation which was corrected
by employing HAC-standard error approach, null hypotheses of ‘no misspecification’ of the
regression model and that of ‘normal distribution’ of the residuals were not rejected at 5%

75
level of significance. Hence the conclusion that the model was well specified and its residuals
normally distributed. These are good indication that results from the model is dependable for
policy purpose.

Research hypothesis 2: The effect of determinants of public expenditure growth in Nigeria

is not significant.

Following above decision rule in hypothesis 1, output of determinants of public expenditure

growth in Table 5.6 shows that p-value of t-statistic for oil revenue (LOG(OLR)), non-oil
revenue (LOG(NOLR)) and population growth rate (D(PGR(-1))) which is 0.0041, 0.0086
and 0.0191 were all less than 0.05. The implication is that above null hypothesis of ‘no
significant effect’ was rejected for OLR, NOLR and PGR. The researcher thereby concludes
that while effects of oil revenue and non-oil revenue have significant impact on public
expenditure growth at 1% level, population growth rate has significant impact on public
expenditure growth at 5% level.

On the contrary, it was found that p-value of t-statistic for public debt (LOG(PD)), official
external reserve (D(OER(-1))), inflation rate (INF(-1)) and lag of government expenditure
(LOG(GE(-1))) is respectively, 0.1542, 0.5455, 0.6089 and 0.0755. Thus, since these p-
values were not less than 0.05, the researcher did not reject the null hypothesis and therefore
conclude that public debt, official external reserve, inflation rate and lag of government
expenditure do not have significant effect on public expenditure growth during the period
under study in Nigeria. When judged with collective level of significance as captured by
estimated F-statistic (865.1956) and its p-value (0.000000), the above null hypothesis was
rejected as 0.000000 is less than 0.05. Hence, conclusion of the study is that collective effect
of determinants of public expenditure growth in Nigeria is highly significant at 5% level.

Once more, diagnostic check after the estimate reveal that the model was well fitted up to
99.56%. Comparable to the result in hypothesis 1, the model has first order serial correlation
which was once again corrected by adopting HAC-standard error. In addition, the null
hypotheses of ‘correct model specification’ and ‘normality assumption’ of the residuals were
not rejected at 5% level of significance due to p-values of their respective statistic (0.6413)
and (0.490032). See Table 4.7 for diagnostic checks.

76
Research hypothesis 3: There is no significant effect of Fiscal Responsibility Act on the
rate of public expenditure growth in Nigeria.

With positive coefficient (0.055295) which shows that period of introduction of Fiscal
Responsibility Act (2007-2014) witnessed higher annual mean growth rate of public
expenditure than the period before (1961-2006) by about 0.0553. Since the p-value (0.6911)
is not less than 0.05, above null hypothesis was not rejected. Based on this evidence, the
study therefore conclude that introduction of Fiscal Responsibility Act has no effect on
growth rate of public expenditure in Nigeria.

Research hypothesis 4: Optimal size of public sector in Nigeria is less than 15%.

Equation 5.5 and Figure 5.1 show that estimated optimal size (level) of public sector in
Nigeria is 26.51%. The null hypothesis that ‘optimal size of public sector in Nigeria is less
than 15%’ is therefore rejected since estimated value (26.51%) is not less than predicted
value (15%).

This decision was confirm by result of model fit and diagnostic check in Table 5.9. Based on
R-squared value, the model achieved about 96.5% goodness of fit. Collective impact of
regressors in the model on real gross domestic product is once again very significant as p-
value of its F-statistic (0.000000) is less than 0.05. Like in previous cases where null
hypothesis of first order serial correlation was not rejected given DW-statistic (1.510105) in
this model, HAC-standard error approach was again adopted. Further investigation shows
that null hypotheses of ‘no misspecification’ of the regression model and ‘normal
distribution’ of the residuals were not rejected at 5% level of significance given p-values
(0.1798) and (0.651835) of their respective statistic.

77
 CHAPTER SIX

SUMMARY, CONCLUSION AND POLICY RECOMMENDATION

6.1 Summary of major findings

The research topic, ‘Economic analysis of public expenditure growth and optimal size of
public sector in Nigeria’ was chosen due to dominant role played by government expenditure
in modern economies, especially after the rise of Keynesian economics in late 1930s.
Interestingly enough, the choice of object of government expenditure and amount of such
expenditure have been a good index for measuring success in provision of public goods. Such
goods which include maintenance of law and order, provision of social security and
regulation of the market economy among others are in no doubt the key facilitators of
economic growth and even distribution of wealth. Economists recently however, observed
that government expenditure does not have perfectly elastic positive correlation with
economic growth. Rather, it was discovered that unregulated growth of government spending
may result to decline in economic growth and rise in inequality in the society. As a result, this
study became interested and considered it important objective to ascertain average growth
rate of public expenditure, investigate determinants of public expenditure growth, examine
effect of Fiscal Responsibility Act (FRA) on growth rate of public expenditure and estimate
optimal size of public sector in Nigeria.

In attempt to realize above objectives, annual time series data was sourced from Central Bank
of Nigeria Statistical Bulletin 2009 and 2014, and World Bank’s World Development
Indicators 2015. It covered observations from 1961 to 2014. In addition, a dummy for Fiscal
Responsibility Act was generated and assumed value of 0 for observations from 1961-2006
(pre-FRA) and 1 for observations from 2007-2014 (post-FRA). Above data were subjected to
Ordinary Least Square econometric technique using Eviews 8.0 econometric software.

In line with objective one of the study, it was found that government expenditure recorded
annual average growth rate of 1.91%. This growth rate is however considered statistically
insignificant in the study. From estimate of determinants of public expenditure growth in
objective two, oil revenue, non-oil revenue and population growth rate distinguished
themselves as key determinants of public expenditure growth in Nigeria. They were
individually found to exert considerable positive impact on public expenditure growth. On
the other hand, other determinants such as public debt, official external reserve (foreign

78
reserve), inflation rate and lag of government expenditure (bureaucratic inertia) also have
positive impact on public expenditure. However, their individual impact on government
expenditure was considered statistically insignificant.

With respect to impact of Fiscal Responsibility Act (FRA) on growth rate of public
expenditure as stated in objective three, the result show that post-FRA period (2007-2014)
contributed about 5.53% more to annual mean public expenditure growth when compared
with the period before FRA (1961-2006). Effect of this differential intercept was however not
statistically significant. This implies that the legal document has no influence on growth rate
of public expenditure in Nigeria. From the result on determinants of real gross domestic
product (RGDP) in Table 5.8, it was found that while relative size of public sector has
positive significant effect on RGDP (showing beneficial incidence of government
expenditure), square of relative size of public sector has negative significant effect on RGDP
(indicating the decreasing marginal productivity of government spending). Furthermore, trade
openness (net export ratio to GDP), tax revenue and broad money supply also counted as key
determinants of RGDP as they all show positive significant impact. On the contrary, while
gross fixed capital formation has positive but insignificant effect on RGDP, inflation rate has
positive but also insignificant effect on RGDP as well. Finally, based on above result, optimal
size of the public sector (optimal level of government expenditure as ratio of GDP) was
calculated to be approximately 26.51%. The implication is that government expenditure
below or equal to 26.51% is acceptable, but any government expenditure above the threshold
should be discouraged as it will likely have negative impact on overall economic growth and
welfare.

6.2 Conclusion

Based on statistical estimates in chapter five, the study concluded that observed 1.91% annual
average growth rate of public expenditure is not detrimental to economic growth in Nigeria.
The reason resonates from the fact that key determinants of public expenditure growth such
as tax revenue (from oil and non-oil) and population growth rate consistently rise throughout
the study period. As such, public expenditure is expected to increase in magnitude so as to
provide for the growing population, insofar as there is commensurate inflow of revenue. In
conformity with most empirical studies in Nigeria, the study also arrived at the conclusion
that tax revenue from oil sector is the most influential determinant of public expenditure as it
accounted for about 32.66% in mean annual growth rate of public expenditure. Oil revenue

79
was previously claimed to have constituted about 80% of government revenue and 90% of
foreign exchange of the economy (Anyaehie & Areji, 2015). In terms of relative importance,
tax revenue from oil sector was closely followed by tax revenue from non-oil sector and
population growth rate which accounted for 24.29% and 22.49% respectively. Furthermore,
the Fiscal Responsibility Act which was introduced to promote transparency and
accountability, and as well promote fiscal discipline has no effect on growth rate of public
expenditure in the country. Hence its core objective of minimizing wasteful spending is yet to
be realized.

Result of this study supported the argument for existence of non-linear relationship between
government expenditure and economic growth in Nigeria as was demonstrated in Armey
curve after Armey and Armey (1995). Thus, government expenditure should not be allowed
to grow incessantly in order to avoid economic downturn from possible government failure.
As a matter of fact, any government expenditure beyond 26.51% of gross domestic product
could be detrimental for economic growth. Succinctly put, the study concludes that
democratic government in Nigeria from year 2000 to date has been on the right track with
respect to relative size of its expenditure to GDP as was depicted by the rising slope of
government expenditure curve in Panel B of Figure 3.1 and 5.1 respectively. This is because
this period witnessed relative size of the public sector accommodated within stipulated
threshold (optimal level) of 26.51% in this study. Consequently, the government may
consider increasing its current share of 10% so as to increase opportunity for sustainable
growth and development through provision of infrastructure and security.

6.3 Policy recommendation

Among key findings of this thesis report is that government expenditure grows by annual
average of about 1.91%, of which was classified ‘not harmful’; tax revenue (from oil and
non-oil sector) and population growth rate are the prime growth drivers of government
expenditure; Fiscal Responsibility Act of 2007 has no influence on government expenditure
growth rate; and optimal level of relative size of public sector (government expenditure ratio
to GDP) in Nigeria is 26.51%. In line with this result therefore, the following policy
recommendations were made.

§ When placed side by side the fact that Nigeria is witnessing severe incidence of
poverty, and result of this study which shows that population growth rate is an

80
 important determinant of government expenditure, it becomes appropriate to reason
that annual government expenditure growth of 1.91% may not be sufficient to contend
annual population growth of over 2.54%. Hence government should consider the
option of either putting up policy to regulate population growth or step up its
intervention in the economy so as to create enabling environment for self-sustaining
growth and development with greater capacity for poverty alleviation.
§ The evidence that tax revenue from oil sector still controls the largest proportion as
determinant of government expenditure implies that Nigeria is yet to achieve its
economic diversification target. As a result, the most crucial lesson from effect of
current deflationary trend in oil price at international market should be that oil
dependent states such as Nigeria live at the mercy of industrialized economies that set
agenda for international oil price. Hence, it becomes pertinent for Nigerian
government to quickly plan its way out of oil-dependency through sound and feasible
investment policy. Such policy should focus on agriculture, manufacturing, solid
mineral development, entrepreneurship and service industry.
§ Having arrived at 26.51% as optimal level of relative size of public sector in Nigeria,
the government may increase share of its annual expenditure insofar as the object of
such expenditure commands higher multiplier effect in the economy. However, care
should be taken to ensure that the threshold is not exceeded. At the same time, it
should be ensure that wastefulness is at minimum. This feat can be achieved through
legislation that promotes transparency and accountability in government business.

6.4 Limitation of the study and recommendation for further research

The effort and success demonstrated in this thesis report does not imply that the study has
been without possible constraints and limitations. Among these include availability of data on
government revenue and expenditure for both federal and state governments given the scope
of the study (1961-2014). This was what informed the choice of restricting the enquiry to
accommodate economic analysis of public expenditure growth and optimal size of public
sector with focus on the federal government of Nigeria alone. The researcher equally
encountered some difficulties in accessing published documents for literature review
especially as regards optimal size of public sector in Nigeria. This is because only very few
reliable studies were done. Lastly, the study was constrained by availability of scarce
resources such as cash requirements for funding the research work. The scarce resources also

81
included time allocated for submission of Master of Science degree thesis report by the
university authority.

In recognition of above constraints, the researcher thereby recommends the following areas
of study for future interested researchers;

§ Analysis of public expenditure growth and optimal size of public sector in chosen
state or region in Nigeria, or in Africa.
§ Econometric investigation of government expenditure, poverty reduction and human
capital development in Nigeria.

82
 REFERENCES

Abdullah, H. A. (2000). The relationship between government expenditure and economic

growth in Saudi Arabia. Journal of Administrative Science, 12(2), 173-191.

Adewole, A. N., & Osabuohien, E. S. (2007). Analysis of the cost of governance and options
for its reduction in Nigeria. Nigeria Journal of Economic and Social Studies, 49(1),
137-159.

Aghonkhese, A. O., & Asekome, M. O. (2014). Impact of public expenditure on the growth
of Nigerian economy. European Scientific Journal, 10(28), 219-229.

Alexiou, C. (2009). Government spending and economic growth: Econometric evidence from
the South Eastern Europe. Journal of Economic and Social Research, 11(1), 1-16.

Alimi, R. S. (2014). Does optimal government size exist for developing economies? The case
of Nigeria. Munich Personal RePEc Archive, 1-18. Retrieved Apirl 2, 2015, from
http://mpra.ub.uni-muenchen.de/56073

Altunc, O. F., & Aydin, C. (2013). The relationship between optimal size of government and
economic growth: empirical evidence from Turkey, Romania and Bulgaria. Social
and Behavioral Sciencess, 92(2), 66-75.

Aly, H., & Strazicich, M. (2006). Is government size optimal in the gulf countries of the
Middle East? An empirical investigation. Economic Inquiry, 39(3), 79-88.

Anyaehie, M. C., & Areji, A. C. (2015). Economic diversification for sustainable

development in Nigeria. Open Journal of Political Science, 5, 87-94.

Appah, E., & Ateboh-Briggs, P. B. (2013). Cointegration of public sector expenditure

patterns and growth of Nigeria. Developing Country Studies, 3(9), 174-188.

Armey, D., & Armey, R. (1995). The Freedom Revolution: the new Republican House
Majority Leader tells why big government failed, why freedom works, and how we
will rebuild America. Washington, D.C.: Regnery Publishing Inc.

Arowolo, D. (2011). Fiscal federalism in Nigeria: theory and dimensions. Afro Asian Journal
of Social Sciences, 2(2), 1-21.

Barro, R. (1989). A Cross-Country Study of Growth, Saving and Government. New York:
NBER Working paper No. 2855, 1-27.

Barro, R. J. (1990). Government spending in a simple model of endogenous growth. The

Journal of Political Economy, 98, 103-125.

Bashir, K. A. (2008). The Mechanics of Revenue Allocation: Understanding the Need for
Effective Data Collection and Management. London: Hutchinson and Co Ldt.

Bird, R. M. (1971). Wagner's law of expanding state activity. Public Finance, 26.

83
Black, D. (1969). On the rationale of group decision making. In K. J. Arrow, & T. Scitovsky
(Eds.), Readings in Welfare Economics (pp. 52-78). London: George Allen and
Unwin.

Blanchard, O., Amighini, A., & Giavazzi, F. (2010). Macroeconomics: A European

perspective. Essex: Pearson Education Limited.

Brown, C. V., & Jackson, P. M. (1990). Public Sector Economics. Oxford: Blackwell
Publishers Limited.

Buchanan, J. M., & Tullock, G. (1962). The Calculus of Consent. Ann Arbor: University of
Michigan Press.

Budget Office of the Federation. (2014). Budget Appropriation Act 2013 & 2014. Abuja:
Budget Office of the Federation.

Central Bank of Nigeria. (2009). Statistical Bulletin 2009. Abuja: Central Bank of Nigeria.

Central Bank of Nigeria. (2012). Statistical Bulletin 2012. Abuja: Central Bank of Nigeria.

Central Bank of Nigeria. (2014). Statistical Bulletin 2012. Abuja: Central Bank of Nigeria.

Central Intelligence Agency. (2014). The world factbook. Retrieved November 21, 2015 from
Centaral Intelligence Agency Web site: http://www.cia.gov

Chobanov, D., & Maldenova, A. (2009). What is the optimum size of government. Journal of
Institute for Market Economics, 1-47.

Chude, N. P., & Chude, D. I. (2013). Impact of government expenditure on economic growth
in Nigeria. International Journal of Business and Management Review, 1(4), 64-71.

Connolly, S., & Munro, A. (1999). Economics of the Public Sector. Essex: Pearson Education
Limited.

Dalton, H. (1920). The measurement of inequality of incomes. The Economics Journal,

30(119), 348-361.

Department for International Development. (2005). Public Expenditure Managment Training,

January 14-18. Retrieved Novermber 16, 2015 from Department for International
Development Web site: http://www.gov.uk/government/organisations/department-for-
international-development

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, 427-431.

Doessel, D. P., & Valadkhani, A. (2003). Theory and econometric evidence explaining public
expenditure: the case of Iran. Research Online, 1-21. Retrieved June 28, 2015, from
http://ro.uow.edu.au/commpapers/406

84
Downs, A. (1957). An Economic Theory of Democracy. New York: Harper and Row.

Durbin, J., & Watson, G. S. (1951). Testing for serial correlation in least-squares regression.
Biometrika, 38, 159-171.

Edame, G. E. (2014). Trends analysis of public expendiure on infrastructure and economic

growth in Nigeria. International Journal of Asian Social Sciences, 4(4), 480-491.
Retrieved January 11, 2015, from http://www.aessweb.com/journals/5007

Egbetunde, T., & Fasanya, I. O. (2013). Public expenditure and economic growth in Nigeria:
evidence from Auto-Regressive Distributive Lag specification. Zagreb International
Review of Economics & Business, 16(1), 79-92.

Ejeh, A. W., & Orokpo, O. F. (2014). Fiscal federalism in Nigeria: an analysis of issues and
challenges. International Journal of Peace and Conflict Studies, 2(1), 37-45.

Ejuvbekpokpo, S. (2012). Cost of governance on economic development in Nigeria. Global

Journal of Management and Business Research, 12(13), 19-23.

Ekeocha, P. C., & Oduh, M. O. (2012). Determining the optimal size of federal government
in Nigeria: an empirical investigation. 68th International Institute of Public Finance
(IIPF), August 16th-19th (pp. 1-18). Dresdan, Germany: International Institute of
Public Finance. Retrieved March 26, 2015, from
http://www.thebrokenwindow.net/papers/1/IIPF68-416.pdf

Ekinci, A. (2011). What is the optimal size of government: a suggestion. International

Journal of Economics and Finance Studies, 3(1), 45-54.

Ekpo, A. (2004). Intergovernmental fiscal relations: the Nigerian experience. 10th Year
Anniversary of the Financial and Fiscal Commission of South Africa, 10-12, August
(pp. 1-19). Cape Town: Financial and Fiscal Commission of South Africa.

Facchini, F., & Melki, M. (2011). Optimal government size and economic growth in France
(1817-2008): an explanation by the state and market failures. Working Paper on
Political Economy and New Perspectives on Public Debt, September 26-28, 2011 at
the Department of Law and Economics of Productive Activities, Faculty of
Economics, Sapienza University of Rome (pp. 1-37). Paris: Centre d'Economie de la
Sorbonne.

Federal Ministry of Finance. (2014). 2014-2016 Medium Term Expenditure Framework and
Fiscal Strategy Paper. Abuja: Federal Ministry of Finance.

Federal Republic of Nigeria. (2007). Fiscal Responsibility Act. Abuja: Federal Republic of
Nigeria.

Fluvian, G. (2006). The rising cost of governance in Africa. Journal of Public

Administration, 22(9), 441-463.

85
Friedman, M. (1997). If only the U.S. were as free as Hong Kong. Wall Street Journal, July
8, A14.

Glomm, J. J., & Ravikumar, D. J. (1997). The growth of public expenditure in selected
developing nations: six Caribbean countries. Public Finance/Finances Publique, 3(3),
57-74.

Greene, W. H. (2012). Econometric Analysis (7 ed.). New Jessy: Pearson Education Inc.

Gujarati, N. D., & Porter, D. C. (2009). Basic Econometrics (5 ed.). New York: McGraw-
Hill.

Gunalp, D., & Dincer, O. C. (2005). The Optimal Government Size in Transition Countries.
Retrieved March 19, 2015 from Economics Today Website:
http://ssrn.com/abstract=832605

Han, S., & Mulligan, C. B. (2008). Inflation and the size of government. Review, 90(3), 245-
267.

Herath, S. (2010). The Size of the Government and Economic Growth: An empirical stduy of
Sri Lanka. Wien, Austria: Institute for the Environment and Regional Development.

Herath, S. (2012). Size of government and economics growth: a nonlinear analysis. Economic
Annals, 58(194), 17-29.

Ighodaro, C. A., & Oriakhi, D. E. (2010). Does the relationship between government
expenditure and economic growth follow Wagner's law in Nigeria? Annals of the
University of Petrosani, Economics, 10(2), 185-198.

Ikeji, C. C. (2011). Polictics of revenue allocation in Nigeria: a reconsideration of some

contending issues. Sacha Journal of Policy and Strategic Studies, 1(1), 121-136.

Jiranyakul, K. (2007). The Relation Between Government Expenditure and Economic Growth
in Thailand. Turdue: National Institute of Development Administration, Purdue
University North Central.

Kale, Y. (2014). Measuring better. Presentation of preliminary results of the rebased nominal
gross domestic product (GDP) estimates for Nigeria 2010 to 2013 (pp. 1-46). Lagos:
National bureau of Statistics.

Karras, G. (1996). The optimal size: Further international evidence on the productivity of
government services. Economicy Inquiry, 34(2), 193-203.

Keynes, J. M. (1936). The general theory of employment, interest, and money. Retrieved
Apirl 16, 2015 from The University of Adelaide Library:
http://etext.library.adelaide.edu.au/k/k44g/k44g.zip

86
Leland, S. (2005). Fiscal characteristics of public expenditures. In D. Robbins (Ed.),
Handbook of Public Sector Economics (pp. 213-239). Florida: CRC Press LLC.

Lewis-Bynoe, D., Winston, M., & Kerry-Ann, A. (2004). An Assessment of the Growth-
Enhancing Size of Government in the Caribbean. Barbodos: Annual Review Seminar,
Research Department of Central Bank of Barbodos.

Liu, C.-H., Hsu, C., & Younis, M. (2008). The association between government expenditure
and economic growth: the Granger Causality test of US data, 1974-2002. Journal of
Public Budgeting, Accounting and Financial Management, 20(4), 439-452.

Louis, N. S. (2012). Governement expenditure and economic growth in Nigeria: an empirical

investigation (1961-2009). The Journal of Economic Analysis, 3(1), 38-51.

Lukpata, V. I. (2013). Revenue allocation formulae in Nigeria: a continous search.

International Journal of Public Administration and Management Research, 2(1), 32-
38. Retrieved January 19, 2015, from http://www.rcmss.com

Mankiw, N. G. (2010). Macroeconomics (7 ed.). New York: Worth Publishers.

Mavrov, H. (2007). The size of government expenditure and the rate of economic growth in
Bulgaria. Economic Alternatives, 2(1), 53-63.

Moreno-Dodson, B. (2008). Assessing the Impact of Public Spending on Growth: An

empirical analysis for seven fast-growing countries. Washington DC: World Bank
Policy Research Working Paper 4663.

Mueller, D. C. (2003). Public Choice III. Cambridge: Cambridge University Press.

Musgrave, R. A. (1969). Fiscal System. New Haven: Yale University Press.

Musgrave, R. A. (1974). Expenditure policy for development. In D. T. Geithman, Fiscal

Policy for Industrialization and Development in Latin America. Gainsville: University
of Florida Press.

Musgrave, R. A., & Musgrave, P. B. (2004). Public Finance in Theory and Practice. New
Delhi: Tata McGraw-Hill.

Mutascu, M., & Milos, M. (2009). Optimal size of government spending: the case of
European Union member states. Annales Universitatis Apulensis Series Oeconomica,
11(1), 447-456.

Nasiru, I. (2012). Government expenditure and economic growth in Nigeria: cointegration

analysis and causality testing. Academic Research International, 2(3), 718-723.
Retrieved March 6, 2015 from http://www.journals.savap.org.pk

Niskanen, W. A. (1971). Bureaucracy and Representative Government. Chiago: Aldine

Publishers.

87
Nurudeen, A., & Usman, A. (2010). Government Expenditure and Economic Growth in
Nigeria, 1970-2008: a disaggregated analysis. Business and Economics Journal, 4(3),
237-330. Retrieved January 19, 2015 from http://astoujournals.com

Nwosa, P. I., Adebiyi, O. O., & Adedeji, A. O. (2013). An analysis of the relationship
between public spending components and private investments in Nigeria. Jounral of
Finance & Economics, 1(2), 14-27.

Okafor, C., & Eiya, O. (2011). Determinants of growth in government expenditure: an

empirical analysis of Nigeria. Research Journal of Business Management, 5(1), 44-
50.

Olaleye, S. O., Edun, F., Bello, H. T., & Taiwo, S. B. (2014). Government expenditure and
economic growth: an empirical analysis of the Armey Curve in Nigeria. the Romanian
Economic Journal, 47- 66.

Olaopa, T. (2013). A critical overview of public sector reform in Nigeria: framework of

transformation for the federal civil service. Retrieved February 7, 2015, from Public
Sector: http://www.tunjiolaopa.com/

Olulu, R. M., Erhieyovwe, E. K., & Ukavwe, A. (2014). Government expenditures and
economic growth: the Nigerian experience. Mediterranean Journal of Social
Sciences, 5(10), 89-94. doi:10.5001/mjss.2014.

Omotoso, F. (2010). Nigerian fiscal federalism and revenue allocation formula for sustainable
development. The Social Science, 5(3), 246-253. doi:10.3923/sscience.2010.246.253

Oriakhi, D. E., & Arodoye, L. N. (2013). The government size - economic growth
relationship: Nigerian econometric evidence using a Vector Autoregressive Model.
International Journal of Business and Management, 8(10), 126-133.

Oyinlola, M. A., & Akinnibosun, O. (2013). Public expenditure and economic growth nexus:
further evidence from Nigeria. Journal of Economics and International Finance, 5(4),
146-154. doi:10.5897/JEIF2013.0489

Peacock, A. T., & Wiseman, J. (1961). The Growth of Public Expenditure in the United
Kingdom. Princeton: Princeton University Press.

Peden, E. A. (1991). Productivity in the United States and its relationship to government
activity: an analysis of 57 years, 1929-1996. Public Choice, 69(4), 74-92.

Pevcin, P. (2004). Does Optimal Size of Government Spending Exist? Ljubljana: University
of Ljubljana.

Piana, V. (2001). Public Expenditure. Retrieved September 13, 2014, from Public Sector
Economics: http://www.economicswebinstitute.org/glossary/pubexp.htm

Pigou, A. C. (1928). A Study in Public Finance. London: Macmillan.

88
Pugel, T. A. (2007). International Economics (13 ed.). New York: McGraw-Hill/Irwin.

Rahn, R., & Fox, H. (1996). What is the Optimum Size of Government. Chicago: Verono K.
Krieble Foundation.

Ram, R. (1986). Government size and economic growth: a new framework and some
evidence from cross-section and time-series data. American Economic Review, 76(1),
191-203.

Ramsey, J. B. (1969). Tests for specification errors in classical linear least-squares analysis.
Journal of the Royal Statistical Association, Series B(71), 350-371.

Ranjan, K. D., & Sharma, C. (2008). Government expenditure and economic growth:
evidence from India. The ICFAI University Journal of Public Finance, 6(3), 60-69.

Rodden, J. (2003). Reviving Leviathan: fiscal federalis and the growth of government.
International Organization, 57, 695-729.

Rostow, W. W. (1971). Politics and the Stages of Growth. Cambridge: Cambridge University
Press.

Rutherford, D. (2002). Routledge Dictionary of Economics. London: Taylor & Francis

Group.

Sagay, I. (2008). How a true federal system should run. The Nation, May 19, pp. 48-49.

Scully, G. (1994). What is the Optimal Size of Government in the US? Texas: National Center
for Policy Analysis.

Scully, G. (1998). Measuring the Burden of High Taxes. Texas: National Centre for Policy
Analysis Policy Report No. 215.

Scully, G. (2003). Optimal taxation, economic growth and income inequality. Public
Choice115, 299-312.

Tarschys, D. (1975). The grwoth of public expenditure: nine modes of explanation.

Scandinavian Political Studies, 26-47.

Thanh, L. (2009). A dual economy model of endogenous growth with R&D and market
structure. In U. Cantner, J. L. Gaffard, & L. Nesta (Eds.), Schumpeterian Perspectives
on Innovation, Competition and growth (pp. 97-114). Berlin: Springer-Verlag.

Tinbergen, J. (1978). Economic Policy: principles and design. Amsterdam: North-Holland .

Udah, E. B. (2010). Macroeconomic reforms, government size, and investment behaviour in

Nigeria: an empirical investigation. Journal of Sustainable Development in Africa,
12(1), 258-280.

89
UNESCO-Nigeria-Project. (2010). UNESCO-Nigeria Project for Revitalisation of TVE in
Nigeria: Instructional/Teaching Material 2010. Retrieved Februray 27, 2015, from
Ministry of Economic Planning Web Site: http://unesco-nigeriatve.org

Vaish, M. C. (2002). Macroeconomic Theory (12 ed.). New Delhi: Vikas Publishing House.

Vedder, R. K., & Gallaway, L. E. (1998). Government size and Economic Growth. Ohio:
Washington Press.

Wooldrige, J. M. (2013). Introductory Econometrics: A modern approach (5 ed.). Ohio:

South-Western Publishers.

World Bank. (2015). World development indicators. Retrieved November 12, 2015 from The
World Bank: http://data.worldbank.org

World Economic Forum. (2014). The global competitiveness report 2014-2015. Geneva:
World Economic Forum.

Worldometers. (2015). Nigeria Population. Retrieved March 24, 2015, from

www.worldometers.info: http://www.worldometers.infor/world-population/nigeria-
population

Yasin, M. (2003). Public spending and economic growth: empirical investigation of Sub-
Saharan Africa. Southwestern Economic Review, 30(1), 59-68.

Zhou, Y. (2007). “International reserves and fiscal policy in developing countries.” SCCIE
Working Paper No.06-18 (pp. 1-33). Califonia: Santa Cruz Centre for International
Economics.

90
 APPENDIX

Appendix 1: Stationarity test

Government expenditure (GE) Gross fixed capital formation (GFCF)

Null Hypothesis: GE has a unit root Null Hypothesis: D(GFCF) has a unit root
Exogenous: None Exogenous: None
Lag Length: 10 (Automatic - based on SIC, maxlag=10) Lag Length: 4 (Automatic - based on SIC, maxlag=5)

t-Statistic Prob.* t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -3.002409 0.0036 Augm ented Dickey-Fuller test statistic -5.395870 0.0000
Test critical Test critical
values: 1% level -2.619851 values: 1% level -2.614029
5% level -1.948686 5% level -1.947816
10% level -1.612036 10% level -1.612492

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

Augmented Dickey-Fuller Test Equation Augm ented Dickey-Fuller Test Equation

Dependent Variable: D(GE) Dependent Variable: D(GFCF,2)
Method: Least Squares Method: Least Squares
Date: 11/14/15 Time: 00:23 Date: 11/18/15 Time: 15:59
Sample (adjusted): 1972 2014 Sample (adjusted): 1967 2014
Included observations: 43 after adjustments Included observations: 48 after adjustments

Variable Coefficient Std. Error t-Statistic Prob. Variable Coefficient Std. Error t-Statistic Prob.

GE(-1) -0.909131 0.302800 -3.002409 0.0052 D(GFCF(-1)) -7.451901 1.381038 -5.395870 0.0000
D(GE(-1)) 0.825822 0.379057 2.178623 0.0368 D(GFCF(-1),2) 7.607592 1.559154 4.879307 0.0000
D(GE(-2)) 2.439844 0.391149 6.237637 0.0000 D(GFCF(-2),2) 8.130172 1.661690 4.892712 0.0000
D(GE(-3)) 1.821361 0.409233 4.450675 0.0001 D(GFCF(-3),2) 8.184332 1.623600 5.040855 0.0000
D(GE(-4)) 0.513121 0.470489 1.090612 0.2836 D(GFCF(-4),2) 9.383075 1.870268 5.016967 0.0000
D(GE(-5)) 0.731233 0.527833 1.385348 0.1755
D(GE(-6)) 1.210139 0.550733 2.197322 0.0354 -
R-squared 0.620084 Mean dependent var 11504.06
D(GE(-7)) 0.467764 0.520086 0.899398 0.3752
Adjusted R-squared 0.584743 S.D. dependent var 1181358.
D(GE(-8)) 0.135598 0.494855 0.274015 0.7858
S.E. of regression 761272.4 Akaike info criterion 30.02170
D(GE(-9)) 2.133055 0.503952 4.232654 0.0002
Sum squared resid 2.49E+13 Schwarz criterion 30.21662
D(GE(-10)) 2.184842 0.399247 5.472411 0.0000 Hannan-Quinn
Log likelihood -715.5209 criter. 30.09536
R-squared 0.826381 Mean dependent var 106.4433 Durbin-Watson stat 1.674926
Adjusted R-squared 0.772125 S.D. dependent var 260.5182
S.E. of regression 124.3616 Akaike info criterion 12.70043
Sum squared resid 494906.1 Schwarz criterion 13.15097
Hannan-Quinn
Log likelihood -262.0592 criter. 12.86657
Durbin-Watson stat 1.743079

91
Inflation rate (INF) Broad money supply (M2)

Null Hypothesis: INF has a unit root Null Hypothesis: D(M2,2) has a unit root
Exogenous: None Exogenous: None
Lag Length: 0 (Automatic - based on SIC, maxlag=10) Lag Length: 8 (Automatic - based on SIC, maxlag=10)

t-Statistic Prob.* t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -2.251870 0.0248 Augm ented Dickey-Fuller test statistic -4.146951 0.0001
Test critical Test critical
values: 1% level -2.609324 values: 1% level -2.619851
5% level -1.947119 5% level -1.948686
10% level -1.612867 10% level -1.612036

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

Augmented Dickey-Fuller Test Equation Augm ented Dickey-Fuller Test Equation

Dependent Variable: D(INF) Dependent Variable: D(M2,3)
Method: Least Squares Method: Least Squares
Date: 11/14/15 Time: 00:26 Date: 11/14/15 Time: 00:29
Sample (adjusted): 1962 2014 Sample (adjusted): 1972 2014
Included observations: 53 after adjustments Included observations: 43 after adjustments

Variable Coefficient Std. Error t-Statistic Prob. Variable Coefficient Std. Error t-Statistic Prob.

INF(-1) -0.178172 0.079122 -2.251870 0.0286 D(M2(-1),2) -8.129988 1.960473 -4.146951 0.0002
D(M2(-1),3) 8.069279 2.134505 3.780398 0.0006
R-squared 0.088847 Mean dependent var 0.034340 D(M2(-2),3) 8.328431 2.240798 3.716726 0.0007
Adjusted R-squared 0.088847 S.D. dependent var 13.87643 D(M2(-3),3) 8.884181 2.271136 3.911778 0.0004
S.E. of regression 13.24565 Akaike info criterion 8.023904 D(M2(-4),3) 8.882449 2.346413 3.785545 0.0006
Sum squared resid 9123.261 Schwarz criterion 8.061079 D(M2(-5),3) 9.170426 2.231512 4.109512 0.0002
Hannan-Quinn D(M2(-6),3) 6.980207 2.450832 2.848097 0.0074
Log likelihood -211.6334 criter. 8.038199
D(M2(-7),3) 11.01964 2.826917 3.898112 0.0004
Durbin-Watson stat 1.816701
D(M2(-8),3) 17.11031 2.075100 8.245535 0.0000

R-squared 0.991823 Mean dependent var 55.64093

Adjusted R-squared 0.989899 S.D. dependent var 968.9418
S.E. of regression 97.38240 Akaike info criterion 12.17893
Sum squared resid 322433.3 Schwarz criterion 12.54756
Hannan-Quinn
Log likelihood -252.8471criter. 12.31487
Durbin-Watson stat 1.831531

1
92
Non-oil revenue (NOLR) Official external reserve (OER)

Null Hypothesis: D(NOLR,2) has a unit root Null Hypothesis: D(OER) has a unit root
Exogenous: None Exogenous: None
Lag Length: 1 (Automatic - based on SIC, maxlag=10) Lag Length: 1 (Automatic - based on SIC, maxlag=10)

t-Statistic Prob.* t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -9.437909 0.0000 Augm ented Dickey-Fuller test statistic -5.695329 0.0000
Test critical Test critical
values: 1% level -2.612033 values: 1% level -2.611094
5% level -1.947520 5% level -1.947381
10% level -1.612650 10% level -1.612725

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

Augmented Dickey-Fuller Test Equation Augm ented Dickey-Fuller Test Equation

Dependent Variable: D(NOLR,3) Dependent Variable: D(OER,2)
Method: Least Squares Method: Least Squares
Date: 11/14/15 Time: 00:33 Date: 11/14/15 Time: 00:34
Sample (adjusted): 1965 2014 Sample (adjusted): 1964 2014
Included observations: 50 after adjustments Included observations: 51 after adjustments

Variable Coefficient Std. Error t-Statistic Prob. Variable Coefficient Std. Error t-Statistic Prob.

D(NOLR(-1),2) -2.408617 0.255207 -9.437909 0.0000 D(OER(-1)) -0.772590 0.135653 -5.695329 0.0000
D(NOLR(-1),3) 0.358202 0.135605 2.641507 0.0111 D(OER(-1),2) 0.483522 0.132049 3.661697 0.0006

R-squared 0.901093 Mean dependent var 0.054800 -

R-squared 0.402833 Mean dependent var 170.1688
Adjusted R-squared 0.899033 S.D. dependent var 272.8366
Adjusted R-squared 0.390646 S.D. dependent var 4681.437
S.E. of regression 86.69473 Akaike info criterion 11.80184
S.E. of regression 3654.383 Akaike info criterion 19.28367
Sum squared resid 360766.9 Schwarz criterion 11.87832
Sum squared resid 6.54E+08 Schwarz criterion 19.35943
Hannan-Quinn
Log likelihood -293.0460 criter. 11.83097 Hannan-Quinn
Log likelihood -489.7335 criter. 19.31262
Durbin-Watson stat 2.087816
Durbin-Watson stat 1.899190

2
93
Oil revenue (OLR) Public debt (PD)

Null Hypothesis: D(OLR) has a unit root Null Hypothesis: D(PD,2) has a unit root
Exogenous: None Exogenous: None
Lag Length: 0 (Automatic - based on SIC, maxlag=5) Lag Length: 10 (Automatic - based on SIC, maxlag=10)

t-Statistic Prob.* t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -8.491403 0.0000 Augm ented Dickey-Fuller test statistic -6.468044 0.0000
Test critical Test critical
values: 1% level -2.610192 values: 1% level -2.622585
5% level -1.947248 5% level -1.949097
10% level -1.612797 10% level -1.611824

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

Augmented Dickey-Fuller Test Equation Augm ented Dickey-Fuller Test Equation

Dependent Variable: D(OLR,2) Dependent Variable: D(PD,3)
Method: Least Squares Method: Least Squares
Date: 11/15/15 Time: 06:45 Date: 11/15/15 Time: 05:32
Sample (adjusted): 1963 2014 Sample (adjusted): 1974 2014
Included observations: 52 after adjustments Included observations: 41 after adjustments

Variable Coefficient Std. Error t-Statistic Prob. Variable Coefficient Std. Error t-Statistic Prob.

D(OLR(-1)) -1.171435 0.137955 -8.491403 0.0000 D(PD(-1),2) -25.59904 3.957771 -6.468044 0.0000
D(PD(-1),3) 23.66342 3.848000 6.149537 0.0000
- D(PD(-2),3) 22.33190 3.678125 6.071545 0.0000
R-squared 0.585716 Mean dependent var 0.298269
D(PD(-3),3) 20.96182 3.568700 5.873797 0.0000
Adjusted R-squared 0.585716 S.D. dependent var 1374.274
D(PD(-4),3) 19.45177 3.343779 5.817300 0.0000
S.E. of regression 884.5499 Akaike info criterion 16.42708
D(PD(-5),3) 17.53257 3.112053 5.633765 0.0000
Sum squared resid 39903853 Schwarz criterion 16.46460
Hannan-Quinn D(PD(-6),3) 14.91029 2.691942 5.538859 0.0000
Log likelihood -426.1040criter. 16.44146 D(PD(-7),3) 11.39134 2.161149 5.270966 0.0000
Durbin-Watson stat 2.091625 D(PD(-8),3) 8.507656 1.536819 5.535885 0.0000
D(PD(-9),3) 4.593618 0.982509 4.675394 0.0001
D(PD(-10),3) 1.989592 0.497571 3.998605 0.0004

-
R-squared 0.953523 Mean dependent var 8.594390
Adjusted R-squared 0.938030 S.D. dependent var 326.5763
S.E. of regression 81.29686 Akaike info criterion 11.85830
Sum squared resid 198275.4 Schwarz criterion 12.31804
Hannan-Quinn
Log likelihood -232.0952 criter. 12.02571
Durbin-Watson stat 1.427278

3
94
Population growth rate (PGR) Real gross domestic product (RGDP)

Null Hypothesis: D(PGR) has a unit root Null Hypothesis: D(RGDP,2) has a unit root
Exogenous: None Exogenous: None
Lag Length: 4 (Automatic - based on SIC, maxlag=10) Lag Length: 1 (Automatic - based on SIC, maxlag=10)

t-Statistic Prob.* t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -8.080581 0.0000 Augm ented Dickey-Fuller test statistic -8.747016 0.0000
Test critical Test critical
values: 1% level -2.614029 values: 1% level -2.612033
5% level -1.947816 5% level -1.947520
10% level -1.612492 10% level -1.612650

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

Augmented Dickey-Fuller Test Equation Augm ented Dickey-Fuller Test Equation

Dependent Variable: D(PGR,2) Dependent Variable: D(RGDP,3)
Method: Least Squares Method: Least Squares
Date: 11/20/15 Time: 21:02 Date: 11/15/15 Time: 05:39
Sample (adjusted): 1967 2014 Sample (adjusted): 1965 2014
Included observations: 48 after adjustments Included observations: 50 after adjustments

Variable Coefficient Std. Error t-Statistic Prob. Variable Coefficient Std. Error t-Statistic Prob.

D(PGR(-1)) -4.286091 0.530419 -8.080581 0.0000 D(RGDP(-1),2) -1.981009 0.226478 -8.747016 0.0000
D(PGR(-1),2) 2.501386 0.452962 5.522286 0.0000 D(RGDP(-1),3) 0.385460 0.133861 2.879554 0.0059
D(PGR(-2),2) 1.512172 0.343226 4.405762 0.0001
D(PGR(-3),2) 0.739704 0.227328 3.253910 0.0022 R-squared 0.756042 Mean dependent var 21.12940
D(PGR(-4),2) 0.299278 0.121154 2.470221 0.0175 Adjusted R-squared 0.750960 S.D. dependent var 3580.754
S.E. of regression 1786.937 Akaike info criterion 17.85357
- Sum squared resid 1.53E+08 Schwarz criterion 17.93005
R-squared 0.838456 Mean dependent var 0.007292
Hannan-Quinn
Adjusted R-squared 0.823428 S.D. dependent var 0.477443 Log likelihood -444.3393criter. 17.88270
- Durbin-Watson stat 2.093876
S.E. of regression 0.200624 Akaike info criterion 0.276439
-
Sum squared resid 1.730744 Schwarz criterion 0.081523
Hannan-Quinn -
Log likelihood 11.63454 criter. 0.202780
Durbin-Watson stat 1.971735

4
95
Size of public sector (SPS) Square of size of public sector (SPS2)

Null Hypothesis: SGS has a unit root Null Hypothesis: SPS2 has a unit root
Exogenous: Constant Exogenous: Constant
Lag Length: 0 (Automatic - based on SIC, maxlag=10) Lag Length: 0 (Automatic - based on SIC, maxlag=10)

t-Statistic Prob.* t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -3.481481 0.0123 Augm ented Dickey-Fuller test statistic -4.302924 0.0012
Test critical Test critical
values: 1% level -3.560019 values: 1% level -3.560019
5% level -2.917650 5% level -2.917650
10% level -2.596689 10% level -2.596689

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

Augmented Dickey-Fuller Test Equation Augm ented Dickey-Fuller Test Equation

Dependent Variable: D(SGS) Dependent Variable: D(SPS2)
Method: Least Squares Method: Least Squares
Date: 11/15/15 Time: 05:43 Date: 11/23/15 Time: 13:58
Sample (adjusted): 1962 2014 Sample (adjusted): 1962 2014
Included observations: 53 after adjustments Included observations: 53 after adjustments

Variable Coefficient Std. Error t-Statistic Prob. Variable Coefficient Std. Error t-Statistic Prob.

SGS(-1) -0.371188 0.106618 -3.481481 0.0010 SPS2(-1) -0.528214 0.122757 -4.302924 0.0001
C 6.346358 1.935417 3.279064 0.0019 C 175.1765 50.90968 3.440927 0.0012

R-squared 0.192024 Mean dependent var 0.064528 R-squared 0.266347 Mean dependent var 1.116340
Adjusted R-squared 0.176182 S.D. dependent var 5.615429 Adjusted R-squared 0.251962 S.D. dependent var 260.1855
S.E. of regression 5.096813 Akaike info criterion 6.132113 S.E. of regression 225.0324 Akaike info criterion 13.70737
Sum squared resid 1324.853 Schwarz criterion 6.206464 Sum squared resid 2582618. Schwarz criterion 13.78172
Hannan-Quinn Hannan-Quinn
Log likelihood -160.5010 criter. 6.160705 Log likelihood -361.2453criter. 13.73596
F-statistic 12.12071 Durbin-Watson stat 2.342031 F-statistic 18.51515 Durbin-Watson stat 2.223469
Prob(F-statistic) 0.001031 Prob(F-statistic) 0.000076

5
96
Tax revenue (TAX) Trade openness (TO)

Null Hypothesis: D(TAX) has a unit root Null Hypothesis: D(TO) has a unit root
Exogenous: None Exogenous: None
Lag Length: 0 (Automatic - based on SIC, maxlag=5) Lag Length: 0 (Automatic - based on SIC, maxlag=10)

t-Statistic Prob.* t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -8.052821 0.0000 Augm ented Dickey-Fuller test statistic -9.997272 0.0000
Test critical Test critical
values: 1% level -2.610192 values: 1% level -2.610192
5% level -1.947248 5% level -1.947248
10% level -1.612797 10% level -1.612797

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

Augmented Dickey-Fuller Test Equation Augm ented Dickey-Fuller Test Equation

Dependent Variable: D(TAX,2) Dependent Variable: D(TO,2)
Method: Least Squares Method: Least Squares
Date: 11/15/15 Time: 06:48 Date: 11/15/15 Time: 05:51
Sample (adjusted): 1963 2014 Sample (adjusted): 1963 2014
Included observations: 52 after adjustments Included observations: 52 after adjustments

Variable Coefficient Std. Error t-Statistic Prob. Variable Coefficient Std. Error t-Statistic Prob.

D(TAX(-1)) -1.120449 0.139137 -8.052821 0.0000 D(TO(-1)) -1.323085 0.132345 -9.997272 0.0000

R-squared 0.559761 Mean dependent var 5.937500 R-squared 0.662126 Mean dependent var 0.054423
Adjusted R-squared 0.559761 S.D. dependent var 1515.668 Adjusted R-squared 0.662126 S.D. dependent var 16.75761
S.E. of regression 1005.653 Akaike info criterion 16.68371 S.E. of regression 9.740687 Akaike info criterion 7.409544
Sum squared resid 51578274 Schwarz criterion 16.72123 Sum squared resid 4838.930 Schwarz criterion 7.447068
Hannan-Quinn Hannan-Quinn
Log likelihood -432.7764 criter. 16.69809 Log likelihood -191.6481criter. 7.423930
Durbin-Watson stat 2.049352 Durbin-Watson stat 1.963858

6
97
Appendix 2: Cointegration test

Null Hypothesis: RESIDUAL has a unit root

Exogenous: None
Lag Length: 0 (Automatic - based on SIC, maxlag=10)

t-Statistic Prob.*

Augmented Dickey-Fuller test statistic -1.864065 0.0599

Test critical values: 1% level -2.609324
5% level -1.947119
10% level -1.612867

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

Augmented Dickey-Fuller Test Equation

Dependent Variable: D(RESIDUAL)
Method: Least Squares
Date: 12/24/15 Time: 07:51
Sample (adjusted): 1962 2014
Included observations: 53 after adjustments

Variable Coefficient Std. Error t-Statistic Prob.

RESIDUAL(-1) -0.107858 0.057862 -1.864065 0.0680

R-squared 0.062128 Mean dependent var 51.08000

Adjusted R-squared 0.062128 S.D. dependent var 2213.829
S.E. of regression 2143.957 Akaike info criterion 18.19738
Sum squared resid 2.39E+08 Schwarz criterion 18.23456
Log likelihood -481.2306 Hannan-Quinn criter. 18.21168
Durbin-Watson stat 1.516276

0
98
Appendix 3A: Average growth rate of public expenditure

Dependent Variable: LOG(GE)

Method: Least Squares
Date: 12/22/15 Time: 13:13
Sample (adjusted): 1970 2014
Included observations: 39 after adjustments
HAC standard errors & covariance (Bartlett kernel, Newey-West fixed
bandwidth = 4.0000)

Variable Coefficient Std. Error t-Statistic Prob.

T 0.019059 0.018435 1.033831 0.3085

LOG(TAX) 0.725239 0.074296 9.761447 0.0000
LOG(PD) 0.093175 0.037503 2.484453 0.0181
PGR 0.045280 0.126858 0.356936 0.7233
C -37.30600 36.33886 -1.026615 0.3119

R-squared 0.993489 Mean dependent var 4.990352

Adjusted R-squared 0.992723 S.D. dependent var 2.624110
S.E. of regression 0.223857 Akaike info criterion -0.036407
Sum squared resid 1.703811 Schwarz criterion 0.176870
Log likelihood 5.709936 Hannan-Quinn criter. 0.040115
F-statistic 1296.905 Durbin-Watson stat 1.304686
Prob(F-statistic) 0.000000 Wald F-statistic 1559.575
Prob(Wald F-statistic) 0.000000

1
99
Appendix 3B: Regression Specification Error Test for average growth rate of public
expenditure

Ramsey RESET Test

Equation: EQ01
Specification: LOG(GE) T LOG(TAX) LOG(PD) PGR C
Omitted Variables: Squares of fitted values

Value df Probability
t-statistic 0.104429 33 0.9175
F-statistic 0.010906 (1, 33) 0.9175
Likelihood ratio 0.012886 1 0.9096

F-test summary:
Mean
Sum of Sq. df Squares
Test SSR 0.000563 1 0.000563
Restricted SSR 1.703811 34 0.050112
Unrestricted SSR 1.703248 33 0.051614
Unrestricted SSR 1.703248 33 0.051614

LR test summary:
Value df
Restricted LogL 5.709936 34
Unrestricted LogL 5.716379 33

Unrestricted Test Equation:

Dependent Variable: LOG(GE)
Method: Least Squares
Date: 12/22/15 Time: 13:28
Sample: 1970 2014
Included observations: 39
HAC standard errors & covariance (Bartlett kernel, Newey-West fixed
bandwidth = 4.0000)

Variable Coefficient Std. Error t-Statistic Prob.

T 0.019760 0.021288 0.928232 0.3600

LOG(TAX) 0.729393 0.085583 8.522617 0.0000
LOG(PD) 0.092310 0.040215 2.295416 0.0282
PGR 0.042357 0.125139 0.338482 0.7371
C -38.69212 42.00201 -0.921197 0.3636
FITTED^2 -0.000756 0.008310 -0.090936 0.9281

R-squared 0.993491 Mean dependent var 4.990352

Adjusted R-squared 0.992505 S.D. dependent var 2.624110
S.E. of regression 0.227186 Akaike info criterion 0.014545
Sum squared resid 1.703248 Schwarz criterion 0.270477
Log likelihood 5.716379 Hannan-Quinn criter. 0.106371
F-statistic 1007.344 Durbin-Watson stat 1.306382
Prob(F-statistic) 0.000000 Wald F-statistic 1195.406
Prob(Wald F-statistic) 0.000000

2
100
Appendix 3C: Normality test for average growth rate of public expenditure
6
Series: Residuals
Sample 1970 2014
5 Observations 39

4 Mean -2.18e-15
Median -0.039179
Maximum 0.450229
3 Minimum -0.397295
Std. Dev. 0.211748
Skewness 0.583035
2
Kurtosis 2.712899

1 Jarque-Bera 2.343488
Probability 0.309826

0
-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5

Appendix 4A: Determinants of public expenditure growth

Dependent Variable: LOG(GE)

Method: Least Squares
Date: 12/17/15 Time: 16:50
Sample (adjusted): 1970 2014
Included observations: 39 after adjustments
HAC standard errors & covariance (Bartlett kernel, Newey-West fixed
bandwidth = 4.0000)

Variable Coefficient Std. Error t-Statistic Prob.

LOG(OLR) 0.326753 0.105287 3.103467 0.0041

LOG(NOLR) 0.242974 0.086348 2.813887 0.0086
LOG(PD) 0.062094 0.042475 1.461897 0.1542
D(PGR(-1)) 0.224977 0.090815 2.477318 0.0191
D(OER(-1)) 2.63E-06 4.31E-06 0.611498 0.5455
INF(-1) 0.001315 0.002543 0.517110 0.6089
LOG(GE(-1)) 0.288735 0.156794 1.841485 0.0755
FRA 0.055295 0.137813 0.401235 0.6911
C 0.759440 0.084975 8.937235 0.0000

R-squared 0.995684 Mean dependent var 4.990352

Adjusted R-squared 0.994534 S.D. dependent var 2.624110
S.E. of regression 0.194014 Akaike info criterion -0.242601
Sum squared resid 1.129240 Schwarz criterion 0.141298
Log likelihood 13.73072 Hannan-Quinn criter. -0.104861
F-statistic 865.1956 Durbin-Watson stat 1.868273
Prob(F-statistic) 0.000000 Wald F-statistic 1900.077
Prob(Wald F-statistic) 0.000000

3
101
Appendix 4B: Regression Specification Error Test for determinants of public
expenditure growth

Ramsey RESET Test

Equation: EQ01
Specification: LOG(GE) LOG(OLR) LOG(NOLR) LOG(PD) D(PGRR(-1))
D(OER(-1)) INF(-1) LOG(GE(-1)) FRA C
Omitted Variables: Squares of fitted values

Value df Probability
t-statistic 0.470770 29 0.6413
F-statistic 0.221625 (1, 29) 0.6413
Likelihood ratio 0.296914 1 0.5858

F-test summary:
Mean
Sum of Sq. df Squares
Test SSR 0.008564 1 0.008564
Restricted SSR 1.129240 30 0.037641
Unrestricted SSR 1.120676 29 0.038644
Unrestricted SSR 1.120676 29 0.038644

LR test summary:
Value df
Restricted LogL 13.73072 30
Unrestricted LogL 13.87918 29

Unrestricted Test Equation:

Dependent Variable: LOG(GE)
Method: Least Squares
Date: 11/19/15 Time: 16:55
Sample: 1970 2014
Included observations: 39
HAC standard errors & covariance (Bartlett kernel, Newey-West fixed
bandwidth = 4.0000)

Variable Coefficient Std. Error t-Statistic Prob.

LOG(OLR) 0.327088 0.102820 3.181157 0.0035

LOG(NOLR) 0.289859 0.181699 1.595276 0.1215
LOG(PD) 0.057946 0.043838 1.321828 0.1966
D(PGRR(-1)) 0.235323 0.104962 2.241987 0.0328
D(OER(-1)) 2.24E-06 4.15E-06 0.540398 0.5930
INF(-1) 0.001070 0.002751 0.388970 0.7001
LOG(GE(-1)) 0.283291 0.158654 1.785590 0.0846
FRA 0.089369 0.105988 0.843206 0.4060
C 0.756676 0.091326 8.285462 0.0000
FITTED^2 -0.004627 0.011311 -0.409029 0.6855

R-squared 0.995717 Mean dependent var 4.990352

Adjusted R-squared 0.994388 S.D. dependent var 2.624110
S.E. of regression 0.196581 Akaike info criterion -0.198932
Sum squared resid 1.120676 Schwarz criterion 0.227622
Log likelihood 13.87918 Hannan-Quinn criter. -0.045888
F-statistic 749.1334 Durbin-Watson stat 1.891903
Prob(F-statistic) 0.000000 Wald F-statistic 1645.743
Prob(Wald F-statistic) 0.000000

4
102
Appendix 4C: Normality test for determinants of public expenditure growth
8
Series: Residuals
7 Sample 1970 2014
Observations 39
6
Mean -8.94e-16
5 Median -0.008923
Maximum 0.532320
4 Minimum -0.304444
Std. Dev. 0.172386
3 Skewness 0.946301
Kurtosis 4.452030
2
Jarque-Bera 9.246789
3.932887
1 Probability 0.009819
0.490032

0
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5

Appendix 5A: Determinants of real gross domestic product

Dependent Variable: RGDP

Method: Least Squares
Date: 12/17/15 Time: 17:42
Sample: 1961 2014
Included observations: 54
HAC standard errors & covariance (Bartlett kernel, Newey-West fixed
bandwidth = 4.0000)

Variable Coefficient Std. Error t-Statistic Prob.

SGS 874.6039 191.4117 4.569228 0.0000

SPS2 -16.49570 4.699807 -3.509868 0.0010
GFCF -0.000827 0.000566 -1.461854 0.1506
TO 83.88714 38.30862 2.189772 0.0337
TAX 2.607451 1.054108 2.473609 0.0171
M2 2.097880 0.588602 3.564172 0.0009
INF 20.35633 20.63339 0.986572 0.3290
C 3194.016 1962.226 1.627751 0.1104

R-squared 0.965116 Mean dependent var 24558.19

Adjusted R-squared 0.959807 S.D. dependent var 15257.36
S.E. of regression 3058.810 Akaike info criterion 19.02539
Sum squared resid 4.30E+08 Schwarz criterion 19.32006
Log likelihood -505.6856 Hannan-Quinn criter. 19.13903
F-statistic 181.8072 Durbin-Watson stat 1.510105
Prob(F-statistic) 0.000000 Wald F-statistic 394.1793
Prob(Wald F-statistic) 0.000000

5
103
Appendix 5B: Regression Specification Error Test for determinants of real gross
domestic product

Ramsey RESET Test

Equation: EQ03
Specification: RGDP SGS SGS^2 GFCF TO TAX M2 INF C
Omitted Variables: Squares of fitted values

Value df Probability
t-statistic 0.897808 45 0.1795
F-statistic 0.953106 (1, 45) 0.1798
Likelihood ratio 36.54131 1 0.0000

F-test summary:
Mean
Sum of Sq. df Squares
Test SSR 2.12E+08 1 2.12E+08
Restricted SSR 4.30E+08 46 9356317.
Unrestricted SSR 2.19E+08 45 4861464.
Unrestricted SSR 2.19E+08 45 4861464.

LR test summary:
Value df
Restricted LogL -505.6856 46
Unrestricted LogL -487.4150 45

Unrestricted Test Equation:

Dependent Variable: RGDP
Method: Least Squares
Date: 11/20/15 Time: 20:45
Sample: 1961 2014
Included observations: 54
HAC standard errors & covariance (Bartlett kernel, Newey-West fixed
bandwidth = 4.0000)

Variable Coefficient Std. Error t-Statistic Prob.

SGS 1338.935 171.1618 7.822628 0.0000

SGS^2 -23.64495 4.626658 -5.110589 0.0000
GFCF -0.001096 0.000294 -3.727636 0.0005
TO 85.19821 32.34112 2.634362 0.0115
TAX 5.124686 0.733073 6.990687 0.0000
M2 5.029105 0.303992 16.54352 0.0000
INF 40.29602 19.86396 2.028599 0.0484
C 1506.633 1320.640 1.140836 0.2600
FITTED^2 -1.75E-05 2.71E-06 -6.484005 0.0000

R-squared 0.982269 Mean dependent var 24558.19

Adjusted R-squared 0.979116 S.D. dependent var 15257.36
S.E. of regression 2204.873 Akaike info criterion 18.38574
Sum squared resid 2.19E+08 Schwarz criterion 18.71724
Log likelihood -487.4150 Hannan-Quinn criter. 18.51358
F-statistic 311.6073 Durbin-Watson stat 1.043427
Prob(F-statistic) 0.000000 Wald F-statistic 562.6932
Prob(Wald F-statistic) 0.000000

6
104
Appendix 5C: Normality test for determinants of real gross domestic product
14
Series: Residuals
12 Sample 1961 2014
Observations 54
10
Mean 6.94e-12
Median 206.7345
8 Maximum 7952.735
Minimum -7169.894
6 Std. Dev. 2849.663
Skewness -0.091566
4 Kurtosis 3.588961

Jarque-Bera 0.855929
2
Probability 0.651835

0
-8000 -6000 -4000 -2000 0 2000 4000 6000 8000

Appendix 6: Multicollinearity test on sample data

GE RGDP GFCF INF M2 NOLR OER OLR PD PGR SGS TAX TO

GE 1.000000 0.967492 0.925361 -0.152957 0.975960 0.979455 0.910485 0.959192 0.922913 0.097276 -0.325314 0.973808 0.218033
RGDP 0.967492 1.000000 0.872903 -0.071091 0.941039 0.950068 0.926963 0.944592 0.864602 0.179641 -0.203134 0.955252 0.316799
GFCF 0.925361 0.872903 1.000000 -0.133074 0.955078 0.953471 0.735484 0.859244 0.950093 0.085302 -0.296616 0.896337 0.003004
INF -0.152957 -0.071091 -0.133074 1.000000 -0.152155 -0.153137 -0.176448 -0.156658 -0.119568 0.170665 0.318995 -0.157298 0.262416
M2 0.975960 0.941039 0.955078 -0.152155 1.000000 0.992135 0.864417 0.915748 0.922207 0.098769 -0.323762 0.943607 0.102728
NOLR 0.979455 0.950068 0.953471 -0.153137 0.992135 1.000000 0.862774 0.924034 0.924011 0.102559 -0.330140 0.953219 0.136972
OER 0.910485 0.926963 0.735484 -0.176448 0.864417 0.862774 1.000000 0.921498 0.713535 0.131457 -0.287409 0.913539 0.325807
OLR 0.959192 0.944592 0.859244 -0.156658 0.915748 0.924034 0.921498 1.000000 0.819570 0.040767 -0.353510 0.995726 0.247989
PD 0.922913 0.864602 0.950093 -0.119568 0.922207 0.924011 0.713535 0.819570 1.000000 0.109499 -0.236413 0.851798 0.105958
PGR 0.097276 0.179641 0.085302 0.170665 0.098769 0.102559 0.131457 0.040767 0.109499 1.000000 0.154587 0.053711 0.162366
SGS -0.325314 -0.203134 -0.296616 0.318995 -0.323762 -0.330140 -0.287409 -0.353510 -0.236413 0.154587 1.000000 -0.352107 0.212748
TAX 0.973808 0.955252 0.896337 -0.157298 0.943607 0.953219 0.913539 0.995726 0.851798 0.053711 -0.352107 1.000000 0.219670
TO 0.218033 0.316799 0.003004 0.262416 0.102728 0.136972 0.325807 0.247989 0.105958 0.162366 0.212748 0.219670 1.000000

Note: Values on shaded background have correlation coefficient greater than 0.8 (r > 0.8)

7
105
 Appendix 7: Sample data

YEAR FRA GE GFCF INF M2 NOLR OER OLR PD PGR RGDP SGS SPS2 TAX TO
1961 0 0.16 258.0000 6.28 0.30 0.22 307.07 0.00 -0.06 2.04 8835.93 6.94 48.1636 0.22 25.49
1962 0 0.17 305.0000 5.27 0.33 0.48 289.02 0.00 -0.31 2.09 9198.47 6.45 41.6025 0.48 21.83
1963 0 0.18 393.0000 -2.69 0.36 0.50 210.03 0.00 -0.31 2.12 9987.57 6.66 44.3556 0.50 22.32
1964 0 0.22 503.0000 0.86 0.43 0.55 228.05 0.00 -0.33 2.67 10482.00 7.61 57.9121 0.55 24.62
1965 0 0.24 615.2000 4.10 0.47 0.65 239.05 0.00 -0.42 2.70 10994.00 7.60 57.7600 0.65 26.91
1966 0 0.26 601.6000 9.69 0.52 0.61 215.35 0.00 -0.36 3.01 10526.70 7.56 57.1536 0.61 25.04
1967 0 0.26 483.6000 -3.73 0.45 0.65 112.36 0.00 -0.40 2.18 8869.45 9.37 87.7969 0.65 28.49
1968 0 0.35 437.8000 -0.48 0.52 0.57 121.19 0.00 -0.22 2.21 8758.73 13.17 173.4489 0.57 25.66
1969 0 0.56 550.0000 10.16 0.66 0.76 132.46 0.00 -0.20 2.61 10878.10 15.67 245.5489 0.76 26.82
1970 0 0.90 882.7000 13.76 0.98 0.47 223.50 0.17 0.46 2.54 13598.40 17.12 293.0944 0.63 19.62
1971 0 1.00 1282.500 16.00 1.04 0.66 431.94 0.51 -0.17 2.14 15534.50 14.99 224.7001 1.17 24.46
1972 0 1.46 1401.400 3.46 1.20 0.64 390.71 0.76 0.06 2.37 16057.10 20.36 414.5296 1.41 22.76
1973 0 1.53 2506.000 5.40 1.37 0.68 622.85 1.02 -0.17 2.45 16923.10 17.72 313.9984 1.70 31.27
1974 0 2.74 2956.000 12.67 2.59 0.81 5708.97 3.72 -1.80 2.57 18811.80 14.56 211.9936 4.54 39.75
1975 0 5.94 5019.800 33.96 4.04 1.24 5665.69 4.27 0.43 2.72 17828.30 27.67 765.6289 5.51 41.17
1976 0 7.86 8107.300 24.30 5.71 1.40 5256.78 5.37 1.09 2.89 19440.40 29.47 868.4809 6.77 42.14
1977 0 8.82 9420.600 15.09 7.68 1.96 4335.99 6.08 0.78 2.15 20611.60 27.99 783.4401 8.04 47.40
1978 0 8.00 9386.000 21.71 7.52 2.82 2028.81 4.56 2.82 2.16 19423.50 23.16 536.3856 7.37 43.31
1979 0 7.41 9095.000 11.71 9.85 2.03 5899.64 8.88 -1.46 2.99 20736.40 17.65 311.5225 10.91 43.88
1980 0 14.97 10976.00 9.97 14.39 2.88 10639.80 12.35 1.98 2.86 21608.30 30.16 909.6256 15.23 48.57
1981 0 11.41 18220.59 20.81 15.24 4.73 4168.45 8.56 3.90 2.72 18771.60 23.97 574.5609 13.29 48.29
1982 0 11.92 17145.82 7.70 16.69 3.62 1926.43 7.81 6.10 2.52 18573.90 24.30 590.4900 11.43 37.75
1983 0 9.64 13335.33 23.21 19.03 3.26 1251.99 7.25 3.36 2.53 17635.80 18.15 329.4225 10.51 27.04
1984 0 9.93 9149.760 17.82 21.24 2.98 1674.11 8.27 2.66 2.62 17279.30 16.65 277.2225 11.25 23.61
1985 0 13.04 8799.480 7.44 23.15 4.13 1891.87 10.92 3.04 2.57 18717.50 19.20 368.6400 15.05 25.90
1986 0 16.22 11351.46 5.72 23.61 4.49 1349.90 8.11 8.25 2.61 17078.90 23.46 550.3716 12.60 23.72
1987 0 22.02 15228.58 11.29 28.90 6.35 1497.83 19.03 5.89 2.63 15242.60 20.93 438.0649 25.38 41.65
1988 0 27.75 17562.21 54.51 38.41 7.77 932.99 19.83 12.16 2.52 16392.30 19.95 398.0025 27.60 35.31
1989 0 41.03 26825.51 50.47 43.37 14.74 2041.08 39.13 15.13 3.05 17452.40 18.92 357.9664 53.87 60.39
1990 0 60.27 40121.31 7.36 57.55 26.22 4128.79 71.89 22.12 2.58 19680.40 22.53 507.6009 98.10 53.03
1991 0 66.58 45190.23 13.01 79.07 18.33 4678.02 82.67 35.76 2.55 19558.80 21.33 454.9689 100.99 64.88
1992 0 92.80 70809.16 44.59 129.09 26.38 1196.05 164.08 39.53 2.64 19643.60 17.42 303.4564 190.45 61.03
1993 0 233.81 96915.51 57.17 198.48 30.67 1640.44 162.10 107.74 2.50 20054.30 34.19 1168.9561 192.77 58.11
1994 0 160.89 105575.5 57.03 266.94 41.72 1649.17 160.19 70.27 2.51 20236.70 17.88 319.6944 201.91 42.31
1995 0 248.77 141920.2 72.84 318.76 135.44 1709.11 324.55 -1.00 2.50 20174.50 12.87 165.6369 459.99 59.77
1996 0 337.22 204047.6 29.27 370.33 114.81 4329.39 408.78 -32.05 2.50 21181.90 12.48 155.7504 523.60 57.69
1997 0 428.22 242899.8 8.53 429.73 166.00 7781.25 416.81 5.00 2.50 21775.50 15.28 233.4784 582.81 76.86
1998 0 487.11 242256.3 10.00 525.64 139.30 7298.55 324.31 133.39 2.77 22366.90 17.99 323.6401 463.61 66.17
1999 0 947.69 231661.7 6.62 699.73 224.77 5649.73 724.42 285.10 2.50 22472.90 29.67 880.3089 949.19 55.85
2000 0 701.06 331056.7 6.93 1036.08 314.48 10099.50 1591.68 103.78 2.51 23668.10 15.30 234.0900 1906.16 71.38
2001 0 1018.03 372135.7 18.87 1309.36 523.97 10646.60 1707.56 221.05 2.25 24712.10 21.55 464.4025 2231.53 81.81
2002 0 1018.16 499681.5 12.88 1555.80 500.99 7566.81 1230.85 301.40 2.79 25647.30 14.73 216.9729 1731.84 63.38
2003 0 1225.97 865876.5 14.03 1766.01 500.82 7415.09 2074.28 202.72 2.54 28302.90 14.45 208.8025 2575.10 75.22
2004 0 1426.20 863072.6 15.00 2131.17 565.70 17256.50 3354.80 172.60 2.57 37851.10 12.50 156.2500 3920.50 48.45
2005 0 1822.10 804400.8 17.86 2612.89 785.10 28632.10 4762.40 161.40 2.60 39155.00 12.50 156.2500 5547.50 50.75
2006 0 1938.00 1546526. 8.24 3562.70 677.54 42735.50 5287.57 101.40 2.64 42370.00 10.44 108.9936 5965.10 64.61
2007 1 2450.90 1936958. 5.38 5857.67 1252.55 51907.00 4462.95 117.24 2.51 45263.20 11.86 140.6596 5715.50 64.46
2008 1 3240.82 2053006. 11.58 8983.33 1335.96 53599.30 6530.63 47.38 2.51 48101.30 13.34 177.9556 7866.59 64.97
2009 1 3452.99 3050576. 11.54 10289.80 1652.70 45509.80 3191.90 810.02 2.72 51436.80 13.93 194.0449 4057.50 61.80
2010 1 4194.58 9424691. 13.72 11315.50 1907.60 35884.90 5396.10 1105.38 2.79 55469.40 12.34 152.2756 7303.67 42.65
2011 1 4712.06 10180859 10.84 12802.70 2237.90 36263.70 8879.00 1158.52 2.29 58180.40 12.60 158.7600 11116.90 52.79
2012 1 4605.32 10618680 12.22 15032.40 2628.77 47548.40 8025.95 975.72 2.33 60670.10 11.36 129.0496 10654.70 44.38
2013 1 5260.77 11723098 8.48 15160.30 2950.56 46254.80 6809.23 1153.49 2.79 63942.90 11.00 121.0000 9759.79 31.03
2014 1 4578.06 11170889 8.10 17680.50 3275.12 37497.20 6793.72 978.43 2.75 67977.50 10.36 107.3296 10068.80 30.20

Note: Above data were measured as defined in Table 5.1, page 55.

8
106