THE IMPACT OF GOVERNMENT EXPENDITURE ON ECONOMIC GROWTH IN KENYA: 1963-2008

The rapid growth in government expenditure in Kenya has caused concern among policy makers on the implication of such growth. Over the three decades, government expenditure in the country grew at a faster rate than the growth rate of GDP. Given this fiscal scenario, an explanation of this requires studying the impact of government expenditure on economic growth. The specific objectives of the study were to: investigate the relationship between the components of government expenditure and economic growth; examine the effects of components of government expenditure on GDP growth rate; analyze the effects of government expenditure reforms on economic growth; and to draw policy implications from the findings. The data used were government expenditure components that included expenditure on government investment, physical infrastructure, education, health care, public debt servicing, economic affairs, general administration and services, defense, public order and national security, and government consumption. Sources of data were Kenya government documents and international financial statistics publications. The study applied Vector Auto Regression estimation technique using the annual time series data for the period 1963 to 2008 to evaluate the impact of government expenditure on economic growth. The Johansen cointegration tests revealed a long-run relationship between GDP growth rate and the selected components of government expenditure. Further, the Granger- Causality test indicated bi-directional causality between GDP growth rate and components of government expenditure. The results of impulse response functions and variance decomposition revealed that government expenditure on investment, physical infrastructure, education, health care, public debt servicing, economic affairs, general administration and services, defense, public order and national security and government consumption have effect on economic growth. Furthermore, the study established that expenditure reforms of budget rationalization, expenditure downsizing, privatization and governance affect economic growth. The study concludes that the composition of government expenditure and public expenditure reforms matter for economic growth.

The relationship between economic growth and government expenditure is an important subject of analysis (Barro, 1990;Easterly and Robelo, 1993;Barro and Sala-i-Martin, 1992;. A central question is whether or not government expenditure increases the long run steady state growth rate of the economy. The general view is that government expenditure, notably on physical infrastructure and human capital, can be growth-enhancing although the source of financing of such expenditures can be growth-retarding (Landau, 1983;Devarajan, 1993;Cashin, 1995;Kneller, 1999). The growth retardation is experienced because of disincentive effects associated with taxation (Musgrave and Musgrave, 1989). Government expenditure may directly or indirectly increase total output through its interaction with the private sector.
Lin (1994) examined some of the ways in which government expenditure can increase growth. These included the provision of public goods and infrastructure, social services and targeted intervention such as export subsidies.
The nature of the impact of government expenditure on growth depends on its form. According to Barro (1990), expenditure on investment and productive activities including state-owned production should contribute positively to growth, whereas government consumption expenditure is expected to be growth-retarding. However, in empirical studies, it is sometimes difficult to 2 determine which particular items of public expenditure should be categorized as investment and which as consumption.
Several analytical and empirical studies have focused on the traditional and new channels through which different types of government expenditure can affect growth (King and Robelo, 1990; Barro, 1990; Barro and Sala-i-Martin, 1992;Bleaney et al., 2001). A direct effect relates to an increase in the economy"s capital stock (physical or human) reflecting higher flows of government funds, especially when they are complementary to those privately financed. Government expenditure on education and health, for example, contribute to an increase in the stock of human capital. Similarly, to the extent that they trigger an accumulation of physical capital, most government expenditure on infrastructure falls in the category of having a direct impact on growth (Barro and Sala-I-Martin, 1992).
In addition, government expenditure can also contribute indirectly to economic growth by increasing the marginal productivity of both government and private supplied factors of production. Government expenditure on research and development, for example, provides higher productivity in the interaction between physical and human capital factors. Similarly, other components of government expenditure related to enforcement of property rights and maintenance of public order can exert a positive indirect effect on growth by contributing to better use of existing capital and labour assets (Trotman, 1997). In countries where crime and violence are endemic, 3 increased government expenditure on security can lead to lower production costs by reducing the need to protect employees and physical assets, hence increasing worker productivity and stimulating private physical investments.
There is growing evidence that suggest that in developing countries, externalities associated with infrastructure expenditure may be important in enhancing growth (Landau, 1985). Indeed, it has been found that infrastructure may have an impact on human capital as well. According to Age"nor and Moreno (2007), government expenditure on infrastructure affects growth not only through its direct impact on investment and the productivity of factors in the private sector, but also through health and education outcomes.
Government expenditure that facilitates access to clean water and sanitation helps to improve health and thereby labour productivity. These expenditures can be in the form of provision of electricity, which is essential for the functioning of hospitals and the delivery of health services, and better transportation networks, which contribute to easier access to health care, particularly in rural areas. In addition, there is evidence of direct linkages between infrastructure and education. Education allows for more training and greater access to learning technologies. Enrollment rates and the quality of education tend to improve with better transportation networks, particularly in rural areas. Greater access to sanitation and clean water in schools tend to raise attendance rates (Stiglitz, 1989).

4
There are two traditional approaches used to analyze the effects of government expenditure on growth. One is the Monetarist approach and the other is the Keynesian approach. Proponents of the Monetarist approach led by Milton Friedman argued that sustained money growth in excess of the growth of output produces inflation (Branson, 1989). The proponents re-evaluated the quantity theory of money and argued that to reduce inflation, the growth in the money supply needs to be controlled and thus the need to control or reduce government expenditure (Brunner and Meltzer, 1992). Proponents of this school of thought further argued that in examining the effects of disaggregated government expenditure on investment using fixed and random effect methods, tax financed government expenditure crowds out private investment (Ahmed, 1999). This is because when government expenditure is tax-financed, any extra expenditure calls for more taxation. A higher tax burden reduces the disposable income for individuals, which results to a reduction in consumption, lower savings and hence lower investment. On the other hand, higher tax burden on corporations and businesses result to decreased profits and thus reduces expansion and development aspects. If the government decides to borrow from money or capital market to finance its expenditure, it has a future obligation to repay the loan and its interest, which places a burden on the future generation, hence reducing their resource envelop. These factors result to crowding out of private investment in the course of funding government expenditure (Ahmed, 1999).
It is also important to note that in a market economy, the basic rule for growthpromoting public sector is that its activities should complement rather than compete with those of the private sector (Stiglitz, 1989).Thus, an important role for the government is to provide certain investments in human capital (particularly in primary education, and public health care) and in physical infrastructure. Certain level of expenditure on the legal system, public order and civil administration is necessary to ensure a stable environment in which increased economic growth is promoted (Mackenzie et al., 1997).
The other approach is the Keynesian. The proponents of this approach argued that markets would not automatically lead to full-employment equilibrium, but in fact the economy could settle in equilibrium at any level of non-full employment income. When this happens, the economy would need prodding to spur growth. This means active intervention by government to manage the level of demand, thus encouraging government expenditure. An increase in government expenditure would mean an imbalance between demand and supply of goods and services. Aggregate demand will be greater than aggregate supply. As a result of the extra aggregate demand, firms would employ more people. This would mean more income in the economy, some of which would be spent and some saved or paid in taxes. The extra expenditure on goods and services would prompt the firms in the economy to produce more, which leads to even more employment and therefore even more income and ultimate growth in the economy (Keynes, 1936).
Thus, the traditional Keynesian macroeconomics of growth maintains that many kinds of government expenditure, especially on recurrent nature, contribute to economic growth. A high level of public consumption is likely to increase employment, profitability and investment via the multiplier effect on the aggregate demand. Government expenditure raises aggregate demand, leading to an increased output, depending on the size and effectiveness of expenditure multiplier (Branson, 1989). The theory develops a rationale for government provision of goods and services based on the failure of markets to provide public goods, internalize externalities and cover costs when there are significant economies of scale (Stiglitz, 1989). This theory, therefore, is progovernment expenditure, which conflicts with the arguments of the proponents of the monetarist school of thought.
The new approach through which government expenditure affects economic growth is endogenous. This approach highlights the fact that if productivity is to increase, the labour force must continuously be provided with more resources. Resources in this case include physical capital, human capital and knowledge capital (technology). Therefore, growth is driven by accumulation of the factors of production while accumulation in turn is the result of investment in the private sector. This implies that the only way a government can affect economic growth, at least in the long-run, is via its impact on investment in capital, education and research and development. The approach makes improved education (and indeed any kind of training or research that 7 adds to human knowledge in any country) the key to achieving economic growth (Folster and Henrokson, 1997).

Trends and Composition of Government Expenditure
In order to explain the growth in the overall government expenditure, it is helpful to consider its breakdown by expenditure categories. The expenditure can be broadly classified in terms of purpose as recurrent and development expenditure. Recurrent expenditure refers to expenditure of recurrent expenses that are less discretionary and are made on ongoing programmes or activities.
It constitutes of wages and salaries, administration, transfers payment, debt repayment and welfare services. Recurrent expenditure may affect economic growth through its effects on people"s ability and willingness to work, save and invest. Development expenditure refers to expenditure that is generally more discretionary and is made on new programmes and activities that are yet to reach their final desired state of completion. It constitutes of investment in such schemes as construction of railways, roadways and communication systems, irrigation and power projects, which raise economic growth both directly and indirectly through encouragement of further private investment (Ag"enor, 2007). 100 120 1 9 6 3 1 9 6 5 1 9 6 7 1 9 6 9 1 9 7 1 1 9 7 3 1 9 7 5 1 9 7 7 1 9 7 9 1 9 8 1 1 9 8 3 1 9 8 5 1 9 8 7 1 9 8 9 1 9 9 1 1 9 9 3 1 9 9 5 1 9 9 7 1 9 9 9 2 0 0 1 2 0 0 3 2 0 0 5 2 0 0 7 Year Percentage Recurrent Expenditure as a % of Total Expenditure Development Expenditure as a % of Total Expenditure.  I table A1). This increase was attributed to increase in the construction costs (Republic of Kenya, 2003). During this period, the country was rebuilding and large amounts of money were spent on infrastructure and 9 services. There was huge expenditure on electricity, roads, telecommunications and airport expansion. A lot of money was also spent on resettlement, nationalization and agricultural development.

11
The government expenditure can further be classified into various subcategories as follows: general administration services, health care, education and training, defense, economic affairs, infrastructure, public debts repayments and others.  1963-1972 1973-1982 1983-1992 1993-2002 2003-2008 Year

Government Expenditure Reforms in Kenya
Since independence, various government expenditure reforms have been implemented. The reasons for the reforms were to raise and sustain the economic growth rate of the country. The public sector contributes to GDP growth rate through provision of government services such as education, health and administration, and productive activities in areas of agriculture, manufacturing, transport and communication and trade (Republic of Kenya, 1979). The government plays a leading role in determining the pattern of economic growth through public sector reforms, which determine directly how much of the country"s resources to divert to its own use, and how those resources should be allocated in order to increase economic growth. The contribution of each sector to the growth rate of GDP in Kenya has been affected by the government expenditure reforms that have been undertaken by the government. The main government expenditure strategy has been restructuring overall expenditure by directing more resources to activities that promote faster economic growth. To achieve this goal, various policy reforms have been implemented, which include: rationalizing government expenditure, with more resources being channeled to development and recurrent non-wage operating and maintenance expenditure in order to stimulate economic growth (Republic of Kenya, 2002).
In the plan period 1974-1978, the policy target was to increase development expenditure by 9 percent in order to expand output. Total capital formation was expanded three times that of the preceding five years. These investments were in agriculture, forestry, manufacturing, electricity and government capital formation. In terms of allocation, priority was given to rapidly expanding education programme and economic and social services, while growth in expenditure on administration, new buildings and main trunk roads was restricted (Republic of Kenya, 1974).
During the planning periods 1979 -1983, 1984-1988, 1989-1993 and 1997-2001, the government undertook rationalization of government expenditure, with more resources being channeled to development and recurrent non-wage operating and maintenance expenditure in order to stimulate economic growth (Republic of Kenya, 1997 (Republic of Kenya, 1986). The central thrust of the policies was to rely on market forces to mobilize resources for economic growth and development, with the role of government increasingly confined to providing an effective regulatory framework and essential public infrastructure and social services. The changes in allocation of budget resources were implemented by government. The government spent proportionately more on immediately productive services. It also increased its outlay on infrastructure to promote smaller towns and rural centers to improve overhead facilities, including roads, power and water supplies. In agriculture, more money was channeled to research, extension services including tea and coffee planting programmes and other projects to raise agricultural production.
The government spending on polytechnics and credit programmes to assist small scale industries in both rural and urban areas were also increased. These expenditures received the first allocation in the budgets that followed (Republic of Kenya, 1986). As a result, the share of formal education, health and other basic needs expenditure was reduced (Republic of Kenya, 1986).
The other major change in budget allocation involved a concerted effort to make all government outlays more efficient and productive through budget rationalization (Republic of Kenya, 1986). To achieve rationalization, the following measures were taken: projects with potentially high productivity were identified and their completion was advanced with an infusion of funds; projects with low potential benefits were identified and postponed or cancelled to free up funds for projects with higher returns; resources were shifted toward operation and maintenance expenditure of existing public facilities and away from investments in new projects; and new development projects were to be funded only if they were productive investment with very high priority (Republic of Kenya, 1986). The general approach then was that available resources for development budget were concentrated on few projects to shorten the construction or implementation period. At the same time, recurrent allocations were diverted to improve the utilization of existing capacity in order to raise productivity of public investments. The goal was to ensure that all government investments became productive as soon as possible through a programme of budget rationalization (Republic of Kenya, 1986).
In order to reduce the rate of growth of expenditure on salaries and allowances, several measures were adopted in 1990, which included the freezing of recruitments into job groups A to G and the ban on filling of posts that were vacant for more than six months (Republic of Kenya, 1994 (Republic of Kenya, 1994). There was a re-allocation of budget resources towards the core functions of government.
These included maintenance of law and order, the administration of justice, the provision of broad-based education and health services, the provision of economic infrastructure and the protection of the environment. To spur economic growth, the development expenditure and recurrent non-wage operating and maintenance expenditure were increased as a share of GDP. The budget rationalization measures aimed at maximizing the productivity of 20 public expenditure. In particular, objective technical and economic criteria were to be applied to project selection, with priority given to projects in the areas of health, education, infrastructure and environment (Republic of Kenya, 1994).
In the plan period 2002-2008, Kenya"s fiscal strategy aimed at increasing the level of economic activity by enhancing the role of private sector as the leading sector in wealth creation. The objectives were: to sustain reduction in the level of government expenditure as a percentage of GDP; to change the composition of government expenditure to focus more on efficient public investment and operations and maintenance in the long-run; and to strengthen the budgeting process. This was to be achieved by rationalizing allocations to recurrent expenditure, especially on wages, interest payments and transfer, while allowing development expenditure to grow (Republic of Kenya, 2002).
There has been increased development expenditure, especially that targeting government investment in core social expenditure in education and health. The expenditure strategy adopted in the Economic Recovery Strategy (ERS) document was to restructure overall expenditure by gradually reducing the level of recurrent expenditure. This was aimed at facilitating a rapid increase in development expenditure within a sustainable macroeconomic framework (Republic of Kenya, 2004).

21
In the Vision 2030, the government is targeting an economic growth rate of over 10 percent by 2030 (Republic of Kenya, 2007). To achieve this, the government has proposed the control of government expenditure to ensure that it does not lead to crowding out of private investments. The key element of the fiscal strategy includes containing growth of total expenditures while creating fiscal space through expenditure rationalization to shift resources from nonpriority to priority areas, including expenditure on the flagship projects that are critical to achieving Vision 2030 (Republic of Kenya, 2007). In this context, the wage bill is expected to decline gradually to 6 percent, suggesting the need for civil service reforms that would facilitate higher remuneration for smaller and more efficient civil service. The increasing requirements for operation and maintenance for the expanded infrastructure has been catered for. The share of development expenditure in total outlays is to be increased from 18 percent in 2007 to 38 percent in 2012 and thereafter. Most of the increase in development expenditure is to benefit the priority sectors such as the infrastructure (expansion of road networks, energy and water supply capacities, and information and technology), agricultural sector and social sectors such as health and education. The country is to scale-up resources towards the above sectors in order to ensure efficiency and effectiveness in their use and management (Republic of Kenya, 2007).
Given the trends in the government expenditure and GDP growth rate, the government expenditure reforms that have taken place so far, and the aspirations of the vision 2030, it is important that a study on the effects of government expenditure and government expenditure reforms on economic growth be carried out. This will unearth the necessary government expenditure reforms that are still needed in order to achieve the Vision 2030.

The Statement of the Problem
The causes of much of the variations in economic growth over time are not well understood. In particular, the effect of government expenditure on economic growth has not been explored exhaustively. Several studies have attempted to investigate the channels through which different types of government expenditure can affect growth (Landau, 1983;Diamond, 1984;Barro, 1990;Davarajan et al. 1993;Kweka, 1995;Colombier, 2000;Njuguna, 2009a). From these studies, the effects of government expenditure on economic growth appear to be inconclusive. Despite this uncertainty, theory suggests that government expenditure has a positive effect on economic growth (Keynes, 1936;Solow-Swan, 1956;Musgrave and Musgrave, 1989;Barro, 1990;Salai-i-Martin, 1992, and. In Kenya, economic growth has been fluctuating despite the government expenditure increasing over time. The Kenyan government spends substantial amounts of money annually on physical infrastructure, education, health care, economic services, public order and national security, defense and general administration. From theory, when there is an increase in government expenditure in these sectors, it is expected that the economy will exhibit a positive economic growth, but this does not seem to happen in the case of 23 Kenya (See figure 1.3). This could be due to non growth-enhancing expenditures that crowd-out outlays that are meant to boost economic growth (Colomber, 2000). Therefore, the issue of which government expenditure can foster permanent movements in economic growth becomes important.
The Kenyan government has undertaken various budgetary rationalization and reforms aimed at curbing unproductive government expenditure, which has been rising over the years. Government expenditure has also been restructured to enhance economic growth by increasing development expenditures, especially those targeting public investments, such as those on education and health. However, despite the reforms, economic growth has not kept pace with government expenditure growth. Therefore, there is need to investigate the impact of government expenditure and its reform on economic growth. In particular, understanding the impact of the different components of government expenditure and reforms on economic growth is crucial to policy makers.

Research Questions
The study sought answers to the following questions (i) What is the relationship between the components of government expenditure and economic growth in Kenya?
(ii) What are the effects of the components of government expenditure on economic growth?
(iii) What is the effect of government expenditure reforms on economic growth?

24
(iv) What are the policy implications of the study findings?

Objectives of the Study
The main objective of this study was to analyze the impact of government expenditure on economic growth in Kenya for the period 1963 to 2008. The specific objectives were: (i) To investigate the relationship between the components of government expenditure and economic growth in Kenya.
(ii) To examine the effects of the components of government expenditure on economic growth.
(iii) To analyze the effects of government expenditure reforms on economic growth.
(iv) To draw policy implications from the findings.

The Significance of the Study
Due to substantial policy and structural changes that have taken place in the Kenyan economy over the period 1963-2008, this study attempts to provide an empirical analysis of the impact of government expenditure on economic growth. More specifically, the impact of the various components of government expenditure on economic growth is analyzed. This is important to policy makers because it enables them to identify the inherent drive of the expenditure growth and consequently be able to effectively target the relevant expenditure components for any fiscal action in line with both medium and long term growth objectives of the country.
Equally, since most adjustment programmes stress the need to reduce government expenditure or undertake expenditure switching, this study contributes to policy discussion on which expenditure to maintain and how expenditure switching is likely to affect economic growth in Kenya. The study further enhances the understanding of the short term and long term effects of the government expenditure reforms undertaken by the government.
Furthermore, the study makes a contribution to both theoretical and empirical literature on the effect of government expenditure and reforms on economic growth, thereby paving the way for further research.

Scope, limitation and organization of the Study
The study was limited to the period 1963 to 2008, since this period was characterized by substantial growth in government expenditure, and because time series data was only available for this period of time. Economic growth can be affected by both fiscal and monetary policies, but this research only looked at the fiscal policy. The fiscal policy constitutes government expenditure and government revenue. The research, however, concentrated on the central government expenditure in terms of the level, composition and reforms, and its effect on economic growth, without addressing the means of financing. Government expenditure was disaggregated into various components in terms of investment and consumption expenditure.

26
The thesis is organized in five chapters. Chapter one introduces the study and its objectives. Chapter two presents the theoretical and empirical literature together with the theoretical frame work. Chapter three focuses on research design and methodology. Empirical results are presented and discussed in chapter four while summary, conclusions and policy implication are presented in chapter five.

Introduction
In this chapter, both theoretical and empirical literature on government expenditure and economic growth is reviewed. The first section reviews the theory and exposes the theoretical foundations that underlie the effects of government expenditure on economic growth. The theoretical representations of the models are described. The second section reviews studies carried out on the subject, and the final section deals with the critic of the literature.

Theoretical Literature
There are several theories advanced on government expenditure. The following is a brief discussion on each one of them.

Wagner's Organic State Theory
The German economist Adolf Wagner (1835-1917) advanced a law of rising public expenditure by analyzing trends in the growth of public expenditure and in the size of public sector in many countries of the world. This theory is primarily concerned with the explanation of the growth of the share of GNP taken up by the public sector. This theory, popularly known as Wagner's law, states that as per capita income grows, the relative size of the public sector will grow also. This is because the state would need to expand administration and law and order services; increased concern for distributional issues; and a greater need to control private monopolies and other forms of market failures.

28
Thus, the state grows like an organism reflecting changes in the society and economy and making decisions on behalf (and to the benefit) of its citizens (Brown et al. 1996).
The flaw in Wagner"s theory is that it does not contain a well articulated theory of public choice. The law assumes the problems of public choice by employing an organic theory of the state. Thus the state is assumed to behave as if it were an individual existing and making decision independently of the members of society. Expansion of public sector also cannot be explained in the absence of industrialization, and finally, the law concentrates upon a demand side explanation of government expenditure growth without considering the supply side explanations. In many ways, Wagner's law provides a good explanation of public sector growth. Its main limitation is that it concentrates solely on the demand for public sector services. What must determine the level is some interaction between demand and supply.

Peacock and Wiseman's Political Constraint Model
Analysis of the time path pattern of public expenditure by Peacock andWiseman (1890-1935) established the displacement effect. This model is based upon a political theory of government expenditure determination, namely that governments like to spend more money; that citizens do not like to pay more taxes; and that governments need to pay attention to the wishes of their citizens. The model assumes that there is some tolerable level of taxation that acts as a constraint on government behaviour. As the economy (and thus income) grows, tax revenue at constant rate would rise, thereby enabling government expenditure to grow in line with GNP (Peacock and Wiseman, 1961).
During period of social upheaval such as war, famine or some large-scale social disaster, the gradual upward trend in government expenditure would be distorted (displaced upward). In order to finance the increase in government expenditure, the government may be forced to raise taxation level, a policy which would be regarded as acceptable to the electorate during period of crises. This is called the displacement effect (Peacock and Wiseman, 1961).
Besides, there is also the inspection effect. This arises from people's keener awareness of social problems during the period of upheaval. The government, therefore, expands its scope of services to improve these conditions, since people's perception of tolerable levels of taxation does not return to its former level, the government is able to finance these higher levels of expenditure originating in the expanded scope of government and debt charges. The net result of these two effects is occasional short-term jumps in government expenditure within a rising long-term trend (Peacock and Wiseman, 1961).
The theory is very relevant to the Kenyan situation because the economy has experienced so many displacements in terms of tribal clashes, famine, and revenue boom like coffee boom of 1976/77. In all these, the government expenditure increased drastically without it falling in the subsequent periods (Republic of Kenya, 2003). The theory is not comprehensive since there are other periods where the government expenditure has gone up and yet there were no war or famine.
The flaw in Peacock and Wiseman"s political constraint model is that if the situation after crises is not explored, the growth in government expenditure may change. It is also important to note that government finances its expenditure from internal and external borrowing, aid, income from sale of goods and services produced by government agencies and also from abroad.
Thus taxation is not the only form of financing government expenditure (Brown et al., 1996). The theory has also been criticized for giving insufficient weight to political influences on the level of public expenditure. Moreover, the theory does not isolate all relevant causes at work. Furthermore, the critics of this theory are based on answer to the question: What happens to expenditure in the post war period?. There is no long-run displacement effect in the case where civilian public expenditure in the post war period return to their original growth path or in the case where there is only a temporally increase in post war civilian public expenditure until the old trend line is reached. There is evidence that after deferred civilian public expenditure has taken place following the war, public outlays return to the pre-war level (Brunkhead and Miner, 1979).

Keynesian Theory
This theory is based on Keynes (1936). Keynesian economics promoted a mixed economy in which both the state and the private sector were considered 31 to play an important role. Keynesian economics sought to provide solutions to what economists believed to be the failure of laissez-faire economic liberalism, which advocated that markets and the private sector operated best without state intervention (Trotman, 1997).
In Keynesian theory, macroeconomic trends could overwhelm the micro-level behavior of individuals. The theory is based on the assumptions of: The economy is operating in the short-run, wages and prices are fixed, money market is not important, taxation is in form of lump-sum taxes only and planned consumption and planned saving are both related to income. Keynes asserted the importance of aggregate demand for goods as the driving factor of the economy, especially in periods of downturn. The theory argued that government policies could be used to promote demand at a macro level and to fight high unemployment and deflation (Branson, 1989). Keynes believed that the government was responsible for helping to pull a country out of a depression. If the government increased its expenditure, then the citizens were encouraged to spend more because more money was in circulation. People would start to invest more, and the economy would go back to normal. If the change in government purchases is denoted by G  and the corresponding impact on output by Y  , then the net impact of an increase in government expenditure on output (Y) is given as: Where: b stands for the marginal propensity to consume. So an increase in government purchases causes a large increase in output and vice versa. A central conclusion of Keynesian economics was that there was no strong automatic tendency for output and employment to move towards full employment levels (Keynes, 1936).
The flaws of Keynes theory are: The theory tended to give rise to the phenomenon known as "stop-go". That is, in periods of high unemployment, the government would expand aggregate demand. This would reduce the unemployment but at the same time tend to create inflationary pressure so that eventually the government would have to reduce aggregate demand again.
Thus, all "go" period tended to be followed by "stop" period and it became difficult to achieve long term economic growth. A second limitation of the Keynesian model is that it fails to take adequately into account the problem of inflation. Third, it tends to understate the influence of money on the real variables in the economy. A change in the money supply, only affects national income through its effects on the rate of interest.

Monetarist Theory
This theory stresses the primary importance of money supply in determining nominal GDP and the price level (Ahmed, 1999). Friedman (1956) argued convincingly that the high rates of inflation were due to rapid increases in the money supply. The key to good policy was therefore to control the supply of money. The foundations of the model were: There is a close relationship 33 between the changes in the money supply and changes in national income in the long-run, without government interference the economy will tend towards its "natural" rate of unemployment, velocity of circulation of money is predictable, money changes will only affect real national income indirectly and the economy is in equilibrium at full employment Monetarists disliked big government and tended to trust free markets. They did not like government expenditure and believed that fiscal policy was not helpful in bringing about economic growth. Where it could be beneficial, monetary policy could do better. Excessive government expenditure only interferes in the workings of free markets and could lead to bloated bureaucracies, unnecessary social programmes and large deficits (Cullison, 1993). The short comings of the model include the following. First, the monetary theory does not offer a complete explanation of the complex phenomenon of changes in the making of which the non-monetary factors also significantly matter.

Crowding Out Theory
This theory as developed by Bacon and Eltis (1970) states that government intervention leads to reduction in private investment activities and this is known as 'crowding out'. The assumptions on the model are: The income generated in the market would equal the value of output; there is no nonmarket (government) sector in the economy, taxes form the channels of resources from market sector to non-market (government) activity. There are two forms of crowding out. The first form is the direct crowding out, whereby public sector production uses resources that could otherwise be used by the private sector (Trotman, 1997). If the public sector replaces the private sector, it is expected to constrain economic growth. This displacement effect occurs directly as the public sector use tax revenues to buy resources that would otherwise be used by the private sector.
The second form is the indirect crowding-out which occurs when government expenditure, taxation and government borrowing cause disincentives to productive effort, namely to work, to save and to invest (Trotman 1997).
Disincentives to invest occur when borrowing lead to higher interest rates or inflation. The sale of government debt in order to finance public sector borrowing could lead to a rise in interest rates, which indirectly crowds out private investment. This would happen if either the increased public sector borrowing leads to higher interest rates or the private sector investment is highly interest-elastic that it results in the fall in private investment. The limitations of this model include: The theory neglects the importance of public sector services as inputs to the private sector especially education which is important in increasing the skills of the work force, and there are, of course , several other determinants of interest rates in the economy in addition to public sector The theory is very applicable in Kenya since there is a coexistence of both public and private sectors in the economy. The proportion of private sector 35 depends on the portion of the public sector. If public sector is to increase its productive investment, then private sector will be crowded out (Bailey, 2002).

Musgrave-Rostow's Theory
This theory takes government expenditure as a prerequisite of economic development, its level being directly related to the stage of development that a country has reached. In the early stage of economic growth and development, public investment as a proportion of the total investment of the economy is found to be high. The public sector provides social infrastructure overheads such as roads, transport infrastructure, sanitation services, law and order, health, education and other investments in human capital, which are all necessary to gear up the economy for takeoff into the middle stages of economic and social development (Musgrave and Musgrave, 1989).
In 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. During the two stages of development, markets failures exist, which can frustrate the push towards maturity, hence increase in government involvement in order to deal with these market failures. In the mass consumption stage, income maintenance programmes and policies designed to redistribute welfare grows significantly relative to other items of government expenditure, and also relative to GNP (Musgrave and Musgrave, 1989).
The theory ignores the productive expenditure of the public sector and assumes that government plays the major role in development, which may not be the case always (Brown et al., 1996).

Neo-Classical Theory of Growth
Most ideas concerning economic growth starts from the aggregate production function where factors of production determine the national output. According to the Neo-classical theories as advanced by Solow-Swan (1956), growth comes about in three ways if land is held fixed: increase in the labour supply; increase in the capital stock; and increase in productivity. Increasing labour supply generate a larger output. Real output rises if more people take part in a country"s production, that is through immigration, or if people who are not a part of labour force begin working. Capital increase can be divided into two parts; increase in physical and human capital. Physical capital increases output because it enhances the production of labour and provides valuable services directly. A productive increase can for instance take place when there is investment in equipment like computers and machinery which can for example reduce labour hours.
Human capital promotes economic growth because people with skills are more productive than those without them. Investment in human capital is made through university studies and on the job training. Productivity increases explain the increase in output that can be explained by the input increases (labour and capital). This is called the productivity of input and can be 37 affected by a number of factors: By either financing or supplying directly the investments that the private sector would not supply in adequate quantities because of various market failure in certain kind of infrastructure projects and basic education and health expenditure, which could directly boost private sector productivity; by efficiently supplying certain basic public services that were necessary to provide basic conditions for entrepreneur activity and long term investment; and by financing its own activities in the manner that minimizes distortions to private sector savings and investment decision and to economic activities more generally (Burda and Wyplosz, 2001). Within this framework, government expenditure could in principle impact growth by affecting capital and/or labour as well as the generation and/or assimilation of technological progress reflected in total factor productivity (TFP). However, since it is assumed in the model that the long-run growth rate is driven by the population growth and the rate of technical progress, which is considered to be exogenous, the effect of government expenditure on growth through production factors is considered to be only transitional.
The theory has some short comings which include the following. First, it provides an inadequate explanation of economic growth. Second, the theory does not give clear understanding of differences among nations-why some are rich and some remain poor and why some grew rapidly while others stagnate. 38

Endogenous Growth Theory
The chief inventors of endogenous growth theory are Paul Romer and Robert Lucas (1990). This theory highlights the fact that if productivity is to increase, the labour force must continuously be provided with more resources.
Resources in this case include physical capital, human capital and knowledge capital (technology). Therefore, growth is driven by accumulation of the factors of production while accumulation in turn is the result of investment in the private sector. This implies that the only way a government can affect economic growth, at least in the long-run, is via its impact on investment in capital, education and research and development. The approach makes improved education (and indeed any kind of training or research that adds to human knowledge in any country) the key to achieving economic growth.
Faster economic growth is associated with a higher rate of investment by the private or government sector, a lower share in GDP of government consumption spending, higher school enrollment rates, and greater political stability. Unlike neo-classical growth theory, technical change is no longer based to chance, but can be fostered and promoted by appropriate policies.
Moreover, as the foundation for innovation and entrepreneurship are secured, the probability of further technical change and associated economic growth occurring, rise significantly. Technical change is no longer regarded as unexplainable and due to chance as in neo-classical theory, but in endogenous theories becomes itself a variable which can be influenced by policy decisions and should now be included within production functions, alongside the 39 conventional inputs of labour and capital. Government policies can affect economic growth rates by taxing consumption, subsidizing investment and research, and shifting resources from government consumption to government investment.
Reduction of growth in this models occurs when government expenditure deter investment by creating tax wedges beyond what is necessary to finance investments or taking away the incentives to save and accumulate capital (Folster and Henrokson, 1997).

Summary of theoretical literature
In summary, Wagner"s organic state theory can only be applicable in Kenya to a limited extent for it only looks at the demand side of the economy. Peacock and Wiseman"s theory assumes that there is some tolerable level of taxation that acts as a constraint on government expenditure and behavior. On the other hand, Keynesian theory fails to account for the inflation in the economy resulting from the increased government expenditure, monetarist theory is against government intervention in the economy, crowding out theory show that government intervention leads to reduction in private investment activities which adversely affect economic growth, Musgrave-Rostow theory is limited by the fact that it ignores the productive expenditure of the public sector and assumes that government plays the major role in development, which may not be the case always. Neo-classical theory modeled growth through exogenous technical progress. By definition when technical progress is exogenous, it 40 cannot be affected by policy. This theory is also based on decreasing returns to capital.
The development of endogenous growth theory has overcome these limitations by explicitly modeling the process through which growth is generated. This allows the effects of government expenditure to be traced through the economy and predictions made about its effect on economic growth. This model provides a positive role for government in the growth process. The theory highlights that if productivity is to increase, the labour force and capital should be provided with more resources. Therefore, endogenous theory of growth seemed to be the most relevant in Kenya and was the one adopted in this study.

Empirical Literature
There are several empirical literature that have been conducted on the effect of government expenditure on economic growth. These studies have looked at aggregate and disaggregated levels. The following is a brief discussion on each one of them. Landau (1983) used panel data of 27 Less Developed Countries (LDCs) to investigate the relationship between government expenditure components and economic growth. The methodology used by the study was Ordinary Least Squares (OLS). The variable used was government expenditure, which was broadly categorized as productive and consumption expenditure.

41
The model used by the study was given as: where Y was the GDP, I was productive government expenditure, G was the consumption government expenditure,  was the marginal productivity of capital in nongovernmental sector,  was elasticity of the nongovernmental output with respect to labour and L was labour input. Landau"s model, measured the size of the government by the ratio of government expenditure to GDP.
The results were that consumption expenditure had a negative impact on economic growth, while productive expenditure had a positive effect on economic growth. In particular, the study found the following results: First, public investment on transport and communication was positively associated with economic growth. This finding was supported by the result of Canning and Fay (1995) on infrastructure and growth, which revealed a strong relationship between the physical stock of roads and growth. Secondly, general investment was positively correlated with growth. Finally, public enterprises investment was negatively correlated with private investment. The strength of the finding is that it offers moderate support for the view that infrastructure investment foster growth, but that public investment in general does not. The weakness of the study was that it did not conduct causality tests.
The use of OLS was not backed by any economic theory.
42 Koori (1984) studied the existence and nature of the crowding effect in Kenya using Ordinary Least Squares and found that phenomenon growth in domestic deficit financing of government expenditure crowded out private borrowing.
The study used the time-series data. Particularly the study pointed that the pattern of public sector investment expenditure completely crowded out the private investment in the manufacturing industry, electricity and water sectors.
The study considered only one side of the government expenditure that is public investment. It did not consider public consumption. The main weakness of the study was failure to test for long-run relationship between the variables using cointegration analysis. Landau (1985) using cross-section data for 104 countries, including 96 developing countries used panel data to analyze the effect of government expenditure on economic growth. The variables used were current expenditure and infrastructural expenditure. The study found that the big government measured by the share of expenditure in consumption reduced growth of per capita income. This method could not be used to separate short-run and longrun effects on government expenditure. The study concluded that total government expenditure, current expenditure and infrastructure had a dampening effect on growth. Landau (1986) extended the earlier study using sixty five countries and related growth in per capita income to several set of independent variables that included; human and physical capital, structure of production, historical and political factors, resources, population and geo-climate, a measure of international economic conditions, and a three year lagged averages of the share of the government spending in GDP, which was disaggregated into education, defense, and transfer payments. By disaggregating government expenditure while holding other determinants of economic growth constant, it was found that government consumption had a negative and statistically significant influence on growth, whereas the influence of spending on education was positive but statistically insignificant. The results also showed that the influence of military expenditure net of the effect of taxation to finance it was essentially zero, as was the effect of transfers. Thus the earlier findings were re-affirmed (Landau, 1983). The methodology used was Generalized Least Squares (GLS), which may not be appropriate for timeseries data in the present study. Ram (1986) investigated the government size and economic growth using a cross-section and time-series data. The variables used were private investment, government expenditure and labour force growth rate.
is the growth of labour force, and ( Y Cg ) is the ratio of government consumption to GDP.

44
The methodology used by the study was Ordinary Least Squares (OLS). The study noted that a large government size was likely to be detrimental to efficiency and economic growth because: government operations were often conducted inefficiently; the regulatory process imposed excessive burdens and costs on economic system; and many government"s fiscal and monetary policies tended to distort economic incentives and lower the productivity of the system. At the same time, the study highlighted that a large government size was likely to be a more powerful engine of economic development. This is due to various reasons such as: role of the government in harmonizing conflicts between private and social interests; prevention of exploitation of the country by foreigners; and securing an increase in productive investment and a socially optimal direction for growth and development.
The study is relevant to the Kenyan case since in a developing economy, both the level and composition of government expenditure are important. The model was based on a mixed economy where both public and private sectors coexist, which is the case in Kenya. However, the results of the study may be unreliable because they are subject to an endogeneity bias, resulting from the failure to account for the endogenous nature of some of the explanatory variables. The specification bias may also have resulted from the fact that the growth regress takes the relationship between the explanatory variables and growth to be linear, whereas in fact it is nonlinear. Furthermore, the researcher could not conduct causality and cointegration tests.

45
Diamond (1989) studied government expenditure and economic growth using a sample of 42 developing countries in Asia and Africa. The study used panel data. The variables were education, health and capital expenditure. The study found that social expenditure (expenditure in education and health) exhibited a significant growth in the short-run, while infrastructure on capital expenditure had a negative effect on growth. Diamond attributed this negative relationship to the long gestation period before such expenditure could be productive. By using cross-section data the study may be suspect due to the fact that the countries pooled differed markedly in their economic structures. The current study will use time-series data.
Grier and Tullock (1989) used a panel data of 113 countries for the period 1951 to 1980. The variable used was government consumption. The finding was that government consumption had significant negative effect on economic growth, but positive effect for the Asian sub sample. The study did not disaggregate government consumption to find out the effects of each component on economic growth. The current study will consider disaggregated components of government consumption. The study will also consider public investment in addition for it to be more informative. Romer (1990) used cross-section data of 112 countries from 1960 to 1985.
The variables used were government expenditure, government consumption, government investment and human capital. The study found a significant and negative correlation between government consumption and economic growth, 46 but a positive effect and significant relation between government investment and human capital. The methodology used was GLS based on panel data. The study suffers from the heterogeneity of the underlying data set. Countries differ from each other in many respects such as in their political, economic system, culture, histories, demographics, resources and locality. As a result, it has so far proved quite difficult to empirically show a robust long-term correlation between government variables and economic growth. To avoid parameter heterogeneity, time-series study should be carried out. Barro (1990) investigated the relationship between public finance and endogenous growth. The model used in the study to estimate economic growth based on government expenditure was given as: where  and u were parameters in the assumed utility function,  was the tax rate, g was productive government expenditure,  was long-run growth rate, A was the productivity factor, y was GDP and  was the elasticity.
The study found that the growth rate of real per capita GDP was positively related to initial human capital while consumption to real GDP had a negative association with growth and investment. The argument was that public consumption had no direct effect on private sector productivity but it however lowered savings and growth through its distortional effects on taxation and government expenditure programmes. In later works, Barro (1991) found that on the overall, government expenditure had a negative impact on growth.
A more general issue of how fiscal policy affects economic growth was considered by Barro (1991), Dowrick (1993) and De la Fuente (1997 where Y was the GDP, Ip was the private investment, IG was the government investment, L was labour input, K was the capital input and G was the government expenditure. The estimation technique used by the study was The study assumed that a given population of identical-economical agents maximized a constant inter-temporal elasticity of substitution utility function of the form: where c was consumption per person and ρ was the constant subjective rate of time preference. The growth model estimated by Cashin (1995) was where A was a parameter that represented the level of technology,  was the output elasticity, 1 T was marginal tax rate used for the provision of public capital and on output, 2 T was the tax used for transfer, y was the implicit growth rate of GDP and  was the constant subjective rate of time preference.
where IGOV was the ratio of public investment to (GDP), SOCSEC was the ratio of transfer to GDP, and CURREV was a measure of the ratio of current tax revenue to GDP.
The estimating technique used by the researcher was Ordinary Least Squares based on the panel data. This method assumed linearity of variables, which might not be the case. The method cannot be used to separate short-run and long-run effects of government expenditure on economic growth. The research provided additional information by including the financing side of the public sector. It also tested the Granger-causality between government expenditure and national income.
The study found that increased government expenditure on those items that entered private production functions as inputs enhanced economic growth.
Examples of such productive expenditures included public investment and intergenerational transfer payments, both of which generated positive growth.
However, the size of government was limited by the need to fund such government expenditure by the levying of distortionary taxes, which reduced the marginal return to private capital and so dampened economic growth. A clear implication of the theoretical and empirical work presented by the research was that there was significant trade-off involved in considering the various contribution of government to economic growth of countries. The study could have generated more information had it included other variables such as defense, health, national security and economic affairs, which could affect economic growth. Devarajan et al. (1996) used the cross-section data for 43 less developed countries for the period 1970 to 1990 to investigate the relationship between public expenditure and economic growth. The variables used were government consumption, government investment and functional categories of public expenditure. The study found that government consumption had positive effect on economic growth, government investment had a negative effect in less developed countries but the results were reversed in the case of advanced countries. The study divided expenditure into productive and unproductive categories, taking into account the levels and mixes of both resources absorbed and output produced by different programmes. The usefulness of productive and unproductive classification for growth was apparent in a dynamic context because it focused on the impact of expenditure on savings and investment and hence capital accumulation.
There were three dimensions of this impact. First, government expenditure needed to be financed and therefore reduced resources for private savings.
Second, to the extent that it improved productivity, it stimulated private savings. The combined impacts of these effects on private savings suggest that the relationship between the levels of government expenditure and growth was 54 typically not monotonic. The methodology used was Ordinary Least Squares (OLS). This was a basic panel data regression that could not separate the short-run and long -run effect of government expenditure and growth. The study did not consider the causality of variables. However, the current research will use the Vector-Auto Regressive (VAR) model to establish the interaction and causality of government expenditure and economic growth.
Cointegration analysis will be used to establish the long-run relationship between government expenditure and economic growth.
Kocherlakota and Yi (1997) analyzed how public capital and taxes affected economic growth in the United States and the United Kingdom in the period 1891 to 1991, and from 1831 to 1991, respectively. It was found that public capital boosted economic growth and taxes hindered economic growth as was predicted in endogenous growth theory. The research only took into account physical capital variable. By focusing primarily on the impact of total public capital expenditure on economic growth, the study largely disregarded the differences in the impact of the various components of public capital such as human capital. By not considering the human capital, the study was not comprehensive, since economic growth is brought about by accumulation of both physical and human capital. The methodology used was Ordinary Least Squares (OLS). This method assumed linearity, which might not have been present.

55
Singh and Weber (1997) did a study on the composition of government expenditure and economic growth. The variables used were infrastructure, justice, defense, education and agriculture expenditure. The study analyzed Swiss time-series data from 1950 to 1994. The estimation technique used by the research was polynomial distributed lag model. The research found that outlays for transport, infrastructure, justice and defense were growth enhancing. Agricultural expenditure as a whole needed further examination, and education expenditure evidence was mixed. This was because of complex relationship between education expenditure and economic growth. The results emphasized that the composition of government expenditure was crucial for the economic growth. The weakness of this study is that due to short period, the estimation technique used may lead to a considerable loss of the degrees of freedom. Ghali (1999) did a study on the effect of government expenditure on economic growth. The study used time-series data for OECD countries from 1970 to 56 1995. The variables were government investment, exports and imports. The results were that government expenditure Granger-causes growth directly for most of the countries. The study only considered the Granger-Causality test but not the effects of government expenditure on economic growth. Tanninen (1999) used panel data of 52 countries for the period 1970-92. The method of estimation employed by the study was General Methods Moments (GMM). The variables used were investment, categories of government expenditure and income inequality. The study found that government expenditure and consumption had negative impact on economic growth, public spending on public goods was growth retarding for large government expenditure but not for small government expenditure, while social security spending was positively related to economic growth. Njuguna (1999b)  Muhlistal (2000) in the study on the relationship between government expenditure and economic growth in Turkey found no co-integration between 58 GDP and government expenditure. This means that there was no long-run relationship between government expenditure and GDP. On the basis of the Granger Causality Tests, the research also found that growth of government expenditure had no effect on economic growth. The methodology used was that of OLS based on the time-series data. This model is not appropriate since it assumed linearity of variables which may not be the case as far as government expenditure variables are concerned.
Colombier (2000)  where y was the per capita output, k was the per capita private capital, A was the production technology, and g was the government provided goods and services. The study employed a Cobb-Douglas-type production function, with government provided goods and services (g) as an input to show the positive effect of productive government spending and the adverse effects associated with distortionary taxes.
The technique used to estimate the equation was autoregressive distributed lag (ADL). The equation was estimated using time-series data. The study found that fiscal policy mattered for economic growth. Productive consumption and public investment had a role in determining growth of real per capita income in Kenya. Productive consumption seemed to have a strong negative effect on growth, suggesting that composition of this expenditure category needed to be re-examined with a view to re-organize it so that it contributes to economic growth. On the other hand, the study suggested that boosting public investment could enhance its complementarity role to private investment and hence enhance growth.
The lesson learned from this study is that government should increase its own investment in areas that are beneficial to the private sector and reduce those that compete with or crowd it out. The research also concluded that any austerity measures aimed at reducing government expenditure should not be achieved by budgetary cuts on development budget as is often the case in Kenya, for this reduced public investment. Consistent with theoretical prediction, unproductive consumption expenditure and distortionary taxes had neutral effects on growth. The study suffers from misspecification because taxes were used and yet they represent the financing of government expenditure. The study was very aggregative since all government expenditures were grouped into productive and unproductive, of which it is not possible to know which is productive or unproductive before estimation.
The study used non fiscal variables such as private investment, school enrolment and foreign aid in form of grants. The study would have been more informative had it included variables such as government expenditure on education and health, which was included in the present study. Njuguna (2009a) studied the government expenditure and economic growth in

Overview of Literature
It is evident from the foregoing empirical literature that most studies were cross-sectional and specific country studies were rare. This meant that their general conclusions could not be useful for policy decision in individual countries because of diversity of their experiences and policies taken.

62
The current study has attempted to overcome the shortcomings of To overcome these shortfalls, this study tested the time series data used for the presence of unit roots, cointegration analysis were performed from which inferences on the long-run relationships between the variables were derived.
Granger-causality tests between the components of government expenditure and economic growth was conducted before finally using the Vector  (1984) and Njuguna (2009a).
Overall, the reviewed literature has provided evidence that government expenditures on physical infrastructure, education and health care are growth enhancing. Also some evidence has been found that expenditure typically not characterized as productive such as certain kinds of social benefits and justice and defense could as well be conducive to economic growth. This study has considered additional components of government expenditures that previous studies omitted. This includes economic affairs, general administration and services, public debt servicing, public order and national security, and government expenditure reforms.

Introduction
The chapter presents the empirical model adopted for the study. The variables used in the study are defined. The data, the data sources and the methods used in data analysis are explained.

Research Design
This study aimed at establishing the effects of government expenditure on economic growth in Kenya. Quantitative data were used in the study to answer the research questions posed in chapter one. The study used data for the period 1963 to 2008 for the components of government expenditure, namely: government investment, physical infrastructure, government consumption, public debt servicing, general administration and services, defense, public order and national security, education and training, health care and economic affairs. The VAR model was used for estimation after undergoing time-series property tests.

Theoretical Framework
This study used a modified version of Ram (1986)  Assuming a constant productivity differential between labour in both sectors: where  > 0 implies lower productivity in the public sector (the reverse would be the case if  < 0) and  0  Totally differentiating (3.1) and (3.2), given that national income Y = D + G, Substituting (3.5) into (3.4) and rearranging: Using (3.5) then: This implied: Substituting (3.7) into (3.6) and collecting terms: Assume the existence of a linear relationship between the marginal products of labour in each sector and the average output per unit of labour in the economy, and including a coefficient for First, government investment in infrastructure is believed to have a direct effect on economic growth through increasing the economy"s capital stock.
The second channel is the externality effect of government spending that alters economic growth indirectly by raising the marginal productivity of privately supplied factors of production through expenditure on education, health and other services, which contribute to the accumulation of human capital. The third channel is government expenditure on goods and services that increases the aggregate demand in the economy. The fourth channel is intersectoral productivity differentials which makes some sectors to be more productive than others (Ag"enor, 2007).

The Empirical Model and Estimation Technique
In order to estimate the growth effects of the composition of government expenditure and take care of the intersectoral productivity differentials, equation (3.10) was modified by disaggregating investment into government investment and physical infrastructural. There was no time-series data on Y dL D . Therefore, the study used human capital development. This is because it captured the changing quality and stock of the labour force, and as such, was preferred to where: I is the government investment, PI is the physical infrastructure expenditure, ED is education expenditure, HT is health expenditure, PD is the debt servicing expenditure, EA is expenditure on economic affairs, GA is general administration and services expenditure, DE is the defense expenditure and NS is expenditure on the national security and public order by the government.
To capture the effects of government expenditure reforms on economic growth, a vector of dummy variables D j was added to the above specification so that: where D j are dummy variables representing j = 1, 2, 3…N government expenditure reforms.
Due to insufficient theories linking the above variables, the growth rate of GDP was not functionally explained by the explanatory variables on the right hand side of equation (3.12). Growth effects of government expenditure components could also emerge rather gradually overtime because they are complementary to private investments. These prompted the use of (VAR) method. This was because VAR model is a theory-free method used for the estimation of economic relationships (Sims, 1980 andKosimbei, 2009). VAR captured the evolution and the interdependence between multiple timeseries, generalizing the univariate Auto Regressive (AR) models (Stock and Watson, 2001). All the variables in a VAR were treated symmetrically by including an equation explaining evolution of each variable based on its own lags and the lags of all the other variables in the model. The VAR model was preferred because all variables in the model were endogenous and each variable were expressed as a linear function of its own lagged values and the lagged values of all other variables in the system (Cheng and Lai, 1997). VAR was also used to test for causality between two or more variables.
Three different types of VAR models exist: The reduced form VAR, the recursive VAR and the structural VAR. The recursive and structural VAR had the same form at the level of matrix equations. The reduced VAR overrun the need for structural modeling by modeling every endogenous variable in the system as a function of the lagged values of itself and of all the endogenous variables in the system (Engle and Granger, 1987). The reduced form and the recursive VAR models were statistical models that utilized no economic structure beyond the choice of variables. The compact form of a VAR is represented as: where A 0 is n x 1 vector of constant terms, A 1 , A 2 , ……….., A p are n x n matrices of coefficients, X t is n x 1 vector of the endogenous variables and e is a vector of serially uncorrelated error terms that have a mean of zero and a covariance of matrix  .
The use of structural VAR was justified because of the possibility to stimulate the response over time of any variable in a set to either an own disturbance or a disturbance to any other variable in a system of equations (Stock and Watson, 2001). A structural VAR was used to examine the relationships among a set of economic variables, and analyze the dynamic impact of random disturbances on the system of variables in this study. In the framework, each variable irrespective of whether it was measured at levels or first differences was treated systematically. Hence, all variables in the system contained the same set of regressors (McCoy, 1997). There were no exogenous variables and no identifying restrictions. The only role for economic theory was to specify the variables to be included. The structural VAR, therefore, estimated the structural coefficients by imposing contemporaneous structural restrictions based on the economic theory.
From equation (3.12), a system of reduced form structural VAR to test for the impact of government expenditure on economic growth was estimated. In this type of VAR model, each variable was regressed on a constant variable Cij, p lags of itself, p lags of each of the other variables and the disturbance term et.
The choice of the lag length (p) was determined using the Akaike (AIC) and the Schwarz Information Criterion (SIC). There was a preference for longer lag lengths because they captured the dynamics of the system being modeled. However, they reduced the degrees of freedom and increased data requirements. This called for tradeoff between having sufficient number of lags and a sufficient number of parameters to estimate or need for more data.
The estimated coefficients of the VAR were meaningless because they lacked the theoretical underpinning (Enders, 1995). However, the coefficients of VAR were used in the derivation of impulse responses and to forecast variance decomposition. Impulse response analysis linked the current value of the error term to the future values of the variables included in the VAR or equivalently, the current and past values of the variables included in the VAR. The forecast error decomposition measured how important the error in the jth equation was for explaining unexpected movements in the ith variable (Enders, 1995).

Definition and Measurements of Variables
GDP growth rate ( Y dY ): Is the average annual growth rate of real GDP. It was measured by change in GDP at constant prices as share of GDP.

Government investment expenditure (I):
is the payment for acquiring land, buildings and other non financial assets to be used for more than one year in the process of production, including transfers for capital assets.
It was measured as the total capital expenditure including gross fixed capital formation and capital transfer as a ratio of GDP.

Data Source, Collection, Cleaning and Refinement
The research used secondary data for the period 1963 to 2008 to analyze the effects of government expenditure on GDP growth rate. Data for the study was collected through analysis of documents. Before data collection was done,  (Branson, 1989 andWawire, 2006).
Furthermore, it measures inflation correctly, since it is a weighted average of the changes in all prices of newly produced goods in the economy. Hence, "it has the advantage of incorporating all the newly produced goods in the economy and allows for changes in composition of output" (Wawire, 2006: 50) The reason for the conversion of nominal average GDP to real average GDP was that the nominal average figures did not reflect changes in production and income caused by inflation that leads to prices rising when the quantities are  (Trotman, 1997).

Stationarity of Data
The first step involved testing for stationarity of the series. This is a standard procedure performed to ensure that the series have a constant mean and variance, so that the resultant regression results would be meaningful (Tsay, 2001). Otherwise, if stationarity of the series is present and not checked, the presence of trend in the data series would mean that the regression results are spurious.
Two main methods for testing stationarity or the presence of unit roots that were applied are the Augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) tests (Perron, 1989). The ADF procedure attempts to retain the validity of the tests based on white-noise errors in the regression model by ensuring that the errors are indeed white-noise. On the other hand, (PP) procedure corrects for serial correlation through a non parametric correction to the standard statistic (Stock, 1994). PP test acts to modify the statistics after the estimation in order to take into account the effect that auto correlated errors have on the results. Therefore, PP test is desirable because it does not require estimation of additional parameters that would require additional data and also may not exhaust degrees of freedom.

Cointegration
The use of cointegration technique allowed the study to capture the equilibrium relationship between non-stationary series within a stationary model, following Adam (1998), andJohnston andDinardo (1997).
Furthermore, it avoided both spurious and inconsistent regression problems, which would have otherwise occurred with the regression of non-stationary data series. It also permitted the combination of the long-run and short-run information in the same model and overcame the problems of losing information that could have occurred from attempts to address non-stationary series through differencing (Adam, 1998). Cointegration technique made it possible to capture the information of non-stationary series without sacrificing the statistical validity of the estimated equation (Stock and Watson, 1988).
Two main tests for cointegration, namely Johansen cointegration test and the Granger two step methods were used. Johansen"s methodology, which was expressed as a VAR of order p is given by: where y t is an n x 1 vector of innovations. This VAR can be re-written as difference. However, the long-run characteristics of the data will be lost. Therefore, the study used the Johansen cointegraion method to test for the long-run relationship between the variables and not ECM.

Granger Causality Test
Granger causality was used to determine whether one time series was useful in forecasting another (Enders, 1995). The VAR equations were used to perform hypothesis is rejected when F-statistics is greater than the p-value.

Data Analysis
The study addressed four objectives. The first objective was to investigate the relationship between government expenditure and economic growth. This was

Introduction
This chapter presents the findings of the study. First, the relationship between economic growth and components of government expenditure, namely government investment, expenditure on physical infrastructure, expenditure on health care, expenditure on education, public consumption, general administration and services, defense, public debt, economic affairs, public order and national security is explained. Second, the effects of the components of government expenditure on economic growth are estimated. Finally, the chapter explains the impact of budget rationalization, downsizing, outsourcing, privatization policies and governance on economic growth.

Government Expenditure and Economic Growth
The first objective of this study was to investigate the relationship between the components of government expenditure and economic growth. The study was expected to establish whether there was a short run or long run relationship between government expenditures on investment, physical infrastructure, education, health, public debt, national security, economic affairs, general administration, public consumption, defense and economic growth.

Findings of stationarity tests
The data series were tested for stationarity using the Augmented Dicky Fuller (ADF) and Phillips-Perron (PP) tests. The reasons why the two tests are 85 required are: The ADF procedure attempts to retain the validity of the tests based on white-noise errors in the regression model by ensuring that the errors are indeed white-noise. On the other hand, (PP) procedure corrects for serial correlation through a non parametric correction to the standard statistic (Stock, 1994). PP test acts to modify the statistics after the estimation in order to take into account the effect that auto correlated errors have on the results. The results are presented in table 4.1. The results of unit roots tests showed that GDP growth rate, ratio of government expenditure on investment to GDP, ratio of physical infrastructure to GDP, ratio of education expenditure to GDP, ratio of health care expenditure to GDP, ratio of public debt servicing expenditure to GDP, ratio of public order and national security expenditure to GDP, ratio of expenditure on economic affairs to GDP, ratio of general administration and services expenditure to GDP, ratio of public consumption expenditure to GDP and ratio of defense expenditure to GDP were stationary and integrated of order I(0).
This suggested that there was a long-run relationship between ratio of government investment, ratio of physical infrastructure, ratio of education, ratio of health care, ratio of public debt servicing, ratio of public order and national security, ratio of economic affairs, ratio of general administration and services, ratio of public consumption and ratio of defense and GDP growth rate variables (Engle and Granger, 1987).
Furthermore, most results and the t-statistics for constant and trend were very close and sometimes the same for the PP. This shows that PP test is more consistent and powerful in testing for stationrity as compared to the ADF.
Johansen test was then carried out to investigate whether there was more than a single cointegration relationship between economic growth and the government expenditure variables. The results of the Johansen tests of variables are reported in Table 4.2. showed that the VAR system was stable at both lag 1 and 2. However, since there was serial correlation at lag of order 2, VAR system of lag of order 1 was preferred. The lag exclusion test indicated lag one as important in the VAR system which supports the lag selection criteria. The Granger causality test results revealed that there was bidirectional causality between government expenditure on defense, economic affairs, education, health care, public order and national security, government investment, physical infrastructure and economic growth. This means that these set of variables predicted each other and hence could be on either side of the equation, (either as dependent or as an independent variable).
The variables that had a unidirectional causality included: government expenditure on consumption, general administration and services, and public debt servicing. This implies that only one variable could explain the other. In this case, they were required to be on the right hand side of the equation as independent variables.
The Granger causality tests revealed important information on the relationship between the variables of interest in this study. Government expenditure on investment, physical infrastructure, consumption, defense, economic affairs, education, general administration and services, public order and national security, health care and public debt predicts economic growth.
These findings confirms the use of VAR model, given that there was bidirectional causality between government expenditure on defense, economic affairs, education, health care, public order and national security, government investment, physical infrastructure and economic growth.
In a nutshell, there was a feedback effect between government expenditure components and GDP growth rate, which supported the Wagner"s hypothesis that states that increase in GDP causes growth in the government expenditure, and the Keynesian hypothesis that states that increase in government expenditure causes GDP to increase. This suggests that allocation of government resources should be designed carefully in order to spur economic growth of the country.

Effects of Government Expenditure on Economic Growth
The second objective of the study was to examine the effects of the various components of government expenditure on economic growth. This was done by the estimation of the Vector Auto Regression (VAR) model and the subsequent use of impulse responses and variance decomposition analysis.
The VAR results presented in table A6 in appendix IV were not interpreted like ordinary regression equations because they were not derived from structural equations (Enders, 1995). They were used for the generation of both the impulse response functions and for conducting the variance decomposition analysis, which depicted the effect of various components of government expenditure on economic growth.
The impulse response analysis traced the effects of one standard deviation shock to the innovation on current and future values of all the endogenous variables of the system. A shock to the j th variable affected the same variable and was also transmitted to all other endogenous variables in the system through the dynamic structure of the VAR (Enders, 1995).
The ordering of variables used in this study was based on the relationship presented in equation (  The response to one standard deviation to government investment expenditure resulted in a stable time path, which declined to zero with respect to economic growth as shown in figure 4.1. This effect lasted for sixteen years on the positive territory before fizzling out. The evidence on the relationship between government investment and economic growth remained inconclusive. Knight, Loayza and Villanueva (1993) and Nelson and Singh (1994) found that the level of government investment had a significant effect on economic growth during 1980s, while Khan and Kumar (1997) and Peter (2003) found government investment less effective on growth compared with private investment. Milbourne, Otto and Voss (2003) found evidence of a positive correlation between public investment and economic growth.

97
The reasons for the increase in GDP growth was because government expenditure on buildings, plant, machinery and equipment helped to increase the productivity of the inputs used by private sector. By increasing aggregate demand, public investment could have also stimulated private investment through the accelerator effect. The increase of what was also due to increase in aggregate demand through investment.
The decline could be due to the increased fiscal deficit or the crowding out of private capital formation by reducing credit available to the private sector or by raising interest rates.

(b)
The Impact of Government Expenditure on Physical Infrastructure on GDP Growth Rate.
The impact of one standard deviation shock to physical infrastructure expenditure on GDP growth rate is shown in figure 4.2. The response to one standard deviation to government infrastructural expenditure resulted in a stable time path, which declined to zero with respect to economic growth as shown in figure 4.2. The effect of a one standard deviation shock on physical infrastructure investment on economic growth lasted for twelve years on the positive territory, and then remained at the equilibrium in the long-run. This positive effect between infrastructure and economic growth finds support in studies by Aschauer (1989) andEasterly andRobelo (1993).
Additional expenditure on infrastructure in such areas as roads, railways, ports, communication, water and electricity could have contributed to economic growth by increasing the productivity of inputs in the private sector.
High government expenditure on transport and communication and energy created an enabling environment for businesses to strive through reduced cost of production.

(c)
The Impact of Government Expenditure on Education on GDP Growth Rate.
The impact of one standard deviation shock to education expenditure on GDP growth rate is shown in figure 4.3. Government expenditure on education improved the economic growth initially for nine years, before dampening it to the negative side for another 13 years.

99
The response to one standard deviation to education expenditure resulted in a stable time path, which declined to zero with respect to economic growth. The effect lasted for twenty two years before fizzling out.
This phenomenon could be attributed to the increased marginal productivity of privately supplied factors of production. This could have occurred due to the provision of free primary and secondary education, expansion of the universities and middle level colleges, and bursary given to the needy students. Provision of free basic education and creation of subsidy schemes created positive externalities and raised the steady state rate of economic growth. However, there was a decline in its positive effect possibly due to the fact that the rate of unemployment is very high in Kenya, and therefore, there was no room for the majority of graduates in the Kenyan labour market.

100
Further, government investment on building of schools, colleges and universities were expenditure on the core functions, and therefore were expected to have a positive effect on the economy. The slight negative effect in the long-run could be because funds meant for the development of the education sector had not been properly utilized and in most cases embezzled, thus precipitating the incessant strike by academic staff union of the university and national union of teachers.
The results differ from conclusion drawn by previous studies (Landau, 1986;Barro, 1995Barro, , 1999and Devarajan, 1996). These earlier studies indicated that the association between education expenditure and economic growth was negative. The departure of the current finding from the previous studies is because of high level of unemployment in Kenya due to high misappropriation of government funds in education and due to change of education system that have experienced in Kenya.

(d)
The Impact of Health Care Expenditure on GDP Growth Rate.
The impact of one standard deviation shock to health care expenditure on GDP growth rate is shown in figure 4.4. This phenomenon could be due to the fact that health expenditure by the government raises the health status and productivity of the people, thereby promoting economic growth. The increased expectation of a longer life could affect the intertemporal discount rate and therefore savings. Increased health expenditure could increase the participation of women in the labour market, and affect fertility, which has effect on demographic transition and therefore on the economy. Further, government investments on buildings of hospitals represent expenditure on the core functions and therefore are expected to have a positive effect on the economy.

(e) The Impact of Public Debt Servicing Expenditure on GDP Growth
Rate.
The impact of one standard deviation shock to public debt servicing expenditure on GDP growth rate is shown in figure 4.5.

(f) The Impact of Expenditure on Economic Affairs on GDP Growth
Rate.
The impact of one standard deviation shock to economic affairs expenditure on GDP growth rate is shown in figure 4.6.  The government expenditure on economic affairs exhibits a positive short-run impact on the GDP growth rate that takes about fifteen years to translate fully into a permanent shift in the GDP growth rate. This result is supported by Romer (1990), who found a positive relationship between government expenditure on economic affairs and economic growth. Nevertheless, when considering the impact in the long-run growth, it is zero. The reason for the positive effect could be because economic affairs provides a direct provision of productive activities through its expenditure in areas of agriculture, manufacturing, trade, mining, fisheries, forestry, tourism and constructions.
The findings of the study supports the finding of the study Njuguna (2009a) which noted that sectors such as trade, tourism and manufacturing are very critical areas of a nation"s development and therefore the allocations to these sectors could make significant contribution to the economy if the resources are channeled to the right areas of development.

(g)
The Impact of General Administration and Services Expenditure

on GDP Growth Rate
The impact of one standard deviation shock to general administration and services expenditure on GDP growth rate is shown in figure 4.7. The response to one standard deviation to general administration expenditure result in a stable time path, which declines to zero with respect to economic growth as shown in figure 4.7. The effect of a one standard deviation shock on general administration on economic growth lasted for twelve years on the positive territory. This is in contrast with the findings of Ranjan (2008), who found a negative effect of general administration expenditure on growth.
General administration encompasses all expenditures that are of recurrent in nature. These are expenditures channeled to pay staff, buy office supplies, personnel management, purchase equity and transfer to households, public enterprises and the rest of the world. The positive effect could be attributed to the increased government expenditure on capital and transfer to households and enterprises, which act as a productive input and also help to increase the after -tax private return on capital. Transfers generated positive externalities that raised marginal product of private capital. Although a typical theoretical result is that for a given tax rate, economic growth is adversely affected by an increase in redistribution through public transfer, the result for this study is consistent with Barro (1989), Sal-i-Martin (1992 and Cashin (1995). Further, general administration could have led to an improvement in the efficiency in the markets for goods and services, factor markets and asset markets. Asset  As shown in figure 4.8, it would take eighteen years for the effect on economic growth due to fizzle out if a one standard deviation shock on defense expenditure is affected. The effect is positive in the short run but in the long run it has no effect. In the existing literature, this effect has sometimes been reported as positive and significant for example by Benoit (1978);and Frideriksen and Loony (1982). At the same time, other studies have found it to be negative for example Bils and Kienow (1998);and Knight et al. (1996), while in yet other studies, effect of defense expenditure on GDP growth has been found to be neutral by for example Ram, (1986). The attributes of this phenomenon could be due to improved security within the economy thereby increasing economic activities such as tourism, and private investment.

(i)
The Impact of Public Order and National Security Expenditure on GDP Growth Rate.
The impact of one standard deviation shock to public order and national security expenditure on GDP growth rate is shown in figure 4.9. The response to one standard deviation innovation in public order and national security and public order resulted in a stable time path, which declined to zero with respect to economic growth as shown in figure 4.8. The effect of one standard deviation shock on national security on GDP growth rate took twenty two years to fizzle out. The effect was initially on the positive side for a period of five years, and then moved to negative territory for seventeen years before moving to the equilibrium. The long period on negative side is because on unproductive nature of expenditure on public order and national security. This suggests that national security has a mixed effect on GDP growth rate and the effect is felt for a long time in the economy. The mixed effect of public order and national security is because of its efficiency of markets on mobilization of 109 resources. This result is in support of Romer (1990), who found a negative relationship between this variable and economic growth.
There is no existing literature of the positive effect and therefore this study makes an important contribution to the existing literature in this area. The positive effect could be attributed to the increased competition in the economy due to increased protection and enforcement of government legal structure.
The government spending more on the rule of law helped to improve the rights ownership of possessions and property as well as basic human rights to life and the right not to be kidnapped and enslaved. As a result, there was an improvement in the efficiency in the markets for goods and services, factor markets and asset markets. Asset markets helped to mobilize resources for investment. However, this turned into negative effect perhaps due to loss of confidence in the local courts by the public. As a result, the competition was curtailed and there was inefficiency in the above mentioned markets, which may have led to the decline in economic growth.  The response of economic growth to one standard deviation to government consumption expenditure resulted in a stable time path, which declined to zero. The results show that a one standard deviation shock on government consumption has a mixed effect on GDP growth rate that lasted for fifteen years.
In the short-run, government consumption expenditure has a positive effect on economic growth but in the long-run its effect was negative. This suggests that government consumption expenditure had a mixed effect on economic growth both in the short-run and in the long-run. In the existing literature, the effect of government consumption expenditure has produced mixed results for various studies. Barro (1997) found that government consumption expenditure measured as a proportion of GDP was negatively correlated with GDP growth.
Devarajan, Swaroop and Zou (1996) in contrast found a positive relationship between GDP growth and government consumption expenditure. Caselli, Esquivel and Lefort (1996) also found a positive effect on growth of government expenditure as a share of output. Easterly, Loayza and Montiel (1997) found no significant effect of the share of government consumption expenditure on GDP on economic growth in Latin America.
The positive effect could be as a result of increase in aggregate demand through the multiplier effect. The negative effect could have resulted from the crowding out effect due to a fall in disposable income of the households.
Financing government consumption expenditure is done through taxes or by borrowing. Hence, increased taxes lower disposable income for households and private consumption may fall accordingly. Government consumption expenditure could also have a crowding-out effect on private sector by causing positive effect on interest rates, which in turn could decrease private investment.
In summary, impulse responses traced out the responses of current and future values of each of the variables to a one unit increase in the current value of one of the VAR errors, assuming that this error returned to zero in subsequent periods, and that all other errors were equal to zero. These estimated impulse responses showed patterns of persistent common variation. The effect of the shock in government expenditure components was realized for a period of between ten and fifteen years. This means that government expenditure components have long-run impact on economic growth in Kenya.

Growth
The fourth objective of this study was to analyze the impact of government expenditure reforms on economic growth. To attain this objective, the dummies for budget rationalization, expenditure downsizing and outsourcing, privatization and governance were included in the VAR model estimation. The impulse responses and variance decomposition for the given dummies were analyzed and interpreted.

(a) The Impact of Budget Rationalization on GDP Growth Rate.
Budget rationalization was a dummy representing the reallocation of government expenditure from less productive to more productive projects of the government. It assumed the value of one during the years of budget rationalization and zero otherwise. The impact of one standard deviation shock to policy on budget rationalization on economic growth is presented in figure   4.11.

Variance Decomposition Analysis (VDA)
The VDA presented a further step of establishing the percentage of the variation in a series that was due to its own shocks and that which was due to shocks of other variables in the model at a given period following Enders (1995) and Stocks and Watson (2001). VDA determined the proportions of a variance in a series that was due to its own shock, the other variable"s shocks and other identified institution shocks.
The variations in GDP growth rate brought about by changes in government expenditure components were analyzed. This was an alternative method to impulse response functions for examining the effects of shocks to the GDP growth rate. This technique determined how much of the forecast error variance for any variable in the system was explained by innovations to each explanatory variable over a series of time horizon (Enders, 1995). The own series shocks explained most of the error variance, although the shock also affected other variables in the system. It was also important to consider the ordering of the variables when conducting VDA. This was because in practice the error terms of the equations in VAR were correlated, so that the result depended on the order in which the equations were estimated in the model. The effects of public expenditure policy reforms have also far reaching effects on economic growth but their effect is only in the short-run.
The own series shocks explained most of the error variance, although the shocks also affected other variables. The largest proportion of the variance was taken by education, economic affairs, physical infrastructure, investment, health, general administration and budget rationalization. However, their proportions declined over time. Defense expenditure, national security, government consumption, public debts, budget downsizing and outsourcing, privatization and governance did not affect the variance very much within the 30 years.

Introduction
This chapter summarizes the study and makes conclusions based on the results. The policy implications from the findings and areas for further research are also presented

Summary
The rapid growth in government expenditure in Kenya has caused concern among policy makers on the implication of such growth, especially to the whole economy in general, and the private sector in particular. Studies have noted that the allocation of financial resources through various policies is neither reflected in the government priorities nor adequately promoted growth in the past. Over the three decades, government expenditure in the country grew at a faster rate than the growth rate of GDP. Given this fiscal scenario, an explanation of this requires studying the impact of government expenditure on economic growth. The specific objectives of the study were to: investigate the relationship between the components of government expenditure and economic growth; examine the effects of components of government expenditure on GDP growth rate; analyze the effects of government expenditure reforms on economic growth; and to draw policy implications from the findings. The findings support the Rams (1986) model which indicated that large government size was likely to be a more powerful engine of economic growth.
This is due to various reasons such as: role of government in encouraging private investors and securing an increase in productive investment and a socially optimal direction for growth and development.
The third objective was to analyze the impact of government expenditure reforms on economic growth. Four government expenditure reforms were identified and included in the VAR model as dummies for estimation purposes. The study showed that the impacts of budget rationalization, downsizing and outsourcing and privatization on economic growth were positive. Governance had initially a negative effect and then a positive effect on economic growth, which means that during an electioneering period, economic growth declined before it started responding favourably once an elected government was in place and its policies were implemented and understood by the public.

Conclusion
On the basis of the empirical results, the study concludes that the composition of government expenditure matters for economic growth. In the long-run, expenditure on economic affairs, defense, education, government investment, general administration and services and physical infrastructure have positive impacts on economic growth. In the short run health care, public order and national security have positive impact on economic growth, whereas, public debt servicing has negative impact on economic growth.
The government also plays a leading role in determining the pattern of economic growth through public expenditure reforms, which determine directly how much of an economy"s resources to divert to its own use and how those resources should be allocated in order to increase economic growth. The results reveal that government expenditure reform on privatization should be given priority for a government interested in promoting long-run growth.
Budget rationalization, governance and downsizing and outsourcing appear to be appropriate policies for increasing economic growth in the short-run.
Therefore, the results of this study emphasize the fact that the composition of government expenditure and public expenditure reforms are important in determining economic growth.
In summary the contributions of this study to the knowledge include: First, the study disaggregated government expenditure into various sectors and reforms and found that each component of government expenditure and reforms had different impact on economic growth. Second, the study found a bi-directional causality between components of government expenditure and economic growth. This indicated that there was feedback effect between components of government expenditure and economic growth. Finally, the study found that a shock in government expenditure is unlikely to lead to a shock in economic growth instantaneously, but rather the effect was gradual and even out after a short period.

Policy Implications
Several policy implications can be drawn from the research findings: The government should increase its investment in areas that are beneficial to the private sector and eschew from those that compete with or crowd it out. It should increase its expenditures on those items that enter private production functions as productive public inputs that enhance economic growth. Such productive government investment expenditure includes expenditure on buildings, plant, machinery and equipment all of which generate positive externalities that raise private investment and thus economic growth. The increase in investment would increase economic growth by four percent on average per year.
The government should allocate more resources to areas of physical infrastructural development in order to stimulate economic growth as envisaged in the vision 2030. This is because according to the study finding, additional expenditure on physical infrastructure in such areas as roads, railways, ports, communication, water and electricity contribute significantly to the economic growth by increasing the marginal productivity of inputs in the private and public sectors. Furthermore high government expenditure on transportation and communication and energy create an enabling environment for business to thrive through reduced cost of production. The increase in physical infrastructure will affect economic growth by forty two percent.
The government should increase it its expenditure allocation to the education and human capital development. This is because the study found that education affect economic growth positively. This could be done through provision of education facilities, training and employing more teachers, ensuring access to education to all citizens, reduction of the cost burden to the parents/guardian and expanding education to the marginalized groups. This is because quality education creates positive externalities and increases the productive capacity that helps to raise the steady state rate of economic growth. Increase in education expenditure is expected to affect economic growth by six percent.
The government should allocate more funds for the development of the health care sector. This is because the study found that health care contributes positively to the economic growth. This should be achieved through investment in capital equipments, health facilities and provision of quality medical supplies. The government should also consider human capital investment through training of the human resources relevant to this sector such as doctors and nurses. This is because, when there is an increase in expenditure allocation to the health care sector, the level of economic growth increases since a health nation is a wealth nation. This increment is expected to influence growth by two percent.
The government should streamline its expenditure allocation to the debt servicing. This is because it was found that public debt servicing affect economic growth negatively. Public debt servicing reduces the resources that could otherwise have been allocated to more productive sectors of the economy. Furthermore, public debt servicing crowd out private investment which affect economic growth adversely. The reduction in public debt can be achieved by reducing government borrowing and ensuring that borrowed loans are concessional in nature. This means that since the government would have a long repayment period at a lower interest rate, the burden on public debt would be lesser. The reduction in public debt servicing would affect growth by two percent on average annually.
The allocation to economic affairs should be increased by the government.
Economic affairs were found to affect economic growth positively. This would affect economic growth by four percent annually. This would be achieved by ensuring that the sectors that are productive in nature under this allocation are accorded the right attention while those that are non-productive in nature are rationalized. This is because economic affairs provide a direct provision of productive activities through its expenditure in areas of maintenance and operations, agriculture, manufacturing, trade, mining, fisheries, forestry, tourism and construction.
The government should increase its expenditure allocation to the general administration and services and also increase its efficiency in service delivery.
The study established that general administration and services affect economic growth positively. This would go a long way in saving funds that could be used in other priority sectors. This implies that the government should have sustainable policies for development to avoid crowding out the private investors who play a vital role to the growth and development of the economy.
The efficiency in government sector could also lead to improved efficiency in the markets for goods and services, factor market and financial asset markets that help to mobilize resources for private investment. The effect would affect economic growth by three percent annually.
The government should increase its expenditure allocation to defense and public order and national security. This is because the study found that defense and public order and national security have a positive effect on economic growth. When the allocation to these sectors is increased, there is a positive change in economic growth. These sectors help to improve security within the economy thereby increasing economic activities in areas of tourism and private investment. In addition the sectors help to increase competition in the economy due to increased protection and enforcement of government legal structure. This would affect growth by one percent per year.
Effort should be made by the government to reduce its consumption. This is because consumption on goods and services has a negative effect on economic growth. Increased government consumption seems to crowd out private investments by reducing the disposable income of the people which result in reduced household consumption in the economy. The reduction in government consumption would increase growth by one percent per year.
The government should restructure government expenditure through budget rationalization in order to achieve an effective public sector. This would affect economic growth by two percent per year. The study found that budget rationalization had a positive effect on economic growth. It is important for policy makers to pay attention not just to the levels of government expenditure, but also to its composition. Besides, prioritization of government expenditure should be judged not only by virtue of its economic returns, but also on the technical, administrative and financial feasibility front. Measures for cost and benefits of various government expenditures are essential in this respect. A clear set of specified criteria for deciding the allocations of resources should be followed to avoid arbitrary allocation and rent seeking by promoting transparency and accountability. This requires long-term programmes of budget rationalization focusing on elimination of wasteful and unproductive expenditure thus improving equity through distributional impacts and maximizing economic growth by ensuring that fiscal operations are conducted in the least cost manner.
In order to reduce the rate of growth of expenditure on salaries and allowances, the government should streamline its civil service to the minimum by freezing recruitments and increasing wage in line with the economic growth. This is because the study found that expenditure downsizing and outsourcing has a positive effect on economic growth. The government should also adopt the advanced technologies in its service delivery in order to cut down the size of civil service. This is because reduced civil service helps in the diversion of resources from unproductive expenditure on wages and salaries, to a more productive expenditure in form of infrastructure and education. This is would affect economic growth by 1 percent annually.
The government should reduce its size to an optimal one by adopting a policy on privatization and expenditure downsizing and outsourcing to cut its expenditure and in turn ease public debt. This is because privatization was found to effect economic growth positively. The result showed that this will increase economic growth by one percent per year. Running a large public debt for a long-period of time could have an adverse effect on the economic growth since borrowing might crowd out private sector investment.
Donors and development partners should partner with government in creating a platform for investment and development and support the initiatives 131 spearheaded by the government. This is because study found that government investment contributes positively to economic growth Regional integration should therefore be explored to ensure that gains that could be obtained regionally such as trade and industrialization are maximized.
The private sector should partner with government in provision of certain services through Public-Private Partnership (PPP). This could be achieved through joint efforts in provision of services such as infrastructure, energy, health and education. This is because physical infrastructure was found to affect economic growth positively.

Areas for Further Research
In view of political challenges such as the demographic burden and climatic change, it becomes increasingly important to explore further what portfolio of government outlays is optimal in economic growth and welfare terms.
Although the focus of this study was solely on measuring the impact of government expenditure on economic growth, an important issue to address in future research is what determines government decision to allocate expenditure among various components?. In particular, the role of demographic factors and the nature of the political process is important.    (18.47 (33.91 (39.91 (18.21 (14.56 (11.97)