EFFECT OF FUEL TAXATION ON FUEL PRICE IN KENYA BY LAWRENCE NKOROI GITONGA A DESSERTATION SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE AWARD OF MASTER OF SCIENCE DEGREE IN COMMERCE (FINANCE AND ACCOUNTING) IN THE SCHOOL OF BUSINESS AND PUBLIC MANAGEMENT AT KCA UNIVERSITY NOVEMBER 2018 EFFECT OF FUEL TAXATION ON FUEL PRICE IN KENYA BY LAWRENCE NKOROI GITONGA A DESSERTATION SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE AWARD OF MASTER OF SCIENCE DEGREE IN COMMERCE (FINANCE AND ACCOUNTING) IN THE SCHOOL OF BUSINESS AND PUBLIC MANAGEMENT AT KCA UNIVERSITY NOVEMBER 2018 DECLARATION STUDENT’S DECLARATION I declare that this dissertation is my original work and has not been presented for any degree in any other university. Student Name : Lawrence Nkoroi Gitonga Reg. No 11/01871 Sign____________________________ Date___________________ SUPERVISOR’S DECLARATION This dissertation has been submitted with my approval as the university Sign____________________________ Date___________________ Dr. Njogo Lecturer School of Business and Public Management KCA University DEDICATION I dedicate this research work to my wife Virginia Kawira, our son Allan Mugendi, daughters Precious Kagwiria and Martha Nkatha and our parents for their love, support and inspiration to excel and further my studies. They have seen me through this studies, genuine love, encouragement and unwavering support. ACKNOWLEDGEMENT I wish to thank KCA University for giving me an opportunity to undertake a Master of Science degree in commerce. This study could not be possible without understanding, prayers and encouragement from my lovely wife Virginia Kawira and children Allan Mugendi, Precious Kagwiria and Martha Nkatha, who missed my company and denied themselves other needs to make my studies a success. My sincere appreciation goes to Dr. Njogo who provided academic guidance, constructive criticism and supervision throughout my study. Finally, I am grateful to the lecturers at KCA University, School of Business And public Management for their encouragement and support during the course work. ACRONYMS AND ABBREVIATIONS ERC : Energy Regulatory Commission KRA : Kenya Revenue Authority KRB : Kenya Roads Boards IEA : International Energy Agency UNEP : United Nations Environmental Program VAT : Value Added Tax TABLE OF CONTENTS DECLARATION ...................................................................................................... ii DEDICATION ......................................................................................................... iii ACKNOWLEDGEMENT ....................................................................................... iv ACRONYMS AND ABBREVIATIONS ................................................................. v TABLE OF CONTENTS ......................................................................................... vi ABSTRACT ........................................................................................................... viii LIST OF TABLES .................................................................................................... x LIST OF FIGURES ................................................................................................. xi DEFINITION OF TERMS ..................................................................................... xii CHAPTER ONE ....................................................................................................... 1 INTRODUCTION .................................................................................................... 1 1.1 Background of the Study .................................................................................... 1 1.2 Statement of the Problem .................................................................................... 7 1.3 General Objectives .............................................................................................. 9 1.4 Specific Objectives ............................................................................................. 9 1.5 Research Questions ............................................................................................. 9 1.6 Delimitation of the study .................................................................................... 9 1.7 Limitation of the Study ..................................................................................... 10 1.8 Basic Assumptions ............................................................................................ 10 1.9 Significance of the Study ................................................................................ 10 CHAPTER TWO .................................................................................................... 11 LITERATURE REVIEW ....................................................................................... 11 2.1 Introduction ....................................................................................................... 11 2.2 Theoretical Review ........................................................................................... 11 2.2.3 The Modern Theory of Taxation.................................................................... 13 2.3. Empirical Review............................................................................................. 15 2.4 Conceptual Frame Work ................................................................................... 21 CHAPTER THREE ................................................................................................ 23 RESEARCH METHODOLOGY............................................................................ 23 3.1 Introduction ....................................................................................................... 23 3.2 Research Design................................................................................................ 23 3.3 Data Collection Procedure ................................................................................ 23 3.4 Diagnostic Tests ................................................................................................ 24 3.4.1 Unit Root Test: ............................................................................................... 24 3.4.2 Co-integration Test ........................................................................................ 24 3.5 Optimal Lag Order Selection ............................................................................ 25 3.6 Data Analysis .................................................................................................... 25 CHAPTER FOUR ................................................................................................... 27 DATA ANALYSIS, INTERPRETATION AND DISCUSSION .......................... 27 4.1Introduction ........................................................................................................ 27 4.2 Preliminary Analysis ......................................................................................... 27 4.2.1 Descriptive Analysis ...................................................................................... 27 4.3.1 Stationarity ..................................................................................................... 30 4.5 VEC Model ....................................................................................................... 33 4.7 Impulse response functions ............................................................................... 37 CHAPTER FIVE .................................................................................................... 40 SUMMARY OF FINDINGS, CONCLUSIONS AND RECOMMENDATIONS . 40 5.1 Introduction ....................................................................................................... 40 5.2 Summary of the findings ................................................................................... 40 5.3 Conclusion ........................................................................................................ 43 5.4 Recommendations of the Study ........................................................................ 44 5.5 Suggestions for Further research ...................................................................... 45 REFERENCES ....................................................................................................... 47 APPENDICES ........................................................................................................ 50 Appendix I – Data Collected................................................................................... 50 Appendix II – Trend plots for variables and differenced variables ........................ 51 Appendix III – correlograms differenced variables ................................................ 52 Appendix IV – Impulse Response Graphs .............................................................. 53 Appendix V: Letter of research authorization ........................................................ 56 ABSTRACT The study is done to assess the effect of fuel taxes on fuel pricing in Kenya. In Kenya, taxes as a share of fuel prices are highest for petrol and lowest for kerosene. In Kenya petroleum accounts for 22 per cent of the total primary energy supply, 67 per cent of which is consumed in transport sector. Tax rates are mainly set through state legislation while expenses are driven by market forces such as infrastructure costs, inflation and demographic shifts. The tax rates in Kenya are given in the third schedule of income tax Act (Cap 470). These rates are revised from time to time. Measures to incorporate marketing strategies such as market promotion, advertisement, price-cut, design and improvement of customers comport are necessary. The right price strategy is important for maximizing the revenue. The most common taxes charged on fuel in Kenya include Road Maintenance Levy, Excise Tax and Railway Development Levy. The revenue collected from the fuel taxes is mostly used to construct and repair roads; however, most of the monies collected from tax are not used for the purposes it was levied for. If the taxes charged on fuel are absorbed by customers inform of increased price, their earnings decline by the amount of tax. The raised price causes inflation, lower demand of the fuel which hurt the economy negatively. This is followed by studying the relationship between Road Maintenance Levy and fuel pricing, Excise fuel taxation and fuel pricing and Railway Development Levy and fuel price in Nairobi and other towns within the country. The tax rates of these taxes are altered from time to time to raise the needed revenue to the state, however high tax rates can affect savings, wages and employment in the economy and hence affect the earnings of marketers and the oil dealers in the country. There is a conceptual frame showing the relationship between variables understudy. The study used descriptive survey design which attempted to quantity issues, conditions and collections from Kenya revenue authority (KRA) and energy regulatory commission analyzed and the results represented in line graphs and tables to aid understanding. The auto regression model is used to determine the relationship between variables followed by summary of findings, conclusions of the study and recommendations made and suggestion for further research. Key words: -Fuel taxes, Fuel prices, tax rates LIST OF TABLES Table 2.1 : Operationalization of the Variables ..................................................... 22 TABLE 4.1 Descriptive Statistics ........................................................................... 27 TABLE 4.2: Descriptive Statistics for log variables .............................................. 29 TABLE 4.3: Dickey-Fuller Test for unit root ......................................................... 30 TABLE 4.4: Lag Selection Criteria…………………………………………………………………………………..31 TABLE 4.5: Johansen co-integrating Test.............................................................. 33 TABLE 4.6: MODEL FITNESS............................................................................. 34 TABLE 4.7: Speed of Adjustment .......................................................................... 35 TABLE 4.8: VECTOR ERROR CORRECTION ESTIMATES ............................ 36 TABLE 4.9: IMPULSE RESPONSE FUNCTION TABLES ................................ 38 LIST OF FIGURES FIGURE 2.I: Conceptual frame work ..................................................................... 22 FIGURE 4.1: Trend plot for variables .................................................................... 28 FIGURE 4.2: Trend plot for Log variables ............................................................. 29 FIGURE 4.3: Trend plot for co-integration ............................................................ 32 DEFINITION OF TERMS Fuel Price : Refers to a change in the price of petrol, diesel and other fuels caused by a variety of factors such as market forces (Hindu, 2017). Fuel tax : Refers to a fuel tax (also known as petrol, gasoline or gas tax or as fuel duty) is an excise tax imposed on the sale of fuel (Wikipedia, 2010). Tax rate : Refers to the percentage at which an individual or a corporation is taxed (Investopedia, 2012) Tax : Refers to a mandatory financial charge or some others Levies charged upon taxpayer by a governmental organization in order to fund public expenditure (Wikimedia, 2018). Taxi shifting : Refers to transferring some or all of a tax burden from an entity to another (Business Dictionary, 2014) CHAPTER ONE INTRODUCTION 1.1 Background of the Study Fuel price is a driving factor of fuel consumption in a developing economy. Fuel taxes are responsible for generating a large majority of funds used to build and maintain our transportation infrastructure. Revenue generated from these taxes are increasingly lagged behind expenses and have recently relied on supplemental funding to meet the demands. Tax rates are largely set through state legislation while expenses are driven by market forces such as infrastructure costs, inflation and demography shifts. Fuel excise tax is charged on a gallon of fuel or directly embedded into the price of gasoline fuel. The tax rates in Kenya are given in the third schedule of income tax Act (Cap 470). These rates are revised from time to time (Saleemi, 2000). Over the past few years demand for fuel has been increasing with the need for more cost-efficient means of transport and discovery of new ways of income generating. Because fuel costs had been rising, companies have employed cheaper means of transport. Though existing public means users are usually loyal to their brand, measures to incorporate new marketing strategies such as service promotion, advertisement, price-cut, design and improvement of customers comfort are necessary for the business to survive the challenges posed by environmental, political and economical environments. The right price strategy is crucial for maximum total revenue. Generally, higher prices mean lower volume and small business can often command higher because of their personalized services. In most countries, motor fuel taxes are responsible for generating a large majority of funds used to build and maintain the surface of transportation structures in America. Tax rates are largely set through federal and state legislation while expenses are driven by market forces of infrastructure costs, inflation and demographic shifts. In a recent market trends, including increased vehicle efficiency and declining in per capita miles travelled, are reducing fuel used and further constraining fuel tax revenues (Polk, 2015). It is becoming increasingly clear that the current system of taxing motor fuels does not properly account for the array of technology options that are available to consumers. A study conducted by Workman and Rall (2012) in U.S.A found that seven states charged excise tax on gasoline and diesel fuel to collect revenue for improving and developing the transporting structure. Oil use account for 43.1% of fuel shares of total final consumption (Iea, 2006). Most of oil is used in the transport sector. The total consumption of petroleum products is currently about 73 billion liters and is projected to increase by more than three times in line with vision 2030 which is the blue print document for economic and investment policy. Gasoline tax in Great Britain of 50 pence/litre in 2000(about US$ 2.80/US gallon) is the highest amongst industrial countries taxation level in United States of about 40 cents /gallon is the lowest (Parry &small, 2005). Most common fuel taxes in Kenya are excise duty and road maintenance levy which are used to KRB (Mutua et al, 2008). In most African countries as in Europe, road transport is treated in a distinctive way within the fiscal system, through charging of special excise taxes on fuels. In particular, fuels are taxed considerably more heavily than other goods and services. In particular, high rates of tax are levied on fuels at low administrative cost with limited risks of evasion. Large scale commercial oil importing and refining activities are closely controlled and monitored by the revenue authorities. According to South African conference on excise taxation (2003) on efficient structure of road taxes that are charged on every road user for the precise social costs incurred is as a result of their decisions are reflected on various social costs of road transport. In some countries, the revenue generated from taxes on road transport provides funds used to build and maintain road infrastructure. A Substantial excise duty is generally levied on motor fuels either in form of specific or ad valorem duties. In most of the countries, Railway Development Levy is charged on fuels either where diesel fuel is taxed at lower rate than petrol. According to Hossain (2003) high petrol prices would tend to encourage manufacturers to design more fuel-efficient vehicles and fuel price changes caused by changes in taxation is likely to be quite low. In Kenya, Petroleum account for 22% of the total primary energy supply, 67% of which is consumed in the transport sector while the rest is consumed mainly in industrial processing and power generations (Aligua, 2006). In recent years (2003-2007), Kenya has recorded a substantially increased demand for fuel, both diesel and gasoline. Gasoline and diesel consumption currently stands at about 890 million and 1.4 billion litres respectively (UNEP, 2006). A study by west (2004) showed that a rise in price due to fuel taxes charged had reduced consumption in the economy. In Kenya today, there is an argument that the current fuel taxes are too high. This has been intensified by oil price volatility which has led to high price of gasoline of up to Ksh.110 in November 2008 following high prices of crude petroleum in the international market which had increased to US$ 147 per barrel. The price had since dropped to about Kshs 75 in March 2009 (KRB). Tax burden is too high for marketers and once collected, it is not used well to finance road development and rehabilitation works. The government has introduced Road Maintenance Levy, Excise Tax, gasoline tax and Railway Development Levy on fuel to raise revenue which has resulted to high cost of operating business (Bitenge, 2001). The petroleum companies are required to pay thus amount to the government to finance construction and improvement of roads. The rate of tax was increased from Ksh2 per litre to Ksh2.70 per litre of petrol in 1996 and increased by Ksh0.50 per litre in 1997(Hook, 2001). Because road maintenance tax rate varies from time to time. These taxes are not spent on projects of the people. KRA is a body responsible for collecting direct and indirect taxes in Kenya. It has made tax evasion illegal with fines to defaulters (Saleemi, 2011). Edmonds (1998) found that poor access not only affects trading costs but it is closely linked to diverse aspects of poverty and social exclusion. Fuel costs are accounting to 10%-40% of overall motor vehicle operating costs and a price increase has been the common cause of public protest. A number of fees and taxes have been put in place, which have primarily been levied directly against vehicle owners and drivers. Workman and roll (2012) found that motor fuel taxes cannot capture for these vehicles but with a few exceptions, these taxes and fees are charged in addition to excise taxes. Car taxation is an instrument that is influencing the purchase decisions of consumers and taxes are differentiated to support the market introduction of fuel efficiency (Sterner, 2007). Road side inspections are found to be the major cause of delays and charges and to have done little to improve the dangerous, un –road worthy conditions of many vehicles. A study by UNEP (2006) found that a rapid growth of motor vehicles in Kenyan cities have emerged as a key contributor to regional air pollution. The study found that over 700,000 vehicles had already been licensed to apply to Kenyan roads and the numbers are increasing by 30,000 units per year. Transport services are mostly provided by small operators in the private sector across Africa. Services tend to be least satisfactory in area with low populations, undeveloped markets and poor infrastructure. Hence the motor vehicle transport tends to be expensive, crowded and unsafe with little diversity of modes, little competition Sand no critical mass to make it easy to buy and maintain vehicles (Starkey et al, 2001). Although there was massive expansion of its inter-urban road systems in 1970S and 1980S oil boom, inadequate construction quality and failure to maintain roads led to severe deterioration of road transport system. Recent studies have found that there are rural access problems and linkages between poor access and poverty. Aligna (2006) emphasized on effects of fuel taxes on industrial processing and power generation. Manufacturing and processing industries in rural and urban areas had been declining nationally for more than a decade due to high cost of fuel. The specific problems of remoteness and off-road rural and urban areas access have been emphasized in a number of studies. There is surprisingly little research on the effect of fuel taxes on fuel pricing, the gap that the study is intended to Address. 1.1.1 Fuel Taxation Raising fuel taxes could significantly reduce emissions of greenhouse gases and other pollution from the transportation sector. One of the prime agreements against raising fuel taxes is the perception that they are costly to the poor and other socioeconomic groups. But the recent studies have suggested the opposite, particularly for developing countries. Fuel taxes rates vary widely, from a mere 19 cents per litre in the United States to 81.19 per litre, the average for western European countries. Elsewhere, such as in Japan and Australia, the rates fall in the intermediate range. In the western European countries fuel taxes have been converging over the last 20 years, whereas the spread between Western Europe and other countries has been increasing overtime. Differences in taxation is the main cause of fuel price variation between countries in addition to handling, quality related and refinery costs, product transportation and distribution costs which form a small share of the variation (International Energy Association, 2009). Oil-exporting developing countries have typically had very low domestic prices and often subsidize fuel consumption. One main reason why fuel tax is resisted by countries is that fuel taxes hurt the poor. However, studies have found neutral/weak regressive results in richer countries and quite strong progressive evidence in the developing countries such as China, India, Ethiopia, Indonesian, Ghana, Nairobi, Mali, and several more. The intuition is not surprising since in most developing countries, the very poorest households cannot afford to own a car at all. In many countries, including notably, industrial countries, prices are market determined and subject to only taxes and special levies. However, in other countries, notably developing countries the prices are fixed by the government or state-owned enterprises. According to Hossain (2003) gasoline and diesel prices in Africa varied from as low as 20 cents in Ghana, 27 cents in Nigeria and to about 86 cents in Uganda in the year 2000. In general, the prices of gasoline is significantly higher than that of diesel fuel reflected mainly wide variation in the levels of the taxes and subsidize on petroleum products imposed for various reasons. In Kenya fuel taxes are mixed with market price of fuel to determine the cost of fuel to consumers. VAT is charged to every litre of fuel sold to customers, while excise duty is charged on fuel produced as specific or ad valorem taxes. Railway Development Levy are added to every litre of fuel to be sold and the amount collected are remitted to the Kenya Revenue Authority (KRA, 2018). When fuel prices increased in 2000, there were concerns about the impact on the transport sector. Therefore, it was considered worthwhile to simulate the effects of a significant and durable change in oil prices on transport cost, transport demand, and transport externalities. Upon review of the changes in fuel prices and of its components over the last decade, we observe there is need to undertake a study of the effect of fuel taxation on the price of fuel in Kenya. While fuel prices are mainly driven by the import prices, changes in the foreign market prices as well as government price controls, we also realize the need for a research on the effect of taxation. 1.2 Statement of the Problem Taxes are imposed to implement the protection policy of the government and to maintain economic stability in the country. Imposition of taxes on fuel tends to raise the prices of fuel leading to higher prices which hurt the economy. In United State of America, the president had supported a new tax of $0.25 on petrol to help fund infrastructure plans. Though no one is sure of this concrete proposed or just an off the cut of remark, the mere thought of new fuel taxes makes drivers, industry folks and inventors stop to think because the impact is high to both consumers and businesses (Wald, 2018). However, petrol prices can quickly rise (or fall) making it hard for consumers to afford the added burden of gasoline tax. An addition of $0.25 tax would hurt even more. However, as the price of gasoline rises, eventually, consumers begins to limit their driving, more commuters will carpool. Fewer people drive large cars or use boats on the weekend. More drivers switch to electric vehicles and hybrids. Americana stop taking leisurely drives, and families vacate closer to home and further tax decrease gasoline consumption For automobile travel a 10% increase in petrol and diesel prices impacts leisure and holiday travel most. According to World Bank (2014) the tax rate goes up as the price of crude fuel falls, denying consumers the full benefits of dropping global prices. In Ghana about 70 per cent of the fuel price goes to taxes which hurt the consumers and reduce dealer’s profit margin. In comparison to other major sectors of the economy the profit made by marketers is low. High fuel taxes are common cause of high transport operating costs. An imposition of fuel taxes tends to raise the prices of fuel, thereby leading to higher cost to higher wages and prices which has negatively effect on economy. Kenyan government’s effort to reduce taxes on fuel for power generation and transport has not been successful (Mutua; Sterner & Borjesson, 2008). Because of this tax burden is too high for consumers. A tax increase hurts the economy as high prices charged on fuel have reduced consumption and investment. In Kenya VAT was imposed in 1989 to replace sales tax. In 2015 ERC was urged by government to lower oil prices by cutting marketer’s margin indicating that the government was not willing to reduce taxes which accounts for the largest share of the price. This means that imposition of 16 percent VAT has placed heavy tax burden on fuel prices (Ministry of industry and energy 2015). In Kenya taxes as a share of fuel prices are highest for petrol and lowest for kerosene. This means that on average, 33 per cent of Kenyan’s fuel price goes up than that of Ethiopia (33 per cent) Tanzania 40 per cent and (41 per cent) (Energy Regulatory Commission, 2015). Heavy fuel taxes charged raise the price to marketers, lower demand, and bring a fall to the business. Studies touching on the effects of fees and sales taxes on motor vehicles in California, Florida, Georgia, New York and West Virginia have been done by Workman and Roll (2012). More studies have been done by Starkey (2001) on the effects of fuel taxes on motor vehicles operating costs. Owing to the fact that, there is little literature on the effect of fuel taxation on fuel pricing in Kenya available, the researcher has decided to conduct a study in the area to determine the effect of fuel taxes on fuel pricing in Kenya to fill the gap. 1.3 General Objectives The general objective of this study is to determine the effect of fuel taxation on fuel prices in Kenya. 1.4 Specific Objectives The following are the specific objectives; i) To determine the effect of Road Maintenance Levy on fuel prices in Kenya. ii) To determine the effect of Railway Development Levy on fuel prices in Kenya. iii) To determine the effect of Excise Tax on fuel prices in Kenya. 1.5 Research Questions i) What is the effect of Road Maintenance Levy on fuel prices in Kenya? ii) What is the effect of Railway Development Levy on fuel prices in Kenya? iii) What is the effect of Excise Tax on fuel prices in Kenya? 1.6 Delimitation of the study The study was conducted in Kenya to find out the effect of fuel taxes on fuel pricing. A monthly data of petrol prices and fuel tax rates for 2013, 2014, 2015, 2016 2017 and 2018 are collected from Kenya Revenue Authority and Energy Regulatory Commission and used in the study. This is because these bodies are responsible for determining the amount of tax charged on fuel and setting the price for the fuel in Kenya. These data are used because they are most recent and can give the current information. 1.7 Limitation of the Study The study had used the secondary time series data obtained from Kenya Revenue Authority and Energy Regulatory Commission about fuel tax rates and fuel prices charged on fuel in Kenya. To get up to date and relevant data is a challenge. Adequate finance and the confidential information time series data would be difficult to get. 1.8 Basic Assumptions It is assumed that oil marketers and dealers had knowledge on tax matters, experience and the documents required for them to operate. Kenya Revenue Authority and Energy Regulatory Commission would provide reliable and honest information on petrol prices and fuel tax rates required by the researcher and the tax authorities are constantly monitoring fuel tax operation. 1.9 Significance of the Study The study is useful to scientists since it help them design a more fuel efficiency machines and vehicles and find an alternative energy that can be used instead of fuel. It assists the ministry of energy officers who make policies to govern the prices and flow of fuel within the country. It is useful to commuters and KRA as it eases transport industry and the government to increase revenue collection for the development. It is useful to researchers in taxation and other stakeholders in transport, communication and the energy sector. CHAPTER TWO LITERATURE REVIEW 2.1 Introduction This chapter presents the theoretical reviews, empirical evidence and the conceptual framework for the study. The information is obtained from other pieces of literature and past studies. This chapter is intended to cover various aspect of taxation theory, Kenyan taxation law and practical taxation problems and the gaps to be filled are identified. 2.2 Theoretical Review This section presents theories on taxation and pricing. We shall discuss three main theories among them concentration theory, demand and supply Theory and the modern theory of taxation. 2.2.1 The Concentration Theory Concentration theory advocated by Adams Smith and Recard (18th century) states that there is tendency for all taxes to be concentrated on objects or classes which enjoy a surplus. According to this theory, there are two surpluses in the economy namely rent and the profit. All tax incidents would get concentrated on the two and be absorbed by the two. It is a common belief that the price of gasoline is determined by the supply and demand of crude oil. Depending on the country, fuel taxes can add substantially to the retail price of fuel. Inflation and fuel taxes have contributed for the biggest increase in the price of fuel. Profits made by dealers are likely to decline by the amount of fuel tax added. The supplier of fuel can transfer the tax to the public through price revision depending on elasticity of demand and supply (Saleemi, 1994). Fuel taxes’ rates revision is an issue to consider if the supplier of fuel has to make profit out of the operation. A study of pure theory of taxation advocated by Ysidro (1958) not only attached to the incidence of taxes, but also in a minor degree, doubtless to the distribution of the fiscal burden among the tax payers. Adding fuel taxes such as road maintenance levy, railway development levy and excise tax to pre-tax price of fuel are likely to raise the pump price of fuel to consumers. It is understood that rising fuel price in Kenya have caused high burden to the poor unfairly. It is necessary for the government to exempt tax on fuel to enable the poor afford the commodity and reduce the political discourse. Fuel taxes can be absorbed by dealers or shifted to the customers through price revision depending on elasticity of supply and demand of the fuel. This theory is important because it enable the tax regulatory body know where to place fuel taxes to avoid hurting transport industry. Seligman and Edge Worth applied the theory on price in tax transfer under different market conditions. 2.2.2 Pricing Theory Price theory, advocated by Weber (2012) explains pricing and taxation concepts. According to this theory, tax incidents can be shifted only through sale/purchase transactions and only through price revision. Fuel price revision is determined by the relation value of supply and demand elasticity of fuel in a country. Charging taxes in fuel tends to raise the price of diesel and petrol and other petroleum products. A tax increase in the transport fuel hurt the public due to high progressiveness in nature (Mutua, Borjessons & Sterner, 2008). Rising fuel price due to taxes charged by Energy Regulation Commission can affect the demand for fuel because consumers are likely to turn to cheaper fuels such as charcoal, firewood and electricity. The tax charged is shared between the dealer and consumers in order to stabilize the price. According to Adam (2009), a fuel tax can be shifted to the public in form of increased fare. It is understood that the interaction between the government and the sellers of fuel determine acceptable price by all stakeholders. More tax charged on fuel tend to increase the price of fuel and less tax tend to lower the price. Demand, supply and price theory advocated by Bouchard (2013), state that a competitive market is a market with many buyers and sellers and no single trader can influence the price. In transport industry charging fuel tax on petroleum, diesel and gas increase the cost of fuel which can lead to an increase of fare charged by motor vehicle operators. Because most of the consumers cannot afford the fare they can turn to cheaper means of transport such as bicycles, carts, and motorbikes. This is different when Tax Regulation Commission, Kenya Revenue Authority and other government bodies dealing with tax regulation remove or reduce the amount of tax on fuel, the price of fuel decline. 2.2.3 The Modern Theory of Taxation Modern theory of taxation, advocated by Dalton (1936), shows the relationship between the burden of taxation and the price charged by fuel dealers. According to Dalton since the elasticity of supply and demand of the taxed good has effect on tax burden of the seller and the buyer, there is need to transfer tax to any of them or borne by both. Fuel taxes charged on petrol, diesel and gases are transferred to public, absorbed by the seller and consumers of the service through price variation. This is done by rising and lowering fuel taxes such as value added tax, excise tax and gasoline tax which are likely to affect the price of fuel. This theory is important because it help fuel dealers to set fair prices to customers. In addition, it assists the Energy Regulatory Commission to price the fuel. Diffusion theory advocated by Thuitara (2014), Vidolo (2016) and Kiplagat (2016) state that it become impossible to trace the final incidence of any tax and that in reality all taxes get diffused in the economic system. This theory state that because of the constant interaction by sales/ purchases transactions, a fuel tax imposed at one place can shift to all sectors of the economy through price revision. Tax shifting theory, advocated by Musgrave (1985) states that incidence of tax considers the distributional consequence of budgetary policy change. According to Musgrave incidence is the resulting changes in the distribution of income available for private use which arises as a result of changes in budget policy. This is a change in policy of taxation and public expenditure. Resource transfer can occur without taxes and taxes can occur without resource transfer. According to Musgrave when a tax is imposed on fuel, the resources are transferred to public sector from private use. This theory states that the government may impose taxes that are most easily accessed and collected causing least obstruction to national wealth. Imposed fuel tax can be shifted and re-shifted by increasing and reducing the fuel price in manner that no one can evade it. The process of exchange shifts the fuel tax burden extensively. This can lead to equilibrium when fuel tax burden is distributed equally among all taxpayers through price revision. Dalton (1936) argued that such approach only tries to run away from the basic problem of ascertaining the incidence of fuel tax. This theory has weakness as it is based upon the assumption of perfect competition, which is a myth. Its incidence is shallow and misleading because it avoids the real difficulties in tracing out the incidence of fuel taxes. 2.3. Empirical Review Several empirical studies have investigated the relationship between fuel taxation and fuel prices. These are as discusses below:- 2.3.2 Road maintenance levy and Fuel Prices Energy is a key to economic growth and development as well as improvement of quality of life(mutual,2008). Rural areas have traditionally based their economies on farming, fishing, forestry, and, or mining industries that have been declining nationally for more than three decades. According to sterner (2008) most common fuel taxes which are related to transport service include excise duty, VAT, Road maintenance levy and railway development levy. The amount of revenue collected is used to construct and maintain road infrastructure for economic growth and development. However, high tax rate on fuel has reduced income and savings which have affected investment and economic development. Bogonko (2001) stated that introduction and charging of road maintenance levy on motor vehicles owners by increasing petrol prices and provide revenue to the government that it uses to provide essential services to the citizen. Petrol in Kenya can be as low as Sh55 per litre if the fall in crude oil prices were reflected in refined products such as petrol and diesel (Otieno, 2016). Energy cabinet secretary has urged the ERC to lower oil prices by cutting marketers margins indicating that he was unwilling to reduce the taxes which account for the largest share of the price. In Kenya, taxes as a share of fuel prices are the highest for petrol and lowest for kerosene. Of the super petrol price of Kshs 88, Kshs 41 (47 percent) goes to international buying and freight costs while Sh32 (37 percent) goes to taxes and levies. Eight percent (Kshs 7) goes to wholesale marketers, while retailers at the pump get five percent (Kshs 4). For the diesel price Kshs 79,52 percent (Kshs 41) goes to international buying and freight costs, 29 percent (Kshs 23) goes to taxes and levies, 14 per cent (Kshs 11) to wholesale and retail margins that on average, 33 percent of Kenya’s fuel (Petrol and diesel goes to taxes) (Kenya Revenue Authority,2015). Bogonko (2001) found that, the rate charged as road maintenance tax was increased from Kshs 2 per litre to Kshs. 2.70 per litre of petrol in 1996 and was further increased by kshs.0.50 per litre in 1997. With this high level of fuel consumption, there is need to discourage consumption of fuel. A study by Mutua et al (2008) found that by increasing Road maintenance levy on petroleum products to raise revenue for use in water harvesting and other projects has resulted to high cost of petrol. While most of the money used to build and construct roads comes from fuel taxes, most of it is not used for the purpose. Spending on motor fuels in most African counties is subject to substantially, higher taxation than spending on most other goods and services and taxes on motor fuels often make a substantial contribution to total tax revenues. According to Gupta and Mahler (1994) total revenues contributed by taxes on petroleum products (mainly motor fuel) in developing countries can be as high as 30% with some African countries, including the ivory coast, Kenya, Tanzania and Uganda among those with particularly high revenue shares. More than 20% of the total tax revenue contributed by taxes on petroleum(gasoline) products in Uganda since 1999 was broadly equivalent to the total revenue yield from taxes on income and profits, Metschies (2001) argued that as a result of political and other pressures, it is often difficult for administered prices to be adjusted in line with high inflation and currency devaluations. In countries where large devolutions occur, there is tendency, for administered fuel prices to fall in relation to world prices implying a fiscal subsidy which may cancel out a large part of the impact of fuel taxes and duties. Metschies described the impact of the 50% devaluation of the CFA Franc in January 1994 and observes that in those CFA Franc zone countries that did not immediately adjust local fuel prices there were losses in public revenue from taxation. Countries identified as subsidizing diesel fuel are oil-producing countries and relatively few are in Africa. According to Cohen (2015) the decline in various taxes and fees including Road maintenance levy have contributed to lower priced fuel in the country and vice versa indicating positive relationship between Road maintenance levy and fuel prices. 2.3.3 Railway Development levy and Fuel Prices A steady increase of motor vehicles in the rapidly growing Kenyan cities make transports emerge as a key contributor to regional air pollution (UNEP, 2006). By the end of the year 2006, over 700,000 motor vehicles had already been licensed to operate on Kenyan roads and the number of those vehicles had been increasing by an average of 30,000 units annually. This implied that about 245 million litres of gasoline were consumed in 2004 releasing about 98.1tonnes of lead, 75% of which ends up in air, soils and plants and the rest remaining in engine block and lubrication oil. A study by Mutua et al (2008) found that there is an increasing excise duty on petroleum products to raise revenue for use in water harvesting and other projects. The study found that there is reduction of import duties, excise duty, railway development levy on diesel and residual fuels which enabled oil companies earn profits out of their operations. Although oil companies and commercial vehicles are struggling to make large profits, their dream is affected as more public vehicle agencies adopt more entrepreneurial focus without abandoning interest in non- monetary factors such as customer comfort or the needs of low-income people. In Kenya, fuel taxes such as fuel levy and excise duty are too high. The planned charge of 16 per cent VAT means that the tax burden will weigh even heavier on fuel. The government has moved in to reduce taxes on oil used in power generation but fuel taxes have not been affected (Mutua; Sterner & Borjesson, 2008). Ziramba (2008) argues that the transport modes used in the study is public as well as private transport, since these accounts for use of fuel and consequently are affected by fuel taxes, while non-motorized is here left out. A tax increase on transport fuel hurt the economy if the demand for fuel is elastic and the demand for supply of the same is inelastic. Therefore, there is need to revise the fuel taxes rates in order to give incentives to oil sellers as well as increase revenue potential to the treasury. A study by Nicholgon et al (2003) argued that the impacts of doubling the fuel taxes in Mozambique had affected living standards of poorer households. The study found that charging higher fuel tax would increase the number of people below the poverty line by about 28,500(about 0.15 percent of the total population). However, they present little quantitative detail on the distributional impact of fuel taxes on the poverty rate. He argued that it is a clear implication of other studies that fuel taxes, a part from taxes on domestic Kerosene, have substantially less impact on poorer households than most other potential revenue sources. In developing countries, the direct distributional impact of taxes on motor fuel varies, depending on the level and pattern of car and motorcycle ownership, and the use made of public transport. In most European countries, fuel taxes tend to be mildly progressive, with higher rate of car/motorcycle ownership and use among higher income households and poorer households making relatively more use of public transport. In much of United State, however, private consumption taxes to a single rate broad based railway development levy. Ramsey (2014) found that under certain circumstances, higher taxes on inelastic demanded goods could raise a given revenue requirement at lower economic cost than a uniform sales tax. For certain motor fuel uses, at least, this argument could justify above average taxation, although this may conflict with distributional objectives in tax and social policy. Fuel taxes in Germany are €0.4704 per litre of low sulphur diesel and €0.6545 per litre of unleaded petrol. A railway development levy of 19 percent on the fuel itself is added. This had raised the prices of these commodities up to €1.12 and €1.27 per litre. 21 percent of railway development levy is added over the entire fuel price making Dutch taxes on the highest in the World. In Russia value added tax account for 18 percent on fuel and taxes (Russia Ministry of industry and energy, 2006). The fuel taxes of various countries had been used to mitigate the rise of oil prices in 2003. Heavy taxation has left Kenyan motorists paying higher fuel prices compared with their Tanzania counter parts. Kenya’s pump prices are set to rise even further as the Treasury imposed 1.5 per cent on petroleum later this year (Kenya Revenue Authority, 2018). This shows that there is a positive relationship between railway development levy and the fuel price in Kenya. 2.3.4 Excise Fuel Taxes and Fuel Prices Elasticity of demand and supply determines the distribution of fuel taxes among the energy stakeholders. Saleemi (2001) stated that if the supply of the commodity is highly elastic, given the elasticity of demand, the money burden of fuel tax are more on the buyer of the fuel than the seller. Likewise, if the elasticity of commodity is less elastic, given elasticity of demand, the money is more on the seller than the buyer. This shows that the distribution of fuel taxes to the suppliers and buyers depends on supply and demand. A study conducted by Ziramba (2008) in South Africa shows that the distribution aspect of fuel taxes is borne by the rich customers (commuters) who absorb the tax. It is widely accepted knowledge that some taxes are progressive while others are regressive and the degree of each varies considerably. It is understood that improved income among households encourages them own cars, and these results to high demand, for fuel resulting to a high price changed on the commodity. According to Sterner, Mutua and Borjesson (2008) indicate that tax burden on petroleum products are consumed by users of private transport is mainly borne by the richest exhibiting progressivity in nature. A tax increase in fuel mainly impacts/ hurt the high-income group because of their progressivity in nature. Hanghton (1998) calculated along run revenue maximizing rates of motor and fuel excises in Madagascar as 104% for regular petrol compared to 39% in 1996 and 81% for diesel fuel which were taxed at the rate of 24% in 1996. High rates can yield even higher revenues, but the higher revenue would be eroded over time by behavioral responses of fuel users. In Kenya; Kshs 19,445 was charged on premium petrol, Kshs 19,055 charged on regular petrol and Kshs 10,005 is charged on diesel which had raised the price of fuel to a road user. A study conducted in Mauritius in (2003) showed that an excise tax and road fund duty were charged on oil fuels at tax rate of RS 9.80 Per litre (petrol) and Kshs 3.00 per litre (diesel). Fuel oil used for non- transport purposes is subjected to an excise tax of Rs 2.00 per litre. VAT is applicable on petroleum products at 15% up to retail stage. However, Osoro, Mpango and Mwingimvua (2001) estimated that the revenue maximizing rate of fuel taxes in Tanzania is in excess of 100% in the long run. According to them, only five of the 22 petrol tax rates reported exceeded 100% where three of the diesel excise rates reported exceeded 100% where three of the diesel excise rates are below 80%. A study conducted by Chem, Matovu and Reinikka (2001) in Uganda found, that charging fuel taxes had increased the cost of transportation which is passed forward to consumers. However, the argument is that the burden of fuel taxes on transportation would fall on poor most consumers than better of urban consumers. The sale of fuels in the Netherlands is levied with an excise tax. In 2015, an excise tax on petrol had reached to Euro 766 per litre and a diesel excise tax had reached Euro 482 per litre, while LPC, excise tax is Euro 185 per litre. In 2007 fuel tax was Euro 0.684 per litsre or $3.5 per gallon. This caused the raise in price of fuel (Netherland ministry of energy, 2015). Fuel prices are a mix of market price and taxes and fuel taxes are the simplest method for governments to influence prices. However, there is positive relationship between the price set by the fuel producer and the excise tax imposed. 2.4 Conceptual Frame Work A conceptual framework is the diagrammatic presentation of variables, showing the relationship between the independent variable and dependent variables. In this study, the independent variables were; Road maintenance levy, Excise tax and railway development levy. The relationship between the independent variables and dependent variable is presented schematically in the conceptual framework in Figure 2.1 Independent variables Dependent variable Road maintenance levy . Monthly Average rate per litre. Fuel price . Monthly Average price per litre Excise tax . Monthly average rate per litre. Road development levy . Monthly average rate per litre s Figure 2.1: Conceptual frame work Table 2.1 : Operationalizaton of the Variables Variable Indicator Measure Dependent variable Fuel prices Monthly averages Monthly average price per litre. Independent variable Road maintenance levy Monthly rates Monthly rate per litre. Excise tax Monthly rates Monthly rate per litre. Railway development levy Monthly rates Monthly rate per litre, CHAPTER THREE RESEARCH METHODOLOGY 3.1 Introduction In this chapter we discuss the research design employed, the data collection procedure which involve secondary data obtained from Kenya Revenue Authority and Energy Regulatory Commission. Diagnostic tests such as unit root test, co-integration test, lag order selection and post estimation tests are discussed. This is followed by a data analysis model for analyzing time series data. 3.2 Research Design The research design used in this study is the descriptive research design. According to Kothari (2004) descriptive research design describe the characteristics of a particular phenomenon seeking to describe the relationships and frequencies of occurrence. The variables in this study are described as they are justifying appropriateness of this research design. Descriptive research shows that the variables with greater dispersion indicate disparities within them and provide important clues regarding the issues that the researcher should focus on (Mugenda, 2011). This research design is appropriate to this study as it enables higher level analysis such as correlation and regression analysis that allow for establishing the nature, strength and extent of the association between variables. This research design is suitable for collecting quantitative data useful in the study to establish the effects of fuel taxes on fuel prices in Kenya. 3.3 Data Collection Procedure The researcher had collected secondary data from Kenya Revenue Authority (KRA) and Energy Regulatory Commission (ERC). A monthly data for 2015, 2016, 2017 and 2018 were collected and used to perform the analysis. The data for these years is preferred because they contain the most current information about fuel taxes and prices. 3.4 Diagnostic Tests Diagnostic tests Provides researchers with tools for assessing the quality and reliability of regression estimates. They aid in the systematic location of data points that are either unusual or inordinately influenced, measure the presence of collinear relations and pinpoint the estimated coefficients that are potentially most adversely affected (Belsley, Kun & Welson, 2005). The following diagnostic tests were done. 3.4.1 Unit Root Test: In research, a unit root tests are used to determine whether a time series variable is not stationary and possesses a unit root. The root hypothesis is defined as the presence of a unit root and the alternative hypothesis can either be stationary or trend stationary or explosive root depending on the test used. The test will use Dickey Fuller test can be used to test the presence of unit root in a series and how it can be viewed as a test. This test the null hypothesis that the demand follows a unit root process. The null hypothesis is rejected when the P-value is less than or equal to a specified significance level as 5% (0.05), 1% (0.01) or even 10% (0.10). In d fuller demand above, the P- value is 0.2924 is greater than 0.05, 0.01 or 0.10 the null hypothesis in there three cases is not rejected and the variable is non-stationary. On the other hand, if a test statistic is smaller than a critical value e.g. 5%, 1% or 10%, the null hypothesis is rejected and the variable is claimed to be stationary (Dickey Fuller,2013). 3.4.2 Co-integration Test In research co-integration tests are used to test the hypothesis that there is a statistically significant connection between the futures price and the spot price for combination of the two series. Two series with trends can be co-integrated only if there is a genuine relationship between the two. Co-integration is a statistical collection of X1, X2,…………,XK of time series variables and all of the series must be integrated of order 1. To test hypothesis concerning the relationship between non- stationary variables, is to run ordinary least squares (OLS) regressions on data which had been differenced. If all series in the collection are integrated of order I (contains a unit root and attains STATIONARITY on the first differencing and a linear combination of this collection, that is K=ax+by+cz) is integrated of order zero (stationary) then the collection is said to be co-integrated. A Vector auto regression is carried out when there is no co-integrating relations and vector error correction models are used for time series with at least one co-integrating equation. 3.5 Optimal Lag Order Selection Optimal lag order selection is carried out to gauge whether sufficient lags are included in autoregressive models. By introducing too many lags wastes degrees of freedom, while too few lags leave the equations potentially mis-specified and are likely to cause autocorrelation in the residuals. Three main lag order selection criteria are computed in STATA, after which the optimal lag is highlighted. 3.6 Data Analysis The study used the STATA to help in analysis. An Auto regression model is used to determine the regressions of time series values on previous values from that same series. The order of an auto regression is the number of immediately preceding values in the series that are used to predict the value at the present time. The model is shown below. Yt = ao + ß1 x1t-1 + ß2 x2t-1 + ß3x3t- 1+ ß4 yt-1 X1t = a1 + k1yt-1 + k2x1t-1 + k3x2t-1 + k4 x3t-1 X2t = a2 + m1 yt-1 + m2 X1t-1 + m3 x2t-1 + m4x3t-1 X3t = a3 + Z1 yt-1 + Z2 x1t-1 + Z3xt-1+Z4x3t-1 y = Dependent variable- fuel prices, x1 = Road maintenance levy rate x2 = Railway development levy rate x3 = excise fuel tax rate Data analyzed will be presented using tables and graphs. CHAPTER FOUR DATA ANALYSIS, INTERPRETATION AND DISCUSSION 4.1Introduction This chapter presents the results of the detailed data analysis work. The characteristics of data under consideration were first input into excel and then STATA for analysis. The results were presented using graphs, tables and descriptive statistics. Data was tested for stationary and co-integration and the VEC model was fitted. The data used for analysis is as per Appendix IV, data collected. 4.2 Preliminary Analysis Prior to examining the effect of fuel taxation on fuel prices in Kenya, preliminary data analysis was carried out through the use of tables and graphs. This was primarily to examine the basic characteristics of the data representing the different variables. 4.2.1 Descriptive Analysis Table 4.1 below presents results of descriptive statistics of the data in this study. TABLE 4.1 Descriptive Statistics Petrol Rdl Rml Exc Mean 99.924 .5468333 10.88787 15.99906 Minimum 80.71 .36 5.791 7.203 Maximum 127.8 .83 18.451 19.8954 Std .Dev. 9.271007 .1215088 4.890766 5.266541n Observations 60 60 60 60 The observation for all variables is 6o months from October 2013 to September 2018. Table 4.1contain the statistics used to describe the characteristics of all variables. It shows that the petrol price Nairobi has a mean of Kshs.99.924, RailwayDevelopmentLevyKshs0.5468333, Road Maintenance LevyKshs10.88787 and Excise Tax Kshs 15.99906. As indicated in the table, the highest petrol price charged was Kshs 127.8o and the lowest Kshs 80.71 with an error (std. Dev) of Kshs 9.271007. In addition to this, the highest tax rates are 83 cents and the lowest 36cents for Railway Development Levy, Kshs 18.451 and Kshs 5.791 for Road Maintenance Levy and Kshs 19.8954 and Kshs 7.203 for Excise Tax respectively. It is shown in the table that the Standard Deviation (error) is 0.1215083 for Railway Development Levy, 4.890766 for Road Maintenance Levy and for Excise Tax 5.266541 FIGURE4. 1: Trend plot for variables 0501001500204060MonthPetrol RDL RMLEXC This study considered 60 time series observations for all the variables that is, 60 months from October 2013 up to and including September 2018. The summary of their basic characteristics are as presented in Table 4.1 above. The basic characteristics show that the petrol price is an outlier; it is way above the other variables. The pictorial presentation in Figure 2 shows the combined trend plots for the variables over the period October 2013 to September 2018. The primary purpose of creating these plots was to visualize the trend of each of the variables in the study. We therefore transformed the variables to their logs to give a better fit and distribution; then tested for their basic characteristics as per below table. TABLE 4.2: Descriptive Statistics for log variables Log_petrol Log_rdl Log_rml Log_exc Mean 4.600198 -.6268214 2.286947 2.703183 Minimum 4.390862 -1.021651 1.756305 2.990489 Maximum 4.850467 -.1863296 2.915119 1.974498 Std.Dev. .0925364 .2154534 .4546069 .4002608 Observations 60 60 60 60 The study has 60 months observations for all variables from October 2013 to September 2018. Table 4.2 shows the statistics for all log variables. It shows that the log_ petrol price in Nairobi has the mean of Kshs 4.6oo198, Railway Development Levy Kshs-.6268214, log_ Road Maintenance Levy Kshs 2.286947 and log_ Excise Tax of Kshs 2.703183. The table shows that the highest log_ petrol price is Kshs4.850467 and the lowest is Kshs 4.390862 with an error of 0.0925364. In addition to this, log_ Railway Development Levy is Kshs -. 1863296 as the highest and Kshs-1.021651 as the lowest, log_ Road Maintenance Levy highest Kshs 2.915119 and the lowest Kshs 1.756305 and log_ Excise Tax has Kshs 1.974498 as the lowest and Kshs 2.990489 as the highest. The table shows that log_ Railway Development Levy has standard deviation (error) of 0.2154534, log_ Road Maintenance Levy 0.4546069 and 0.4002608 for log_ Excise Tax respectively. FIGURE 4.2: Trend plot for Log variables 4.3 Pre-Analysis Test for Time series 4.3.1 Stationarity TABLE 4.3: Dickey-Fuller Test for unit root Variables Null Hypothesis: Variable is Non stationary Level First Difference Test statistic p-value for Z(t) Test statistic p-value for Z(t) Log_petrol -1.070 0.7268 -6.079 0.0000 Log_rdl -0.712 0.8436 -10.596 0.0000 Log_rml -0.434 0.9042 -7.854 0.0000 Log_exc -1.711 0.4257 -7.741 0.0000 Critical values Critical values 1% 5% 10% 1% 5% 10% -3.567 -2.923 -2.596 -3.569 -2.924 -2.597 Time series data is assumed to be non-stationary. We therefore tested if the variables are stationary. Majority of economic and financial data is assumed to be integrated of order one I (1), there is therefore need to confirm this before proceeding to fit the appropriate multivariate model. Our Null hypothesis (Ho) is non-stationary while our alternative hypothesis (HA) is stationary. Table 4.3 presents the results of the Dickey Fuller test; the critical values at the different levels of significance (1%, 5% & 10%) are also displayed at the bottom of the table. The correlograms and trend plots for both the Log variables and the differenced log variables are as per Appendix V, VI, VII and VIII. 4.4.1 Lag selection Table 4.4 below presents results of the number of lags to be included in the model as presented by the various information criteria techniques. TABLE 4.4: Lag Selection Criteria Selection-order criteria Sample: 12- 60 Number of obs = 49 Lag LL LR Df P FPE AIC HQIC SBIC 0 285.854 16 1.2e-10* -11.5042 -11.4456* -11.3498* 1 295.39 19.073 16 0.265 1.5e-10 -11.2404 -10.9475 -10.4682 2 299.447 8.1142 16 0.945 2.6e-10 -10.7529 -10.2256 -9.36304 3 312.118 25.341 16 0.064 3.0e-10 -10.617 -9.85535 -8.6094 4 324.855 25.475 16 0.062 3.7e-10 -10.4839 -9.48781 -7.8585 5 337.527 25.343 16 0.064 4.8e-10 -10.348 -9.1176 -7.10491 6 361.457 47.861 16 0.000 4.2e-10 -10.6717 -9.20693 -6.81087 7 377.382 31.849 16 0.010 5.6e-10 -10.6686 -8.96947 -6.19005 8 395.181 35.599 16 0.003 8.0e-10 -10.7421 -8.80856 -5.64577 9 424.11 57.857 16 0.000 9.4e-10 -11.2698 -9.10189 -5.55573 10 475.506 102.79* 16 0.000 7.0e-10 -12.7146* -10.3123 -6.38274 Endogenous: D.log_petrol D.log_rdl D.log_rml D.log_exc Exogenous: _cons LR: Likelihood Ratio FPE: Final Prediction Error AIC: Akaike Information Criterion SBIC: Schwarz Bayesian Information Criterion HQIC: Hannan-Quinn Information Criterion The results in Table 4.4 shows, that the FPE, HQIC and SBIC choose no lag for the model but since this is a time series, then the variables must lag. LR (Likelihood ratio) and Akaike Information Criterion suggest that ten lags be included in the model. The decision criterion usually is to choose and use the number of lags preferred by most criteria but AIC is also considered a more superior information criterion. Therefore ten lags will be used in the model. 4.4.2 Co-integration test Time series are assumed to be co-integrated if they co-move towards long run equilibrium. The determination of stationary of the series is there the first step before co-integration. The Johansen methodology was adopted to carry out co-integration tests and fit the appropriate model which in this case should be vector autoregressive (VAR) model if no co-integration is found and a vector error correction model (VECM) if the series has co-integration. In order to investigate and determine the existence of both short-run and long-run equilibrium relationships among the variables under consideration, the Johansen co-integration test was done as per Table 4.5 below in STATA and using 10 lags as determined above. The null hypothesis (Ho) is that there is no co-integration while the alternative hypothesis (HA) is that there is co-integration. From Table 4.5, the model has 4 lags and 3 co-integrating equations. FIGURE 4.3: Trend plot for co-integration -.20.2.4.60204060MonthD.Log_petrolD.Log_rdlD.Log_rmlD.Log_exc TABLE 4.5: Johansen co-integrating Test Johansen tests for cointegration Trend: constant Number of obs = 49 Sample: 12 - 60 Lags = 10 Maximum rank Parms LL eigenvalue Trace statistic 5% Critical value 0 148 424.79739 - 101.4182 47.21 1 155 450.08634 0.64378 50.8403 29.68 2 160 465.55661 0.46817 19.8998 15.41 3 163 473.9544 0.29020 3.1042* 3.76 4 164 475.50649 0.06139 Maximum rank Parms LL Eigenvalue SBIC HQIC AIC 0 148 424.79739 - -5.583783 -9.129943 -11.29785 1 155 450.08634 0.64378 -6.060011 -9.773894 -12.04434 2 160 465.55661 0.46817 -6.294326 -10.12801 -12.4717 3 163 473.9544 0.29020 -6.398818* -10.30439* -12.69202 4 164 475.50649 0.06139 -6.382744 -10.31227 -12.71455 Figure 4 above shows the trend plots for the integrating series. Graphically, the variables are seen to be moving together towards a long-run equilibrium. 4.5 VEC Model The results of the Johansen co-integration test revealed that the variables are co- integrated and therefore the vector error correction model is the appropriate model to be run for the variables. The Johansen test showed that the variables have 4 lags with three co-integrating equations. The VEC model has been sub-divided into various tables and each of the values generated described after each table. TABLE 4.6: MODEL FITNESS Sample: 4 - 60 No. of obs = 57 AIC = -9.83835 Log likelihood = 307.393 SHQIC = -9.462245 Det(Sigma_ml) = 2.43e-10SBIC = -8.870589 Equation Parms RMSE R-sq chi2 P>chi2 D2_Log_petrol 6 .052318 0.2773 19.56913 0.0033 D2_Log_rdl 6 .050033 0.6963 116.9439 0.0000 D2_Log_rml 6 .093286 0.3901 32.62422 0.0000 D2_Log_exc 6 .123507 0.2616 18.07142 0.0061 The Parms is the number of parameters which for this study is six while the RMSE is the root mean square error which represents the standard deviation. The R-squared explains the proportions; petrol prices this month, is explained by 27.73% of its own lags and the lags of road development levy, road maintenance levy and excise tax. Road development levy this month is explained by 69.63% of its own lags and the lags of petrol prices, road maintenance levy and excise tax. Road maintenance levy this month is explained by 39.01% of its own lags and the lags of petrol prices, road development levy and excise tax. Excise tax this month is explained by 26.16% of its own lags and the lags of petrol prices, road development levy and road maintenance levy. Variables Log_ Railway Development Levy and Log_ Road Maintenance Levy are significant at 1% since the P>chi2 = 0.0000. Variable Log_ petrol is significant at 1% since the P>chi2 = 0.0033 while variable Log_ Excise Tax is significant at 10% since the P>chi2 = 0.0061. TABLE 4.7: Speed of Adjustment _Cell 1 Coef. Std.Dev. Z p>|z| [95% Conf. Interval] D2_Log_petrol -.103363 .055729 -1.80 0.073 -.2162039 .0094778 D2_ Log_rdl .3643478 .0550589 6.62 0.000 .2564344 .4722612 D2_ Log_rml .3140566 .1026565 3.06 0.002 .1128534 .5152597 D2_ Log_exc .0854483 .1359132 0.63 0.530 -.1809366 .3518333 Table 4.7 explains the long-term relationship and the coefficient (Coef.) represents the speeds of adjustment. Log_ petrol is moving upwards at a speed of 10.33% towards the equilibrium, Log_ Railway Development Levy moves downwards at a speed on 36.43% towards the equilibrium, Log_ Road Maintenance Levy is moving downwards at a speed of 31.41% towards the long-term equilibrium while Log_ Excise Tax moving at a speed of 8.5% downwards towards the long-term equilibrium. The speeds of adjustment for Log_ Railway Development levy and Log_ Road Maintenance Levy are significant. The speeds of adjustment for Log_ petrol and Log_ Excise Tax are not significant as their p>|z| values are more than 5%. TABLE 4.8: VECTOR ERROR CORRECTION ESTIMATES (FOR SIGNIFICANT VALUES OF LAGS) Coef. Std.Err. z p>|z| [95% Conf. Interval] D2_Log_ Petrol Log_petrol -.4280297 .1308377 -3.27 0.001 -.6844668 -.1715927 Constant .0022463 .00693 0.32 0.746 -.0113364 .0158289 D2_ Log_rdl Log_rml .1790012 .0896143 2.00 0.046 .0033605 .3546419 Constant .0002617 .0066274 0.04 0.969 -.0127278 .0132512 D2_ Log_rml Log_rml -.4756186 .1670847 -2.85 0.004 -.8030985 -.1481387 Constant .0004035 .0123568 0.03 0.974 -.0238153 .0246223 D2_ Log_exc Log_exc -.5053162 .1695644 -2.98 0.003 -.8376564 -.172976 Constant .0001184 .0163599 0.01 0.994 -.0319464 .0321831 Table 4.8 shows the short-run relationships of the variables. None of the constant values are significant; they are all above 5%. When petrol prices is increased by one unit (1%) last month then petrol prices this month will decrease by 42.8%. If Railway Development Levy increased by one unit (1%) in the last month then Road maintenance levy in this month will increase by 17.9%. If Road Maintenance Levy increased by one unit (1%) in the last month then road maintenance levy in this month will decrease by 47.56%. If Excise Tax is increased by one unit (1%) in the last month then Excise Tax in this month will decrease by 50.53%. The t-statistic values are all significant at 5% (95% confidence level). 4.7 Impulse response functions Immediately there is a shock in petrol prices, the price will increase by 100%and by 46.86% in the first month. A shock in petrol prices has a permanent impact on the petrol prices itself, the road maintenance levy, road development levy but has a transitory effect on excise tax which oscillates around the zero mark.A shock in the road development levy has a permanent positive impact on itself. An impact on the Railway Development levy has a permanent negative effect on Road Maintenance Levy and Excise Tax while it has a transitory effect on the petrol prices. After one month of a shock in Railway Development Levy, the Road Maintenance Levy will decrease by 82%. At the instant when there is a shock in Road Maintenance levy, there is no impact on Road Maintenance Levy itself, on petrol prices and on Excise Tax but it has 100% impact on Railway Development levy. After the first month of a shock on Road Maintenance Levy, the Railway Development Levy increases by 54.84%, there is a permanent positive impact on Railway Development Levy. The shock on Road Maintenance Levy has a transitory effect on itself since it shifts between positive and negative values, fuel prices and on Excise Tax. The moment that there is a shock in Excise Tax, there is no effect on Excise Tax itself, petrol price and railway Development levy. After one month of a shock in Excise Tax cause an increase in Road Maintenance Levy by 47.13% which continues having a permanent positive impact, while there is a transitory effect on Excise Tax itself, petrol price and Railway Development Levy. TABLE 4.9: IMPULSE RESPONSE FUNCTION TABLES Petrol IRFs Step (1) oirf (2) oirf (3) oirf (4) oirf 0 1 0 0 0 1 .468607 .215999 .119275 .003005 2 .710244 .182182 .138609 .026986 3 .612163 .142049 .063169 -.007258 4 .654223 .161393 .115791 .017658 5 .632142 .168791 .095390 .00558 6 .645005 .155403 .100554 .009860 7 .637896 .164022 .098646 .008595 8 .641507 .160205 .100474 .009014 9 .639697 .161873 .098987 .008867 10 .640681 .160840 .099826 .008856 Road Development Levy IRFs Step (5) Oirf (6) Oirf (7) oirf (8) oirf 0 0 1 0 0 1 .317699 -.424304 -.820749 -.294528 2 .128284 .085241 -.480919 -.097068 3 .179435 .102823 -.449403 -.163133 4 .173683 .023093 -.537723 -.150885 5 .180004 .034631 -.503561 -.156803 6 .169608 .057092 -.501759 -.148572 7 .178138 .038775 -.508413 -.155579 8 .173399 .045650 -.506689 -.151592 9 .175597 .044338 -.505581 -.153389 10 .174486 .044748 -.506701 -.152561 Road Maintenance Levy IRFs Step (9) oirf (10) oirf (11) oirf (12) oirf 0 0 0 1 0 1 -.052438 .206868 .548402 .044737 2 .060693 -.170774 .597402 -.053668 3 -.042333 .105196 .637983 .039427 4 .022646 -.021327 .627460 -.020819 5 -.01264 .024007 .614159 .011676 6 .007236 .003530 .627407 -.005741 7 -.005281 .018450 .621343 .004602 8 .00289 .007372 .623067 -.001924 9 -.002215 .014193 .622462 .002083 10 .000805 .010560 .623040 -.000244 Excise Tax IRFs Step (13) oirf (14) oirf (15) oirf (16) oirf 0 0 0 0 1 1 .084253 -.136755 -.06527 .471321 2 -.016825 .035405 .035137 .765071 3 .059620 -.069692 -.026422 .605784 4 .013145 -.027434 -.007485 .687236 5 .039472 -.042617 -.007909 .647038 6 .024067 -.033771 -.010072 .667222 7 .033516 -.041056 -.009296 .656582 8 .027721 -.035757 -.009374 .662368 9 .031156 -.038949 -.009228 .659231 10 .029180 -.037204 -.00951 .660908 CHAPTER FIVE SUMMARY OF FINDINGS, CONCLUSIONS AND RECOMMENDATIONS 5.1 Introduction This chapter highlights major findings obtained through analysis of secondary data collected. From the findings and discussions, the relevant conclusions are drawn and appropriate recommendations made. The general objective of the study was to determine the effect of fuel taxes on fuel prices in Kenya. More so the specific objective of the study includes determination of effect of Railway Development Levy, Road Maintenance Levy and Excise Tax on fuel pricing in Kenya. The Data for this study was secondary data obtained from Kenya Revenue Authority and Energy Regulatory Commission which was analyzed using STATA software and the results were presented using tables, graphs and an auto regression time series. 5.2 Summary of the findings The study was conducted using 60 times series observation for all the variables from October 2013 to September1, 2018. It is shown in table 4.1 in chapter 4 that petrol is an outlier since it is placed above the other variables. In order to give a better fit and distribution the variables are transformed and then tested for their basic characteristics. The data in time series is assumed to be non-stationary and the variables are tested for stationary. Since the majority of Economic and finance data is assumed to be integrated of order 1, this need confirmation before multivariate model is fitted. The study used Dickey fuller test to test the presence of unit root in the series. This tests the critical values at different levels of significance (1%, 5%) and 10% as displayed at the bottom of table 4.3 of chapter 4. The Null hypothesis is not rejected when the p- value is greater than the critical values of 1% , 5% and 10%. The study shows that lags were preferred by most criteria but AIC is considered as more superior information criterion and therefore 10 lags would be used in the model. Time series are co-integrated if they co-move towards long run equilibrium. The Johansen methodology was adapted to carry out co-integration test and fit the model vector autoregressive (VAR) when there is no co-integration found and a vector error correction model (VECM) when the series has co-integration. The result of Johansen co-integration test indicate that the petrol price, Railway Development Levy, Road Maintenance Levy and Excise Tax variables are co-integrated and the model vector error correction (VECM) is suitable for the variables. The Johansen test has shown that 4 variables have 4 lags with 3 co-integrating equations. In model fitness Railway Development Levy is explained by 69.63% of its own lags and the lag of petrol price, lag of Road Maintenance Levy and lag of Excise Tax. On the other hand, Road Maintenance Levy this month is explained by 39.01% of its own lags and the lags of petrol price, Road Maintenance Levy and Excise Tax. In addition, Excise Tax this month is explained by 26.16% of its own lags and the lags of petrol prices, Road Maintenance Levy and Railway Development Levy. The study established the long term relationship one coefficient representing the speed of adjustment. The study found that log_ petrol is moving upward at speed of 10.33% towards the long run equilibrium, on the other hand, log_ Railway Development Levy is moving downwards at a speed of 36.43% and log_ Road Maintenance Levy is moving at a speed of 31.41% towards the long run equilibrium while Excise Tax moves at a speed of 8.5% towards the long run equilibrium. It is found that the speed of adjustment for log_ Railway Development Levy and log_ Road Maintenance Levy are significant while those of log_ petrol price and log_ Excise Tax are not significant because their P>H value are more than 5%. From table 4.8, is shown that none of the constant values are significant as they are above 5%. If petrol prices is increased by one unit (1%) last month then petrol prices this month would decrease by 42.8%, if Railway Development Levy is increased by one unit (1%) in the last month the Railway Development Levy this month would increase by 17.9% , if the Road Maintenance Levy is increased by one unit (1%) last month then Road Maintenance Levy this month would increase by 47.56% and if Excise Tax this month is decreased by one unit (1%) in the last month then Excise Tax this month would decrease by 50.53% .The study found that a shock in petrol prices would increase the price by 100% and by 46.865 in the first month. As shock in petrol prices has permanent impact on the petrol prices itself, the Road Maintenance Levy, Railway Development Levy, but has transitory effects on excise tax which oscillate around zero mark. The study established that an impact on Railway Development Levy has permanent negative effect on Road Maintenance Levy and Excise Tax while has transitory effects on the petrol prices. After one month of a shock in a Railway Development Levy, the Road Maintenance Levy would decrease by 82% and when there is a shock in a Road Maintenance Levy itself, on petrol prices and on Excise Tax but has 100% impact on Railway Development Levy. The study found that after the 1st month of shock on Road Maintenance Levy, the Road Maintenance Levy is increased by 54.84% and there is a permanent positive impact on Railway Development Levy. The shock on Road Maintenance Levy has a transitory effect on itself since it shifts between positive and negative values, fuel prices and on Excise Tax. After one month of shock in Excise Tax cause an increase in Road Maintenance Levy by 47.13% and continues having a permanent positive impact while there was a transitory effect on Excise Tax itself, petrol price and Railway Development Levy. 5.3 Conclusion First, the study established that the government is charging Road Maintenance Levy on petrol to collect revenue that it can use in construction and repair of roads in the country. The study found that, Road Maintenances Levy is charged on petrol at fixed price in one financial year but is subject to change from time to time. The study found that his levy is charged by government to collect about kshs.18.9 billion that can be used to construct and repair roads in the country. This has caused heavy burden on tax payers to maintain roads because high price due to levy is passed on to the public inform of raised fare. In addition to this, the study found that Road Maintenance Levy in the current month is explained by 39.10% of its own lag and the lags of petrol, Railway Development Levy, Excise Tax and when the Road Maintenance Levy is increased by a unit in the previous month, would decrease the levy by 47.56% and excise tax by 50.53% in the current month. Secondly, it was established in the study that Railway Development Levy is charged on petrol to help the government collect revenue that it can use in developing the country. Charging Railway Development Levy on petrol has made the cost of fuel high for the customers which has raised the cost of living and put more economic pressure on the public in the country. The study found that charging Railway Development Levy in the current month is explained by 69.63% of its own lag and lags of petrol prices, Road Maintenance Levy and Excise Tax. The study found that by increasing Railway Development Levy by a unit (1%) in the previous month would increase the levy by 17.9% in the current month. It was indicated that a shock in the Railway Development Levy has a permanent positive impact on itself. It was shown that after a month of a shock in Railway Development Levy, the Road Maintenance Levy would decrease by 82%. Thirdly, the study established that there was positive relationship between Excise Tax and the price of fuel in Kenya. It is found that when Excise Tax is charged on petrol, the price of petrol is raised by amount equivalent to tax. The study found that by charging Excise Tax on petrol can raise the retail pump price of petrol. More so the study has found that high Excise Tax rates on petrol has contributed to raise of retail price of petrol in Nairobi and other towns in the country. It was found that Excise Tax in the current month is explained by 26.16% of its own lag and the lags of petrol prices, Railway Developing Levy and Road Maintenance Levy. In addition, the study found that by increasing an Excise Tax by a unit (1%) would decrease the tax by 50.53% in the current month. More so it was found that a shock in Excise Tax after one month caused an increase in Road Maintenance Levy by 47.13%. When petrol price is increased by one unit (1%) in the previous month, the petrol price would decrease by 42.8%. The study shows that the variables log_ Railway Development Levy and Road Maintenance Levy are significant at 1% since p>chi2=0.0033 and the variables log_ Excise Tax is significance at 10% since p>chi2=0.0061. The coefficient shows the speeds of the adjustment and long term relationship between the variables. 5.4 Recommendations of the Study Fuel price should be the same all over the country irrespective of the distance covered in transportation. The fuel dealers should be encouraged to accept small profit margins. The government through Kenya revenue Authority should ensure that tax is imposed on other sources of energy such as charcoal, solar, gases and electricity to enable the state raise revenue without overtaxing petrol, diesel and kerosene. Fuel tax rates should be easily adjustable by increasing and decreasing the rate on monthly bases to enable the state raise enough revenue for the development of the country. Fuel taxpayers should be informed on how taxes are computed and paid and assured that fuel taxes charged will be spent on projects and services beneficial to all to avoid resistance. Fuel tax officials and all those responsible for collection of fuel taxes should be made accountable for losses of tax revenue in which the punishment should be severe including attachment of their properties to recover the revenue loss. The state should import a lot of crude oil when the price is low, store and release to the market when the market price is high in the exporting nations. This will help to stabilize price of petrol, diesel and kerosene in the country. Oil companies should be encouraged to use cheaper means of transport such as pipeline instead of lorry and tankers in order to lower the cost of fuel which can result to low price. Kenya Revenue Authority should develop supervisory and monitoring mechanism to ensure that all tax revenue collected is accounted for. Legislation should be passed in parliament that will make fuel tax avoiding illegal and such act should be punished by fines and imprisonment of the wrong doer. The government should ensure that vehicles and vehicle spare parts are taxed more than petrol, diesel and kerosene which are mainly used by low income earners and the poor. This ensures that, the rich are taxed more than the poor and the tax system is progressive in nature. 5.5 Suggestions for Further research This study concentrated on investigating the effect of fuel taxation on fuel price in Kenya. From the study’s findings and conclusion, further research can be done on fuel taxation and its effect on fuel sales and consumption in Kenya as this can affect dealers’ margin and the demand for the fuel. Since the study concentrated on effect of Road Maintenance Levy, Excise duty and Railway Development Levy on fuel price in Kenya, further research can be done on the effect of fuel Value Added Tax and, Petroleum Regulatory Levy on price, sales and consumption in Kenya and the world REFERENCES Addison, T.& Osei, R.(2001). Taxation and fiscal reform in Ghana, WIDDER Discussion paper NO 2001/97. Aligula, M. (2006). Police Options for Reducing the Impact on Energy Tariffs in Kenya, National Economic and Social Council, Final Draft. Barthold,A. (2004). Issued in the design of environmental taxes. The journal of Economic perspective vol. 1 (Winter.2004), pp 133-151 Bogonko,J. (2001).Taxation simplified: N.A.Saleemi publishers ,Nairobi Breit consulting. (2008).Analysis of impact of argumentation of commercial Motorcycle operation. Cameroon Chen, D., matovu, J.,&Reinika, R. (2001). A quest for revenue and Tax Incidence in Uganda, IMF WORKING PAPER wp/01/24. Cooper & Schindler. (2008). Business Research Methods 10th edition; McGraw hill International Edition. Singapore Goodwin, P.B.(1992). A review of new Demand Elasticity with special reference to short and long run effects of price changes.” Journal of Transport Economics and Policy (May), pp.155-169. Government of Kenya. (2004. Sessional Paper No. 4 on Energy. Ministry of Energy. Nairobi: Government Printer. Gupta, S.&Mahler. (1994). Taxation of petroleum products: Theory and Empirical Evidence, IMF Working paper 94/32. Washington DC: International Monetary Fund. Gupta, Sanjeev. (2002). Issues in Domestic petroleum pricing in oil producing countries, “IMF working paper 02/140 (Washington: International Monetary Fund). Hassain, Shahabuddin. M. (2003). Taxation and pricing of petroleum products in Developing Countries: A framework for analysis with Application to Nigeria. IMF working paper wp/es/42 (International Monetary Fund). Haughton, J.(1998). Calculating the revenue-maximizing excise tax. Harvard Institute for International Development, African Economic policy Discussion paper NO 12. Hossain, Shahabuddin, M.(1991). “Demand pattern of Energy Products in Bangladesh-Evidence from Household survey data” (unpublished); Cambridge: Cambridge University. _______. (1995). “The Equity impact of Value Added Tax, “IMF staff papers, Vol. 42 (June), PP. 411-30 Hughes, G.A. (1983). The impact of fuel prices in Thailand (Washington: World Bank). Iea. (2006), World Energy Outlook 2006. International Energy Agency/OECD, Paris. Metschies, G.P. (2001). Fuel prices and vehicle Taxation, Second Edition (October). Eschborn: Deutsche Gesellschaft fur Technische Zusammenarbeit (GTZ) GmbH. Mugenda, O.M.,& Mugenda, A.G. (2003). Research Methods Quantitative and Qualitative analysis approaches: Nairobi acts press. Mugenda,G.A. (2008). Social science research theory and practice Nairobi: Applied research and training services. Mutua, J., Borjesson, M. & Sterner, T.(2008). Transport choice, Elasticity and distributional effects of fuel taxes in Kenya, Paper presented at the 9th Global Conference on Environmental Taxation, Singapore, 6th-9th November, 2008. Nasar,Ahmed. & saleemi. (2011). Taxation Simplified (revised and updated) Nairobi: Saleemi publication limited. Nicholson, K.B., O’Laughlin, A., Francisco& Nhate,V .(2003). Fuel Tax in Mozambique. Ordho, A.J., & Kombo, D.K. (2002). Research methods. Nairobi: Kenyatta University Institute of learning. Osoro, N., P. Mpango & H. Mwinyimvua. (2001), “An analysis of excise Taxation in Tanzania.” EAGER African Economic policy Discussion paper No 72. Otieno. (2016). Kenyan’s fuel prices are lower than East African World average: Nairobi Kenya. Otiso, W.& Behneus, R. (2011). Bodaboda taxis and their role in urban Transport case studies at Nauru and Kisumu. Parry,I.W., & Small,K.A. (2005).Does Britain or the United States have the right gasoline taxes? The America economic review, Vol. 95 No. 4 (Sep. 2005) pp.1276-1289. Poverty and social Impact Analysis. The PRSP Monitoring and Synthesis project, OverseasDevelopmentInstitute,London.http://www.prspsynthesis.or/Mozambique final PSIsA.doc Ramsey, F.P.(1927).A contribution to the theory of taxation,” Economic Journal, 37(1), pp.47-61 Sterner, T. (2007). Fuel Taxes: An important instrument for climate policy. Energy policy 35 (2007) 3194-3202. Suits, B. (1977). Measurement of Tax Progressivity, the American economic review, Vol, 67. No. 4 (Sep. 1977), pp. 747-752. World Bank. (1988). Road transport Taxation in Developing Countries; The design of user charges and Taxes for Tunisia, “World Bank Discussion paper No, 26 (Washington: World Bank). APPENDICES Appendix I – Data Collected Month Petrol RDL RML EXC 1 109.52 0.36 5.791 7.203 2 109.07 0.37 5.792 7.204 3 107.92 0.38 5.798 7.208 4 106.3 0.39 5.799 7.209 5 107.92 0.38 5.8 7.214 6 101.58 0.39 5.801 7.255 7 99.59 0.4 5.802 7.256 8 98.73 0.41 5.803 7.268 9 96.17 0.43 5.804 7.275 10 115.62 0.42 5.805 10.306 11 116.62 0.43 5.805 10.307 12 111.64 0.44 5.806 10.308 13 110.89 0.45 5.807 10.309 14 106.8 0.46 5.809 10.310 15 102.01 0.46 5.811 10.311 16 92.88 0.47 5.815 10.312 17 84.71 0.48 5.816 10.312 18 89.46 0.49 5.834 10.312 19 89.35 0.51 5.848 10.314 20 92.89 0.52 5.853 10.314 21 97.28 0.5 5.855 10.316 22 98.59 0.53 8.978 19.50446 23 102.65 0.54 8.99 19.50455 24 102.65 0.54 9.000 19.50477 25 93.29 0.52 9.01 19.50479 26 90.46 0.53 9.064 19.5049 27 90.06 0.51 9.112 19.5051 28 88.64 0.5 9.155 19.50517 29 86.5 0.51 9.175 19.5052 30 85.58 0.48 9.321 19.50521 MONTH PETROL RDL RML EXC 31 80.71 0.52 9.400 19.5053 32 84.25 0.49 9.444 19.50533 33 86.17 0.53 9.454 19.5054 34 92.23 0.54 12.023 19.8478 35 95.13 0.56 12.119 19.887 36 91.56 0.52 12.199 19.888 37 94.94 0.5 12.231 19.888 38 94.2 0.56 12.234 19.8887 39 98.2 0.56 12.311 19.89 40 96.27 0.56 12.321 19.890 41 100.27 0.58 12.365 19.8901 42 101.05 0.52 12.369 19.891 43 98 0.54 12.455 19.892 44 99.59 0.58 12.456 19.8924 45 98.73 0.53 12.47 19.893 46 97.1 0.68 17.856 19.8934 47 96.08 0.69 17.987 19.894 48 98.3 0.66 18.123 19.8945 49 101.67 0.64 18.137 19.8946 50 102.7 0.65 18.147 19.8947 51 104.7 0.7 18.223 19.8949 52 106.3 0.72 18.257 19.8949 53 107.92 0.74 18.258 19.895 54 107.46 0.75 18.261 19.895 55 106.83 0.73 18.315 19.895 56 107.17 0.76 18.333 19.8951 57 108.81 0.76 18.366 19.8952 58 112.2 0.79 18.416 19.8952 59 113.73 0.82 18.432 19.8953 60 127.8 0.83 18.451 19.8954 Appendix II – Trend plots for variables and differenced variables Trend Plots for variable Trend Plots for differenced variables Log_petrol 4.44.54.64.74.84.9Log_petrol0204060Month d.Log_petrol -.10.1.2D.Log_petrol0204060Month Log_rdl -1-.8-.6-.4-.2Log_rdl0204060Month d.Log_rdl -.10.1.2.3D.Log_rdl0204060Month Log_rml 1.522.53Log_rml0204060Month d. Log_rml 0.1.2.3.4D.Log_rml0204060Month Log_exc 22.22.42.62.83Log_exc0204060Month d. Log_exc 0.2.4.6D.Log_exc0204060Month Appendix III – correlograms differenced variables d.log_petrol d.log_rdl d.log_rml d.log_exc -0.40-0.200.000.200.40Autocorrelations petrol0102030LagBartlett's formula for MA(q) 95% confidence bands -0.40-0.200.000.200.40Autocorrelations rdl0102030LagBartlett's formula for MA(q) 95% confidence bands -0.40-0.200.000.200.40Autocorrelations rml0102030LagBartlett's formula for MA(q) 95% confidence bands -0.40-0.200.000.200.40Autocorrelations exc0102030LagBartlett's formula for MA(q) 95% confidence bands Appendix IV – Impulse Response Graphs D_Log_petrol, D_Log_petrol .03.04.050510model31, D_Log_petrol, D_Log_petrolstepGraphs by irfname, impulse variable, and response variable (D_Log_petrol, D_Log_rdl .004.006.008.010510model31, D_Log_petrol, D_Log_rdlstepGraphs by irfname, impulse variable, and response variable D_Log_petrol, D_Log_rml .008.01.012.0140510model31, D_Log_petrol, D_Log_rmlstepGraphs by irfname, impulse variable, and response variable D_Log_petrol, D_Log_exc .02.025.03.035.040510model31, D_Log_petrol, D_Log_excstepGraphs by irfname, impulse variable, and response variable D_Log_rdl, D_Log_rdl D_Log_rdl, D_Log_rml -.020.02.04.060510model31, D_Log_rdl, D_Log_rdlstepGraphs by irfname, impulse variable, and response variable D_Log_rdl, D_Log_petrol D_Log_rdl, D_Log_exc 0.005.01.0150510model31, D_Log_rdl, D_Log_petrolstepGraphs by irfname, impulse variable, and response variable -.020.02.040510model31, D_Log_rdl, D_Log_rmlstepGraphs by irfname, impulse variable, and response variable -.02-.015-.01-.0050510model31, D_Log_rdl, D_Log_excstepGraphs by irfname, impulse variable, and response variable D_Log_rml, D_Log_rml .04.06.08.10510model31, D_Log_rml, D_Log_rmlstepGraphs by irfname, impulse variable, and response variable D_Log_rml, D_Log_petrol 0.001.002.003.0040510model31, D_Log_rml, D_Log_petrolstepGraphs by irfname, impulse variable, and response variable D_Log_rml, D_Log_exc .04.05.06.07.080510model31, D_Log_rml, D_Log_excstepGraphs by irfname, impulse variable, and response variable D_Log_exc, D_Log_rml -.010.010510model31, D_Log_rml, D_Log_rdlstepGraphs by irfname, impulse variable, and response variable D_Log_exc, D_Log_exc D_Log_exc, D_Log_petrol .04.05.06.07.080510model31, D_Log_exc, D_Log_excstepGraphs by irfname, impulse variable, and response variable -.0050.005.010510model31, D_Log_exc, D_Log_petrolstepGraphs by irfname, impulse variable, and response variable D_Log_exc, D_Log_rdl -.01-.0050.0050510model31, D_Log_exc, D_Log_rdlstepGraphs by irfname, impulse variable, and response variable D_Log_exc, D_Log_rml -.0050.0050510model31, D_Log_exc, D_Log_rmlstepGraphs by irfname, impulse variable, and response variable Appendix V: Letter of research authorization