Taxing the Aviation Sector

Transcription

1 Taxing the Aviation Sector A special notion on the 'natural oil tax' Final report Client: ECORYS Research Prograe ECORYS Nederland BV Leon van Berkel Rotterda, 31 st of August 2008

2

3 ECORYS Nederland BV P.O. Box AD Rotterda Wateranweg GG Rotterda The Netherlands T +31 (0) F +31 (0) E netherlands@ecorys.co W Registration no ECORYS Macro & Sector Policies T +31 (0) F +31 (0) DK/RdH CD17500 Pillar 5

4 DK/RdH CD17500 Pillar 5

5 Table of contents Preface 7 Abstract 9 1 Introduction 11 2 Literature review Environental taxes Ipact assessent ethods The critical paraeters of the odel Elasticity of deand The pass-through rate The oil price 27 3 Model The Old BAU scenario Tax scenarios The New BAU with the natural oil tax Tax scenarios on the New BAU 37 4 Analysis The ipact of the three tax scenarios on the Old BAU The ipact of the three tax scenarios on the New BAU The natural oil tax Sensitivity analysis Policy iplications 54 5 Conclusions Liitations 58 6 Appendices Appendix A: Exchange rate and inflation predictions Appendix B: E-views output of several regressions Appendix C: Assuptions of the saple routes 61 7 References 63 DK/RdH CD17500 Pillar 5

6 DK/RdH CD17500 Pillar 5

7 Preface This study about taxes in the aviation sector is the 5 th pillar of the ECORYS Research Project Ipact Assessent Methodologies and Industrial Copetitiveness. In stage one an extensive literature review about ipact assessent ethodologies had to be written. In the second stage one of these ethods had to be used within the, 5 th, environental pillar. After reading into several environental issues the topic was based on the ticket tax in the Netherlands, which has been introduced at the 1 st of July this year. The following question cae to ind: What ipact would a ticket tax have on the aviation sector in the whole EU?, which becae the topic of this study. Although other studies already have been done in this area, there was soe roo left for new insights. For this study it was rather difficult to gather data so several organizations were asked for data. The Association of European Airlines was willing to help and they provided soe useful data and tips, which I a very thankful for. Furtherore, I would like to thank the experts fro ECORYS who supported e by eans of coaching and exchanging ideas. Therefore, I would like to take this opportunity to thank: Mr. Stefan M. Bruehlann, (Manager Strategy & Statistics, AEA), Ms. Sue Lockey, (General Manager Market Research, AEA), Mr. Dario Spila, (Manager Research & Analysis, AEA), Dr. Floor Sakan, (Consultan ECORYS), and Mr. Robert Piers, (Senior Consultan ECORYS). Leon van Berkel Taxing the Aviation Sector 7

8

9 Abstract Environental taxes are becoing trendier and within a few years the aviation sector in the European Union is likely to be taxed extra as well. Such a tax will affect the aviation sector in nuerous ways. The price of tickets will go up leading to a fall in deand which will be accopanied by lower profits, less eployen and a fall in greenhouse gas eissions. This study will quantify these ipacts with a quantitative causal-chain analysis. The following three tax scenarios will be looked at: a ticket tax, a fuel tax, and a cobined tax, which will be iposed on the Old BAU, with low oil prices, and on the New BAU, with higher oil prices. In this ipact assessent there are three critical paraeters that are considered naely the oil price, price elasticity of tickets and the pass-through rate. Especially the higher oil price, a natural oil tax, is interesting as it can diinish the necessity of iposing a tax to reduce greenhouse gas eissions. This research concludes that taxes would daage the aviation sector excessively ore in ties of these high oil prices and that they are not needed to reach old CO 2 eission targets since the natural oil tax already yields the wanted policy results with respect to the CO 2 eissions without distorting the level playing field in the airline industry. Taxing the Aviation Sector 9

10

11 1 Introduction Environental probles have been oving up the European political agenda 1 in the last decennia. One of these environental probles is the issue of aviation and cliate change. The greenhouse gas eissions caused by the aviation sector are increasing rapidly and are responsible for a significant part of the eissions. According to the European Coission (2005) the share of the aviation of overall greenhouse gas eissions is only 3%. However, if aviation will grow according to the trend it follows now, this share will increase and this could offset the reductions in greenhouse gas eissions achieved in other sectors, which would ake it ore difficult for the EU to fulfill the requireents of the Kyoto Protocol and its follow up. This gave reason to any econoists and environentalists to coe up with solutions to reduce these negative environental effects. The proposed solutions range fro eission trading to a tax on tickets to a fuel tax. Especially the fuel tax raised extra attention as kerosene is not taxed ye while, for exaple, the taxes on gasoline for cars are over a 100% in the Netherlands. The effects of these policy instruents have been researched, however the assuptions ade or the ethodology used has not always been optial. This study will challenge other studies on three aspects. Based on ore recent studies a lower price elasticity will be assued. Moreover the pass-through rate will not be 100% as assued in ost other studies. Another and aybe the ost interesting aspect in this study will be the concerns about the high oil prices. Previous studies done in this area assued a constant oil price which is well below today s oil price. If the oil price reains high the necessity of a tax would diinish, because the goal of decreasing the greenhouse gas eissions would be already achieved. In that case the ticket and/or fuel tax would be replaced by the natural oil tax where the tax revenues would go to the oil copanies instead of the governent. This study will analyze the effects of a ticket tax, a fuel tax, and a cobined tax on the aviation sector as well as on the environent. The ain effects that will be looked at are the effects on the price and deand for tickets, the greenhouse gas eissions 2, profits of the airlines, eployen and tax revenues. This will be done based on six scenarios, consisting of three tax scenarios which will be iposed on as well the Old BAU, with low oil prices, as on the New BAU, with higher oil prices. Scenario A will be a ticket tax of for traffic within Europe and 45 for traffic fro Europe to other parts of the world. Scenario B will be a fuel tax of 50% on their fuel costs. Scenario C will be a cobined tax consisting of the ticket tax and the fuel tax as applied to the previous scenarios. The ipacts of each of these scenarios will be assessed with a quantitative 1 2 European Coission (2005), Reducing the Cliate Change Ipact of Aviation. For siplicity this paper will only take CO 2 eissions instead of all greenhouse gas eissions as CO 2 eissions are by far the ost iportant ones. Taxing the Aviation Sector 11

12 causal-chain analysis, which will be translated into a odel built in a spreadsheet progra where all relations between the variables are built on sound theoretical arguents. However it has to be noticed that giving a cost-benefit analysis is not in the scope of this study. Hence a value judgent about whether such a tax should be iposed or not will not be ade. These objectives can be translated into the following research question: What will be the effects of a tax on the aviation sector, and ore specifically on the profits of airlines, eployen CO 2 eissions, and tax revenues? This paper starts with a literature review where the concept of environental taxes (Pigovian tax) will be discussed. Furtherore, the choice for the causal-chain analysis as ipact assessent ethod will be discussed. Nex this paper will discuss the three aforeentioned critical paraeters of this study. In the following chapter, the odel will be explained followed by an analysis of the effects of these taxes. This will be done based on the three aforeentioned scenarios. And finally, an evaluation of all three scenarios will be ade followed by a policy advice as well as recoendations for further research. 12 Taxing the Aviation Sector

13 2 Literature review In this chapter the relevant literature about environental taxes, ipact assessent ethodologies, and the three aforeentioned critical paraeters will be discussed. 2.1 Environental taxes Environental taxes norally have two goals. Firstly, there should be an environental benefit and secondly, the governent can obtain additional tax revenues. Environental taxes belong to the arket-based instruents (MBI) to anage behaviour of people and copanies for environental objectives. By using MBIs the negative environental externalities are internalized into the price of the process, good, or service. Pigou (1920) cae up with the concept of using taxes to internalize these externalities by raising the costs of pollution with a tax. This is in line with the polluters pay principle, where the negative externalities of a polluting product should be paid by the consuer. In Figure 2.1 below the concept of the Pigovian tax is explained. Figure 2.1 The Pigovian tax G Source: Schöb, R. (2003) Taxing the Aviation Sector 13

14 In Figure 2.1, X is a polluting consuption good, and the arginal benefits MB(x) are equal to the private arginal costs MC priv,which are constan at point C in the graph where the aount x 0 is consued. When x 0 is consued the social arginal costs MC soc are at point F which eans that there is a dead weight loss of the area CDF. Moreover it can be seen that Pareto optiality can be achieved only at point D where the aount x p is consued. This equilibriu can be achieved when the private arginal costs of the consuers will be increased by the aount t p. When this Pigovian tax is included the equilibriu can be found at point D where the aount of x p is consued at the price of MC priv + t p. In this case the governent collects taxes that will have the size of the grey area ABDG. Furtherore the area CDF will be the dead-weight loss, which are the cobined losses of the consuer and producer surpluses, as a result of the tax. Although solving environental probles with a Pigovian tax ay appear to be siple there are nuerous coplications to deal with (Schöb, R. 2003). The Pigovian tax concept assues that the MC soc are known, at least when the tax has to be quantified. However it is rather difficult to calculate the MC soc because a significant part of these costs are non-onetary e.g. costs of polluted air or costs of cliate change. Expressing these non-onetary costs in a onetary unit is difficult and soeties even ipossible. Hence the distance fro MC soc to MC priv can only be estiated, which leaves roo for discussion about the height of the tax. Even though the exact aount of the tax is hard to calculate, it can be said that at any point between x 0 and x p we are better of than at x 0, which eans that any tax not higher than t p iproves social welfare. However Meade 3 (1955) argues that in a 2 nd best world a tax would not always iprove social welfare as this could lead to ore distortions. Also iportant is that the tax should not be too high as such a tax will create an additional distortionary effect instead of reducing it. Therefore it could be argued that an environental tax can be beneficial if the optiu tax tariff is known. Another point of concern is the that the arginal cost and arginal benefits curves do not have to be linear, which is often the case in reality. If such a curve is nonlinear then it becoes even harder to calculate the exact aount. A tax on the aviation sector qualifies as a Pigovian tax. In this case the CO 2 eissions are the negative externalities which are not incorporated into the price. However, as entioned earlier, it is hard to put a price to soething non-onetary as air quality. Estiating the height of the tax is rather difficult and therefore the choice of the taxes in this study are based on the taxes that the Dutch governent applies to the Dutch aviation arket. The official tax is for a flight within Europe and 45 for flights going out of Europe (Minfin, 2007), which see to be arbitrarily chosen. The ratio of 1 to 4 is not arbitrarily chosen but copied fro the ticket tax of the UK and France. The fuel tax is deterined on 50%, which would be roughly equal to the proposed ticket tax. A second iplication can be found in the distributional aspect of the Pigovian tax. The ain concern with distributional ipacts is that the poor will suffer disproportionally ore. An OECD report (2006) shows that energy/carbon taxes tend to be oderately regressive, which eans that the poor suffer slightly ore than the rich. The low-incoe households suffer relatively ore because they spend a higher share of their disposable incoe on energy. Although ost energy taxes are regressive there is evidence that 3 Although several econoists wrote about this topic Meade (1955) was the first one entioned it in his work. 14 Taxing the Aviation Sector

15 transport taxes could be progressive if the good is considered to be a luxury good (Barker and Köhler, 1998). Other factors that should be taken into account are the tax incidence and the distribution of environental benefits (Norden, 2006). The tax incidence aspect deals with the question who will effectively bear the burden of the tax (Baranzini et al., 2000). They argue in their paper that the tax could either be passed on fully which eans higher consuer prices or absorbed by the copany so that the latter bears all costs which ay lead to lower profits and/or salaries. Whether the tax will be passed on or not depends on the price elasticity of deand and supply. A high price elasticity of deand and low price elasticity of supply iply that the consuers bear the largest part of the tax burden. The environental benefits could be distributed unequally aong the households. In spite of the iportance of this factor, rather little is known about it. In the case of the aviation sector it can be argued that flying is a luxury good. Therefore a tax on flying could be considered as a progressive tax. However, a ticket tax would relatively be lower for a business passenger than for a leisure passenger. Assuing that passengers flying business class are richer the ticket tax would be regressive. In contrast a fuel tax will be equally high for business and leisure passengers in relative ters and will therefore be not regressive. A third iplication can be found in the copetitiveness effects of a Pigovian tax. Taxes can create a distortion on the level of playing field (ECORYS, 2008) if different tax rates are iposed on specific products and/or services across countries or group of countries. This can have negative ipacts on the arket with the higher tax rate. Soe sectors are ore vulnerable to international copetition than others which depend on a.o. the openness of trade, arket power, and share of energy expenditures. When a sector is vulnerable and is taxed ore heavily than the sae sector outside the legislation area, then it has a copetitive disadvantage which can lead to lower profits and less jobs. These adverse effects are ost felt in the short-run where adoption is rather difficult i.e. changing a production process, which ay take years. In the long-run it can be argued that the Pigovian tax will change incentive structures (aking the process ore efficient), which can lead to less negative externalities. However, it can also be that the polluting industry relocates its business and that the negative externalities prevail, but appear in a different area. Moreover, even when the polluting industry decides to stop producing these harful products it can still iport these products, which eans that the negative externalities are just transferred to the exporting country. These cases can be characterized as a lose-lose situation where the country will have negative econoic ipacts and where the positive environental ipacts are non-existent or disputable. The copetitiveness of the aviation sector is likely to worsen as the sector is vulnerable to international copetition and the tax will only be iposed in the European Union. The aviation sector will be hit hardest at the hub-airports. Passengers travelling between two places outside the EU with a transfer at an EU hub airport will find it ore attractive to travel via a non-eu hub airport e.g. airports in Russia and North Africa. Especially the hub airport just over the EU border will benefit. Therefore, carriers operating at EU hub airports will have a clear copetitive disadvantage copared to carriers operating at non- EU hubs. In turn this could lead to a diversion of traffic fro EU airports to non-eu airports. Another point of concern will be passengers flying between EU and non-eu destinations. Carriers who fly directly will have to copete with carriers who ake a transfer at a non-eu airport. For instance a passenger flying fro Asterda to Tokyo Taxing the Aviation Sector 15

16 can fly directly or ake a stop in Moscow. If the passenger takes the direct flight he has to pay the fuel tax for the whole fligh however if he akes a transfer at Moscow he only has to pay the fuel tax for roughly 1/3 of the flight. Hence carriers whose strategy includes an extra stop will benefit fro the fuel tax and this could lead to a diversion of traffic fro EU airports to non-eu airports. It has to be noticed that this does not occur in the case of a ticket tax which reains 45. A third point of concern are the airports at the border of the EU. Passengers flying fro and/or to those airports ight choose other airports just over the border to avoid being taxed. This could again lead to diversion of traffic fro EU airports to non-eu airports. Hence in general a tax on the aviation sector would lead to a copetitive disadvantage of the European aviation industry. The box below shows a part of a news bulletin of Blooberg, where they explain that Dutch passengers depart fro other airports to avoid the ticket tax. Box 2.1 Part of a news bulletin of Blooberg Flying Dutchen Depart Fro Gerany to Avoid Local Airport Tax By Jurjen van de Pol and Jora Kanner July 25 (Blooberg) -- Asterda's Schiphol Airport expects fewer passengers than it had a year ago today, traditionally the busiest day of the year. It has nothing to do with rising fuel prices or the econoy: Travelers and airlines blae a new tax. This onth the Dutch governent iposed a national departure tax of as uch as 45 euros ($71) on each ticke aking Schiphol Europe's ost expensive airport after London's Heathrow, a governentcoissioned study showed in March. The tax, expected to raise 350 illion euros a year, seeks to slow traffic growth and ake travelers aware of the environental costs of flying. The charge has sent travelers looking for cheaper alternatives in Belgiu and Gerany, about 130 iles away. The levy ay cut Air France-KLM Group's Dutch passenger nubers by as any as 1 illion this year and slow job creation at the state- owned airpor the airline and a governent planning agency say: This is outrageous; this tax goes directly into the Treasury and doesn't benefit the environen. Source: Blooberg website (2008) 2.2 Ipact assessent ethods In the ECORYS research project we covered an extensive literature list of different exante ipact assessent ethodologies 4. Ipact assessents are widely used in the world of policy aking. A policy change should be supported by an ipact assessent that gives insight in the effects and iplications of the policy change in question. To assess such an ipact ex ante several ethodologies have been created. Soe of the provide a quantitative estiation and others a qualitative. The ipact assessent ethodologies in the review have been divided into Applied Siulation Techniques, Regression Analyses Methods, Survey Methods, Descriptive Methods, and Other Methods. For this study three of these ipact assessent ethods of the review were taken into consideration: 4 ECORYS (2008), Structured Overview Ipact Assessent Methodologies, 16 Taxing the Aviation Sector

17 Coputable General Equilibriu (CGE) odel, the Global Siulation Model (GSIM) belonging both to the Applied Siulation Techniques, and a quantitative Causal-Chain Analysis belong to the Descriptive Methods. Furtherore soe relations are estiated with a regression analysis. Coputable General Equilibriu The CGE odel tries to show and explain how policy changes will affect acroeconoic variables by use of a siulation. A CGE odel consists of a set of equations (translating the behaviour of all agents in an econoy) and a detailed database called GTAP (data on all independent variables and elasticities). The odel is based on neoclassical theory, which eans that the arket participants optiize their behaviour. The database consists of data which is presented in tables like the input-output table or a social accounting atrix (SAM) and of a set of elasticities explaining the relations between arket variables. CGE odels are ostly used for coputing effects of iporttariffs or a production subsidy. It is also very useful when tie series data is scarce, because the CGE odel has strong underlying assuptions about an econoy, which can replace scarce data series. The advantage of a CGE odel is that it takes all direct and indirect effects into consideration. Moreover effects crossing sectors are easily coputed. However, the CGE odel is only capable of deterining relative changes. Another disadvantage is the coplexity of a CGE odel. For instance, running a CGE odel can take several hours and preparing the data for it takes several weeks. Furtherore, the traditional CGE odels do not include social and environental variables, which is quite relevant in this study. Although newer extensions on the CGE odel, like the GREEN or TEQUILA odel, are able to stiulate environental variables these odels are not easy to access. Moreover the aviation sector is put into the transport sector which leaves out specific inforation about the airline industry like the deand of air travel and price of tickets for each arket segent. Hence for the purpose of this study a CGE odel will not be used. Global Siulation Model The GSIM developed by Francois and Hall (2003) is a partial equilibriu odel instead of a general equilibriu, which akes it less coplex and easier to use than a CGE odel. It uses less data and a sall GSIM can be used within atters of inutes and preparing and interpreting the inputs and output can be done in a couple of days. The odel is built into a spreadsheet progra which akes it accessible to use for everyone knowing how to handle such a spreadsheet progra. Table 2.1 shows a suary of the GSIM developed by Francois and Hall. Based on a set of equations and a database, just like the CGE odel, a siulation can be run. In this exaple the odel starts with solving it for world prices. The next step is to calculate the aount of iports and exports, which are followed by the calculation for the different (consuer, producer and the governent) surpluses. Taxing the Aviation Sector 17

18 Table 2.1 Suary of the GSIM developed by Francois and Hall (2003) Source: Mutabatsere, E., (2006). GSIM is often used to assess the ipacts of tariffs on the econoy. Its ain advantage is that the odel is publicly available and can easily be used while still looking at the world econoy. However, GSIM is a partial equilibriu odel which does not take all indirect effects into consideration. Moreover, GSIM is not capable of coputing specific inforation about the aviation sector and its different arket segents. Therefore the GSIM will also not be used. Causal-Chain Analysis The CCA is often described as a descriptive and qualitative ipact assessent ethod. Policy akers often use a causal-chain analysis to get a good first overview of how a policy refor will affects certain variables directly and indirectly. Its ain advantage is its power to identify indirect effects. While ost CCAs are qualitative it is possible to quantify the relations between the variables. Moreover, it is possible to incorporate any variable as long as it is related to one of the other variables. However using a causal-chain analysis also has its caveats that we need to be aware of. A researcher needs to be careful not to exclude iportant variables. Soe indirect effects could easily be forgotten which could lead to wrong results. Furtherore, ost econoists prefer a CGE odel to a CCA, because it incorporates the whole world and all indirect effects are taken into consideration. Although the CCA has its disadvantages it can be a good ipact assessent ethod when used correctly. Hence this study will use the CCA to assess the ipact of the taxes on the aviation sector. To quantify the results, a odel in a spreadsheet progra has been buil which includes all relevant variables and paraeters. More inforation about this odel will be provided in Chapter 3. The figure below shows the direct and indirect effects of both taxes in a structured way. The dark green boxes are the dependent variables, the light blue boxes indicate the three critical aspects (independent variables), and the yellow boxes indicate the taxes (paraeters). As can be seen in Figure 2.2 below, the ticket tax will only partly be passed on to the ticket price. This increase of the ticket price is expected to decrease the revenue passenger kiloetres (RPK) depending on the price elasticity of deand. Due to the decrease of the RPK, the aount of CO 2 eissions and the revenue will decrease. For the fuel tax alost 18 Taxing the Aviation Sector

19 the sae reasoning can be applied except the fact that the fuel tax is also expected to give an additional incentive to iprove the efficiency of aircrafts which reduces CO 2 eissions. Moreover a ticket tax will not be raised over cargo whereas the fuel tax does have an effect on cargo traffic. Hence, ex ante, the fuel tax is expected to be ore effective than the ticket tax. Figure 2.2 Causal-chain analysis of a tax on the aviation sector Ticket tax Ticket price Passenger Kiloetres Pass-through rate Price elasticity of deand Oil price Costs Profit and other econoic effects Fuel tax Efficiency iproveents CO 2 eissions Source: PWC (2005) odified with own input Regression analysis As entioned earlier soe relations have been estiated with the use of a regression analysis, where relationships between variables are odelled. The siplest for, which is used in this study, is the linear regression: Y = β 0 + β1x + ε. With this function a linear relationship between the dependent variable, Y, and the independent variable, X, is estiated. The ethod used in this study, and the ost coonly used ethod, is the ethod of Ordinary Least Squares (OLS). OLS tries to iniize the deviations between the data and the estiated linear relation. When the (adjusted) R-square is near 100 we know that these deviations are sall. When we want to know whether a relation has used the right variables we look at the significance of the independent variable. If the independent variable is highly significant (1%) in explaining the dependent variable then the right variable has been chosen. Hence every regression is tested on their (adjusted) R- square and significance. In addition a regression, especially a tie-series regression, should be tested on serial correlation. If a regression is serial correlated then the OLS estiator will not be that efficient because the error ters fro different tie periods are correlated. To test whether a regression is serial correlated the Durbin-Watson (DB) test can be used. The DB value should be around 2 to be significant on a 5% level, but deviations are allowed depending on the nuber of observations and explanatory variables. Taxing the Aviation Sector 19

20 2.3 The critical paraeters of the odel When the effects of a tax on the aviation sector have to be assessed there are three paraeters which need to be considered, which are: the elasticity of deand, the cost pass-through rate, and the oil price Elasticity of deand Elasticity of deand easures how the deand is affected by a specific factor such as price, incoe, and prices of other goods. It is defined as, ceteris paribus, the ratio of the percentage change in quantity deanded to the percentage change in the factor inducing the change. Hence if the deand decreases by 3% because the price increased by 1.5% then the price elasticity of deand would be -3/1.5 = -2. The inus sign indicates a negative relation between price and deand. A deand is said to be inelastic/elastic if the absolute value is saller/larger than 1, which eans that the deand is affected less/ore than proportional. Generally the price elasticity of deand depends on several key factors: The availability of close substitutes; if there are ore close substitutes available there will be ore price copetition which indicates a higher price elasticity of deand; The tie-horizon; in the short-run any decisions are already ade and fix while in the long-run there are ore possibilities to adapt to the changed circustances; The characteristics of the good/service in question; a priary good like bread is less price elastic as people need it to survive. In contrast luxury goods are ore price elastic as the necessity of it is rather low; Whether the price increase is individually or collectively; if only one copany is facing an extra tax he will also face a higher price elasticity of deand than in the case where all copanies face the sae extra tax. Table 2.2 below lists the three ost iportant factors that affect the price elasticity of deand for tickets in the aviation sector. Table 2.2 Price elasticities affected by several factors Factors: Category 1 Category 2 Effects Class Business Leisure Leisure flights have higher price elasticity Length flight Short-distance Long-distance For short-distance flights price elasticity will be higher Tie-horizon Short-run Long-run In the long-run price elasticties will be higher Level Route-arket Pan-national On the pan-national level price elasticity will be lower Source: Brons et al (2002) and own input Business vs. Leisure: Leisure travellers are norally less tie restricted and therefore they are ore willing to use other odes of transportation e.g. cars, train, and ships. In contrast business travellers value tie higher and see these loss-of-tie costs as a substantially part of their total travel costs. Therefore, other odes of transportation are not really considered as close substitutes for business travellers. In addition leisure travellers are ore willing to fly on 20 Taxing the Aviation Sector

21 less attractive ties or dates. For instance, business travellers often like to stay hoe at weekends to spent tie with their failies, and therefore tickets that incorporate a Saturday night are cheaper. Furtherore, leisure travellers have to pay their own tickets while business travellers can pass their costs on to their copanies. Finally, leisure travellers consider flying as a luxury good in contrast with business travellers who see it ore as necessity. Hence the price elasticity is expected to be higher for leisure travellers. This difference is also found in several epirical studies. A survey of these epirical studies (Brons et al., 2002) noticed a difference of around 0.6 between the price elasticities for business travellers, who were relatively price inelastic i.e. 0 < 1, and leisure travellers, who were relatively price elastic i.e. > 1. Short flights vs. Long flights: A traveller would take a car to ake a 200k ride instead of taking the plane; however when a travel of 10000k has to be ade a travel by car is less likely. Long-distance flights have less substitution possibilities especially when a traveller has to cross an intercontinental ocean. Subsequently, it can be argued that short-distance flights will have higher price elasticities. Gillen et. al. (2001) also perfored a survey of epirical studies and found a difference of 0.3 between the short flights and long flights which is line with the theory. Short-run vs. Long-run: In the short-run price elasticities are generally lower due to the process of adaptation which takes tie and the fewer options that are available in the short run. It could be that a fir relocates its industries to reduce the aount of flights due to the higher prices of tickets. Furtherore, it could be that the aount of passengers adapts faster than the aount of flights. As the aount of passengers decreases the load-factor will decrease as well, but it will take soe tie for airlines to decide to reduce the aount of flights. Consequently, the price elasticity is expected to be higher in the long-run. Fewer epirical studies have been done in this area. Route-level vs. Pan-national-level: In the case of an additional tax on tickets it atters whether only one carrier or every carrier will face this tax. When only one carrier will face an additional tax the travellers will easily substitute to other carriers. The price elasticity will be higher in this case because it includes the cross-price elasticity. In contras if every carrier will face an additional tax the price between different carriers will not change, which eans that the cross-price elasticity has no effect on the price elasticity of deand. In a report of InterVISTAS (2007) it is argued that the price elasticities of whole arkets and individual air carriers differ. In order to calculate the right price elasticity of deand they cae up with ultiplicative adjustors: 1,4 for individual air carriers, 0,8 for a whole nation, and 0,6 for a pan-national level like the EU. Hence if the price elasticity of a specific route is -1,2 then the price elasticities would be 1,4 x -1,2 = -1,68 and 0,6 x -1,2 = -0,72 for respectively an individual carrier and the EU. Price elasticity of deand has received a lot of attention in the literature. The found elasticities differ significantly due to different assuptions, research ethods, geographical areas, and segentation of the air travel arket. Gillen et al. (2001) provide an overview of existing studies on price elasticity and give a value judgent of how Taxing the Aviation Sector 21

22 representative they are. They also ake a distinction between leisure vs. business, short haul vs. long haul and doestic vs. international. The last distinction is ade because in the U.S.A. and Canada a long-haul flight can also be ade doestically. Although in Europe none of the countries are sufficiently large to ake use of this kind of distinction it can be argued that when Europe is taken as one country the distinction can be used. In Table 2.3 below all edian own-price elasticity values are given for each category. Table 2.3 Suary of edian elasticity values by type of a survey Category Median Own-price Elasticity Value All estiates All long haul estiates All long-haul international estiates All long-haul international business estiates All long-haul international leisure estiates All long-haul doestic estiates All long-haul doestic business estiates All long-haul doestic leisure estiates All short/ediu haul estiates All short/ediu haul business estiatesc All short/ediu haul leisure estiates All cross-section study estiates All tie-series study estiates All estiates fro studies All estiates fro studies Source: Gillen et al (2001) These results are in line with the theory entioned earlier. Leisure flights do have a higher price elasticity than business flights and short-haul flights are ore price sensitive than long-haul flights. Besides these results there is a rearkable difference between cross-section and tie-series study estiates. This difference is not well explained in their study. However they indicate that a study using panel data or tie-series gets 2 points while a study using cross-section data gets 0 points 5. They argue that policy analysis should be based on long-ter elasticity easures which are better estiated by tie-series data. Consequently, the later elasticities would be better suited for y study. Moreover y study covers only one area over 20 years which indicates the iportance of tie-series effects ore than cross-section effects. A last arguent can be found in the fact that the tie-series estiates are closer to the relevant estiates of the InterVISTA (2007) repor which is one of the ost recent study done in this field. As entioned above, their estiates are based on a syste with ultiplicative adjustors to differentiate between a price increase of an individual carrier and a price increase of all carriers at a pan-national level. It could be argued that the prices of all carriers depend on roughly the sae costs. Consequently, an increase of price of oil would increase the costs, and consequently the price, of all carriers. This would ean that the relative differences between the prices of different carriers would hardly change. Therefore, the cross-price 5 According to Gillen et al (2001) the given points indicate the quality of the study. Based on these points they assigned a weight to each study to coe to a weighted average instead of a siple average for all studies. 22 Taxing the Aviation Sector

23 elasticity will not affect the price elasticity, which eans that the price elasticity will be lower copared to a case where one single carrier increases his price. The price of oil changes over tie is captured in tie-series data but not in cross-section data. Hence the elasticity of a price increase on a pan-national level can be better estiated by using tieseries data. This is also a plausible reason why the elasticities on a pan-national level of InterVISTA are close to the estiates based on tie-series data. However it should be noticed that ost studies use elasticities that are based on cross-section data. Logically this deviant decision will also lead to a different result. Based on the arguents in this section the following price elasticities are used shown intable 2.4. Here a distinction is ade between leisure and business and between short-haul and long-haul. Moreover, it incorporates the arguent of InterVISTA that the elasticity faces by individual carriers is higher than that faced by the whole arket. Table 2.4 Price elasticities Full Charter Service Low Cost Carrier Intra-EU leisure -1,12-1,12 Extra-EU leisure -0,80 Intra-EU business -0,45 Extra-EU business -0,32 Intra-EU cargo -0,92 Extra-EU cargo -0,66 Weighted average -0,87-1,12 Source: InterVista and own calculations The pass-through rate The pass-through rate (PTR) is the share of a cost increase (decrease) that is passed on to the consuers. The PTR is a critical paraeter in this study as the price and therefore the deand of air traffic depends on it. In previous studies soe assued a PTR of 100% and others assued a PTR around 30% 6. In a situation of perfect copetition the PTR will be 100% as copanies have no argin left to absorb the additional costs e.g. price is equal to the arginal costs. However in a situation with iperfect copetition the PTR will be different as onopolists and oligopolists can set their prices higher than arginal costs and can therefore play with the PTR. The aviation sector can be best characterized as an oligopoly or soeties even a onopoly. In a study done by Brueckner and Flores-Fillol (2006) they assue a duopolistic situation which has been considered as a reasonable assuption. Although the nuber of copetitors differs aong routes there is a consensus that it is between 1 and 8 copetitors depending on the popularity of a specific route (Ernst & Young, 2007). As entioned before, the level of copetition has an ipact on the PTR. Ten Kate and Niels (2005) argue that the PTR does not depend on price elasticity but only depends on the 6 CE Delft (1998, 2002) assue 100% pass-through and Ernst & Young (2007) assue pass-through rate around 30%. Taxing the Aviation Sector 23

24 nuber of copetitors. Moreover, they find that PTRs are alost unifor aong firs in the sae arket. The theoretical foundation can be ade clear by using an exaple. Ten Kate and Niels use a situation with 1 onopolist who sells hoogenous goods. The deand is given by ( p) P = a b D, (1) where the price, p, depends on soe constan a, inus a constan b, ties the deand function, D(p). The arginal costs are constant as shown in equation 2 C ( q) c q =, (2) Where the total costs, C, depends on the arginal costs, c, and the quantity, q. The fixed costs are denoted by F. The profit function in this case would be: ( p c) D( p) F π =, (3) where p is the price, c the variable cos D(p) the deand which depends on the price, and a and b are constant. By axiizing the profit with respect to the price, the price will be expressed as: p = 1 a c, (4) When equation 3 is differentiated with respect to the arginal costs we find dp dc = 1 2, (5) which eans that a cost change will be passed on to the price by exactly half e.g. a PTR of 50%. Furtherore, it can also be noticed the PTR is independent of the price elasticity dp as does not depend on the constants a and b. dc 24 Taxing the Aviation Sector

25 Figure 2.3 Pass-through in a onopoly case Price Airport Deand pc 0,1 pu 1 = c 1 Deand pu 0 = c 0 c Quantity qu 1 qu 0 qc 0,1 Source: Ten Kate and Niels (2005) In Figure 2.3 above the sae case is shown where p p is half the size of c. This is due to the arginal revenue curve which is twice as steep as the deand curve. In the work of Ten Kate and Niels (2005) they extend this siple case to other case including an oligopoly. Their conclusion is that the PTR in a arket with n copetitors is equal to: n PTR =, (6) n ( +1) Equation 6 would ean that on a point to point route where 3 copetitors operate the 3 PTR will be = 75%. Generally, while looking at function 3 it can be seen that a arket with less copetitors has a lower PTR. This is, what we expect intuitively, because a copany with fewer copetitors has ore arket power and subsequently will aintain a higher profit argin aking it easier to absorb a cost increase. In contrast when n goes to infinity PTR goes to 1 which is in line with the case of perfect copetition entioned before. According to a study done by Gerardi and Shapiro (2007) the average nuber of copetitors in the aviation sector is 2.61 which translates to a PTR of 72%. According to Oxera (2003) supply constraints should also be taken into consideration. They point out that congested airports have different PTRs than uncongested airports. As deand increases and cannot be et by an increase in supply the price as well as the profit argin increases. The higher profit argin in turn decreases the PTR as there is ore roo to absorb an increase in costs. Also, the airline already charges the axiu price they can charge looking at the willingness to pay expressed by the deand function. In Figure 2.4 below the prices and quantities are given by respectively p and q while the u and c expresses whether the airport is uncongested or congested. The subscripts 0 and 1 Taxing the Aviation Sector 25

26 are the different periods with period 1 having higher arginal costs (c 1 >c 0 ). In the case where the airports are uncongested the increase of c increases the price by the sae aount which leads to a lower quantity. Consequently, the PTR is 100% due to the iplicit assuption of perfect copetition e.g. price is equal to the cost. Figure 2.4 Uncongested vs. congested airports as in the theory of Oxera Price Airport supply Deand uncongested pc 0,1 pu 1 = c 1 pu 0 = c 0 c Deand congested Quantity qu 1 qu 0 qc 0,1 Source: Oxera (2003) In the ore interesting case of a congested airport the increase of c has no ipact on the price and thereby no ipact on the quantity. The PTR in the latter case is 0% which is due to the ability of the profit argins to absorb the additional costs. Hence the PTR depends on the airport supply. Although the theoretical arguents are sound there is a lack of evidence supporting a PTR of 0% at congested airports. Moreover Defra (2007) argues that this odel is based on arginal cost pricing, which iplies perfect copetition, and, as entioned earlier, this is not the case in the aviation sector. Hence if a copany does not price their tickets at arginal costs then there is roo for a profit argin and a cost PTR will not be 0%. Although Ten Kate and Niels (2005) and Oxera (2003) provide a solid theoretical background of a PTR lower than 100% there are other studies that provide theoretical and/or epirical evidence of a PTR close to 100%. In a study done by PriceWaterhouseCoopers (2005), changes in kerosene prices are regressed on changes in an air travel price index for the UK with a one year lag. They found a PTR of 105% for full service airlines and 90% for low cost carriers. However their sensitivity analysis gave an error argin of close to 50% which akes the results less robust. Their explanation was that profit argins were low due to a high degree of copetition which would leave no roo to absorb a cost increase. This PTR is in line with PTRs used in studies done by CE Delft (1998, 2002) and by Defra (2007). Defra (2007) states that the PTR is not only deterined by the nuber of airlines but also by their coercial strategies and by the price elasticity of consuers. When all these characteristics are put in a odel the PTR will be between 80% and 150% depending on the arket segen nuber of firs and nature of deand. The PTR is lowest when a fir has a profit-axiising strategy, 26 Taxing the Aviation Sector

27 facing a linear deand and a sufficiently sall aount of copetitors. These are ore likely to be characteristics of leisure flights. If firs face a constant elasticity deand, then the PTR will be highest which is ore likely for business flights. Hence leisure travellers face a PTR of 80 to 100% and business travellers face a PTR of 110 to 150%. The ost striking notion of this study is that the PTR does depend on price elasticity which is in contrast to the study of Ten Kate and Niels (2005). They argue that the PTR can even exceed the 100% if a onopolist faces a consuer with an inelastic deand. Because the onopolist has arket power an increase in the price will lead to a disproportional decrease in deand, which eans that its profit will increase. This study concludes with a general finding that the PTR will be around a 100%. Taken all these arguents into consideration, the average pass-through ay be assued at slightly less than 100%. In the study done by Defra they re-exaine ost influential studies and provide epirical evidence of their odel. Therefore, this study assues their average PTRs of 90% for leisure flights and 130% for business flights The oil price The price of oil is one of the aforeentioned critical paraeters of this ipact assessent. As the kerosene price is closely related to the crude oil price it akes sense to look at the how the oil price has evolved and to the forecasts that have been ade. Based on these forecasts an attept will be ade to reach a consensus for the oil price for the coing decades. Figure 2.5 The real oil price in euro index oil price (1970 = 100) Years real oil price in (1970=100) Source: econagic (2008) with own corrections on exchange rate and inflation As can be seen fro the figure above the price of oil -corrected for inflation and exchange rate oveents- of 1999 has increased by ore than 400% in less than 10 years. This rearkable surge in oil prices had an ipact on the aviation sector where fuel costs went up and cost savings had to be ade. It can be argued that ost cost savings Taxing the Aviation Sector 27

28 already have been ade and it is not likely that any cost savings still can be ade (HSBC, 2003). Hence higher oil price in the future would ean that profit argins will decrease or that ticket prices have to go up depending on the cost PTR discussed earlier. Subsequently, a higher ticket price decreases deand which can lead to lower profit depending on the price elasticity of deand. Therefore, it is iportant to know what the oil price will do the coing decades. Most forecasts ade in the aviation sector assued a stable oil price ranging fro $50 7 to $70 8 per barrel for the next 20 years while the oil price have reached a $141 per barrel at the 1 st of July 9. Although the oil prices are lower at the oent it is still well above the $70 per barrel as assued in previous studies. Furtherore, the Deutsche Bank expects an oil price of $115 per barrel in 2013 which is well above the values ost forecasts used. Hence it can be noticed that there is no consensus about what the oil price will do the coing decades. However, there are any econoists who think the oil price we face today is well above the price that would be dictated by supply and deand. This could be due to the speculation of investors like Goldan Sachs who buy oil futures in the hope to sell the later on for a higher price. Speculators who have ore oney to invest after they had to pull it out fro the house arket in the U.S.A. which collapsed last year. The Chief econoist of the Air Transport Association of Aerica J. Heilich (2008) argues that it has to be speculators driving up the price, because supply exceeds deand at the oent and is expected to stay that way the coing years. Although these arguents sound rather convincing the International Energy Agency (2008) argues in their newest oil arket report that it is not likely that the large investent flows are contributing to the high oil prices. They argue that prices are rising due to a constrained supply and an increasing deand. The Deutsche Bank (2008) tries to incorporate both sides of the story and they think that part of the high oil price is due to speculators, but that the oil prices will reain high even without the investent flows. They predict that the oil prices are going down to $100 per barrel in 2010 to increase to $115 per barrel in This study will use the latter predictions and extrapolate the accordingly to 2028 which would give the Figure 2.5 shown above. This would ean that the oil price would reach $234 per barrel in This sees really high, but the concerns of Peak Oil 10 should not be underestiated as the forecasted supply does have a hard tie to stay in pace with the forecasted deand. Furtherore, this oil price is not yet corrected for inflation. In order to show the oil prices in euro corrected for inflation additional assuptions needed to ade about the exchange and inflation rates. Table 2.5 below shows the exchange and inflation rates which are used in this study A study done by PriceWaterhouseCoopers in Boeing, Current Outlook Market st of July at Peak Oil refers to the peak of the world s oil production. After this peak has been reached supply will decrease which eans an increasing oil price if deand does not decrease. 28 Taxing the Aviation Sector