Tax Policy in the European Union Notes Lectures by Professor Gaëtan Nicodème Fall 2019-2020 B Disclaimer: Might have mistakes/stu missing. Probably has a bunch of typos. Use at your own risk. Use the slides while reading this since I didn't always put them in. No notes included on the required readings or presentations. Good luck! Contents 1 2 3 4 5 6 Lecture 1 - Evolution of Public Finances in the Member States 2 1.1 Role of the State and Taxation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.2 Tax Developments in Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.3 Tax Policy in Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Lecture 2 - Introduction to Corporate Income Taxation 11 2.1 Determination of Prot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.2 Double-Taxation Relief . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Lecture 3 - Corporate Income Taxation: Domestic Distortions 25 3.1 Distortions in the Choice of How to Do Business: Taxation and Incorporation . . . . . . 26 3.2 Distortions in the Choice of Financing: Debt-Equity Bias . . . . . . . . . . . . . . . . . 27 3.3 Distortions via Lock-In in Size: Taxation of SMEs . . . . . . . . . . . . . . . . . . . . . 31 Lecture 4 - Corporate Income Taxation: International Distortions 38 4.1 Devereux-Mani (2006) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 4.2 Corporate Taxation and International Choices: Taxation and Business Location . . . . . 41 4.3 Distortion in the Amount of Investment: Taxation and Business Investment . . . . . . . 44 4.4 Distortion in the Allocation of Prot - Prot Shifting: Empirical Evidence . . . . . . . . 46 Lecture 5: Corporate Income Taxation - International Distortions (CONT) 50 5.1 Distortions in the Allocation of Prots - Prot shifting: Empirical Evidence CONT . . . 50 5.2 Capital Structure and International Debt Shifting in Europe . . . . . . . . . . . . . . . . 51 5.3 Patent Boxes and Patent Location Decision 56 5.4 Distortions in Shifting of Tax Burden - Tax Exporting . . . . . . . . . . . . . . . . . . . 59 5.5 Distortions in Competition: Tax Competition . . . . . . . . . . . . . . . . . . . . . . . . 61 . . . . . . . . . . . . . . . . . . . . . . . . . Lecture 6: Corporate Tax Competition 6.1 69 Standard Model Of Tax Competition: Zodrow - Mieszkowski - Wilson . . . . . . . . . . 71 7 Lecture 7: Aggressive Tax Planning Structures 76 8 Lecture 8: Policy Responses 86 8.1 Session of 27/11/2019 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 8.2 Session of 04/12/2019 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 8.3 Session of 18/12/2019 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 1 1 Lecture 1 - Evolution of Public Finances in the Member States In this lecture, we will discuss the role of the state and what activities economists and philosophers think the state should handle. We will see that this has an impact on taxation. We will be given some important gures to try and grasp what's going on in the EU. Then, there will be some general words on what we do on the level of EU in terms of tax policy. We will see that it's relatively constraint. Taxation is one of the few remaining policies for which the unanimity rule is still in place. This means that you need the 28 member states to agree when you want to do something at the EU level. This is challenging and may explain why (except for the last few years) there has been relatively little legislation at the EU level compared to other elds (such as nancial markets, agriculture, employment etc.). 1.1 Role of the State and Taxation The reason we have the state is simple, there are some activities that should not be left to the market. The idea that some type of activities should not be left to the market was already thought about in the 19th century, particularly by John Stuart Mill (1848). JSM argued that the following activities should not be left to the market: • Protection of citizens against force and fraud. This includes police activities and defence. • Ownership of vital resources: air, water, forests, fuels, etc. These are what we would call "services of general interest" today. • Common goods: money, measurement systems, public lighting, roads, etc. Some more activities that should be left to the state are now discussed. The state can create positive or negative externalities. The idea is that the state can actually in- ternalize externalities by acting. For example, if we want to reduce carbon emissions, we may use CO2 taxation (not the only solution). We are trying to give a signal, we try to tax the bad or promote the good. One example could be tax incentive to vaccinate. Another example of an incentive could be R&D incentives that we may give via the tax system. We might think R&D is an important activity that is not provided enough by the market, notably because the social return to the public is larger than the private return. We want to incentivize companies to do more research than they normally would if they were to only get the private benet so it would receive an extra tax incentives. Other activities that should be organized by the state are all public services or services of general interest. This includes access to a minimum amount of water, minimum postal services, to a minimum level of knowledge. The healthcare system is also one because of the asymmetry of information. This is referring to the famous problem of the markets for lemons. You know better what your health situation is than the insurers. Thus if you left the selection problem to the market, the insurers will put premiums higher than what the healthy would like to pay because they will not trust you. This results in only the unhealthy staying in the system providing a good reason as to why it should be organized by the state. Another important activity is redistribution. This became a lot more important recently because of the US election campaign: the idea of wealth taxation. Musgrave has also thought about the classication between the allocative, stabilisation, and redistribution functions so one can also see the activities of the state via this prism. The view of whether these activities should be organized by the state will vary among dierent segments of the population and across the political spectrum. There is no rm or well dened answer as to what should be left to the state and what should be left to the market. There is importance to how you nance all these activities and the primary way of nancing these activities is through taxation. The level of taxation and the structure of taxation (tax individuals, tax savings, tax earnings, tax companies, tax pollution) will vary across countries. 2 Now we will discuss the characteristics of taxes because one must be able to distinguish them, for example, from fees or other contributions. Some characteristics of taxes dened by Loeckx, Van Dionant and Neyens (1970): • Transfer to public authority, • In money or in-kind, - One can actually pay their taxes in Belgium with art. • Final - Once you have your tax assessment, you paid, it's nal. We can't come back to it. • Under force/constraint - See what happens when you try and not pay your taxes. • For which the payment is not followed by immediate and apparent counter-action/benet by public authorities, - Not a toll to get access to the highway, for example. It goes instead to a general budget that is unassigned and is redistributed depending on what expenditure one wants to nance. • Established and requested according to xed rules, constitutes a means through which the government will carry out its policies, - These should be known in advance, there should be tax rules that explains how you will be taxed, there is tax certainty. If you don't have tax certainty because it's complex, you can contact the tax authorities and obtain an advance tax ruling. You go to the tax authorities with your scheme, you explain to them what you want to do, and they will provide a written interpretation of the tax laws with the way they see things regarding the scheme and this is binding. The tax authorities must then respect this and not change their mind afterwards which results in tax certainty. • Owed by citizens because they are members of an organized community, • And assigned to the general budget. There are real debates on what should be classied as taxes or fees etc. For example, church taxes. You may pay taxes to nance churches. Is it a tax or is it not a tax? All of this depends on whether the tax is compulsory, or if you have the right to choose whether to pay or not. Now we will discuss the principles. • Legality: There are laws organizing the tax. • Annual: Usually annual, or transaction by transaction. • Equality - Non-discrimination: People in the same situation should pay the same amount of taxes. • General Principles of Law: Reality: Based on the real activities that you do. Territorial Non-retroactivity: You cannot suddenly decide to change the rules for the last ten years. Choice of `least taxed legal way': If you have several ways of organizing your activities, you allow the citizen to choose the one that minimizes your taxes. This leads to a problem of tax avoidance as opposed to tax evasion which is illegal. Public order • International treaties: You have to respect international treaties if you do cross border operations, if your activities depend on dierent jurisdictions. For example, you are a Belgian resident working in France. The way you are taxed between France and Belgium is organized by a treaty. 3 Now onto typologies. There is taxing authority: who taxes? It can be a local authority, a federal authority, a regional authority. There is also the source: what is taxed? It can be wealth, streams of revenues, specic products, specic transfers, ownership (ownership of a car or house, gets taxed every year), expenditure or consumption (typically VAT). There is also the nature of tax base: how is it taxed? We can tax ad valorem (a certain percentage of the amount) or something more specic such as energy taxation or excise duties which may be based on quantities. For example, a tax of 30 euros per 100l of wine. In terms of rates, it can be a lump sum, each of the students pay 30 euros regardless of what your situation is. It can be proportional, I tax you 20% on your income. It can be progressive, each of you pay 10% on the rst 100 euros that you earn, 20% on the next 100 euros, and 30% above that. The unit of taxation also matters. In most European countries we now have separate taxation of couples. If your partner is also earning some money, both will be taxed separately. In Germany, there is still joint taxation. You add the revenues of both couples, and since you add them to the same progressivity system, the revenues of the second one starts where the rst one ends in terms of progressivity so you tax much more. This can cause a disincentive for the second earners to work. The tax can also be direct versus indirect. and is very dicult to make. This is an important distinction that has no legal basis Direct taxation targets a situation, it is asked directly to the citizen through a tax roll. A roll is a document that tax authorities will make where they say what the situation is, for example, you are employed so you are liable to personal income taxes, they want to tax your annual revenues from labour and they established based on your declaration that you earn 100k per year. By applying their progressive system you owe them 25k euros in taxes. This is sent to the citizen directly and the citizen will pay. Indirect taxation is taxing an event, for example a transaction. You go to the restaurant, you get a meal, that's a transaction, you pay value added taxation (6% in Belgium). Example of direct taxes: personal income taxation and corporate taxes are the two main ones. Indirect taxes: VAT, import duties, excise duties, CO2 taxation, nancial transaction taxes. Now we will talk about two principles and two criteria that we apply a lot in economics when we think about taxation. The two principles are the ability to pay and the benet. The ability to pay says that we should tax people in accordance to their ability to pay. If someone has less means than someone else, he or she should pay less than the other person. That's a bit the idea behind the progressive tax system, for example, because the ability to pay of the rich increases with revenues as all their basic needs are fullled and the marginal utility of the last euro that the rich person is earning is probably less than the last euro of a poorer person. The other principle is the benet. This says that people should be taxed in accordance to the benets that they receive from the public authority. If I receive a lot from the state, I should be taxed more than someone who receives less. One can see immediately that these two principles may be in conict with each other. Someone who earns very little may receive a lot of social help so how do we apply the benet principle and how do we apple the ability to pay principle? There is a normative choice to make in the balance between the two. An example of benet principle are road tolls (even though he said earlier it wasn't a tax, this is just an example of the benet principle). You pay to use the highway, that's clearly an application of the benet principle. To some extent also taxation on petrol, it's taxed depending on the volume that you use. There is an indirect link with the number of kilometres that you drive and hence the usage you make of the roads. There is no progressivity in the system, it is not because you earn a lot that you pay more for your petrol but it depends on your usage. If you drive a lot and use the roads a lot, you will have to pay more than someone who uses them less. One of the criteria we look at is eciency. We want taxes to be ecient. They should not create distortions. A tax system is ecient to the extent that it does not create distortions. If we don't have taxes, we know that for private goods, the marginal rate of substitution is equal to the marginal rate of transformation between two products and is also equal the ratio of the prices of the two products. We know it's not the case for public goods, where the marginal transformation is the sum of the marginal 4 rate of substitutions and hence people can hide in the sum (the problem of the freerider). If you impose a tax on the product, you immediately create this type of ineciencies. If I start taxing product b at a tax t, the equality is disturbed and ineciencies are created. It creates deadweight losses in the economy which can be seen as a loss in welfare due to reduced economic activities. M RS = (1−t)Pb Pa < M RT = Pb Pa In gure 1, we have prices, quantities, supply, and demand and we have the optimal price P* and quantity Q* where supply equals demand. Figure 2 shows what happens if we impose a xed tax, T, on the product. The quantities exchanged will decrease because the price that will be demanded for those quantities will be Pd and the price that is supplied at that quantity is Ps . Figure 2: Supply and Demand with Tax Figure 1: Supply and Demand Figure 3 shows the consumer surplus without the tax. Consumer surplus is the amount you are willing to pay for an additional good minus the price that you actually paid. For example, imagine you wake up in the morning and you are very hungry and decide to go to the baker for croissants. Let's say for the rst croissant, you're very hungry and willing to pay 5 euros, you really want this one, and if you buy a second one you're willing to pay 3 euros. For the third one, you're ready to pay 1 euro. You arrive at the baker, and it's a pleasant surprise because all the croissants are at 1 euro. The consumer surplus is the dierence between what you were willing to pay (5+3+1) and what you actually pay (1+1+1). Figure 4 shows what happens to consumer surplus after the tax. The consumer surplus is reduced because the price has increased from P* to Pd and there are two areas (A+B) which represents the loss in consumer surplus. Figure 5 shows the producer surplus after the tax. After the tax, the price decreased from P* to Ps and there are two areas (C+D) which represent the loss in producer surplus. Figure 3: Consumer Surplus without Tax Figure 4: Consumer Surplus with Figure 5: Producer Surplus with Tax Tax There is a total loss in welfare in the economy of A+B (consumer surplus) and C+D (producer surplus) but there is a tax that is collected which the the tax T times the new quantities so that's A + C which is the gain in taxes (gure 6). The total deadweight loss is equal to (A+B) + (C+D) - (A+C) = (B+D) shown in gure 7. In words, we have two losses: the consumer surplus and producer surplus minus the tax collected which leaves B+D which is the deadweight loss/Harberger Triangle. 5 Figure 6: Gain in Taxes Figure 7: Deadweight Loss The aim of taxation is to design a tax system that would minimize those deadweight losses. To the extent possible, we should have a tax system that does not distort choices. Due to the tax, consumers and producers will be forced to decrease the quantities exchanged. The Harberger triangle depends on the elasticities of demand and supply. The more inelastic you are, the less the distortions will be because you will not change your behaviour. For example, smokers: the price elasticity of smoking, that is if we increase the price of smoking by 1%, the quantities that are consumed do not change that much (professor believes it decreases consumption by 0.3 to 0.4). The opposite occurs if we were tax something that is very elastic, for example, brands of water. If we tax one brand of water and not another, people will most likely switch to the non taxed brand. The deadweight loss is represented by the equation: DW L = 1 PQ 2 t 2 1 + η1 It is inversely proportional to the elasticities. It is proportional to the tax base (the revenues: PQ), and it is proportional to the square of the tax rate. This is a very important element because if you have a deadweight loss that is proportional to the base and proportional to the square of the tax rate, a policy that would decrease the rate and collect the same amount of revenues by increasing the tax base is a policy that will bring the same amount of tax revenues but decrease the deadweight loss. The new mantra in tax policy is to have a large tax base and a low tax rate. We should stop giving a lot of exemptions and tax only very small activities at very high rates. The new policies now say decrease the rates and enlarge the tax base. The reason behind that is exactly the explanation above, it collects the same amount of tax revenues but it's decreasing the deadweight loss so the taxes are less distortive to choices. There is the Ramsey-rule which says that we should tax more heavily less elastic products. For example, you should tax more heavily cigarette consumption or some type of alcohol consumption compared to products that are more elastic. The only taxes that are not distortive are lump-sum taxes. If the tax authority say that they will tax your usage at 5%, you might decide to consume less and therefore it is distortive because it impacts your choices. If they decide to tax cars but not public transport, you might switch to public transport, thus it's distortive. Now, if they tax everyone 25 euros regardless of what you do (lump sum tax), it's decreasing your disposable income but it does not aect your choices. One person who knew all of this very well was Margaret Thatcher because her advisor was well versed in these theories. At some stage she implemented a new property tax for houses that was called the poll tax. The poll tax that they implemented was a lump sum tax that said regardless of the worth of one's house, they will pay the same amount. There were a lot of protests. Why? She was certainly fullling the principle of eciency but people thought she was violating the second criteria which is equity. People thought it wasn't fair that someone who has a cheaper house pays the same amount as someone who has a more valuable house. Therefore, the second criteria is equity. There are two ways to divide equity: horizontal equity and vertical equity. Horizontal equity is that two people in the same situation should be taxed or treated 6 in the same way. Vertical equity is two people in dierent situations should be taxed or treated in dierent ways. The notion of equal sacrice should be kept in mind. We need to organize a system where every person has the same equal sacrice when they pay taxes. Now the big question is what do we mean by equal sacrice? Is it equal to marginal sacrice or an equal total sacrice? How do we dene it? There is a big debate going on. Notable John Rawls, the philosopher, published the book A Theory of Justice (1971). He developed the principle of maximin. He says that inequalities should be allowed in society to the extent that it benets the poorest. For example, if you have a situation where society might be more equal but some people are very poor, this may be worse than a situation where there is certain extent of inequality but where the one at the lowest level is actually better o than the previous situation. That's the concept of the maximin, you should maximize the situation of the person at the minimum. Lots of opposition, notably Robert Nozick. Robert Notzick (1974) argued that personal income taxation is like slavery because if the state taxes your income from labour, you are basically working for the state. You're forced to work for the state via taxation. An amount of the work that you do is actually working for the state and you have no choice since you have to pay taxes so it's forced slavery. It's a bit of an argument that still has some resonance in some part of societies (U.S.). 1.2 Tax Developments in Europe Now we'll look at some graphs not on tax revenues but on government spending. Figure 8: Government Spend- Figure 9: Government Spending as Figure 10: ing as % of GDP: US % of GDP: UK & Germany Government Spend- ing as % of GDP: Other Figure 8 has government spending as a % of GDP for the US. Up to the 1920s, the size of the government was very small (a few % of GDP). The state had almost no role except for organizing the army, police, the courts, a rule of law system, but that was basically it. There were huge debates in the US at the time on whether they should increase the corporate income taxes from two to three percent and huge arguments about whether this was fair or not at the time (shows how small the role really was). You can see in the graph that there are two spikes, the two World Wars, and then you have a large increase after the second World War and you see the same picture in most countries (Figure 9 and 10). What we see in all those countries is that up until the second World War, the size of the government was small because people thought of the government as having a certain number of functions and not having to deal with other things like redistribution, for example. After the second World War, we see the emergence of welfare systems that have been delivering a certain number of services to citizens. Also, more and more use of the system to stabilize the economy, more and more public services oered, and a corresponding increase in spending. Those increases also explain how taxation has evolved because the counterpart of this is to nance it via some revenues from tax. Public social spending as a share of GDP has really ramped up after the second World War (Figure 11). To some extent, it has been mirrored with what we see in terms of tax collection. Figure 12 has the total tax to GDP ratio for EU. All the data that we have on taxation always comes with a year and a half of lag. We never have gures that are more recent because, for example the data 2019 in June, the tax declarations are done for the revenues in 2018. This will only processed by the tax authorities by the end of the year and obtain the nal amount/adjustment. There are some statistical breaks that can be explained by the change in denition in the dimensions for Eurostat. It is due to 7 the reclassication of certain taxes redened as fees. What we see is a large increase in total taxes from 1970 up to the 1990s and then a stabilization afterwards with a decrease due to the nancial crisis and a new increase occurring recently. The point here is that the level of taxation will actually depend on social choices. It's ne if you think you need a welfare state but it needs to be nanced. Hence, there is an increase in taxation over the years that it developed. Figure 11: Public social spend- Figure 12: Total Tax to GDP Ratio: ing as a share of GDP EU Figure 13: Total Tax Burden Figure 13 compares total tax burden of dierent countries from 2007 and 2017. By and large, the EU, with an average tax of close to 40%, is a high tax area compared to OECD, US, Japan, Switzerland, other developed economies. The tax collection in Europe is higher than in most parts of the world in percentage of GDP. This hides large variations of member states, going from less than 25% in Ireland to levels of close to 45% in Belgium, Denmark, and France. The reason why is political choices based on what services you want to nance and provide. There is quite a lot of heterogeneity in the services wanted between the dierent members of the EU, which is shown not only in the level of taxation and in the structure of taxation, but also in the way it's taxed. Some countries may want a very progressive tax system for personal income taxation, some implement a at tax. Figure 14: Indirect Taxes Figure 15: Direct Taxes Figure 16: Social Contributions Figure 14 and 15 shows the split between indirect and direct taxes. Ireland, Slovakia, and Germany are countries that collect relatively little indirect taxes compared to Sweden, Croatia and Hungary. If we look at direct taxes, we have very high levels in Denmark, amazingly high compared to others. There are very little direct taxes in Romania, Lithuania, or Bulgaria. The reason Denmark collects so much direct taxes is because Denmark barely collects any social contributions, all the welfare system is nanced via personal income taxation (Shown in Figure 16). This is contrary to other countries such as France and Germany that nance their social welfare system with social security contributions. The choice in how you nance your programs will aect your tax collection. The Danes pay a lot of personal income taxes but pay almost no social security contribution and the situation may be dierent in other countries. It means we have 28 tax systems that are quite dierent across Europe. Figure 17 shows tax to GDP ratio in three categories: labour, consumption, and capital. Impor- tant: capital is the residual, it is everything that is not labour and consumption. It can be taxes on stock of wealth, it can be taxes on ows of dividends, it can be taxes on property, it's a mixed bag of many things but it is not necessarily taxes on capital income. There is still a statistical break because Eurostat has changed its classication a number of times over the years. What we see here is that there is an increase in collection of labour taxes and a stabilization at around 20%. 11% of GDP represents 8 taxes on consumption and remains relatively at. Then, there has been an increase of taxes on capital in 1970 that has climbed to about 9%. We have to be careful here because this graph is not showing that we tax labour more than capital. These are taxes on labour to GDP or taxes on capital to GDP and not taxes on labour and capital on itself. We don't know what the share of labour and capital in the economy is. We have to look at Figure 18 to see whether labour is taxed more than capital. This provides the implicit tax rates of labour, capital, and consumption. This is the tax collected on each economic function to its base, so taxes collected on labour divided by labour income. Labour is still taxed around 35% and higher than capital (30%) on average but the dierence is much less severe compared to the previous graph. Always be careful when looking at graphs - The only statistics you can trust are the ones you have falsied yourself Winston Churchill. Figure 17: Taxes in % of GDP Figure 18: Implicit Tax Rates 1.3 Tax Policy in Europe We didn't have a lot of time left at this point so the professor shortened this section quite a bit. Taxation carries an important element of national sovereignty. When you talk to member states they will say taxation is their prerogative, they don't want any other institution, country, and/or supranational authority telling them what they should do about taxation. That's also subject to the principle of subsidiarity. Why should we tax at the EU level if we can tax well at the lower level? This is reected in the unanimity rule. Any time you want to do something at the EU level on taxation, we must have unanimity. There has been a proposal by the commission last year that proposes to move away from unanimity on some elements (baby steps), for example cooperative administration (not where the big issues are), and then to progressively go towards qualied majority voting using what you call a Passerelle clause. This clause says that if all member states unanimously agree then they can change the rule of unanimity to QMV on specic issues. So essentially we would need unanimity to go away from unanimity but it means that is possible to do it without changing the treaty which is already a step. Every time we want to do something, we also need to make sure that it has a legal base. There are some articles in the treaty that can be legal basis, for example 113 for VAT, 115 for direct taxes. Then, there are a number of principles of law that you need to respect: for example the four freedoms and non discrimination. (Skipped objectives of EU Tax Policy). Some instruments: we have regulations and directives that we can issue. If they are not adopted by unanimity, you can go into enhanced cooperation which means that a number of member states, a minimum of 9, say they want to go further even though there was no unanimity at the EU level. There is currently a discussion under enhanced cooperation with regards to the case of nancial transaction tax (FTT). There was no unanimity at the EU level but 10 member states (originally 11 but one dropped out) are now still discussing and saying they want to apply it between them and there are a number of rules they need to respect to do that. They must keep the same proposals and make sure that it doesn't have a negative impact on the others. Then, an important issue that we will come back to is that all questions on taxation that arises between two countries are always treated bilaterally. There is no multilateral tax treaties in the EU, there is only one that exists in the world: the Nordic one. Besides that, there is an OECD framework/model that most countries follow but the details of the rules are always bilateral. So Belgium has a treaty with France 9 that can be dierent with its treaty with Germany. You can understand that it's already complex when you are in a bilateral situation, so when you are in triangular cases, it may become extremely dicult to understand who has the right to tax and where? Imagine that you are a French citizen residing in Belgium that crosses the border to go to Luxembourg and your Luxembourgish company sends you half of the year to work in Denmark. Who has the right to tax you? The answer is: you must look at all the treaties and hope they make sense all together and hope that you won't be taxed twice and also hope there might be some loopholes where you end up actually being taxed nowhere (if you're lucky). 10 2 Lecture 2 - Introduction to Corporate Income Taxation This lecture provides an introduction to corporate taxation. It is THE tax that is most discussed at the moment in terms of international taxation and it has a lot of economic eect. In Europe, we have a total tax to GDP that is close to 40%. Corporate taxation in terms of GDP is only between 2.5% and 3%. We collect very little in terms of corporate income tax (professor not sure about this gure but believes it's close to 600 billion in Europe). This does not necessarily mean that corporations are not taxed enough. Remember from last lecture: it depends on the size of corporate prots in the economy. It may just be that the size of corporate prots in the economy is not that large and there may be multiple reasons for this. It can be reasons like corporations are a very small size of the economy, it can be that they are not very protable, and/or it can be that they avoid taxes. We try to investigate what are the explanations for this tax collection. It's a tax that is not very important in terms of revenue collection. For example, we collect around 12%-15% of VAT in terms of GDP and 7%-8% of personal income taxes in terms of GDP. Those are much bigger in terms of taxes we collect however there is a lot of discussion about corporate taxation. Why is it the case that we discuss corporate taxation at the international level more than other forms? The main element is that it's not just collecting tax revenues but attracting businesses, investments, and labour to the country. Maybe we don't collect a lot of taxes from corporate income but we get all the advantages that come from these businesses: possible better paid workers, possibly more technical jobs, more investments. Another element is the element of fairness. We will see that the international tax system may oer possibilities for tax planning. This may be available for some companies that are international and not necessarily for domestic companies. There may be an issue of levelling the playing eld between international and domestic companies. One element that is also interesting but where research is lacking: you have, on the one hand, issues of distortion of competition so some companies have the possibility to do tax planning that others don't have (Usually multinationals corporations are those that are able to do tax planning activities - are legal) and, on the other hand, you have the context of rising inequalities that have been made popular by books such as Capital in the Twenty-First Century by Thomas Piketty and some research by authors like Nicholas Bloom who found that the inequalities in terms of wages are more interrms (across rms) than intrarms (within rms). The question is whether the corporations who are able to pay the workers much more than the average companies are the same as those that benet from the international tax system. There is also a lot of debate due to the emergence of the digital economy and the fact that many companies are now able to act remotely from a country: having activities in one country through digital means but not being taxed there. This is seen by some countries as being unfair and there has been a lot of discussion about the international tax architecture, that is, the way it is organized at the world level, with current discussions focusing on how to reform this tax architecture (notably at the G7, G20, the OECD, and in the EU). To understand all of this, we rst need to go back to what is corporate income taxation and what are corporations? Who is Liable to CIT? The rst question we might want to ask ourselves is who is liable to CIT? By denition, corporate taxation is paid by corporations. Corporations are entities that are incorporated so that have a legal status as a company as opposed to the status of sole proprietorship or sole trader. For example, you are self employed and you are running your own activity, you are not a company. You can choose between staying unincorporated as a natural person and pay personal income taxation on your revenues or you can choose to incorporate and become a company. Sometimes in some countries they allow for 1 person companies and you start paying corporate income taxation on the revenues of your business activities. The sole traders or sole proprietorship is basically run by a natural person as opposed to corporations who don't have owners who are either corporate (for example, Siemens Belgium is owned by Siemens Germany) or by individuals (Siemens Germany is owned by many individuals who bought shares on the stock market and are the owners of the company). What distinguishes sole proprietorship from corporations is the legal accountability. Corporations have limited liability while sole proprietorship 11 have liability on the nances of the business which means that the owners will have to cover the losses. Whereas if you are the owner of a corporation, you will lose all your money that you invested (the share that you have of the company may be worth nothing if the company goes bankrupt) but you are not liable to pay the remaining debts of the company. There is a category in between corporations and sole proprietorships which are partnerships. In some countries, some types of partnerships pay corporate income taxes and in other countries they pay personal income tax, it depends on the jurisdictions where they are established. What is important here is that corporations are incorporated entities, they pay corporate taxation while sole proprietorship and sole traders are unincorporated entities and pay personal income tax. What are Corporations? Corporations are legal entities, they are authorized to act as an entity. For example, they can act in justice in their own name. They are private or public. Private means the shares cannot bought, public means that the shares can be exchanged. They can be exchanged either on the stock market (listed) or outside of the stock market (non-listed). Some examples of companies that pay income tax: the AG and GmbH in Germany, Sociétés Anonymes (SA) in Belgium/France, Sociétés de Personnes à Responsabilité Limitée (SPRL) in Belgium, Limited Companies (Ltd) in the UK, Naamloze Vennootschap (VN) in the Netherlands/Belgium, C corporations in the US, etc. Separate Entity Approach Figure 19: Separate Entity Approach Figure 19 is a summary of what we have discussed so far. There is a dual system. Either you're a corporation: the company will make a prot and will be taxed on its prots via the CIT system, they will distribute dividends, and the shareholders will be taxed on those dividends at the personal income tax. The other possibility is a sole proprietorship where all the activity of the sole proprietor is taxed at the personal income tax so there is transparency/pass-through between the activities and the status of the owner. Conditions for Taxation You have some conditions for taxation and may vary depending on the jurisdiction but you have two main principles that you see in the world. There is one which is the concept of residence meaning that you need to legally implement it in a country to be taxable in that country: either you have a legal seat or you have your place of management in the country. That's the denition that the tax authorities may opt for: this company is a resident in their country and as a resident in their country they have a right to tax its activities. Usually it is linked to the worldwide principle, meaning the company can be taxed on its worldwide income with the possibility of some relief with the tax that has been paid abroad (more on this soon). It can also be taxation based on the activity. The revenue that is generated in another jurisdiction applies the territorial principle. You tax on the local income in the source country only. Quick summary - you have the two main systems: the worldwide system and the territorial system. The territorial system basically means you only tax the revenues generated in your jurisdiction and you don't want to tax the revenue that is generated abroad. Worldwide principle means you want to tax the worldwide income of the company and possibly give a relief for the tax that has already been paid abroad. 12 There are very few countries that adopt the worldwide principle, a large country that has just switched is the US. There has been a major US tax reform last year: the Tax Cuts and Jobs Act - Trump Tax Reform Plan. They went from the worldwide principle to the territorial principle. Before, companies based in the US were taxed on all the revenues generated by US Multinationals and the subsidiaries abroad. For example, Apple would be taxed in the US on US activities and on all revenues of its subsidiaries abroad. When the money of the subsidiaries was repatriated to the US, it was taxed in the US and it would give a tax credit for the tax that has already been paid in Ireland, Belgium, France, etc. where they already paid taxes. Now, there has been a switch to the territorial system so now only the US activities are taxed in the US and the activities abroad are taxed abroad. The worldwide system, in economic terms, is more ecient because it leads to capital export neutrality but it was leading to some tax planning activities. Notably because in the US as long as the prots of the subsidiaries were not repatriated to the US, it wasn't taxed so there was a deferral. For example, you park your money in US Virgin Islands and it stays there and you pile up the cash there. As long as it's not repatriated, it's not taxed. In the end, the US just went for a territorial system. Another point that is interesting: the Apple case where there was a dierence in the denition of residence between Ireland and Bermuda. Ireland was saying the residence is actually where you have the place of management (which is Bermuda so Bermuda should tax) and Bermuda was saying the residence is where you have the legal seat (which is Ireland so Ireland should tax). In the end, none of the countries were taxing. 2.1 Determination of Prot The idea is that you need a starting point. In many countries the starting point is the generally accepted accounting principles with some corrections (local GAAP). You could possibly tax based on the IFRS (International Financial Reporting Standards) or specic tax rules such as the US and the UK. The nancial statements of the company based on accounting rules may be extremely dierent to the tax statements. It makes a huge dierence and it is at the expense of researchers. As researchers, you have an increasing access to nancial statements of companies via big databases but it's not the actual tax statements so nancial statements have been used to do research. When they have access to tax returns and tax data, it is obviously much more precise. Some countries like Belgium follow quite closely the accounting rules but you also have some departure from those rules. Taxable Prot Generally, all income is considered as business income and would be taxable. Everything that comes into the company is usually by default considered business income. This does not mean that some part of it will not be exempted. The taxation is done on an annual basis so companies have to do a yearly tax statement and it is based on accrual methods as opposed to cash methods. It means that we don't look at the cash ows (what comes in and out of the companies) but the accrual. The dierence is, for example, you have taxes to pay and you know you will have to pay taxes in two months. We are in December and we have to pay taxes for the year, but you know that you will only pay them in two months: if you look at the cash the transaction will take place in January or February but in accrual method you already registered a liability towards the tax authority in the form of a debt (provision for taxes paid) that will be a deduction in the prot for the year. Accrual method means as soon as you know that something will happen, you need to record it in your accounting system (same thing for provisions or pensions). Another example: you buy something from a supplier and you know the supplier gives you thirty days to pay, well on the day that you buy you don't wait for the thirty days to record it into your accounts. On the day that you buy, you record the fact that you have a debt a debt towards your supplier that will lead to a payment and a deduction in prots. What may dier from the accounting is that for tax purposes you may have some income that's not taxable because the law says so and you also have some expenses that may be non-deductible for tax purposes. An example of this in several countries are the expenses for restaurants: when you go to a restaurant for business, it's not necessarily fully deductible, it may be deductible 25%, 50%, or 75% depending on the law of the country, but part of it may not be deductible. You have some income that the tax authorities may consider 13 not business related so they do not consider it taxable. In accounting, all income and expenses need to be recorded and that's where the accounting statements and tax statements may start to be dierent. Prot Determination Sales + Other Revenues = Turnover Turnover - Cost of Materials - Wages and Compensations = EBITDA (=GOS) EBITDA (=GOS) - Depreciation and Amortization = EBIT (=NOS) EBIT (NOS) - Interest Paid = Taxable Prot = Retained Earnings + Dividends The prot determination is very simple. All your sales (business related) plus other types of business revenues (for example, interest received if you make a loan) results in the turnover of the company. You deduct from that all the materials that you bought from your suppliers and the wage and compensation that you paid for your labour. For example, Coca Cola makes a lot of sales of cans of soft drinks, they may receive some interest from other parts of the group, the cost of all the materials are basically all the cans, the water, the magic formula, advertising, then you pay your workers and arrive at EBITDA. EBITDA stands for earnings before interest, tax, depreciation, and amortization and it's roughly the operating prot of the company. There is a concept in macroeconomics that comes close to it which is the gross operating surplus (GOS) (not always a reconciliation between micro and macro). The gross operating surplus of the corporate sector is very close to the total EBITDA of the corporate sector, it's a good proxy in the national accounts of countries. Then when you are at the EBITDA, you deduct depreciation and amortization to arrive at the which is earnings before interest and tax. In macroeconomic terms, it is very close to the EBIT net oper- ating surplus (NOS) of the corporate sector. The dierence between gross and net comes from the depreciation and amortization part. For example, once again you are Coca Cola. To produce your cans, you need machines. You will buy the machine for a million, you will pay cash to the supplier when you buy it, it will register in your accounts as an asset. You have a cash outow of 1 million but in exchange you get an asset (double entry system). You will be able to deduct a certain depreciation of the machine, yearly. This depreciation will depend on the tax rules. For example, if it's a building it might be depreciated in 25 years so 4% a year. A certain type of building might be 33 years so its 3% per year. Some machines can be depreciated in 8 years, some others in 10 or 15 years. A computer in one year. Back to the Coca Cola example, in accounting the machine might be depreciable in 5 years, and they bought it for 1 million so every year in their nancial statements, they will deduct 200k a year until the moment it reaches 0. Then the machine has been completely depreciated and has disappeared from their accounts, it may still be there physically, there will still be in accounting terms a need to re-evaluate it at the current cost but for tax purposes the machine has been depreciated. It's dierent from a system where you have immediate expensing. Immediate expensing corresponds to depreciation of 100% in the rst year, so the machine in the example will immediately deduct 1 million. Tax authorities usually do not allow that, except for small items. They ask companies to depreciate over time, so if it's 5 years, a fth of the value of the machine is depreciated every year. Once you are at the EBIT, you still have to deduct the interest that you pay on the debt. For example, Siemens Germany or Coca Cola got a loan from the bank to buy the machine and they pay an interest of 5% (50k euros) has to be paid to the bank every year in terms of interest plus the capital they need to reimburse. That's the interest that needs to be paid and it's tax deductible: interest paid for companies is tax deductible. You arrive at the taxable prot which is what the tax authorities are interested in. The tax au- thorities will say you have a prot of 200k and the tax rate is 25%, so they take 50k from your prot and you are left with an after tax prot of 150k. The company has a choice between distributing all or part of it as a dividend to shareholders or reinvest it in the company (retained earnings). For the shareholders, it "should" not matter because either they receive the money through a dividend or make a capital gain because of the increase in the value of the share due to the retained earnings. 14 Treatment of Losses It may well be that the company, after all the costs etc., doesn't have a positive taxable prot. There is nothing to tax for the tax authorities and no tax authority would allow for a system where there is a negative tax. No tax authority will subsidize losses made by companies. Instead, what happens is that tax authorities will say no taxable prot so no tax. The tax liability of companies has a oor of 0zero, there are no negative taxes. They never receive from the tax authorities a tax-back. This can create problems. For example, it can create the following situation shown in gure 20: Figure 20: Treatment of Losses Situation Imagine the tax rate is 25% and you have two companies (Company 1 and Company 2) who are both on the market and they operate from year one to year four. Company 1 is a very regular and stable company, every year they make a prot of 100 so the total over the four years is a prot of 400. Company 2 is a lot more volatile. In year 1: they have a prot of 200, year 2: they have a loss of 400, year 3: a prot of 200, and year 4: a prot of 400. The total prot over the four years is 400. The total prots over the years for both companies are the same but the timing of them is dierent. In terms of taxes, company 1 pays 25 every year for a total of 100 over the four years. You can get the average rate 100/400 which is 25% so the system is fair. Company 2 uctuates a lot more so in terms of taxes, the rst year they will pay 50, the second year they have a loss so they pay zero, the third year they would pay 50 again, and in the last year they would pay 100. Therefore the total tax they paid over four years would be 200. The average rate they paid would be 200/400 leading to an implicit rate of 50%. They have the same prot but pay more in taxes so they would have a disadvantage with this system. To avoid this situation, countries allow companies to carry forward the losses: to take the loss they have on a specic year and to push it to a subsequent period. Figure 21: Company 2 with Loss Carry-Forward Figure 21 has the same situation as company 2 in gure 20 but with loss carry-forward. The tax to be paid in the rst year would be 50 and second year there is a loss so a tax of zero. They have the right to carry forward the loss from that period to the next one resulting in a deduction of the prots that they make. For example in year three, they would have a loss of 200 because its a combination of 15 the prot of the year of 200 plus the loss of the previous year of 400 (which they carry-forward to the next period). So in year three, they have a loss of 200 so they pay 0. This loss of 200 is also allowed to be carried forward so it would go to the next period. The prot of the year is 200 (400-200) which means they pay 50 in tax for the year. Over the four years, the company pays a tax of 100 which leads to an implicit average tax rate of 100/400 = 25%. This is what we call the carry-forward of losses and in many countries it is unlimited in time so you can carry forward your losses as long as you want. In some countries it is limited in number of years, some countries allow for ve years, for example. Some countries, although rare, allow for another system called loss carry-back. For example, you made a loss this year, but you made a prot the previous year. They will impute part of the loss on your prot from last year and they will reimburse you the tax that you already paid last year. This situation is demonstrated in gure 22. Figure 22: Company with Loss Carry-Back Sticking with our previous example, you would end up paying zero in year one because 200 of the 400 loss of year 2 is carried back to year one which leads to a prot of zero in year one. Year two: tax would be zero because you have a loss of 400. In year three, the tax becomes zero because the remaining loss of 200 from year two is carried forward. In year four, it's fully taxed so you pay a tax of 100 leading to a total tax paid of 100. This leads to implicit average tax rate of 25%. It's a system that insures that companies are not disadvantaged because they have uctuating revenues. The carry-forward can be limited in time or unlimited in time. For example, the Netherlands allows for nine years carry-forward while the UK, Germany, and Belgium allows for unlimited carryforward. The carry-back in practice is always limited in time. The UK, Germany, and Netherlands allow for a one year carry-back. Disclosure: professor not 100% sure whether the number of years on the slide are still valid or if they have changed. Note: if your company has a prot, you may want to acquire and merge with a company making losses so that you get compensation in terms of the tax system (if the tax system allows it) allowing you to not pay taxes on that year so in the end you defer your taxation. In other words, you may have some acquisitions that are driven by those tax reasons. You want to buy loss-making companies because you think that they are badly managed and they are worth something but it's an incentive because you know that you can deduct the losses against your own taxes, especially if you are in a system that consolidates the prots at the level of the group. Specic Adjustments The production costs that are tax deductible may or may not include (depending on the countries) all the administration costs and distribution expenses. The inventory valuation is also important for taxation because the value of your inventory, every time you sell something you need to give it a value, and you may have produced things at dierent moments in time, it's in your inventory, you produced a can of coke and at the time it costs you 10 cents and then you produce one at another time and the price of aluminium has increased so the cost was 12 cents etc. The production costs vary but all your cans are in your inventory. When you are selling one, you need to decide in accounting which one you actually sold. The valuation techniques are mainly FIFO (rst-in, rst-out) or LIFO (last-in, rst-out). This doesn't really matter in the long-run, what's important is that you're consistent in your methodology. Some countries allow methods, some others don't. In terms of amortisation and depreciation, in the 16 example of Coca Cola earlier it was a straight line depreciation meaning it was a xed percentage every year. There is also a declining balance depreciation/amortization where you have an acceleration in the beginning. For example, the rst year you can deduct 40% but it's always 40% of the remaining value so the second year it would be 40% of the remaining 60% that's left after the rst year, and so on... Immediate expensing is where you immediately deduct the full amount of your investment and it's usually something that is allowed for low cost assets. Goodwill, all these intangibles, and IP assets may or may not (depending on the tax system) be depreciable. For example, you are an insurance company and you want to buy another insurance company. You assess the value of the company you want to buy, in the end there are two values. You have the real value of the company (the buildings, the number of contracts, the premiums you expect to receive, etc.) but you may be willing to pay more because you know that it's actually a good set of customers and you think those customers may buy in addition some of your products that you have and you can capitalize on this (synergies available). You're willing to pay more for that company and the dierence between the value of what you buy and the price you paid is the goodwill. It's a cost for you, like an investment, some countries allow you to depreciate it and some countries don't. There are a lot of dierences across countries that makes harmonization a bit dicult. 2.2 Double-Taxation Relief Now we will introduce the international tax system and the way it works. The international tax system as we know it today dates from 1928: The League of Nations. The League of Nations commissioned a working group led by a Belgian from the tax administration to devise international tax rules. The year 1928 was a time where economic structures were brick and mortars, it was the beginning of television, it was a time when a lot of the things we know today (communication means, IT, internet, digital, etc.) did not exist. That creates a lot of diculties between what the tax system is and the reality of the world. In 1928, one of the things that the group already recognized is that tax systems were already so dierent between countries that it was dicult to think about harmonization. They also recognized that it was dicult to have multilateral tax treaties, where you would harmonize the way you do things because there was a variety of systems that dier and it was already dicult to have one rule that ts all. So they said countries should deal with their disputes or relationship in a bilateral way via bilateral tax treaties. For example, Belgium has a tax treaty with France that may be dierent than its tax treaty with Germany which will be dierent between the tax treaty between the Germans and Danes, etc. Those tax treaties usually look at a certain number of taxes, mostly direct taxes (corporate taxation and personal income taxation, notably the revenues of labor). For example, you're French and come every day to work in Belgium, the question is: who should tax you? Is it France (where you are a resident) or is it Belgium (where you are working)? The tax treaty should provide an answer to that, otherwise both countries could decide to tax you. Property taxes has the same concept. For example, you are a British citizen and have a house in the south of Spain. The house is your property so probably the British tax authorities may say since you are a British citizen working in the UK so they should tax you on the revenues that are derived from the house in Spain. Spain might say the house is in their country so they are the ones that should tax. Once again, the answer should be in the tax treaty. Inheritance taxes has the same reasoning. For example, you are a French citizen retiring in Portugal. You've been living in Portugal for a number of years and you die. You have a French citizenship but you were residing in Portugal so according to which law should we deal with for inheritance taxation? Do we apply the French system or the Portuguese system? Answer: look at the tax treaty. We can see immediately that some cases may be complex and there has been a need to ensure a certain system to avoid double taxation: that citizens or companies don't get taxed twice. That's notably what's been implemented in corporate taxation, as an example, but the same is there for other taxes, notably the nancial ows. For employment revenues, usually there is a rule that will say who taxes and that's it. Usually it's the place where you work, it's not your place of residence (generally). For property taxation, it's usually the place where your property is located (German owning a house in Belgium that you rent out, revenues from the rent should be taxable in Belgium and not in Germany). 17 Let's take a look at the following example. Figure 23: Hypothetical Double Taxation Example You have a parent company in country P that has a subsidiary in country S and the subsidiary makes a prot. At some stage they want to distribute this prot to the owner, which is the parent in country P, in the form of a dividend. In this example, they have made a prot of 100 and they want to redistribute this prot to the parent. What happens is that country P is a high tax country with a tax rate of 30% and country S has a tax rate of 25%. The dividend ow will go from the country of the subsidiary where the 100 is generated and possibly taxed to the parent where the tax authorities will have to decide what they will do with the 100. There may be also, on top of the taxation in country P and S, a withholding tax but assume for simplicity that it is zero in this example. Let's rst see the case of no relief, meaning there is no tax treaty. No Relief Figure 24: No Relief There is a prot generated in the subsidiary before tax. The tax authorities in the subsidiary will decide to tax this 100 prot at the CIT rate of the subsidiary country of 25% which results in a prot after tax of 75. The subsidiary wants to send their after tax prot to the parent in the form of a dividend. They are sending this dividend of 75 to the parent country. The tax authority in the parent country will have a look at it and will say that company has received a dividend net of tax of 75 but they know it's coming from a prot of 100. The tax authorities will gross it back up to the amount of origin and will apply their own tax rate of 30%. So there is the grossed-up dividend of 100 on which we will apply the CIT rate of the parent country of 30%, which results in 30 in taxes for country P. The subsidiary pays 25, the parent pays 30, the remaining prot after all the taxes is 45, resulting in a total tax paid of 55%. This is the case of no relief and there is double taxation. Intercompany Dividends There are a number of ways to avoid this situation. There is a method called tax credits, so you give credits for the tax that has already been paid. This credit can be direct or indirect. Indirect credit is when you have relief for the corporate tax and the withholding tax that has been paid and the direct credit is only relief for the withholding tax. You have another method to avoid this situation called exemption, which is widely applied in the EU, notably there is the P/S Directive (the parent/subsidiary 18 directive) that says if you have a dividend ow between two companies that are related (25% threshold of ownership), this dividend ow is exempted from taxation in the receiving country. The last system is deduction which is a bit less interesting. We will now go through examples explaining these concepts. Exemption Figure 25: Exemption This is the same situation as before except now there is an exemption at the level of the parent company/country. The tax authorities of the parent say their double tax relief system will be the exemption system. What happens is shown in gure 25. Nothing changes in the subsidiary country S, the prot will still be taxed at 25%. The 75 that will be available for distribution is sent to the parent. The parent does the gross-up of 100 but that's it. The tax authority in country P exempt the incoming dividend so they don't levy any tax. The total prot after tax is 75, resulting in a total tax paid of 25%. Therefore, country P is applying a territorial system since they think the prots should only be taxed in country S - the source country. Summary: territorial system - exemption - receiving country exempts the dividend because they consider it has already been taxed in the subsidiary country. Tax Credit Figure 26: Tax Credit Figure 26 shows the same situation as before except now there is a tax credit at the level of the parent company/country. In this case, nothing changes in the country of the subsidiary. The subsidiary makes a prot of 100, gets taxed at the CIT rate of country S of 25% which results in an after tax prot of 75. The 75 will be distributed to the parent company in country P in the form of a dividend (no withholding tax). The parent country does the gross-up so the amount goes back to 100. The tax authority of country P say they will tax at the CIT rate of country P of 30% on the grossed-up dividend but they will give a tax credit for what they have paid abroad. A tax rate of 30% means the parent company owes 30, but since they already paid 25 in the subsidiary country S, they receive a tax credit and only has to pay 5 in country P. This means the after tax prot is 70 and the total tax paid is 30%. The tax rate at the end is the tax rate of the parent and the parent country applies a worldwide system. The tax authorities in country P are taxing their companies on a worldwide basis, the prot generated in the subsidiary country will be taxed (they will give a tax credit but will still tax it). 19 Now what happens if the tax rate is higher in the subsidiary country S than the parent country P? For example: What if the tax rate in the subsidiary country S was 50? We would be in a situation where the subsidiary would pay 50 in taxes to the country S. The after tax prot of 50 would be distributed as a dividend to the parent in country P. The tax authorities in country P gross it up to 100 again. They say they owe 30 in taxes but they know the subsidiary has already paid 50. In this case, that's it. There is never any country that will accept to pay a company back on the taxes that has been paid abroad. This situation is called being in excess foreign tax credit. If you are in this situation, in a sense you break the logic of the system. This system being: the ultimate rate would be the tax rate of the parent because it's a worldwide system. In the previous example, the ultimate tax rate was 30% which is the rate of the parent, which is true for a tax credit system only as long as the tax paid abroad is lower than the tax due at home. That was often the case with the US tax system, the US had a worldwide system. They had a very high statutory rate, the federal rate was 35% (one of the highest in the OECD countries) plus the state taxation which vary from state to state (from 0-15%). It was not uncommon in the US to pay corporate income tax of 40% which was, in the majority of cases, much higher than what companies would pay abroad. To avoid this, they were going to countries like Bermuda with a tax rate of 0% and they didn't distribute the prots. They deferred the transfer of dividends back to the parent, so there was an accumulation of prots of US nationals abroad because of the tax system. When the US changed to a territorial system last year, they had a one time tax on repatriation of 8-10% so they had a massive repatriation of prots back to the US economies. It's also what drove the switch since all this money is parked abroad, it could actually come back to the US economy and boost the economy. In the example of being in excess foreign tax credit, you have dierent situations (mostly had since there are very few countries that apply the worldwide system). They could say there is a limit per transaction or per company or you could actually mix with other transactions. For example, you mix with dierent types of countries: the excess foreign tax credit that you had from one country can be recuperated on the tax credit that you would have on other countries. There is a blending that could be done per type of transaction, per type of countries, etc. The tax authorities have dierent systems, it's not necessarily that it's only bilateral. For example, if this parent had a subsidiary in a high tax country and in a low tax country, they could try to ask the tax authority to mix the tax credit so that in the end they could recuperate the totality - the excess foreign tax credit on the one hand could be balanced with a tax credit that is available somewhere else. Deduction Figure 27: Deduction The last system that exists (which is not used often) is the deduction system. Same situation once again at the level of subsidiary, nothing new happens. It's a tax relief given by the parent country P, not the subsidiary country S. The company pays 25 in taxes at the subsidiary level and wants to repatriate the remaining 75 as a dividend to the parent company. Once again, in country P, the dividend is grossed-up back to 100. The tax authority gives a deduction: the tax paid by the subsidiary in Country S is deducted from the grossed-up dividend which leaves the taxable dividend of 75 in country P. The CIT rate in country P of 30% is then applied to the taxable dividend which result in a tax payment of 22.5 by the parent corporation. The remaining prot is 52.5 and the total tax paid is 47.5%. The tax 20 is still high and there is still double taxation. The International Tax System If you dene the following variables: • ti is corporate tax rate in subsidiary country i • tp is corporate tax rate in parent country p; • wie • τi is dividend witholding tax rate in country i is eective (two country) tax rate and you apply the following situations: • Exemption • Indirect foreign tax credit for corporate taxes and withholding taxes • Direct foreign tax credit withholding taxes only • Deduction of foreign taxes • No relief You can actually come up with the following formulas (apparently from a paper that the professor is not asking us to read/look at so not sure how important this is): • Exemption ti + wie − ti wie • Indirect foreign tax credit for corporate taxes and withholding taxes max[tp , ti + wie − ti wie ] • Direct foreign tax credit withholding taxes only ti + (1 − ti )max[tp , wie ] • Deduction of foreign taxes 1 − (1 − ti )(1 − wie )(1 − tp ) • No relief ti + wie − ti wie + tp These formulas are used by the law to determine what the eective tax rates are in each of the situations. CEN and CIN Now we will discuss two important concepts: capital export neutrality and capital import neutrality. These concepts are used quite often in economics of taxation. What is capital export neutrality? There is capital export neutrality when coming from the par- ent country, you want to do business in a subsidiary, and you should be indierent vis-a-vis the tax system whether you invest in a country S1 or a country S2 . Figure 28 shows a possible tax situation used to describe capital export neutrality. You are a par- ent corporation with a rate of 30%, you have a subsidiary S1 and a subsidiary S2 . S1 has a tax rate of 10% and S2 has a tax rate of 20%. You invest in those and they will send you back a dividend. You have capital export neutrality if you are completely indierent between investing in S1 or S2 , i.e. the 21 tax system will not aect your choices. Now, consider an example where you have an exemption system at the level of the parent in this situation. You would pay a 10% tax if you invested in S1 , you would be exempted afterwards at the parent level. You would pay a 20% tax if you invested in S2 and you would be exempted afterwards at the parent level. What is your choice if all else equal? You will invest in the country S1 with 10% tax rate because that's where it's the cheapest. Important summary: if you have an exemption system, you do not achieve capital export neutrality. On the contrary, consider an example where you have a tax credit system. What happens? If you invest in S1 , you pay 10% tax, the dividend ows back to the parent and you pay 30% but receive a tax credit of the 10% that you already paid. If you invest in S2 , you pay 20% tax, the dividend ows back to the parent and you pay 30% but receive a tax credit of the 20% that you already paid. In both systems, as long as you're not in excess foreign tax credit, the amount that you pay in total will be the same which is the parent tax rate of 30%. A worldwide tax credit system ensures capital export neutrality. As an investor you will be indierent, all other things being equal, between investing in S1 where you pay 10% and an additional 20% at the parent level and investing in S2 where you pay 20% and an additional 10% at the parent level. Figure 28: Capital Export Neutrality Figure 29: Capital Import Neutrality What is capital import neutrality? It means that if you invest in a country S, it should have the same tax burden as a domestic investment. Figure 29 shows a situation used to describe capital import neutrality. You have two parent cor- porations: P1 and P2 and you have a potential subsidiary S. Both parents invest in the subsidiary (50/50) and the subsidiary sends back a dividend. There is a rate of 25% at the subsidiary level, 30% for P1 and 50% for P2 . If you have a system of tax credits, the investment in the same company will be more expensive for P2 than for P1 . This is because P1 will pay 25% at the subsidiary level and an additional 5% at the parent level (30% in total) while P2 will pay 25% at the subsidiary level and an additional 25% at the parent level (50% in total). Under the worldwide tax credit system (as long as you're not in excess foreign tax credit), the tax rate is that of the parent: it means that there is a disadvantage for parent companies from high tax countries. The same investment in the same company will be more expensive from a tax point of view for P2 than for P1 . If, on the contrary, you have an exemption system for both companies: they will pay 25% in the source country and the dividend will be repatriated and exempted in both countries. This, therefore, insures capital import neutrality from the perspective of that subsidiary. Past Exam Question (QCM) Does a worldwide tax system ensure capital export neutrality? (a) Always (b) Sometimes (c) Never Solution on page 24. 22 These are important concepts that we use quite often in economics. Usually, we have a tendency that tax systems go towards capital import neutrality even though economists prefer capital export neutrality. Capital export neutrality is a concept that is much closer to eciency for economists, it's the one that ensures eciency in production (Diamond/Mirrlees Eciency Theorem). In the absence of international rules, the system has drifted progressively to an exemption system. Part of the international discussions is to go back to a system that would be closer to the credit system and capital export neutrality. Methods in Force We have ve potential systems for tax relief: direct tax credit, indirect tax credit, exemption, deduction, or no relief. They are determined by a tax treaty, for example the tax treaty between France and Belgium might say we apply exemption but within the treaty may be conditions. One of those conditions could be a certain holding threshold. For example, the tax treaty would say we apply exemption but only between companies that have a sucient relationship (holding threshold), so those companies that own more than 20%. If it's between companies that hold below this threshold, we apply something else (could be deduction, no relief, etc.). Usually, when it has to do with qualied shareholders (means that it's a parent that holds a substantial share in the subsidiary) and they receive a dividend, it's considered as an internal transaction and countries oer exemptions or a tax credit. When it's not a qualied shareholder, sometimes it's exemption, sometimes it's credit, but it can also be no relief or deduction system. Tax Treaties There are model tax treaties. The OECD has a model tax treaty that is more of a framework on which most countries actually align themselves but the interpretation of those tax treaties may actually dier across countries. EU Legislation EU legislation has two important directives. The parent-subsidiary directive says that there is an exemption and no withholding tax on dividends that are paid from a subsidiary to a qualied shareholder parent. We have implemented this system that is also based on freedom of movement of capital (one of the four freedoms in the EU) and to ensure that there is a compulsory exemption system between companies and no withholding tax on dividends. We have something similar to this for interest and royalties - interest and royalties directive. This is a taxable income but there is no withholding tax that can be applied. A summary of this can be found in gure 30. Figure 30: EU Legislation Summary There is a way to shift prots within groups by paying interest from high tax countries to low tax countries because you gain the dierence in tax rates (same thing for royalties). This will be seen more in detail later on in the course. Contractual Relationships Group taxation: some countries can say they tax the company individually or based on the consolidated prots of the group (Belgium has started applying this). For example, there is AB InBev and imagine they have another subsidiary in Belgium called InBev Logistics that distributes the kegs. InBev logistics is a dierent legal entity. Belgium can say they looked at prot of InBev Logistics independently and tax InBev Logistics. Then, they also looked at AB InBev independently and tax AB InBev. The other possibility is that they may decide to consolidate them, cancel all the internal transactions, and tax based on the consolidated prots of the group. 23 Another important point here is the application of anti-avoidance rules (We will see it in more detail later in the course). If we go back to the tax credit example earlier, there was a tendency for companies to park their prots in the subsidiary if it's a low tax country (imagine Bermuda 0% tax, keep all the prots there and don't distribute). US tax authorities get nothing because there is no prot distributed. You have anti-avoidance rules that you can apply in the form of what we call a CFC rule (controlled foreign corporation). The US tax authority will say there is some prot that is made in the subsidiary abroad and that has been lightly taxed and not repatriated. They know that this prot belongs to, ultimately, the parent company that is active in their country. What they will do is apply a rule that says that if the taxation in the foreign country is lower than a certain threshold (usually if the tax rate in that country is lower than 50% of their own tax rate), they apply the CFC rule meaning they do as if the prot had actually been generated in their country. They disregard that it hasn't been distributed, they assume it's taxable prot in their own country and tax it at their own rate. They will give a credit for the tax that has possibly already been paid but apply the CFC rule to avoid companies trying to park the prots. The US had this rule (there is a dierent system now called GILTI - pronounced guilty) but what happened is that there were loopholes in the tax system. There was a system called check-the-box which was a loophole in the legislation where if the company in the tax statement was checking a box, they were exempt from the CFC legislation. Solution to Past Exam Question on Page 22 Answer: (B) Sometimes. What is a worldwide tax system? It's a system of credit ensuring capital export neutrality be- cause it's always the tax rate of the parent that counts wherever you invest AS LONG AS you are not in excess foreign tax credit. There is an exception where if the tax rate in the subsidiary country is higher than the parent country, you will be in excess foreign tax credit and you don't have capital export neutrality. The answer is sometimes. 24 3 Lecture 3 - Corporate Income Taxation: Domestic Distortions In this lecture, we continue the discussion of corporate taxation on two grounds. The rst one is why do we tax companies? Companies are abstract concepts. You always have a personal shareholder behind every company. Even if it's a company owned by another one, there is an individual that you can tax at the end of the chain. So why do we tax companies if we could tax the shareholder? The second question is linked to why do we have CIT given that we collect very little as a percentage of GDP (2.5%-3% of GDP on average in the EU)? Corporate tax is not a major tax in terms of revenue raising. Given all these diculties, we may want to reect on the type of distortions this type of taxation can create. Why Do We Have CIT? We see that CIT distorts a lot of choices. It will distort the choice of where you locate yourself, it will distort how much you will invest, it results in prot shifting, so it's not necessarily an ecient way of taxing. We will see some models of tax competition later during the course. What those models are saying is that under the hypothesis that capital is mobile, the optimal tax is zero. The predic- tion of those models: for eciency reasons, if you tax capital it moves away so we should not tax it. Therefore, why do we tax corporations? is a legitimate question. There are a couple of answers to this and there are not actually many reasons, for economists, on why to tax corporations. There are a couple fundamental reasons but there are not a lot compared to the many distortions that CIT creates. One element of the answer that you can bring into the debate is to consider that these companies are an entity by themselves (if each company enjoys limited liability) that consume several goods that are supplied by the government: use of roads, an educated workforce, rule of law, respect of contracts, IP rights, etc.. If we think about the benet principle, it's actually normal that those companies would be taxed because that would be the compensation for the type of services that they receive. Indeed, someone needs to be taxed for those services that are provided as those services help the companies make prots. This prot should be taxed but that does not necessarily solve the question of whether those prots should be taxed at the company level or taxed directly through the owners of the company. If you want a pass-through approach or transparency approach, the owner of the company (if the company makes a prot) will receive dividends or will benet from a price increase of their shares (due to retained earnings) that could be taxed via capital gains. You could tax people through those means instead of taxing the company. Another element you could bring to the debate, which is more of a strategic one, is that when you tax companies, you are not only taxing domestic companies but also foreign companies: companies that are owned by foreigners. Belgium could say they want to tax companies because they can tax a brewer that makes beer that is owned by Brazilians. Therefore, they do not raise taxes on their own citizens, they are raising taxes on the foreign owners. That is the tax exporting argument. We will be shown later on some research that has been done on this that tends to accredit that something like this may actually happen (the Wildasin model). Main important one: There is a more fundamental argument that says that corporate income taxation is a backstop for personal income taxation. It's a prepayment for personal income taxation that should normally be levied on the owner of the company. Why would a government want to do that? Imagine a world that does not tax companies but only individuals. What would you do as an individual? You would incorporate, you would create a company. You create a company, you enjoy a company car, you enjoy many other services provided by the company, you pay yourself a low salary but you don't pay taxes. To avoid the type of system where we all incorporate, there should be a corporate income tax. It's sort of an anti-avoidance element, it's a rule that makes sure that the prot 25 is taxed at least somewhere. In economic terms, that's the main argument for economists as to why we should tax corporations. There is another element of to) as labour. political economy: you should tax capital on the same level (or close A large part of the population would see it as unfair if you would tax labor but not corporations. It would be seen as capitalists vs the workers. 3.1 Distortions in the Choice of How to Do Business: Taxation and Incorporation De Mooij and the professor were intrigued by the question: does corporation taxation and does the dierence between CIT and PIT lead to a specic choice of the way you do business? What they looked at is a dataset from Eurostat that looked at data for 17 countries in 60 sectors during seven years. The data looked at the way people do business when they create a new company. Eurostat distinguishes three forms of companies: limited liability (LL paying CIT), sole proprietorship (SP paying PIT), and partnerships (PA - depends on country). The researchers did a regression that tried to see whether the rate of incorporation CORP1 = LL (LL + SP ) can be explained by the following regression: CORP = β0 + β1 (Tp − Tc ) + β2 X + They tried to estimate β1 which is the coecient of the dierence of what you would pay if you pay PIT and what you would pay if you were paying CIT. They tried to see whether this dierence could explain the rate of incorporation. You can see that the dierence is always positive and always highly signicant. This means that this dierence in tax rate has an explanatory power to explain the incorporation rate of new businesses. They did it by sectors, with dierent types of regression (structure, cluster the errors), with dierent concepts for incorporation, etc. but whatever they did, they still had the positive and signicant coecient. They found that the semi-elasticity of the tax base is at the level of 1.0 for employment (more on this in a bit). The main result is that when you reduce CIT by one euro (in terms of the amount of taxes that you pay) you have a part that you regain by the incorporation element (24 euro cents), which is what is called income shifting. That is: if the government decides to decrease its corporate income tax to promote something (for example, entrepreneurship), they will lose a certain amount due to the reduction in the rate of corporate tax but they will regain part of this because people will actually start incorporating so you have an increase in the number of companies. This is not a net eect. It means that it's actually people who used to pay PIT (at potentially higher rates) that now shift to corporate sector and are paying lower CIT. 26 Figure 31: Dierence Between Personal Income and Reduced Corporate Tax Figure 31 demonstrates the gap between PIT and CIT from 1990 to 2006. You can see that this gap is on average 17%, historically. It means that the top personal income tax rate is 17 percentage points higher than the corporate income tax rate. This increase in the dierence can explain the expansion of the corporate sector in the economy. People tend to incorporate. For example, in Belgium, at some stage you had the possibility to do a one person company where you could incorporate and have limited liability. This was one of the elements where corporate taxation will distort your choices: the choice of how to do business. If you want an ecient system (remember the rst lecture: DWL), you already see that CIT creates distortions. As soon as a tax distorts some economic choices, it leads to a distortion that will lead to a deadweight loss that will lead to an economic loss in terms of welfare. Now onto the second element of distortion. 3.2 Distortions in the Choice of Financing: Debt-Equity Bias The second element of distortion is the choice of nancing your activities. What can you use to nance your investments? Three choices: debt, equity, or use your retained earnings. The problem is the following: the tax systems have not been designed by economists. They have been designed by legislators. When a company needs to pay the creditors or the nancing providers, on the one hand they will pay an interest when they have debt, on the other hand they will pay a dividend (assume it's a dividend) to the shareholders. You need to remunerate the people who lend you some money, otherwise they would not lend you money and invest. Traditionally, what the legislator said was when you pay an interest on your debt, it's like a cost of doing business. Like any other cost of doing business, it should be tax deductible. When you derive the prots from your company: Sales - Cost of Materials - Wages & Compensation - Depreciation - Interest Paid = Taxable prot Tax = Net Prot > can either be Dividend or Retained Earnings The legislator has said that the interest you pay is tax deductible because its a cost of doing business but the dividend you pay shareholders is not a cost of doing business: it's just remuneration for the shareholders so it should not be tax deductible. It comes after you've been taxed. This is creating a distortion. If you nance through debt, it's tax deductible. If you nance through equity, it's not tax deductible. This is quite a large distortion and it creates the corporate debt bias. It means that companies, in their choice, will tend to be over-indebted. They will prefer to use debt compared to equity because it's reducing their tax burden. It creates a system of companies that are over-indebted which makes them more fragile in cases of adverse economic conditions. This is because if you don't want to provide a dividend to shareholders because you're making losses then so be it but if you tell a lender that you cannot pay back the interest, they may seize your assets. You have to pay back debt, it's compulsory. You don't necessarily have to distribute dividends, it's a policy decision of the company. There has been a movement in recent years to try and address this debt bias, notably after the nancial crisis. The problem is exactly the same for the nancial sector except for the nancial sector, 27 the rate of indebtedness (it's their business model) is much higher than non-nancial companies. The EU Commission has done some simulations with a model called SYMBOL that shows: if you could reduce the debt bias in corporate taxation, you would reduce the indebtedness of nancial companies and you would reduce both the risk that they go bankrupt (there is a cascading eect that we've seen during the crisis) and also the cost of repairing this. At the end of the day, it's the state that needs to put the additional money when there is no money left to pay for the losses. They've tried to devise a tax system that would move away from the traditional system of deductibility of interest payment/non deductibility of dividends and to try to change the system so that it addresses this debt bias: either through limited reforms (anti-avoidance rules) or more fundamental reforms (reshape the tax system). We will discuss the possibilities. There is actually no valid economic reasons as to why we should give a deduction for interest and not for dividends. For an economist, those are two ways of nancing a company and if you want to avoid a distortion in the choice of the nancing instrument, then you should have an equal treatment from a tax perspective. Even if you look at the legal aspect, there is no characteristics that are intrinsic to either dividends or interest that would call for a dierentiated tax treatment. They are dierent, one is compulsory, one is not; one is xed, the other is not; they have dierent terms: debt is xed term, equity is not. None of those characteristics could explain why we should have dierentiated tax treatment. Actually, with the emergence of hybrid instruments, it's more and more dicult in the legislation to determine what is considered a debt and what is considered equity. Take the example of convertible bonds, it's a bond so you pay interest and at some stage there could be a decision that converts it into shares and start paying dividends. You see that this example is blurring the frontier between what is debt and what is equity (could go even further with nancial derivatives). Empirical Evidence There is empirical evidence that this debt bias is aecting the indebtedness of companies. Notably, De Mooij (2011) did a meta analysis of 267 regressions and found that we have a mean tax elasticity of 0.65. A meta analysis is the following. You have a lot of papers in the literature that looks at something. For example, in this case, you try to explain the debt-to-equity ratio (a measure of indebtedness). You try to explain it for a certain country at time t via some coecients and a set of variables i, t (for example, size of company, maturity of the company, etc.) and your variable of interest is the tax rate in this case (T) plus the error term. You have a lot of studies that look at this and nd that indeed the higher the tax rate, the higher the debt level. There is a positive D E i,t = α + β1 Xi,t + βˆ2 Ti,t + i,t βÌ‚ : beta estimated (positive coecient). where βˆ2 is greater than 0 This is because the higher the tax rate, the higher the rate against which you can deduct interest, so the higher the tax advantage. If you deduct your interest against a tax rate of 5%, you pay 100 of interest, your tax advantage is just 5. If you can deduct it against 50%, your tax advantage is 50. What we call the debt-shield is much higher in the second case. You have a positive relationship between the level of the rate and the level of indebtedness of the companies in your country. Meaning countries that have high corporate tax rates are also those where there is more incentive to use debt than to use equity (more incentive to decrease the tax burden). With meta analysis, you will take all the regressions in papers doing this and try to explain the results that is found by the characteristics of the study. β2s = γ + θXs + s They are trying to explain βˆ2 (β2 estimated) by a set of variables. For example, the study was maybe using debt-to-equity as the denition (or not) or it was using debt-to-asset which would give you a dierent beta. You try to clean the estimated coecient that you have by the characteristics of the studies. You try to look at what could explain this beta (when you remove everything), sort of the real average eect that you nd in all the studies. This is what we do in a meta analysis. The mean tax elasticity means the following. If we say the tax elasticity of debt-to-equity is 0.65 it 28 means: if you increase your tax by 1%, there will be an increase in the debt-to-equity ratio by 0.65%. Here our tax went from 20% to 20.2% and the debt-to-equity ratio went from 100 to 100.65% (for example). ∆D/E D/E T ED/E = 0.65 = ∆T T so ∆D/E D/E = 0.65% and ∆T T = 1% If the tax rate is at 20%, and increase it by 1%, it goes from 20% to 20.2%. If we increase it to 21% then it's an increase of 1 percentage point. That's another concept called semi-elasticity which is the following. For example, if we say the tax semi-elasticity of debt-to-equity is 1.2, it means: if we increase the tax rate by 1 percentage point, we have an increase in the debt-to-equity ratio by 1.2%. So here our tax rate went from 20% to 21% and the debt to equity went from 100 to 101.2% for example. T SED/E = 1.2 = ∆D/E D/E so ∆T ∆D/E D/E = 1.2% and ∆T = 1pp Don't forget what we are talking about: elasticities and semi-elasticities are not the same thing. There are so many mistakes, even in the profession, of people who misinterpret these two. Also worth a reminder, the following is an example of the marginal eect: ME = ∆D/E where ∆T ∆D/E = 2pp and ∆T = 1pp For example, we increase the tax by one percentage point (from 20% to 21%) which results in an increase our debt-to-equity ratio by 2 percentage points (from 100 to 102). This is the marginal eect. When we read economic papers, people may talk about semi-elasticities, they may talk about elasticities, they may talk about marginal eects, and they are all very dierent concepts. There is a relationship between them because the semi-elasticities, elasticities, and marginal eects that we estimate are usually estimated at the average of the central. We nd the estimation in our regression, usually it's done at the average value so it's the average T and the average D/E. EÌ‚ = ∆D/E D/E ∆T TÌ„ If we know the values from the study (usually have a table of summary statistics), we can usually work our way back. Back to the empirical evidence. Feld et al (2013) found a marginal eect of 0.3. This means that when you increase your tax by one percentage point (from 25% to 26%), you have an increase in the debt-to-equity ratio by 0.3 percentage point (from 60 to 60.3). What is interesting is that this type of study has also been done for nancial institution, by Keen and de Mooij (2011) and Hemmelgarn and Tecihnmann (2013). You would think that nancial institutions are dierent bodies because they are heavily regulated. Normally, they are subject to capital asset ratio requirement, they should have a minimum level of capital on their nancial statements (Basel rules). You would think that they would be less responsive to dierences in taxation because they have to have a certain amount of equity. When you look at the studies, you nd a similar size of response. The mean tax elasticity is the same for nancial institutions than for non-nancial institutions. The problem is that it's based on higher average level of leverage. Non-nancial companies usually have a debt-to-asset ratio of about 60% on average while banks are 80% on average (some much higher). You see that they are more risky institutions than non-nancial corporations. What should we do to deal with this debt bias? One option is to disallow interest deductibility: interest is not tax deductible anymore. It was a proposal made by the US Treasury in the 1990s called the comprehensive business income tax model (CBIT). The other option is to allow a dividend deductibility called allowance for corporate equity (ACE): we allow the deduction of interest paid 29 and, on top of this, we allow you to deduct a notional percentage of your equity (not a deduction of the whole dividend). For example, every year we allow you to deduct interest paid and 3% of your equity. Belgium had this system called the notional interest system (système d'intérêt notionnel) (also in Italy, Portugal, Turkey). These countries have/had this in place precisely aimed at dealing with the debt bias. It is to put equity nancing and debt nancing at par by allowing for a deduction of both debt nancing and equity nancing. We will now look at the economic eect and the notions we have behind this. Disallowing interest and dividends deductions altogether or allowing a deduction does not have the same impact on the eciency or economics of the company. Figure 32: Normal Return and Economic Prot When you have a company, you have your revenues. You try and derive your prot by deducting the cost of materials, the cost of labour, depreciation, the interest that is paid to your creditors, and you have the remaining which is your prot. This prot can be split into two parts: the minimum return required by shareholders (risk-free interest rate + the risk premium) and the economic prot/rent (everything that is above that). In the current system, what you do is tax a denition of the prot that derives from accounting that is larger than what economists consider as being the prot. Companies pay corporate taxes on economic prot plus an amount that shareholders actually require to continue to invest in the company. We have two concepts that are derived from this: we have the economic prot/rent and the normal return to capital. The current system is actually taxing both the rent and the normal return to capital. Remember that prot will be zero in perfect competition: the intuition being that if there are losses, companies exit; if there are prots, companies enter until there is a balance at zero. As long as there is an economic prot/rent to be made, companies will continue to do business and to invest. Taxing the economic rent does not change the incentives at all: companies continue investing in the market, continue doing business, continue locating in specic places. Taxing the economic rent is not distortive but taxing the normal return to capital is distortive. (This can be shown mathematically but he's not doing it so guess the proof isn't necessary). The story is that taxing the normal return to capital will increase the cost of capital of companies, will increase the marginal eective tax rate, and marginal average tax rates that those companies will pay. If you derive a system where you will only tax the economic rent, (once again can be shown mathematically but he's not doing it, just says to believe him) the consequence is that it will not be distortive and the marginal eective tax rates will be zero. That is, the tax rate that you pay on the last euro that you will invest will actually be zero; it will not distort the choices of the company. A notional interest system achieves this because a notional interest system will actually give a deduction for interest payment but also a deduction for a notional amount on equity. If we had chosen the other system, the CBIT, where we disallow the deductibility, you have a much larger prot to tax but you see that you tax the total return to capital. 30 A word on cost of production: you could see it another way. If you want to have a system of tax- ation, what you should tax from the point of view of economists is all the revenue that you generate via your activities minus all the costs that you incur while doing these activities. That's the denition of prot for economists. We have cost: the costs of materials (deductible), the cost of labour (deductible), the depreciation (deductible), the interest (deductible), but currently there is one cost of doing business that we disallow in terms of deductibility which is the minimum return required by shareholders. Financing costs, for economists, whether they come from debt or equity is a production cost. The following gures compare the taxable prot under under traditional corporate income tax (CIT), comprehensive business income tax (CBIT), and under allowance for corporate equity (ACE). You have the choice between those two systems if you want to address the debt-bias but they have a lot of dierences in terms of the incentives. On the one hand, you only tax the economic rent so it's not distortive, it will not aect the choices of the companies, it's more ecient, does not create deadweight losses. On the other hand, you will actually have a much larger denition of what is prot, your tax base is much larger, you could possibly decrease the rate if you want to raise the same amount of revenue, but you will aect company location and will aect the level of investment of the companies. Figure 33: Traditional CIT Figure 34: Taxation Under CBIT Figure 35: Taxation Under Ace One of the problems of Europe today, which is why we have a plan called Capital Market Union(CMU), is that equity markets are underdeveloped compared to the US. The US also has a debt bias but one of the elements of the CMU was indeed addressing the debt bias. That's why, in one of the proposals that was made at the EU to harmonize the way we compute corporate taxation, it included a system that looks a lot like notional interest because the EU commission thought it was something important to address with a correction for some of the avoidance techniques that we saw in Belgium. 3.3 Distortions via Lock-In in Size: Taxation of SMEs Now we will discuss the tax treatment of small to medium-size enterprises (SMEs). In many countries, SMEs are tax favoured. One typical way of giving tax incentive for SMEs is, for example, to give them reduced rates on corporate taxation. It's the case in Belgium, SMEs have a lower corporate tax rate than larger companies. It can also be an advantage given in terms of tax incentives. For example, you may have super deductibility if you are a small company on some depreciation or an R&D cost etc. to try and help the SMEs. It could also be a dierent type of reporting standards: if you are an SME, you have a way of reporting your prot to the tax authorities that is simplied compared to larger companies etc. What are SMEs? There is a denition at the European level. An SME is a company with less than 250 employees AND has a turnover below 50 million euros or a balance sheet that is below 43 million euros. You can distinguish between medium sized companies, small companies, and micro companies. Small companies are those with less than 50 employees AND less than or equal to 10m euros in turnover or less than or equal to 10million on the balance sheet. A micro company is a company with less than 10 employees AND less than or equal to 2m euros in turnover or less than or equal to 2m on the balance sheet. This is the advised denition by the EU but member states, notably in their tax systems, may apply other denitions. 31 How Important are SMEs? They are important economic actors. They constitute 99.8% of companies in Europe, 67% of employees and 58% of value-added (in 2014). The majority (93%) are micro companies. They are mainly active (3/4) in ve sectors: wholesale and retail trade, manufacturing, construction, business services, and accommodation and food services. Most Pressing Issues for SMEs SMEs have some issues (2014 SAFE Survey). 1. They have problems in nancing customers (20%) and availability of skilled sta or experienced managers (17%): If we (the students) are asked at our graduation whether we want to work for Coca Cola or the Big 4 or for the local plumber, our answer will be pretty obvious. 2. SME's complain about regulation (16%). 3. They have issues of competition (15%). 4. They have problems with access to nance (13%): if they go to a bank to borrow money, the bank will say what is your collateral? They want an asset to secure the loan. Usually if it's a small business, the person will provide their house but beyond that it's very dicult. Additionally, they barely have access to nancial markets. They have diculties to raise money, to invest and in particular have diculties in obtaining bank loans. Banks may be more reluctant to a small company that doesn't have a lot of business, doesn't have collateral compared to lending to a larger company. 5. They have cost of production or labour (12%) that is quite high: it is costly to hire an additional person. SMEs Receive Specic Treatment Policy makers have always considered that given their importance in terms of economic activity, given what is considered as being issues that tend to be unfair to them (notably the access to nance), and given the fact that SMEs vote compared to large companies (If they see that the policy of the government is against them, it's much more identiable in terms of who they will vote for than if they work for a large company) that we needed to think about giving them advantage in terms of the tax system: to help small companies be (at least) on equal par with the large ones. Instead of going into this and trying to address these distortions (for example, you could say they tried to have programs to have better name brand recognition for small companies or they tried to help SMEs have access to nancing via venture capital etc.), the decision in countries is often to add another distortion to correct the rst one which is to change the tax system in favour of the small companies. They often get tax incentives, often in terms of reduced corporate tax rate. It's not optimal because maybe the rst best policy would be to address the distortions in the rst place and not add a new one but as a second best option, it could be interesting to try and correct the distortion by having a tax system that tilts towards SMEs. Taxation of SMEs It's often seen as a good idea for politicians. There is a market failure in some part of access to nance. There is the political economy argument that SMEs vote so having them on your side is great if you're a politician. But there is also the issue of the incorporation eect: if you give more people incentives to be a small company, they will incorporate instead of staying self-employed or staying employees and paying personal income taxes. There is also the issue of bunching eects of tax measures that is not known to most policy makers. It's actually something that economists see a lot but policymakers fail to see this. Evidence Via Bunching Techniques The bunching techniques are relatively recent and there are a few economists worth noting that have looked into this. There is Emmanuel Saez (2010) and Chetty et al. (2011): (professor bets both will 32 get a nobel prize). Emmanuel Saez and Garbeil Zucman are currently involved in the US campaign: they have designed the economic program of Elizabeth Warren - the wealth tax. Two other economists who have popularized this technique is Henrik Kleven and Mazhar Waseem (2013). The idea that they have is that tax systems are not linear systems. When you have tax systems, you have what are called kink points and notch points. A kink point is described with the following example (gure 36). If you look at the tax you pay and how it varies with income: you will have the rst part which is untaxed, then you will have a certain percentage that is applied so the tax that you pay will increase. Then, you hit another threshold and it will increase again even more, etc. It's not something linear, at some stage you have dierences in rates that will make the slope of the curves change. Kink points: discrete changes in the slope of choice sets. Figure 37: Notch Point Figure 36: Kink Point Notch points are described with the following example (gure 37). The gure contains a graph of VAT reporting and sales. The tax system may say when your tax system reaches one million, you have to report. If it's below one million, you don't have to report. You have a change at some matrix of the system. Notch points: discrete changes in the level of choice sets. Kink point is a change in the rate while the notch point is a change of the system. The economists have been looking at administrative data with very large datasets. They have access to a wealth of datasets coming from the tax administrations. Empirical evidence in Public nance is in the lead on many grounds. They were the rst in the profession to have commercial access to databases of nancial statements. The professor was one of the rst to use the Amadeus database (now called Orbis). It was a breakthrough for research because for the rst time you could rely on rm level data instead of macroeconomic data. You had much more variation which means your estimates are more precise. The problem with those commercial databases is that they are geographically biased. There is a lot of reporting for Europe and the US, not for other countries. There is a lot of reporting for large companies, not for the smaller ones. Your sample is a bit biased. Also what you see as tax paid in Orbis or Amadeus is not necessarily the tax that has been paid. Sometimes it's the tax that should be paid that is booked for which there is a provision but it's not clear what actual tax is paid. The new breakthrough that happened is some tax authorities (notably the Nordic ones) have given access to administrative data. The big reason why they would not give it is anonymity: protection of the citizens. You should not be able to check the tax of your neighbour. They now try to provide anonymized data to researchers because it's a wealth of data that is even more precise. Bunching Techniques (Kleven, 2016) What they do with this is that they try to take those points and see whether the behaviour is changing at precisely when you have a kink or a notch and see what is happening at the margins of those changes in the tax system. An example is provided to explain the concept. Figure 38 has a graph of the number of companies and the size of the company. We have a system where we collect the data and see the actual distribution: we see something like the graph in gure 38. You can see something quite bizarre is happening. The distribution is a certain slope, changes, and then goes back to the normal slope. 33 Let's assume it happens at around one million in assets. For example, it could be that it's exactly the threshold that you have in the tax system for beneting for a reduced corporate tax rate: if you are below this you have a reduced tax rate and if you are above it you have the normal CIT rate. Then, they try to estimate the loss that occurs because of this by having an estimated distribution next to actual distribution. For the estimated distribution, we arbitrarily take out some part around the change and try to estimate what should be the normal distribution if we would keep the slope before and after the change. Figure 38: Bunching Example What you see is that compared to the simulated distribution, you have a number of companies below the threshold that is higher than what you would expect, and you have a number of companies that is lower than what you would expect. This is because companies want to keep the reduced CIT rate and they bunch next to the limit. You have what we call the excess bunching which is a measure of the number of companies or the size of the market that is abnormal (that is too low) compared to what it should be if you would have the normal distribution. Excess bunching: the dierence between what is observed and counterfactual bin counts in the bunching range. Figure 39: UK CIT Figure 40: Spain Auditing Figure 41: Finland VAT Figure 39 has the the UK CIT from 2001-2007: Brockmeyer (2013). You have the frequency of companies and the taxable income. There is a slope with a small jump, then you have a huge jump of companies at ten thousand pounds of taxable income followed by going back to the previous slope. An income of ten thousand pounds is exactly the threshold at which the rate is changing, where you benet from the reduced rate and you move to the normal CIT rate. You see that there is a huge number of companies that actually bunch right before the threshold. They try to avoid going over because it's not something progressive. It's not that you pay the reduced rate on the rst ten thousand and the normal above, it's a real change of rate on all your income. Figure 40 shows a similar story that can be observed in Spain (data from 1999-2007): 34 Almunia & Lopez-Rodgriguez (2014). You see a little bunche but the second one is a lot bigger. It's due to the eect of auditing, in this case. You have a dierent system of auditing for companies that are above the six million euros of revenues threshold. Companies, to avoid being subject to the more stringent auditing system, bunch right below the threshold. Figure 41 shows the Finland VAT threshold (data from 2000-2013): Harju, Matikka, & Rauhanen (2016). The threshold is set at zero here but it's in terms of the reporting of the VAT. You see that you have a distribution of the reporting that has a certain slope but right below the threshold (from 0 to -10), when compared to the estimated distribution, you have a large spike and missing companies just above. Those are companies that are in the 0 to -10 range that should be in the 0 to 30ish range if they would follow the normal distribution. There is bunching right before the threshold. Figure 42: Georgia CIT Figure 44: UK VAT Figure 43: UK CIT Example 2 Once again, gure 42 shows Georgia CIT: Bruhn & Loperick (2014). They look at the distribution of the number of companies and declared revenues. The distribution is decreasing, jumping up right before the threshold, and it falls back down. This is for 2008-2009. Figure 43 shows CIT 10,000 pounds and 300,000 pounds in the UK: Devereux & Loretz (2011). There are two thresholds: one at 10,000 pounds and one at 300,000 pounds. Again, big bunching at those levels. Figure 44 shows the VAT threshold in the UK from 2004-2010: Lockwood & Liu (2015). You see that there are many more companies below the threshold than right after the threshold. All of these are evidence for bunching techniques of companies. The problem that you have with systems that try to favor SMEs via the tax system is that you actually prevent those companies from growing. Those companies do not grow because they want to try to keep the benets of having a reduced rate or more favourable tax treatment. End of Slides From here, we were done with the slides but still had 15-20 minutes left. What remains of the lecture is not necessarily linked to any slides but it's still relevant to what was being discussed during the lecture. Student question: why don't we have some progressive laws or other laws to encourage companies to break this glass ceiling and become larger? It's a question to ask policy makers. The professor does not have the answer but it's a good point. A progressive system would actually be better than this so it's an interesting question. On another note, you have the same story for VAT. You have the normal VAT rate which is 21% in Belgium. You have restaurants complaining that the prot margin is very low and it's very dicult for them: they have diculties with hiring people, it's very costly, hence you have part of the profession 35 that has shifted to the black economy with people that do not declare etc. The way the government acted is that they gave a tax incentive to promote people going to the restaurant by oering them a reduced VAT rate. Therefore, in Belgium, if you go to a restaurant, instead of paying 21% (as it used to be) you will pay 12% on the drinks, 6% on your food, and 0% for take-away. Interesting sidenote: Take the example of Exki (restaurant), the rst thing they ask is whether you will eat-in or take-away. This is because the VAT they need to compute is changing, the price doesn't change because you are using the facilities etc. but because the VAT is dierent. If you buy the same items but request one for take-away and one for eat-in, you will see that the price is the same but the VAT is dierent. In one case, you will pay as a restaurant service at 6% and the other case you will pay as take-away food at 0%. Reduced VAT rates for restaurants results in restaurants being very happy. You may think it's a good idea because it's probably bringing more people to restaurants. It's arguably cheaper and it's also therefore creating employment. The professor looked at the data for France. The professor looked at the cost of the measure (the loss in VAT collection for the state) and divided it by the number of new jobs that the federation of hotel and restaurant claimed it had created. The cost per job created was 100,000 euros. It's pretty high for low qualication jobs (waiters, waitresses, etc.). If you wanted to help the sector, the smarter policy would have been, for example, to use the 100,000 euros to decrease social security contributions on the people that you hire. That would have been a much more ecient policy. It looks like a good idea from the start but when you start looking into it more, it's not very ecient. On top of this, the main demographic of restaurant goers are high-income people. You're giving a subsidy to rich people to go to the restaurant and those people are probably the least responsive to those incentives. They will not necessarily go more often just because it's cheaper (they have the means anyway). Why would you give a reduced VAT rate to someone who goes to a three star michelin restaurant to eat lobster? It doesn't really make sense. Same thing happens for a reduced rate on food, a zero percent rate on food in supermarkets. Great policy on paper because everybody needs to get access to food, it's a basic need. It works, some people do benet from it but who benets the most? If you look at consumption of food, it benets proportionally to low income people because a large part of their expenditure is on food but in absolute terms it benets the higher income side of the population because they are the ones, in absolute terms, that are buying much more food than all the others. The economic center of ministry of nance computed this and found that 60% of the tax advantage was actually going to the upper income side of the population. It's something you have to be aware of and you must be very prudent when there is something that looks like a good idea: think about the economics of it. Think like an economist! There was another case: tax incentives for energy ecient appliances. There are a lot of people at home who have very old fridges so we would like to give them an incentive to buy new ones of category A++ or A+++ because those are very ecient. We want to give an incentive to people so they switch from their current inecient fridge to the new one. What we will do is give them an incentive via a reduced VAT rate. The following gure demonstrates the example we are discussing. Figure 45: Refrigerator Example We give a reduced VAT rate for categories A++ and A+++; for the other ones, it's the normal VAT 36 rate. The price of the more ecient may decrease, this depends on the elasticities of supply and demand. Quick tangent demonstrating eect of elasticities in a similar case: to encourage people to have more ecient heating systems, the Belgium government wanted to incentivize people to do annual maintenance. The maintenance at the time cost 100-120 euros, they gave an incentive and said it would be tax deductible. What happened is that the price increased by the same amount because the supply was inelastic. The people doing the maintenance were getting all the benet because the supply was inelastic and they increased the price. Back to the fridges. Let's say you have an old fridge of 30l, you decide to take advantage of the tax incentive and upgrade to a more ecient refrigerator (A++ or A+++). You decide to not only upgrade to a more ecient fridge but also to a much bigger fridge because the price decreased there as well (60l). The problem is that the energy consumption of a 60l fridge is much higher than the energy consumption of a 30l fridge. For example, a B class fridge in terms of eciency has the same energy consumption as a 60l A++ fridge. This doesn't solve the energy consumption problem. On top of this, consumers most likely put the old fridge in their garage for extra storage. You have this policy that was there to address this environmental problem, people switched to bigger fridges, doubled the number of fridges at home, and you have a so-called rebound eect. You wanted to decrease energy consumption and, because of the bad design of the policy, you have a rebound in terms of energy consumption: people are actually consuming more energy than they used to. What should they have done? One policy is to only provide the rebate if you recycle the old one. The second policy is instead of basing themselves on articial classes (which by the way are not ocial classes - classes made by producers) is to look at actual energy consumption of the fridges. Something that looks like a wonderful idea turns out to be something that has the exact opposite eect of the intended one. Be extremely careful of these very simple good on paper ideas. Same thing with the rent. There is always a policy that tries to tax more housing: to tax land- lords more on the rent. It's a policy that could make sense because maybe property taxation is more ecient than other types of taxation. If you think of the renter market, the supply is pretty inelastic and so immediately any tax that will fall on the landlord will end up in the rent. The policy that you wanted to do to help with rent controls etc. end up being detrimental in the end for the people renting. You must always think: what is the economics behind the policy? What are the elasticities? Who will have the economic incidence of the tax? Remember the dierence between legal incidence and economic incidence. The legal incidence is who pays the tax according to the law. There is a property tax meaning the landlord pays the tax because the law says so. The economic incidence may be extremely dierent and the economic incidence is who at the end pays the tax and it can be that because the market supply of housing is inelastic, we have the possibility to shift this cost towards higher rents and economically this will be the people renting that will pay the tax increase that was legally on the landlord. 37 4 Lecture 4 - Corporate Income Taxation: International Distortions In this lecture, we will move from domestic distortions and instead talk about the international aspects of taxation. We've seen how corporate taxation works and some of the distortions corporate taxation may create at the domestic level. We've also seen the economic arguments for why we should tax corporations and not tax the owners directly. We'll continue going into the details of why corporate taxation is so important, notably why so many people are bothered with it given that it's a very small tax in terms of tax revenue collection. Just an example of the debates that is currently being held: the OECD has just released today (09/10/2019) the document explaining where they are in the negotiation to reform the international tax system so there are a lot of articles in the press on this current event. Why is corporate taxation important? It's because it has some spillovers and it creates distortions elsewhere in the economy. Policymakers can actually do things about that and can (at least) understand these aspects when they decide on corporate tax policy. 4.1 Devereux-Mani (2006) One way to see how corporate taxation works is to try to think about corporate decisions as a decision tree. In reality, all these decisions are made at more or less the same time. It's not sequential like in the upcoming decision tree but it's sort of a nice way to think about it and it's to try to understand not only by separating the issues but also for research. We will see that some economists are using the properties of a decision tree when they are doing research to see how each level will impact the next one. The decision tree by Devereux-Mani (2006) is demonstrated in the following gure. Figure 46: Decision Tree You have a multinational corporation (MNC) that makes a rst decision. The rst decision is whether to produce at home and export their products or whether they should produce abroad. For exam- ple, a US multinational must decide whether they want to stay in the US and export their products to Europe or want to locate their production somewhere in Europe. The second step is conditional on having made the choice to produce abroad: you need to decide the location of production. You have dierent options: A, B, C, D or a combination of them (a variety of possibilities). step is conditional on having chosen a location: choose the The third level of investment. For example, a MNC has chosen location A and C, they must now choose their level of investment. How big does the company want their investment to be so that they can produce at a level that ensures prot optimization. The fourth step is where do they relocate the prot articially. Does the company want their prot to be taxed in location A, C, or somewhere else? Where is the most interesting location for the company to locate their prots regardless of where they have their production? The company will try and use possibilities oered by the tax system and the international tax system to shift their prot from one location to another. For example, the company might not want to be taxed in A (assuming it's high-tax country) but wants their prot taxed in country C (assuming it's low-tax country). What we know from research is that taxation will aect some of those decisions. 38 What research has shown is that most studies that looked at this nd that step one is not inuenced by taxation (only marginally). The tax aspects do not necessarily come into the decision of whether you should produce at home or produce abroad. However, the step two is aected by taxation. It does not mean that taxation is the only determinant of where you locate. There are obviously other elements that you take into consideration that may be much more important in terms of eect. For example, the size of the market, the fact that the country is in Europe because you have access to a larger market without trade barriers, labour costs, infrastructure. There are many elements that could explain why a company chose location C instead of location A but taxation is one of those elements. The third step is also aected by taxation. This is because you want to invest a certain level that will maximize your prot. As long as your marginal prot on the last euro that you invest in this plant (for example) is higher than the marginal cost, you should continue to invest: up to the moment they equalize (MR = MC). You see immediately that in a world without taxes, you may equate the marginal product to the marginal cost but if you take into consideration corporate taxation, this will aect the balance between the marginal return and the marginal cost. You might invest less: you may actually invest at a level that is lower the higher the tax rate that you have on the last euro that you invest. The fourth step is also aected by taxation. This is obvious because if you do prot shifting for tax reasons, you want the prots that are ocially registered in a low tax country rather than a high tax country. The higher the dierence between the two locations, the higher the incentive to shift profits because the higher the gain. A lot of papers have used this decision tree (implicitly or explicitly) to try to separate those steps and to try to understand what are the elasticities to taxes of each of these decisions: how do taxes aect those three decisions of companies and which type of taxation indicator is important? For each of these steps, we are talking about corporate taxation and we will try to understand which type of tax indicator we must look at. Researchers have looked at dierent types of tax indicators to try to see which one explains best the decisions of companies. We will discuss step four now: reallocating prot among locations. In step four, the one that is found to be instrumental to the decision of where to locate your prot (where to make your prot appear/shift prot) is the statutory tax rate (STR). The statutory tax rate is the nominal rate: the rate that you have in the tax code. For example, it used to be 33.99% in Belgium, 35% in the US (plus the state rates), 33.3% in France. Why? You have a prot that is already there and the gain that you have is really a dierence in statutory rates. If you tax 100 euros at 50% or at 10%, the dierence is just the dierence in nominal rates. There is no other element that will come into play. Your prot has already been determined according to the rules of those locations and then you try to shift it somewhere else. The marginal gain is explained by a dierence in statutory rates. We will now discuss step two: choose a location. We try and wonder what type of tax indica- tor would best explain the location of companies. We know that the statutory rate will matter because where you locate is where you will be taxed but the way you compute the tax base is also important. If you have some additional incentive, for example: you have a higher depreciation rate for your assets or a special deduction for R&D, all of this will come as a deduction to your taxable prot and there will also be some advantages from that. We want a measure of taxation that combines the statutory rates and the computation of the tax base. Economists have come up with a measure of prot that is called the eective tax rate (ETR). What do we do to compute an eective tax rate? Figure 47 demonstrates an investment model. In this investment model, you invest one euro in some asset and nance this investment via debt/equity/or retained earnings. Remember the eect will be dierent: debt = interest is tax deductible while equity = no deduction. Then, we see what the return is without taxes on this investment. We assume a certain level of protability: we do an investment that has a protability of 20% (rÌ‚ ). Then, we do exactly the same investment with taxes so we have a 39 certain return (λÌ‚). Figure 47: Investment Model When you consider the taxes, your return is lower so λÌ‚ is lower. The proportional dierence is the eective tax rate on this investment. What was found during research is that on step two, the best predictor of location is the eective average tax rate (EATR). You have a number of investments that in general will be protability, and you x the level of protability arbitrarily at 20% (can be 10%, 15%, it's an assumption), and you have a certain return on these investments with and without taxes. It's supposed to mimic an average situation of a bunch of investments that you do in a country and that gives you, on average, the eective tax rate that those investments will carry. In a sense, it's a form of average taxation in the country on an investment but it derives, not from the actual data, from an investment model. Obviously, these eective rates are heavily correlated with the statutory rate because that's the one you imply into the model. You have an investment, you perform all the possibilities that the tax authorities allows for (deduct interest, super deduction, super allowance, etc.), you compute your tax base, you apply the statutory rate and nd what has been the return for your one euro investment. It depends on how your base has been computed but it also depends on the statutory rate that you applied. There is a correlation between the two. In general, when the statutory rate increases, the eective average tax rate will increase as well because that's the statutory rate that you imply into the model. Then, there is another family of eective tax rates that is found to be the best predictor for step three: the level of investment. This is the eective marginal tax rate (EMTR). It's exactly the same concept. You have an investment model, you choose a basket of assets in which you want to invest, you choose how you nance it, then instead of considering a protability of 20%, you will take the investment that will exactly break even where your additional return on this additional euro invested is equal to the marginal cost. You nd, for that specic investment, the one that breaks even, the one where the last euro that you invest is really the one that breaks even your investment, which tax rate that is applied to this last euro that you invested using the base and the statutory tax rate. It can be dierent from the average (usually is). It makes sense that it's the best predictor of investment because, given that you invest until the moment where the last euro that you investment exactly breaks even, you want a measure of eective taxation that is applied to an investment that exactly breaks even. On the one hand, you have something that is considered protable and you apply an arbitrary protability that is positive. On the other hand: For EMTR, you want to look at an investment that exactly breaks even so that you know what the eective rate is on the last euro that you invest and this will be the best determinant of investment because the higher the EMTR the lower your investment. The level of investment will be inuenced by the EMTR. These three concepts are quite important because we use them a lot in research and as indicators of the level of taxation. If you want to have an analysis or prediction of what will be the impact of a tax reform on the attractiveness of location for companies, you know that the one you want to use is the eective average tax rate. If you want to know the impact on investment, you know to use the eective marginal tax rate. If you want to know the impact on prot shifting, use the statutory rate. These indicators may tell a dierent story, at least in terms of the size of the impact because they change in the same way when the statutory rate changes but not with the same amplitude. Figure 48 40 provides a quick summary. Figure 48: Summary 4.2 Corporate Taxation and International Choices: Taxation and Business Location There are many empirical papers on the eect of taxation on business location. We will now discuss a paper by Devereux-Grith (JPubE, 1998) that has looked at this eect. It looked at a similar decision tree as the one discussed above. They looked at US MNC: these MNCs had the choice between investing in the US and exporting or locating themselves in three European countries: Germany, France, and the UK. What they tried to do with this model is to look at how taxation will impact the probability of a rm locating in a specic location. You have a special category of econometric model that you can use for this: the Logit model. It's a model that is used for categorical dependent variables. For example, if the variable you want to explain is not on a continuum: the size of the investment is a continuum (from 0 to innity) but the location is not a continuum, either you locate and takes a value 1 or you don't locate and it takes a value of 0. You need a special model to treat that because OLS econometrics will not be able to capture this. Figure 49: Logit Model Figure 49 shows the concept of a Logit model. You have all your observations and you have a measure of one on the tax rate (1/t). You want to see how a change in the tax rate will either result in locate (1) or not to locate (0). When taxation is very high, you may choose not to locate while when taxation very low you may choose to locate in the country which results in observations as shown in the gure above. As long as taxation decreases, you will see that you have more and more companies that will actually decide to locate in the specic location. If you apply a normal OLS to this (if you try to have a linear t), you would have something like the red line in the gure. You will have a regression and the parameters of your regression will say 1/t changes the probability of location. The problem is that as soon as you go over a certain value, your predicted outcome will be above 1 which conceptually is not possible: it's either 0 or 1. When it's below a certain value, you will have a negative predicted outcome which is also conceptually impossible. At the extremes, you have predicted values that are not in the normal set because you should have either a 0 or 1. Logit model has the other form (the type of curve) and it is better to predict the outcome of independent variables that are located in these extreme cases. You see at a certain stage it goes up immediately, doesn't take many of the parts in the middle, and it immediately goes to t either 0 or 1. This is the type of model that you use for this type of study. 41 They use a Logit model and what they found is that taxation does not aect the decision to invest in Europe compared to the US but the average eective tax rate will play a role on where to locate within Europe conditional on having decided to locate in Europe and not in the US. Conditional on Step 1, your decision on Step 2 will be inuenced by the average eective tax rate. Conditional because it's also something that you use in econometrics. You need to use that in the model, you don't consider the US as an option, it's sort of nested decisions in a sense: you have a rst nest, then you have another nest that is once you have decided to go into Europe, you rerun a conditional logit on the location and you will nd that the AETR is a good predictor of where to locate. International Tax System Another paper that we will now discuss is International Taxation and Corporate Location Decisions by Barrios, Huizinga, Laeven and Nicodeme (JPubE, 2012). What they did in this paper is to try to reproduce what Devereux-Grith have done but with much more options than just a few countries. Furthermore, instead of having decision trees that goes US - Europe (and where in Europe), they have the possibility to include any country on the top: invest at home or invest elsewhere in the world. They don't have a 1xN relationship but a NxN relationship (multilateral approach). The other thing that they did is that Devereux-Grith look at the statutory rate or the average eective tax rate in the country where you want to locate and the authors of this paper decided what they did was not enough. You know that if you invest in a country your prot is taxed at the statutory rate but ultimately this prot will be redistributed back to the parent, be taxed again, and there will be a double tax relief that will be applied (exemption, credit, deduction, no relief, whatever). Instead of doing what D-G did (looking only at the tax in the subsidiary country), they look at the entire tax system when the dividend is repatriated. Value-Added of the Paper So the summary of the value-added of this paper is the following: • Structure of the rm: Paper can distinguish domestic and foreign subsidiaries and parent companies. • Multilateral approach: First NxN countries study (variation in data) • International tax system and double tax relief: Full description of international taxation variables and systems. • Estimation techniques: McFadden versus Chamberlain estimations. International Tax System We have exactly the same structure as what has been shown in lecture 2. • ti is corporate tax rate in subsidiary country i • tp is corporate tax rate in parent country p; • wie is dividend witholding tax rate in country i • τi is eective (two country) tax rate > Comment: it's not eective in the sense of average eective tax rate from earlier (it's a misnomer) but eective after applying the computations that they do (taxation in the sub-country, transfer as a dividend to the parent company with a withholding tax, taxation at the parent, and the relief that is applied). Double Tax Relief Conventions: • Exemption • Indirect foreign tax credit for corporate taxes and withholding taxes 42 • Direct foreign tax credit withholding taxes only • Deduction of foreign taxes • No relief They apply these formulas for each bilateral relationship. • Exemption ti + wie − ti wie • Indirect foreign tax credit for corporate taxes and withholding taxes max[tp , ti + wie − ti wie ] • Direct foreign tax credit withholding taxes only ti + (1 − ti )max[tp , wie ] • Deduction of foreign taxes 1 − (1 − ti )(1 − wie )(1 − tp ) • No relief ti + wie − ti wie + tp For example, if a Belgian parent needs to choose to invest in 50 countries, they compute this for every year and for every bilateral relationship, and then they do the same for each of the countries. They have a network of countries, NxN relationship times the number of years. Basic Regressions Figure 50: Basic Regressions Figure 50: you can see is that the tax variables are negative and signicant. It means corporate taxation is a deterrent to locations of companies. If you decompose the eective tax rates that we've computed into its dierent components, you see that each of the components will have an inuence. The international tax system and the way it has been build really has an inuence, it's not only having a low tax rate in the destination country but also how the bilateral relationship between the parent and that country will play in the double tax relief that will also be a determinant of where you locate. Interpretation of Results Figure 51 shows that they computed a few elements for each of the regressions. They computed the eective tax mean, the eective tax marginal eect, and the eective tax semi-elasticity. In this case, the independent variable is the eective average tax rate and the dependent variable is the probability for a company to locate in that country. The eective marginal tax eect of the basic regression (2) is -0.615. This means that if you increase the average eective tax rate by 1 percentage point (35.3% 43 > 36.3%), the probability to locate in that country decreases by 0.615% (percentage point). The eective tax semi-elasticity of the basic regression (2) is -0.761. The semi-elasticity is the decrease in percent: when the tax rate goes up by 1 percentage point, there is a decrease in percentage of the actual value. Review page 28-29 if you need a reminder/better explanation of these. Figure 51: Interpretation of Results Interpretation of Results The elasticity that they found is -0.25. If we increase the tax rate by 1 percent (from 20 to 20.2%, for example), the probability of locating in that country would decrease by 0.25%. This elasticity is three times higher than what previous studies had found (-0.08). The reason for this is that they take into account the entire tax system, not taking into account this (like in previous studies) leads to an underestimation of the eect of corporate taxation on the location of companies. 4.3 Distortion in the Amount of Investment: Taxation and Business Investment We nished looking at step 2 in the decision tree, we will now look at step 3: the amount of investment. Most of the studies look at the eect of corporate taxes on foreign direct investment (FDI). There are very few studies that look at the eect of corporate taxes on domestic investment: for example, if my rate in Belgium changes, what will be the eect of this change on the investment of Belgium companies? There have only been a few papers for the US (one is by Kevin Hassett - Trump's former economic advisor). What you need as data is the tax rate (how it has changed) and the level of investment of those companies, and then you want to control for some of the characteristics of those companies (size, number of employees, the structure, the level of debt). The reason why there have been so few studies is because the most important thing that you want in econometrics is variability. The variable of interest in this case is the tax, we want to know what the eect of a change of tax is on the level of investment. What we observe in the US between 1986 and last year is a constant tax rate. To have good econometrics, we want a variable that changes because otherwise we will never have an ecient estimate. You want your variable to vary to see what the eect is on the level of investment. Otherwise you see a variable that is constant and you try to explain something that is changing from year to year, the only thing you will see is the impact of other variables but not the one that is constant because you cannot measure it. Studies that only look at one country have diculties to be carried out because statutory corporate tax rates rarely change. There is not a reform every year, there is a reform maybe every ten years. If you expand to the international level, the advantage is that you have dierent levels of rates. You have a within estimator that is within each country that you look at that will not tell you a lot but the between estimator, which is the one that they compare between dierent locations, will have some explanatory power. That's why you have more international studies on this than national studies. Eects of FDI What De Mooij and Ederveen (2006) have done is a meta-analysis of 427 regressions: from a bunch of studies (probably around 30-40) that has a number of regressions trying to explain FDI with the tax 44 rate. Then, they try to make a summary of it. If we look at the eect, it has the following distribution: Figure 52: Semi-elasticity Distribution That's the type of elasticity that we nd. For example, we nd a median semi-elasticity of -2.9. This means that if you increase the tax by one percentage point (from 20% to 21%, for example) then you decrease the FDI by 2.9%. The FDI is referring to inward FDI, not outward FDI: how much do you attract FDI from abroad? How much will the parent that has decided to invest in your country and locate a subsidiary will try to invest? What they nd is that half of the studies do not nd a signicant eect. A signicant eect is statistically dierent from zero: statistically, at a certain percentage of certainty, it's dierent from zero. They nd a wide range of semi-elasticities locating most of them just below zero but some of them nd very extreme values of -30 and some of them are even positive (maybe not signicant but positive). Then, they tried to explain the variation of results that they found in those studies. One way to explain is to look at the characteristics of the studies. They try to explain the outcome, which will become the y in the regression(dependent variable), and explain it by the characteristics of the study. The rst characteristic is which tax are they using in the studies: the statutory rate? The eective marginal tax rate? The eective average tax rate? Also, what is the eect of choosing one instead of the other? How does it aect the semi-elasticity that is found in the study? The second characteristic is which country do they use? The third characteristic is which period? 80s? 90s? 2000s? The fourth characteristic (very important one) is what do they mean by FDI? It's important when you do studies involving FDI because we tend to confuse FDI with investments.There are three main categories of FDI: 1. Portfolio Investment 2. Merger & Acquisitions 3. Greeneld Investments There is no reason to believe that taxation aects those three categories in the same way. Still, most economists use FDI as if it was a single category that looks at the same thing. Figure 53: Eects of FDI 45 You can see that they did some regressions to see what are the variables that explain the dierences. You have the dierent tax rates that are used, the type of regression (discrete choice models, panel data, cross-section). The main nding is that there seems to be a statistically signicant negative eect of the eective marginal tax rate on the investment of companies. So when we design the tax system, we may want to think about a tax system that tries to minimize the eective marginal tax rate: the rate that applies to the last unit of investment. There are ways to do it: for example, you can show that systems that do not tax the normal return to capital (from previous lecture/gure 35: a notional interest system or cash ow system, for example) are systems that, even though you apply a statutory tax rate, will not tax the normal return to capital and will lead to an eective marginal tax rate of zero. In the US with the current system, they have immediate expensing (cash ow system), you can show that this does not tax the normal return to capital, only taxes the economic rent, it's not distortive because it does not distort the investment level. You can show that even though the tax rate is 21%, the rate that applies to the last euro that you invest is zero. The eective marginal tax rate in the US is zero so it should boost investment. This will be shown later. 4.4 Distortion in the Allocation of Prot - Prot Shifting: Empirical Evidence We will now discuss step 4: prot shifting. You have some empirical evidence of taxation being sensitive to FDI. In de Mooij and Ederveen (2006), we had a semi-elasticity of about 3. If you look at another meta study more recently by Feld and Heckermeyer (2011), they now nd a semi-elasticity of around 1.68. Possible Exam Question What could explain the smaller eect of the sensitivity of tax to FDI? Part of the answer is that it's not the same sample. Is it because there is a higher share of the digital economy? You don't need to invest physically into a country where you are a subsidiary. Now, if you want to serve a country and your activities are digital, you don't have to locate there. It's part of the problem of taxation, you don't have to be located there. Normally with the openness of the border, with the increased mobility of capital, it should be more sensitive so the sensitivity should have increased. The professor would expect with time, the sensitivity of FDI to taxation increases because you have much more choices of where to invest, much more possibilities. Is it because there has been conversions between the rates? The professor is not sure that we actually see conversions in the rates worldwide. The two possibilities we have is that either we really have a decrease of sensitivity of FDI to taxation (that indeed when de Mooij and Ederveen did the study it was much higher than when Feld and Heckermeyer did it) while the second possibility is that it has not decreased but the estimation techniques have changed. They are much more precise than they used to be because the econometrics has evolved. In some areas like conditional logit and special econometrics (more in context of tax competition) is something that has made huge progress, we are able to have estimators that are much more precise and the access to the data has improved/increased as well. We had, for example, datasets that were macro data aggregated at the country level which is not the same as if you use FDI at company level macro data. Assuming the sum is the same, it is still better to have data at company level than aggregate because there is much more variation, meaning much more precise estimates. You see this, for example, in general equilibrium models so it was not rare (quite a few years ago) to see in general equilibrium models eects of GDP of 3-4-5% sometimes up to 10% when you were running those simple models. Now, when you have 0.5% of something, you already say that it's actually a big eect. You need to take this into account: the techniques have evolved so it's sometimes dicult to compare very old studies from the past with the new ones because it's a completely dierent landscape. You have to think not only of the results but also of the techniques and the way economics has evolved. There doesn't seem to be a concrete answer here so maybe/probably have to gure out some more possibilities for this before the exam. Seems pretty open-ended. 46 We also expect the burden of CIT, because capital is mobile, should actually fall on the imobile factor: being usually, in our model, the labour. The legal incidence, if investment/prot/corporations are mobile, is on the corporation but probably someone else will pay it: we will see that workers are actually paying part of it. We have evidence of prot shifting. There is a meta study by Heckemeyer and Overesch (2017) that shows that there is prot shifting because they found that the semi-elasticity of the tax base is 0.79. This means that if we increase our corporate tax rate by one percentage point, the corporate tax base (prots reported) decreases by 0.79 percent. That's evidence that prot has shifted. They are able to use a technique to split this eect between what we call the non-nancial channels and the nancial channels. They nd that in the prot shifting techniques: about 20% is the nancial channel (shifting debt and the advantage of being indebted) and 80% of the prot shifting is due to the non-nancial channels (transfer pricing or the location of IPs). We also have some more precise evidence with di-in-di. For example, for Germany: Egger et al. (2010) and Finke (2013) took a sample of MNC and another sample of domestic companies and tried to match those two samples on a number of characteristics. For example, the size of the company, the structure of the asset, anything that could inuence the tax burden of a company except for the rates. There are techniques for this in econometrics. They take a sample of multinationals and try to nd a sample of purely domestic countries (companies that do not have access to subsidiaries abroad), tried to match those two samples and try to look for those two samples at the dierence in eective taxation. What they found is that multinationals have a tax burden that is 30% lower than domestic companies with the same characteristics. It's evidence that companies that have access to group members abroad have the possibility to play the international system, to lower their tax burden to almost one-third. Prot Shifting: 3 Channels Figure 54: Prot Shifting: 3 Channels There are three main ways of proting shifting (Hecemeyer and Overesch, 2017): 1. Transfer Pricing 2. Location of IP 3. Debt-shifting Transfer pricing is the price that you will charge for activities within your group. For example, you are Coca Cola and you have one company that is in the group that is actually producing the aluminium cans. You are buying the cans directly from the company which is within the group. Of course, there are other companies in the world that are also producing aluminium cans and from whom you could also buy those cans. If you would go on the international market and nd any company, you would have a market price - the price that is charged in the market - for an aluminium can: for example, 5 47 cents on the international market. But you are Coca Cola Belgium, you know Belgium is a high tax country, and the can company is located in a low tax country. Is there a way to charge a price for that can when you buy it within the group that would be dierent from the market price but that could be advantageous for the group because you could shift the prot out of Belgium to this low tax country where the company is located? The way you could do it is that Coca Cola Belgium would pay a price that is higher than the market price so that they would have a business cost that is higher, resulting in a deduction of prots in Belgium that is taxed at a higher rate. Instead, this price would be the prot for the company located abroad in a low tax country and would be additional prot for the company but taxed at a lower rate. Coca Cola Belgium would pay less taxes, the aluminium can company in the group would pay more taxes, but when you net out at the level of the group, it's a pure gain for the group. That's a way to shift prots: charge higher or lower prices than the market price to try and benet from prot shifting. What do tax authorities do to counter it? They normally have one rule: there is a price of comparison: the arm's length price. That is the price that is normally applied for a similar product. The tax authorities when they will look at the account of Coca Cola Belgium and the tax authorities of the country where the can company is located will ask that company why they charge 10 cents while the world price is around 5 cents. They are suspicious that they are trying to overcharge so they ask for all the operations to provide a transfer pricing documentation where you explain to the tax authorities the reason why you charge higher: what is special about the can? Is there is a specic issue they should know? Could you nd arguments that show that the world price is not 5 but maybe 9? So it's to try and see if you can explain the 10 instead of 5, it's very costly series of events. Then, there could be disputes between the tax authorities because the tax authorities to the can companies are happy because they have more prot booked with them while the Belgian tax authorities will say they disagree with it so the tax authorities need to get into an agreement together to nd the right price. It's easy to do when you have a standardized product like a can but it's more dicult for a product that is dicult to price. For example, you are selling a patent for an invention of a new medicine: what is the price for this? Obviously, if you have a patent, no one else has this product so it's very dicult to nd an arm's length price for those products. That's where diculties arise. The second way of shifting prots is location of IP. An IP stands for intellectual property. That's an invention for which you have a patent. You go to the patent authorities and they will, against payment, protect your invention. You've invented this new process/design, or a name, or a brand, or a logo and they will protect these against competitors/imitators for a period of X years in this specic geographical zone. You're protected by this and you can actually allow other companies to use your patent/invention against the payment of a royalty. This is what Ikea seems to be doing, for example. Ikea is a Dutch company, the HQ is registered in the Netherlands (even though originally from Sweden). Ikea Netherlands, according to some reports made by the green political group in parliament, has a system where Ikea Belgium pays to Ikea Netherlands a royalty for the use of its design. Ikea Belgium is making sales in Belgium, Ikea Belgium makes a prot and then pays a royalty to Ikea Netherlands that is located in the Netherland - the parent company. The advantage of this is that Ikea Belgium can deduct the royalty that it pays against its prot (so it's decreasing the prot booked in Belgium). It's the same for any Ikea in Europe. This royalty instead goes to the Netherlands where it will be taxed. The Netherlands is one of those countries that has a specic tax incentive for patents, they have something called a patent-box. The patent box says if the income that the company is earning does not come from normal prot but from royalties: we will not tax it at the standard rate but a reduced rate because we think it's a promotion of research (patents = research). You have this system where you try to attract the location of patents to try and ocially attract research but what you end up doing is attracting the tax base. From the perspective of the country, it's some sort of method of attracting some research activities. From the perspective of the company, it's being able to use patent-boxes to shift prot. There has been a mushrooming of patent-boxes in the world, in particular in Europe, for the past ten years. This was to the point that the OECD and the EU had to change the rules and to agree that the benet of a patent box can only be granted if there is a minimum level of research done in that country. There is a formula that is applied to see whether the cost of doing the research activity 48 that is linked to the patent is basically occurring in the country where the tax benet is granted (in an attempt to moderate the tax competition a bit). The third one is the location of debt. An example follows. You have Coca Cola Belgium (high- tax country) and Coca Cola Ireland (lower-tax country). What you can do if you want to decrease your tax rate: you know that if you create some debt within the group, the interest that you will pay on the debt will be tax deductible for the one paying, and taxable for the one receiving it. You can ask Coca Cola Belgium to borrow some money from Coca Cola Ireland: Coca Cola Ireland will make a loan to Coca Cola Belgium. Coca Cola Belgium will pay interest to Coca Cola Ireland. This interest is tax deductible in Belgium at a high rate and will be taxable in Ireland at a lower rate. At the level of the group, there is a gain that corresponds to the interest paid times the dierence in rates. You can also manipulate the interest rate that you charge. If you know that when you go on the international market you would pay 3% and within the group you charge 15%: you say that the company is risky, you know the account within the group well and it's risky so you want to ask an additional risk premium. The tax authorities will observe this and you will need to have a discussion as to why the company is risky and justify the risk premium etc. There are a lot of discussions that may be going on. All these techniques are completely legal. It's just using the tax system to shift prots. Transfer Pricing There are forums where we discuss these cases at the level of the EU (notably between the tax authorities). There are also some arbitration conventions that exist at the level of the EU on how to deal with those disputes (Professor doesn't want to go into the details of this but you have to know that both of these exist). It's not completely left to each member state, we try to organize these discussions within an EU framework. We will now talk about some evidence of transfer pricing. There have been extremely few studies looking at transfer pricing, notably because you need good data. You need data at the level of the product with the price that has actually been charged and you need to be able to compare the same products for transactions that are within the group (related parties) and unrelated parties (arm's length). It's very dicult to nd the right data to do this. The following paper is now discussed: Clausing (2000). She looked at some trade data in the US. She found that the balances between the US and the low-tax countries in terms of intrarm trade was potential evidence of transfer pricing. Meaning that, in that case, the ow of trade was going from lowtax countries to the US so that the US can actually buy more expensive goods from low tax countries so that it could deduct it from its tax base: those prots taxed in the low-tax countries. The level of taxation was able to explain the intrarm balance. She found an eect of the dierences in taxation between the two countries (the lower rate in the aliate country) to explain the change in the intrarm trade balance. She has two papers that basically say the same thing (according to the professor). The last paper discussed is by Davies et al. (2018): topic 5 of group work. They looked at French data and look at the exports by French companies. They found evidence of transfer pricing. Quote from the paper: Back-of-the envelope calculations suggest that tax avoidance through transfer pricing amounts to about 1% of the total corporate taxes collected by tax authorities in France. The lion's share of this loss is driven by the exports of 450 rms to ten tax havens. It's not huge but what is important is that it's highly concentrated in a few rms. Most of it is in the top 450 that export to tax havens. Not only does it seem to be concentrated with a few rms but also in some bilateral relations: France vs Tax Havens and not France vs other high tax countries. You see that there is indeed evidence of tax avoidance via transfer pricing. 49 5 Lecture 5: Corporate Income Taxation - International Distortions (CONT) 5.1 Distortions in the Allocation of Prots - Prot shifting: Empirical Evidence CONT In this lecture, we will go further in the tax aspects of corporate taxation and notable the issue of prot shifting. Last lecture, we discussed the three channels of prot shifting: transfer prot, location of IP, and debt-shifting. Transfer pricing is altering the price of transactions within the group because you can choose at which price you sell or buy within the group. This can aect where prot will be booked on the one hand and cost on the other. We also saw that another channel is the location of intellectual property, notably via patent boxes. The third channel is the use of debt: it's to create debt in high tax countries so that they will have a burden in terms of interest payments that they can deduct against their high rate. The nancial channels is estimated to be 20% of the total and about 80% is the non-nancial channel (location of IP and transfer pricing). Debt-Shifting We will now discuss debt-shifting: how this works and how can you transfer prot if you use debt. Take the following situation: Figure 55: Debt-Shifting Situation There is a parent company that makes a prot before tax of 100 and is located in a high tax country at 25%. The parent has a subsidiary that also makes (for the purpose of the example) 100 in prot and is located in a low tax country with a CIT rate of 10%. If we look at what each company would pay in its own jurisdiction: the parent company would pay 25% on 100 which is a net after tax prot of 75 while the subsidiary would pay 10% on 100 which results in an after tax prot of 90. This results in a group prot after taxation of 165. Figure 56: Debt-Shifting Situation The CFO of the company has an idea and says why don't we create a situation where the part of the group that is located in the high tax country borrows money from the subsidiary that is in the low tax 50 country? In this example, the parent takes on a loan: it's a debt of 1000 and the subsidiary applies an interest rate of 5%. This results in the parent having to make an annual payment of 50. The situation therefore is as follows. At the level of the parent, the interest payment is tax-deductible so the tax base on which the 25% applies is 100 (the original prot) minus the interest payment of 50. That's a tax base of 50 and if you apply the CIT rate of 25%, what is left is an after tax prot of 37.5. At the level of the subsidiary, the tax base is the original prot of 100 and the interest rate received of 50, which is a taxable amount. That's a tax base of 150 and after applying the CIT rate of 10%, that's an after tax prot of 135. Under this arrangement, the parent will pay less taxes and the subsidiary will pay more taxes because the tax base increases at the level of the subsidiary and decreases at the level of the parent. If we look at the after-tax prot at the level of the group, we now have 172.5 compared to 165 previously. There has been additional net of tax prot created by the fact that we've shifted debt from the low tax country to the high tax country. Simply, by creating this internal debt to nance a purpose, one is able to shift prots from high tax jurisdictions to low tax jurisdictions. 5.2 Capital Structure and International Debt Shifting in Europe The professor looked at this in a paper on debt-shifting (Huizinga, Laeven, Nicodeme). The idea comes back from what we talked about earlier regarding domestic distortions and at the debt bias: debt is actually tax favored in most countries. In most countries, the interest is tax deductible, whereas there is no such deduction for dividend payments. Recall the Modigliani-Miller theorem (1958): in the absence of taxation, the value of an unleveraged company is equal to the value of a leveraged company. The value of the company does not depend on its sources of nancing. VL = VU However, if you introduce taxation into the system, you will have the apparition of what we call a tax shield. VL = VU + tD where tD is the tax shield. You can show that if there is a tax aspect that comes into account, in this case the deductibility of interest payments, the value of the leveraged rm will be equal to the value of the unleveraged rm plus the value of the tax shield. The tax shield is the tax advantage of using debt which can be shown to be t (the tax rate) multiplied by the value of the debt: net present value of being in debt. This breaks the Modigliani-Miller theorem. If we take that into consideration and international taxation will take a role via the double tax relief, we have a suspicion that there is an optimal debt policy for the company that will depend on taxation. International Tax System Same slide once again. We can actually model the international tax system: • ti is corporate tax rate in subsidiary country i • tp is corporate tax rate in parent country p; • wie is dividend witholding tax rate in country i • τi Comment: the eective tax rate is the rate that applies to a prot that is created in a subsidiary and send back to the parent, taking into account corporate taxation at the level of the sub, the withholding tax that applies to this transfer of dividend, the way the dividend is taxed at the level of the parent and the double tax relief system that is applied. 51 You have ve possible systems for relieving double taxation: • Exemption • Indirect foreign tax credit for corporate taxes and withholding taxes • Direct foreign tax credit withholding taxes only • Deduction of foreign taxes • No relief You can model this and compute the eective rate on the basis of the statutory rate for each of the systems: • Exemption ti + wie − ti wie • Indirect foreign tax credit for corporate taxes and withholding taxes max[tp , ti + wie − ti wie ] • Direct foreign tax credit withholding taxes only ti + (1 − ti )max[tp , wie ] • Deduction of foreign taxes 1 − (1 − ti )(1 − wie )(1 − tp ) • No relief ti + wie − ti wie + tp Optimal Leverage Figure 57: Optimal Leverage You can now think about the optimal leverage be in debt since there is a tax advantage. λ* of a company. High taxes will give an incentive to You can model the leverage of a specic company i as a function of several elements The rst one is the optimal leverage ratio that would apply in the absence of taxation. If there would be no taxes, there is probably, given the structure of the company, an optimal leverage ratio and the company will deviate from this optimal leverage ratio because of tax considerations: domestic tax considerations (you anticipate that you will have a positive relationship between the leverage ratio and the domestic tax rate: the higher the domestic tax rate, the higher you want to be in debt to benet from interest deductibility) but there will be an additional element that 52 is a function of the dierences between your own rate and the rate in other jurisdictions. We have here a weighted average of the dierences between your rate and other jurisdictions. So if your rate is higher than the rate in other jurisdictions, you expect that the relationship will be positive: that is, subsidiaries that are located in other jurisdictions will try to send you debt to benet from prot shifting. If, on the contrary, your rate τi is lower than the rate τj of other jurisdictions, you will be the one that wants to send debt to the high tax jurisdiction within the group to benet from prot shifting so you would have a negative eect. So, a domestic eect and international debt shifting eect, that is, the dierences in the rates multiplied by a ratio that reects the size of assets in jurisdiction i over the total assets of the group. Questions by student: At the end of the day, if you're part of the same group, your accounts will be consolidated and the debt that you transfer from one part of the group to another part of the group will cancel out. Why is this an advantage? This is because in most countries taxation is based on separate accounting. You do not tax on consolidated accounts. You tax on unconsolidated accounts in your jurisdiction. Some countries allow for a taxation based on consolidation but only within the country: group taxation within the country. Immediately, it gives you a hint on a policy that could defeat prot shifting. If you would consolidate accounts at a higher level, the EU level, there would be no point in playing with debt shifting within Europe because in the end your accounts will be consolidated. The proposal of the commission, 2011 and 2016, is exactly this: consolidation at the EU level for taxation purposes and then a reallocation according to a key across the dierent countries. So the reason why this works is because most countries apply taxation based on separate accounting, they look only what's in their jurisdiction, they tax in their jurisdiction, and each jurisdiction does this so the consolidation for accounting purposes is not present for tax purposes. Obviously if all countries have a common rate, there is no point in doing prot shifting. You would be taxed at exactly the same rate in any country. The only reason why you may still want to do prot shifting in that case is to play with the timing of the recuperation of the losses. If you have one part of the group that has losses, you may want to shift some of the prot there to recuperate the loss so that there is an immediate benet of the loss. You would cancel out your losses today against the losses somewhere else and not pay taxes on this prot so you gain a year in terms of timing and time is money. So the equation in gure 57 is the expression for the leverage of a company that depends on an optimal theoretical leverage and tax elements (the domestic part and the international part). Company Data What they did is try and estimate this using company level data with a database called Amadeus by Bureau van Dijk (two brothers). The brother that created the company basically created the rst database of nancial accounts of companies and it was a real breakthrough for research: more variation means much more precise estimates than using macro indicators as they did in the past. It's a database that was not built for the purpose of doing research. The main purpose of the database in the beginning was usually for the banking sector to look at credit rating. The database is presented in such a way that for each of the company, you click on the link and it gives you the nancial accounts, the ownership of the company, the ultimate owner, the location, who is the auditor, a whole bunch of information that is useful for people in the banks to assess the risks of the company and be able to price the interest rate/premium in case you want to give credit to the company (if it has nancial diculties or not, bankrupt, etc..). Academia has taken on the database but the interface was not what they wanted. The professor was one of the rst to talk to Bureau van Dijk at the time and to try to get at les, when you do research, you are not interested in the interface and click by click, what you want is a at le with denitions of the variable, a tax le that you can put either into Excel or into Stata or SAS. The database has for each company information on the company that they own with the percentage of 53 shareholding, so it says Coca Cola Belgium owns Chaudfontaine and they have 95% of the shares, you can click Chaudfontaine and they can give you additional information all the way down the road. Also, upstream: it could say Coca Cola Belgium is 100% owned by Coca Cola USA, which you can click on etc. . . You can recreate the structure of ownership of all the companies. It's very bothersome so usually what you do is limit yourself to direct ownership. They dened a company A to be a subsidiary of a parent B if it owned at least 50% in terms of shares in the study. They had data for 10 years: 1994-2003. Basic Regressions Figure 58: Basic Regressions What they did was some regressions and they tried to explain the nancial leverage and some adjusted values of it: the debt-to-equity ratio or debt-to-asset ratio. They had the main variable of interest: the eective marginal tax rate. That is the domestic tax and the tax incentive to shift debt which is the weighted sum of dierences between your rate and the rates of the others. Then, they had a number of controls that they used: one being tangibility. Tangibility is the share of tangible assets on your total assets. The company has a number of assets and those assets can be either tangible or intangible. Intangible is all the intellectual property rights, patents, goodwill. The dierence between tangible and intangible is that the tangibles if they fall on your feet it hurts while if the intangibles fall on your feet it does not hurt. Intangibles are all these rights that you own and the values of those. The tangibles are all the machinery, factories, computers, etc. Why do we think the level of tangibility may aect the nancial leverage? It's because it can be a substitute. Tangibles are tax depreciable, it's something that you can amortize over time. If you have a machine you can amortize your machine over 5 years, so every year you can deduct 20% of the value from your prot. It's a tax deduction that you have, so the more tangible assets that you own, the more deductions that you get. They can be seen as a substitute to getting deductions via the debt. Log of sales is a measure of the size of the company, why do they think it may aect the leverage? Probably because the larger the company, the more in debt it can be because it has access to credit. It will not be constraints by the banks or other creditors, they will rely on the fact that the company is large, has a reputation etc., compared to smaller companies that may have more diculties to go into higher indebtedness. Protability: same idea. Companies that are more protable should normally be able to get access to more credit. Then they have some other elements. Creditor rights is a measure of the rights of creditors in the country. The higher the creditor rights, the easier the is to get credit. The political risk: it should aect your wish to be in debt or not. Ination: it can erode the value of debt. Growth opportunities: this was a variable that was computed as the average growth of the sales over the time period which should also give access to credit. Then, they had a number of xed eect dummies: they control for the country, the industry, and the year in which the observation is reported. You see around 50,000 observations for the last three, more than 5000 parent companies which means that for each parent they have approximately ten subsidiaries on average and an R-squared of 5-8%. Not so great at rst sight: when you do a regression with ordinary least squares, you want to get an R-squared that is pretty high, close to 50% or 100%. When you do panel regression (data for all countries and all years, none missing, you try to have a structure of your data that is full and rectangular, 54 that is you have an observation for all the years and all your companies, none missing), usually you get an R-squared that is close to 10% and it is considered as quite good because you have many elements that are inuencing the results. A lot of the controls are statistically signicant, signicant at a 1% level. What is interesting is that you see the two tax variables have the expected positive sign: the higher your rate on the one hand and the higher the dierence between your rate and the rate of the other, the more in debt you will be. We are able to disentangle/dierentiate the two factors: the one that is the domestic factor, that is the level of indebtedness that depends on your own rate, from the second one, which is the part of the debt that is brought to your country via prot shifting (via the other companies in the group). Basic Regressions What we have based on regression number 3 is that we see that if the rm was a standalone (no subsidiaries abroad) and only this aect would be working, the number 0.184 means that if we increase the tax rate by 10 percent, we would increase the nancial leverage by 1.84 percent. For the second one, it's a bit more complicated to compute this eect because it's an average of the dierence in the rate and the weight. It's a weighted average, where the weight is the assets of the company i on the total assets of all the other companies in the group. What we can do is assume that you have two establishments that are equal in size (same level of assets) so that the relationship would get A on the sum of the A would become a half (A on 2A, which becomes 1/2). You apply the percentage change and when you do all these computations, you nd that an increase in 10% in the eective tax rate will increase the local leverage by 2.44% and reduce foreign leverage by 0.6%. The main conclusion: there is both a domestic and a prot shifting eect that will aect the indebtedness of a company. Huizinga et al (2008, JFE) is exactly the same story. Slides About the Research From Another Presentation That is Not Provided The professor opened another lecture at this point that has not been provided. They talk about the previous literature that looked at the issue, mainly data for the US. They talk about the value-added of the paper itself: the structure of the rm, international tax system, NxN countries so a lot of variations: parents from all countries and subsidiaries from all countries. You start your research by looking at the data and those types of empirical work is very data intensive. What you need to do is for all your countries in your sample, you need to nd the statutory tax rate applicable in the country for all the years observed (1994-2003) so you go to handbooks. The professor gives an example of a previous paper on foreign ownership that he worked on where they went to the IBFD in Amsterdam and spent two days (no laptop) with large sheets of papers with tables and went manually through all the handbooks writing down the rates by hand so when they were back at the oce they could type the information into Excel. Then, you need to check what the applied double tax relief is in those countries. You know that the countries usually apply two dierent systems at most. They apply a system to their treaty partners (the country with whom they have a double tax treaty) and then they apply a separate system for the countries with which they don't have a tax treaty. For example, Czech Republic used to have an indirect credit system with treaty partners and a deduction system with non-treaty partners. You need to do this for all the years. That's the rst step. Then, once you have this, you need to know who was the treaty partner for each of the countries and at which year. You rst collect data, the data on the withholding tax and you look that when the withholding tax the prot is distributed from Belgium to Bulgaria, the withholding tax is 10%. You do this matrices for all the ten years you have of your data, the data you need to collect. Then, you have a table with the bilateral treaties and you know that in 2003 there was a treaty between Czech Republic and Austria but there was no treaty between Czech Republic and Malta, for example. You need to look at all those treaties to know when they started to reinforce. Why? Because this normally should be symmetric: if there is a tax treaty between Malta and Czech Republic, it's likely that there is a treaty between Czech Republic and Malta but the starting date may dier. You may have some years for which the tax treaty between Czech Republic and Malta was enforced but not between Malta and Czech Republic where there is an amendment that says it will only start in a year. You need to collect all this information and 55 then impute it into the system. That's one big part of the research: collect the data and double check afterwards (look at outliers, whip the data until they tell you something). Then, you apply your model. You have two strengths in the profession. You have those who are theoretical economists, they will basically build models, they will optimize, maximization under constraints, develop things. Then, you have people on the empirical side who try to validate/invalidate the models with the data. They take the existing model, think about the research question, try to think about an already existing model or to create one, you always need an underlying model that will guide your choice of regressions. The way you will structure your regression, the type of variables you choose, whether it's a log model, a log-non log mode, linear model, should you take the square of some variables, all of this will be guided by the theoretical model that you have in mind. In this case, they had to build their own model where they thought through the issue and said we think this is what theoretically should happen and this is what we will try to regress: try to empirically validate. For example, they wanted to model a cost of bankruptcy. If you have a theoretical model that says the higher the tax, the higher you want to leverage and there is a tax advantage, what is the optimal policy? As an individual in your model, you will be leveraged until innity (a corner solution). People know that for every euro of debt that you take, you get a tax advantage. If you want to maximize your tax advantage, your incentive is to go into debt until innity. You need something in the model to counterbalances this. What counterbalanced this in the model is to introduce a cost of bankruptcy: cost of nancial distress. This will depend on the size of your company and the square of your leverage ratio. The more in debt you are, the higher the tax advantage but in parallel there is a cost of being indebted that grows exponentially. That's what makes the model converge to a solution that is not a corner solution. You also do then all the additional regressions to make sure that your model is robust. You do dierent types of clustering of the errors, you look at the manufacturing sector only, multinationals, only companies with positive prots, exclude Eastern Europe which may be on a dierent growth path, etc. Quite often those additional regressions will be assessed by the referees when you send in your paper to a journal. They will in the comments say you should look at some regressions where you exclude this, include this, restrict sample to this, etc. If you want to be published, you have little choice but to accept what the referees are suggestions. They had dozens of regressions to show that the model was robust. 5.3 Patent Boxes and Patent Location Decision Now we will discuss the paper Patent Boxes and Patent Location Decisions by A. Alstadsaeter, S. Barrios, G. Nocodeme, A. Skonieczna, and A Vezzani (Economic Policy, 2018). What is the issue at stake? The issue at stake is that we've seen in recent years a booming in the use of a tax advantage that is commonly called Patent Box. What is a patent box? It's an advantage that is given to com- panies that are receiving revenues in the form of royalties. Royalties is the compensation for the use of intellectual property rights. Basically, you are a pharmaceutical company, you have an invention: you've created a new molecule that could be used in medicines/drugs. You patent it to protect your discovery against the fact that some competitors may copy you. You go to the European Patent Oce or to the national patent oces and ask for the patent (there is a fee) that will be given to you. It protects you against any copy from a competitor, nobody can actually use the molecule that you invented. It can be a molecule, a process, a way of building a machine, a machine, a design, the shape of something. You invent something that is protected and insures that you are the only one that can use this process/molecule and you can sell it. Typically, for example, in the pharmaceutical industry, you may have patents for 10/20/30 years. If you want to rent your discovery, what you can do is provide the rights/concessions/a license to someone that can use your discovery against the payment of a royalty. For example, you invented a new machine to produce soda cans, Coca Cola comes to you and says they believe you're a genius and would like to use the process, could they buy your patent? You would end 56 up giving your rights. Another option is to give a license and ask for a payment of ve million euros per year in exchange for the license. You have an income stream that comes from the possibility that you oer to others to use something that is protected. Some countries on the argument that they want to develop/promote research and hence the number of patents: they want to give a tax advantage to the revenues that are derived from patent licenses. If a company receives royalties, what they will do is not include those royalties into the normal tax base for taxation, they will tax it separately. In reality what they say is that a certain percentage of the revenues is exempt from the tax, meaning it corresponds to a lower tax rate. Quite a few times, they will present patent boxes as a lower tax rate on royalties, which is true, but in reality the way the system works is to say we tax at the nominal rate but 80% of the revenues is exempted so it's like taxing at 20% of the nominal rate. A question that arises is that you see that more and more you have a strategic location of those patents. Companies/Groups may try to say well they have an interest from a tax perspective to locate all their patents in a country with a patent box instead of a country where there is no patent box because the revenues that they will derive from those patents will be taxed at a lower rate in a country with patent boxes then in other countries without. You have the question of whether or not countries that use the patent boxes system is actually a new way of doing tax competition in a way that could be harmful (in the sense of the denition of what is harmful competition for the OECD or the EU: a system that departs from normal ways of taxing). On top of this, many of the patent boxes do not include the necessity for the patent to have been invented/developed locally. For example, a company buys a patent from somewhere, acquired from a third company, takes it, moves it to some location where there is a patent box and benets from the patent box system without having ever done any research in the country where the patent box advantage is provided. There has been the start of discussions at the OECD under the program BEPS (Base Erosion and Prot Shifting). It is at the same time the name of a phenomenon, the erosion of the tax base and prot shifting, and also the name of the program that tries to counteract this phenomenon. Also, there is a similar discussion at the level of EU on the need to create a so-called Nexus Approach. What is a nexus? It is to create a link between the tax advantage that is granted and the research activity underlying the patent. The idea there is to say well we would like a system where you can only grant the tax advantage if at least a certain percentage of the research that has led to the patent has been done in the country that grants the tax benet to avoid that there is this tax competition. They have developed at the level of the OECD and at the level of the EU new patent boxes, countries were forced to revise their patent boxes to be nexus compliant - they comply with the new rules and there are complex formulas based on the cost of the companies that determine whether or not you're above the threshold that allows you to have the nexus between the place of the research/development that led to the patent and the tax advantage that is granted. What they wanted to do is to look at the issue of whether those patent boxes were indeed leading to additional research, whether it was really a tool to promote research or whether it is just a tax competition device and what is the balance between the two. Also, to check whether the nexus approach, that is, a form of conditionality that would be imposed on the researcher would actually solve any of the problem. Patent Boxes You benet from a reduced corporate tax rates when you have a patent boxes. Also, it's interesting to see that patent boxes cover actually more than patents. We use the nickname patent boxes but it's usually called IP boxes (intellectual property boxes). It's also another name that we give them because it covers patents but also covers trademarks, designs and models (last lecture, we talked about Ikea using patent boxes in .the Netherlands for the royalties that they receive from their subsidiaries in the world), copyrights, domain names, trade secrets and know-how. Then, you have a bunch of examples of countries that have recently or had recently introduced patent boxes: the Netherlands, Belgium, Luxembourg, France, Portugal, China, UK, Spain, Hungary, Malta, Cyprus, Lichtenstein, Switzerland, 57 Italy, Ireland. Ireland being an interesting case because they introduced a patent box, removed it, and reintroduced a new one. It's interesting for research there because you have some variation: it's not just a cross country issue, you have variation even within the same country. Model What they tried to do is to look at a model - a negative binomial logit - and tried to explain that the prot of a company that is located in a jurisdiction depends on the tax rate of the jurisdiction, the tax rate that is oered under the patent box (the additional tax advantage that is given via the patent box), a dummy showing whether or not there is a patent box and that should cover all the non-tax aspects of the patent box. We have then the tax advantage of the patent box via an interaction: via a way of computing the tax advantage. Imagine, for example, there is a CIT rate of 20% in a country and that via the patent box, you could reduce it to 5%. In that case, you can go from a normal rate of 20 to a reduced rate of 5, you have a tax advantage of the patent box (the dierence) of 15. That's what they compute there, they have an advantage. If there is no patent box, it's 0. They also have a part that is simply the presence or not of a patent box washed from the tax advantage that will be captured by this variable. What is left in this variable, because we wash it (remove the tax advantage and only leave the non-tax aspects of the patent box), you may expect to have only the administrative issues. That is, the procedures that you need to go into to get a patent box, to patent, to get the advantage, to deal with the authorities and so we expect the sign of the variable to be negative. You can decompose the eect of the patent box between its tax eect that should have a positive eect on the attractiveness of prot (you try to attract more prot in this jurisdiction because there is a patent box) and the non tax eect which will be all the administrative burden which we expect to be small but negative. So they have the following model that they do in the paper: Figure 59: Model They have regressions with a lot of controls, with about 50,000 observations but it depends on the variables that they use in the regressions. What you nd is the tax advantage of the patent box appears indeed as being positive and signicant. That is, there is a positive eect to attract the prots/attract the activities of other companies in a country, the higher the tax advantage via the patent box. If you look at the non-tax eect of the patent box, you see that indeed it's negative and signicant. You get exactly in the regressions the expected eect that they thought they would nd. Conclusion There is a strong eect to attract patents. They make a distinction between high quality and low quality patents that is based on the number of citations - the patent can be cited. What is a high quality patent? It's a patent that is quoted quite often and hence has a potential for prot that is higher than the others. They also found that the larger the scope of the patent box, the fact that you can also add to this patent box previous patent that was not just developed but that you have acquired that was already existing, that was embedded in additional previous research - basically the higher the scope of the patent box, the higher the positive eect on the location of patent. Furthermore, they tried to do some regressions looking at whether additional patents would lead to additional research, they did not nd such an eect. Quite on the contrary, what they found is that the fact they have more patents via a patent box has a tendency to not increase the local research. The only thing that increases the local research is whether you could apply development conditions, that is a nexus approach. They found that 58 some of the patent box that they look at had the feature of the nexus approach, they had development conditions, they force that the research has been done locally to be able to obtain the tax advantage. What they see is only if you have this, you are able to mute the pure tax eect of the patent box and you are able to get additional local research. It validates in the sense the policy actions that both the OECD and the EU have been doing. That is, they need to impose a nexus condition to avoid that patent boxes become pure instruments of prot shifting/pure instruments of tax competition between the countries. 5.4 Distortions in Shifting of Tax Burden - Tax Exporting The next topic is the issue of tax exporting. What is tax exporting? Tax exporting is the following story. In theory, if you are a country you may want to attract business and so you will try to do that by decreasing your tax rate. The lower the tax rate, the lower the average eective tax rate, the higher your attractiveness, the more it has an impact on the location of the companies: the more likely the companies will settle in your country. However, this is not the end of the story and there is an additional factor that may actually disturb this. It is when those companies that come in your country and that will be taxed in your country and that you want to tax at the low rate so that they stay and develop their economies, is when they are owned by foreigners. When they are owned by foreigners, you may think that it could be interesting for you in your mix of taxes that you need to collect to get your public expenses, to get your budget, to have a higher tax rate on companies, because ultimately the one that will be paying it are not your residents but the foreign owners of companies that will receive less dividends, receive a lower price for the share because the tax will be higher but you're able to shift some of the tax burden to foreign owners of the companies that are in your country. There was some model and they started to work on the paper, it was the rst time they were using ORBIS/Amadeus database. The professor went on a training that was given by David Wildasin and he realized that Wildasin was working on a similar story from a theoretical perspective. When the professor talked about his research, he said he was writing the model at the time. The idea is the following. Imagine that you have two things: you have companies in your country that has a certain theta which is the share of foreign owners. There is a cost of relocation and speed of relocation (1/t where t is the time that you spent). Figure 60: Cost vs Speed of Relocation Imagine that this cost is increasing and convex: if you have time to relocate, it's not so costly. If you say you want to move out of the country but you will do it within the next 10 years it's not costly. If you say you want to leave and you want to leave within the month, it's very costly. The cost of moving away from the country increases with the speed. What David Wildasin (2003) shows is that if you have some companies that have this feature of cost of relocation: for which it is very costly to move/they are captive in your economy, in the long run they will move but in the short run it's very costly for them to move. Imagine, for example, that they have a nuclear power plant, you cannot dismantle a nuclear power plant in a month. You need several years to move out. If those companies 59 are partly foreign owned, then the optimal rate that you may want to apply to them is positive. It's not necessarily 0, you think that they are captive, you think that they also are foreign owned, so you have an incentive to have a rate that may be higher than if they were not foreign owned and if they were able to relocate. You know that you will be able to shift the burden to foreign owners of companies that are captive within your market. It develops a theoretical model of this where the equilibrium tax rate on the capital is proportional to the degree of foreign ownership. The higher the degree of foreign ownership, the higher the optimal equilibrium rate and inversely proportional to the mobility of capital. If the capital is immobile, you will choose a higher equilibrium tax rate in the model. The idea is to say what you want to check is that if you tax today, you can attract some additional tax revenues because those companies are captive and you know they will move out in the future, maybe in 5 years, 10 years, 30 years, the speed at which they are able to relocate, you will lose this tax revenue as compared to this current tax situation. What you want to do is to balance out the NPV of the gains of additional revenues that you have today against the losses in less revenues that you will have in the distant future when those companies will move out. That's the idea behind the model: if there is foreign ownership, if companies are captive, you may want to have a higher equilibrium rate on those companies because you can extract today additional revenues even if you know in the future you will lose revenues of those companies that will relocate abroad. They were looking at the model and trying to test it and one of the rst things that they did was to show the graph internally. Huizinga and Nicodeme (EER, 2006) plotted the following graph: Figure 61: Tax Burden and Foreign Ownership It is a measure of foreign ownership of the company (the average value of foreign ownership for the ownership of their sample) and a measure of the tax burden which is the tax paid divided by total assets. Surprisingly, there seems to be a positive relationship (Hungary is an outlier): the higher the foreign ownership of companies, the higher the equilibrium tax burden on companies. Nobody believed them at the time. People thought that they were either crazy, or wrong. There was a director at DG ECFIN commenting on their paper: he was doing a parallel with the smoking gun and the evidence of weapons of mass destruction in Iraq. He said it was exactly the same proof that you have: basically nothing. It was a big drama, the ORBIS database was obviously awed, etc. But the results were there. The story of tax exporting was of some relevance. What they did is to work the database for 21 European countries from 1996-2000 and they did two measures of foreign ownership: one at the level of the company (they know who are they owners from Amadeus and they know if they are the same country) and at the country-level (the foreign ownership of individual companies weighted by the assets of those companies). Some statistics at the time: there were 15,361 rms in the sample, they had a measure of foreign ownership in 2000, which was the last year of their observation, of 21.5%. If you look at the dierence between Western and Eastern Europe, you see that at the end of the sample year, they had a degree of foreign ownership in Western Europe of 19.4% and 32.9% in Eastern Europe. If you look at the change in ownership, there is a small increase for 60 the total sample which results from a small decrease in Western Europe but a large increase in Eastern Europe. You have a degree of foreign ownership from 1996 to 2000 in Eastern European countries that was booming. The Regression Figure 62: The Regression This is the regression they looked at. They explained the tax burden of a company by some characteristics of the company, some characteristics of the country where they are located, by the tax rate of the country where the company is located, and by the two measures of foreign ownership: the one at the company level and the one at the country level. What we nd indeed is that the estimated coecient for those two was actually positive and signicant. The benchmark is that you have in this case a marginal eect: if you increase foreign ownership by 1 percentage point, you actually increase the average corporate tax burden by 0.43 percentage point. That means that if in the EU we wouldn't have any foreign ownership, that all companies would be domestic, the rate would go down from 33% at the time to 24%. All of these results show that there is some evidence of tax burden being shifted to foreign owners because you nd a positive relationship between foreign ownership and the tax burden of companies. Robustness Checks What's interesting is if you look at the results per sector. You need to have a degree of foreign ownership but you need also to have companies that are captive. That is, you have an elasticity of moving the location that is low or 0. If you look at the decomposition by sectors and that you try to see the eect of foreign ownership, you consider some sectors like agriculture and utilities which are two sectors that are extremely immobile, those are two sectors that are captive and what you nd indeed is that you have a very high coecient for the foreign ownership variable. It means that the eect of foreign ownership is higher in those sectors than in the average sector for the economy because they are indeed the ones that are captive. If on the contrary you like at nancial services, you see that it's not signicant. The eect is 0 statistically, but you even nd a negative sign which shows that this sector, that is not captive, that can move elsewhere, you don't nd the eect. 5.5 Distortions in Competition: Tax Competition There is an issue of dening tax competition which has been something quite complex for economists. In the professor's view, tax competition is dened in the following way. It is an setting (the way they decide on their taxes: the base, the rate, etc.) to insists on the two terms. The rst is interdependent tax attract a mobile tax base. He interdependent in the sense that the way they decide on those rates depends on the way others decide on the rate: if they increase the rate, you will have an incentive to increase yours. If they decrease the rate, you will have an incentive to decrease yours. If you decrease your tax rate, they have an incentive to decrease theirs, etc. but interdependence is not enough. You must also have a reason for that and the reason is to attract a mobile tax base: attract companies, attract nancial investment, attract FDI, attract labour, attract highly paid people, it can be a lot of things. It's not just about corporations, it can be tax competition in personal income taxation. Spain at some stage had a law called the Beckham Law because it was law that was apparently designed to make sure that David Beckham when he went to play for Real Madrid could benet from a tax advantage, that all his revenues would not be taxed at the progressive rate. You have a lot of other examples, all the expat regimes that are available in many countries: it is to attract expats, a labor force that is from abroad. You have several papers in academia, notably some papers by Emmanuel Saez Taxation and International Migration of Superstars where they look at the eect of those specic regimes targeted 61 at specic sports. Those two terms are quite important because some authors (Devereux, Lockwood, and Redoano) tried to look at what we call tax reaction functions. The tax reaction function is actually a tax equation where my rate in country i at time t is a function of some characteristics of my country i at time t and of the rate in country j dierent from i at time t: ζi,t = α + β1 Xi,t + β2 ζj,t + i,t The rate in Belgium at time t depends on characteristics of Belgium in time t (that can be the number of companies, foreign ownership, level of wages, ination, decit, public debt, etc.) but also depends on the rate of Germany, France, Spain, Portugal, Russia, etc.. We expect to have a positive estimated coecient. There is an interdependence on tax setting, my rate depends on your rate. That was something that was extremely tedious to do from an econometric perspective because the tax rate of country i depends on tax rates of country j etc.. but tax rates of country j etc.. depend on tax rate of country i. You have a circular relationship which created a lot of problems with the errors. The errors were not i.i.d. (independent and identically distributed). The estimations were usually not correct if you use the normal econometric techniques and it's only in the early 2000s that theorists have developed a branch of econometrics that is called spatial econometrics. They found a way to do this correctly with software. What is the problem with the above? So if you have this and you nd 0.1, it says that if Germany increases its rate by 1 percentage point, I increase my rate 0.1 percentage point. It also says that if France increases its rate by 1 percentage point, I increase my rate 0.1 percentage point. That if Russia increases its rate by 1 percentage point, I increase my rate 0.1 percentage point. If Somalia increases its rate by 1 percentage point, I increase my rate 0.1 percentage point. The problem is that you treat any bilateral relationship in the exact same way. What you want to have is something that is a weighting metric, that doesn't treat all relationships in the same way. That doesn't say the inuence of the tax rate of Portugal in Belgium is the same as the inuence of the German rate in Belgium. Obviously Belgium cares much more about the tax rates of Germany than the one in Portugal. There is no reason to believe that Estonians care exactly the same way about the tax rates of Greece than the tax rates of Latvia. You want to put a weight (we will see it at some point in one of the papers presented by a group) that there are various ways to choose this weighting matrix. You can base it on distance, on contiguity (neighbors), GDP per capita (you want to have the same levels of development), wage costs, whatever you want. The only thing that you need is to structure in your regression the weighting matrix so that the sum of each row and the sum of each column is equal to one. For example, if you use distance and a matrix like this: Figure 63: Matrix You need to divide everything by 5000 (the sum) so you have the sum at the level of the column is equal to 1. What Devereux et al. found is indeed a positive and statistically signicant relationship. tax competition, is that true? There is The only thing they found with the regression is that there is inter- dependence in tax setting. They have no way to prove the reason why there is this interdependence. They may have a suspicion that this is done to attract a mobile base but economists (so far, maybe someday) don't have any possibility to prove econometrically that this interdependence is created by a desire to attract a mobile base. We have suspicion, an indirect body of evidence: for example, we nd 62 that this relationship is stronger the higher the mobility of capital. So if you have these regressions with countries where you have capital barriers etc., you see that the relationship is stronger the more mobile the capital. However, it can be tax competition but it could also be tax mimicking, notably some papers have applied this technique at the local level (in Scandinavian countries there are a lot of people looking at this). They look at the tax reaction functions for municipalities and what they found is that indeed there is tax mimicking. You have a phenomenon whereby politicians try to imitate the tax rate that is available in neighbouring cities for reaction purposes. There is a paper by Besley and Case where they take a sample of a governor race (race to become governor of the state) and they are able to distinguish in the panels people who are not up for re-election (that already served the limited number of terms) and also this phenomenon during the years of election and during the years of no election. What they nd is that the eect is there and strong when the person is both during the year of re-election and up for re-election otherwise the phenomenon is less present, if not absent. There is an indication that it's actually only an issue of tax mimicking. Another possibility that we have is that we have convergence in the rate because we may have convergence in economic structures. If all economies become alike (that the share of consumption converges, share of capital converges, share of labour converges, share of public spending converges, etc.) it is not impossible that there would be somehow some convergence also in the structure of taxation and of the rates. At least that was what Slemrod found in a 2004 paper using a sample with a lot of developing countries. The professor tried to replicate this, to have a look at least at some basic statistics of coefcients of variations, you do not nd this at the level of the EU. You do not nd this convergence at the level of the EU, at the moment at least. Quick recap: Tax competitions entails two elements that are extremely important. The rst one is the interdependence in tax settings meaning that the way you decide on your tax system (the features of it, not only the rate but the other parts of it too) is dependent on what the others are doing. The most obvious example is an interdependence of the rates. If one country changes its rates, the other country will be incentivized to go the same direction. The second point is that we need to do that for a purpose, the purpose in this case to attract a mobile tax base. What can a mobile tax base be that you can attract? You can attract the business itself (a relocation of the company), more investment (your country is chosen over others), prot (shifted to your country), any type of mobile tax base concepts that you can attract. Those two concepts are important because so far in the economic empirical studies, people have been able to demonstrate the interdependence (the rate of the country depends on the rate in another country) we nd a positive and signicant coecient for the tax rate as an independent variable so we can see there is interdependence (the tax reaction functions) but it's impossible to formally prove that this is done to attract a mobile tax base. There is a large presumption but formally and empirically, this is not something that has a way to be achieved. There is no technique to prove its to attract a mobile base instead of doing it for another reason. There are signs that it can be the case because this interdependence is higher if the capital is mobile. There are some suspicions that indeed we are on the right track but it could also be other phenomenon that would be there. Tax mimicking: mentioned the fact, notably at the local level, that you want to mimic what neighboring cities are doing. Also, convergence in the tax structures due to economic convergence that could potentially explain that if our economies look more alike, our tax system could have a tendency to be more alike and also hence the rates and the way we dene the base. The mainstream thinking of taxation: you have this phenomenon in the minds of policymakers all of a sudden there is an idea and that seems to be completely evident. If you were in the 1980s it was a at tax, all of a sudden, everyone in Europe started to talk about at taxation for personal income taxes. Flat tax is a tax with one single rate so instead of having a proportional system that increases with the level of income, you would tax everyone at 25% or 15% or 10%. All of a sudden you had these discussions that at tax is something everybody needed to implement. Then, if you're in the 1990s, that was the OECD ranking of taxes, the OECD had a study that looked at the impact of taxation on growth. What they found at the time was that they could establish a type of ranking of taxes according 63 to how disruptive or harmful they are for growth. They found that corporate taxation and personal taxation were very damaging to growth and environmental taxes, consumption taxes, and property taxes (housing) were less detrimental to growth so you needed to shift from those taxes on capital and labor towards the other tax base. Also, the mantra of low rates and a large base, we've seen that in the beginning of the course. For eciency reasons, you want to collect the same amount by decreasing the rate but increasing the base on which you tax and you see that this should normally reduce distortions so increase economic eciency. We know that the counterpart of this may be the issue of equity/fairness - the ability to pay. At some stage, that was in the mind of all policy makers: around the table in Europe, everyone was saying we need to decrease all the rates and enlarge the base, that's the solution to all our problems. The fact that you have at some stage some sort of economic thinking that becomes mainstream may also explain convergence in the way we tax. Professor has never seen a study on this, says that it's probably because it's dicult to measure how much this economic thinking is present even though you may ways with techniques of identication of words/speeches/conclusions of councils or whatever. In summary, there is empirical evidence of interdependence with big suspicions that this is to attract a mobile base. We have corporate income tax collection to GDP that is not zero, as can be seen in the gure below. It's small though, on average we are just below 3% of GDP but you also see a large variation of the corporate tax collection that goes from 1.5% in the Baltics to a tax collection in Luxembourg, Cyprus, and Malta that is over 5% of GDP. Figure 64: CIT to GDP Figure 65: Statutory Corporate Tax Rates What we observe though is a gradual decrease over time of statutory corporate tax rates, as can be seen in the gure above. We had in the 1980s corporate tax rates on average in the EU, the above line is the EU 15 average while the one below is the EU 28 average, you see that we had averages that were very close to 50%. The EU used to tax countries at 50% in terms of corporate tax rates while today we are below 25% in the EU 28. We see that there is a gradual decrease over time that's called the race to the bottom. You have this phenomenon where every year you see some countries decreasing the rates and some other countries following suit. Figure 66: CIT in % GDP Figure 67: Current Taxes on Corporate Income and Wealth to GDP The question however is despite the fact that we have rates that more or less halved over the last 30 years, we don't see a decrease in the corporate tax collection (gure 66 above). Corporate tax to GDP 64 are cyclical, it depends quite largely on the business cycle. When you have years of very high growth, corporate prots are also increasing but what you see is that the trend seems to be stable or even increasing. It's always somewhere between 2.5% and 3% of GDP, on average. That's a big puzzle for policy makers, how do we explain that despite the fact that we have those declining rates, we have a tax collection that remains more or less constant? Referring to gure 66 above, corporate taxes to GDP on the scale is a bit distorted, it's between 1% and 3.5%. The black line is France and it started in 1960, you see that the evolution has been a bit erratic but on the long term you see a positive trend. The blue line is the EU, Germany is in red, and the US is the orange one. You see that you have falling revenues in 1982, that was the big US tax reform where they have cut drastically the corporate tax rate. Then an increase, a new drop in 2001, there was also a year of economic recession. Large fall for all the countries in 2008, the nancial crisis. You see that it's uctuating with the business cycle but you don't see a sharp indication that it's decreasing over time. How do you explain this? Nicodeme, Caiumi and Majewski (2018) looked at this using a method of CIT decomposition. The idea comes from a paper by Peter Sorensen (2007), you can actually decompose your corporate tax rates to GDP ratio into three components (the professor's paper decomposes into four components while Sorenson does three). What Peter Sorenson was showing is that you can decompose the corporate tax collected to GDP by the following three factors: Figure 69: Professor Decomposition Figure 68: Sorenson CIT Decomposition The rst component is the corporate tax collected to the gross operating surplus of corporations. The gross operating surplus of corporations is what corresponds to EBIT of companies in national accounts. It's an operating prot: your sales minus cost of materials, wages, and depreciation & amortization. It's all the non-nancial prots, prots that you derive from your operations. Sort of: the tax collected on your operating prot is like an implicit tax rate, the rate at which on average you are taxed. The second element that you can have in your equation is the ratio of gross operating surplus of corporations to the gross operating surplus of the total economy. It indicates the share/size of the corporate sector in the economy. Then, the last element is gross operating surplus of the total economy to the GDP is a form of measure of protability in the economy. The higher this gross economic surplus of the economy, the higher the prots to GDP. You can try to see which of these elements explains the ratio. Peter Sorenson looks at the evolution of the indicators over time and the motivation of the paper is exactly that from above: CIT rates have declined but CIT collection has not. You have some elements that might actually explain the evolution of this. We have indeed some studies that what happened is that we have falling rates but we have a base broadening that could explain why revenues did not fall (Devereux, Grith, and Klemm, 2002). Some people (De Mooij and Nicodeme, 2008) showed that if corporate tax rates are declining, more and more people try and incorporate: instead of being self-employed they try to go into the business sector because they pay less taxes. It would show up in the size of the corporate sector in the economy, you should see an increase in the size of the corporate sector because you have more corporations on which you can collect taxes. Then, the last element that's brought by some authors (Auerbach, 2007) is the issue of loss carry-forward. Many companies are carry-forwarding their losses and so if you look at the aggregate number, because you have a lot of losses from the past that are cancelled against prot, you have a measure/share of taxable prot that is lower but you still have companies that pay taxes. What you see is the eective tax rate has a tendency to increase. An example: imagine that you have three companies A, B, and C that are the three companies in your economy and they have a prot in year t of 100 each. The tax rate is 25%. Under a normal situation, all companies would pay 25 and you would have an implicit tax rate in your 65 economy that would be the tax rate paid by the two: so (25+25+25) / (100+100+100) = 75/300 = 0.25. Imagine now that company C has some losses that are carried forward from previous periods and they have losses of 200 that they can carry forward. So instead of having a prot of 100, actually what they do is they bring back from the past the loss of 200 and so their prot becomes now -100 so they don't pay taxes. The implicit tax rate in this case will be (25+25+0) / (100+100-100) = 50/100 = 0.50. The point is that we should be extremely prudent when we look at those evolutions and ratios because you have an asymmetry in the system, you never pay negative taxes, you are never reimbursed by the state so it creates a distortion in the sense that the taxes that you pay will never become negative but the prot that you register in your denominator may become negative. Hence, it can actually make your indicators, notably the one on the implicit tax rate, a bit dubious. So the paper by the professor had a look at all of this. They looked at a dataset of national ac- counts from Eurostat that has a quite ne decomposition so they really have for each of the European countries a distinction between the operating prot, the nancial ows, the type of nancial ows. They are able to look at quite a number of things. What they did is re-use the formula of Sorenson but at an additional step: shown in Figure 69. Instead of having the corporate tax rates on the gross operating surplus of companies, they introduce an element which is the theoretical base on which companies would be taxes. They don't tax them on the gross operating surplus because companies also have interest deductibility and interest that they receive, rents, premiums for nancial corporations for insurance, etc... So they are able with the data they have to compute the theoretical base based on the actual data of the net operating surplus of companies, the interest that they pay, so they are able to introduce this new step and they look at the evolution of this over the last 20 years. Figure 70: Graphical Evolution If we rst look at the graphical evolution of this, we see that the corporate tax rate to the base has sort of an evolution that is at over the long term or slightly decrease. The denition of the taxable base to the gross operating surplus of corporations is also evolving with the business cycle and is also relatively at or decreasing. Figure 71: Graphical Evolution The size of the corporate sector in the economy is increasing over time and the protability of GDP seems to be decreasing. This is replicating what Sorensen has done. The question is what you see visually, can it explain scientically what are the contributions of each of the elements to the total? You may have a large increase of one ratio but maybe this ratio doesn't come to much in the total equation. You need to weigh it and see to what extent it contributes. 66 Figure 72: Dierentiation What you do is you simply dierentiate. If you want to have a way to see if you have a product, if you want to see to what extent each of the terms of the product contributes to the total, you apply a simple dierentiation (product rule) and arrive at that formula. What you have is the explanation of how each of the elements in the formula (the rates, the CIT to the base, the base to the gross operating surplus, gross operating surplus of corporate to the gross operating surplus of the economy, and the gross operating surplus of the economy to the GDP) when they vary and how do they inuence the change in the total. You can see this because for each of these terms you now have the weights by dierentiating. Figure 73: Findings This is what they found. First, we will concentrate on the rst column which is the years between 1995 and 2015. What we see there is that between 1995 and 2015, the CIT ratio to GDP has increased by .2% of GDP. We see a slight increase. If we decompose, we see that we had a decrease in the rates that contributed -0.8 percentage point. If everything stayed the same and rates had decreased without anything else changing, we would have seen a decrease in the tax to GDP ratio by -0.8% of GDP. Of course, other things are changing: the reactions of the company, the policies of the government, etc. . . What you see is the taxable base has increased and has more than compensated the decrease in the rates. The increase in the taxable base contributed for 0.899% of GDP. There was a compensation and that's evidence that you have indeed a rate lowering, base widening for corporate taxation in Europe. The base-to-GOSc is irrelevant. Protability is relatively irrelevant as well. The other relevant element that counted is indeed the size of the corporate sector that by itself contributed for 0.178 percent of GDP. You had those two factors that were cancelling out each other and then an increase in the size of the corporate sector that led to an overall increase in 0.222. If you decompose between the two decades so 1995-2005 and 2005-2015, you see a dierent story between the two periods. During the period 1995-2005, you had quite a large increase in the CIT to GDP. This period also did quite well in terms of the economy. You see that the rate was decreasing and contributed for -0.53% of GDP, this was more than compensated by an increase in the taxable base (0.845% of GDP). An increase in the size of the corporate sector as well. If you look at the more recent period, the CIT to GDP has decreased between 2005-2015. The contribution of the rate is almost similar to the one in the previous period but you see that the increase in the tax base hasn't been able to compensate for this change. The policy question there: is the low rate/large base policy coming to an end? Is it coming to a level where you can actually widen the base only to a certain extent? At some state, there are no exceptions, exemptions, or special regimes to cut. You arrive at a stage where if you continue to decrease the rate, you don't have any more room for manoeuvrers for enlarging the base anymore. In gure 74, you can see a summary chart of all the eects that we've seen. It's a bit selective, you may nd other papers that might have other estimates. It's to give you a sense of the semi-elasticities on the organizational form, the nancial policy, the intensive and extensive margin of the investment. It's an interesting concept that we use quite often in economics: intensive and extensive margin. It's notably 67 used on labor markets. If you change a tax rate, you actually have two eects, you have the extensive margin and the intensive margin. The extensive margin is the extent to which you will choose to work or not to work. In this example here, it's whether or not companies will decide to locate in your country or not. The intensive margin is the eort/investment that you do. You are in and you may decide to work more or work less. In this example, it's to invest more or invest less. An example to remember the concept. In terms of drinking, the professor has a low extensive margin but high intensive margin. He doesn't drink often but when he does he drinks a lot. We also have the semi-elasticities of prot shifting, the elasticity of the tax base to the rate is around 0.8 and it's split into both nancial and non-nancial channels, as we've seen so far. Figure 74: Summary Chart 68 6 Lecture 6: Corporate Tax Competition Before we go into the model, we will talk about positive and normative analysis because it is important to make the distinction. Positive analysis is a description of the eects that you observe. You increase your tax rate, the supply is decreasing or demand is decreasing, the interest rate moves up, all these eects that you simply observe. The normative analysis is: given a certain objective function for the policy maker, what is the optimal policy? Again, it's dependent on the objectives. You may be a policy maker that says that your objective is to maximize the welfare of your citizens. You can be a policy maker that says that your objective is to maximize the tax revenues. You may be a policy maker with the objective to minimize unemployment rate. Depending on the objective function that you set, the normative analysis will tell you what the optimal policy is. You may also have normative predictions which is: given the circumstances which policy will be chosen. That's another branch of the normative analysis. It might be useful to start with a basic model. This one is adapted from Tiebout 1956. It's a partial equilibrium model of incidence of local property tax. Figure 75: Tiebout Model You have an amount of capital and on the vertical axis the interest rate. You have a demand for capital that is downward sloping. You have an equilibrium E without taxation that is determined by the international rate of return because we are in a completely open economy. The economy is a price taker, it's a small economy. They have an international rate of return and the economy is too small to inuence this rate, they take this rate as given. It's the concept of a large versus a small country. It's not necessarily the size, it's how large or how small are you on this market. So Belgium is a small country for coee because the fact that Belgium will increase or decrease their coee consumption has probably no impact on the international price of coee. Belgium is a large country for pharmaceuticals or diamonds, these are the two main items that we trade in terms of values. The US may be a large country for some things but a small country for other things. This is a small country in that sense and we have an equilibrium that is determined by the international rate of return that determines K* - the quantity of capital. rÌ„ which is We have here capital income that is earned by the capital suppliers: they earn on this amount of capital a return rÌ„. Then we have a surplus that could be seen as the same sense as consumer surplus because each of the people demanding capital were actually willing to pay for each of these additional units of capital a higher rate or return and they don't have to pay it. So they have a surplus that is in the sense the return to the other factors: those who do not supply capital (labour for example). Refer to gure 76 for the following paragraph. A tax t: What happens when we introduce a tax on this? local source-based property tax - an additional lump sum amount for the purpose of the demonstration. It's not a proportional tax. In that case, we have a new rate of return, the one that is gross of taxes, rÌ„ + t. It will move the equilibrium from E to B. What do we observe from a positive analysis perspective? We see that there is an increase in the price of capital from 69 rÌ„ to rÌ„ + t. There is a decrease in the equilibrium capital that goes from K* to K', there is indeed capital outow. There is less capital that is demanded, capital has own out. We have a tax revenue that is collected that is K', the amount of capital that has been exchanged in the economy, times the level of the tax. There is another thing that changed in the economy is that the gross return to the other factors has actually decreased, it used to be the triangle (r to rÌ„ to E) and is reduced now to the triangle (r to r-rÌ„ + t to B). There is a loss in this surplus that is ((r+t) to B to E to r). If you do the netting, you have positive inow, the tax that you collect and redistribute to your citizens, and a gross loss of surplus in the economy which is the triangle (BCE) + rectangle in the middle. The net loss is the loss of surplus to other factors (BCE). Distortion: You actually tax capital which is mobile, there is a capital outow, and the consequence is that you collect the tax on the new level of capital but you have a loss of surplus for the economy that is larger than the monetary value of the tax you collect. Figure 76: Tiebout Model with tax So normative analysis: what is the optimal tax rate? If you want to minimize the loss of net income to others (BCE), the tax rate must be zero. Any positive rate will actually create this triangle. The optimal tax rate if you want to minimize the loss, or maximize the welfare of the citizens, will be zero under the conditions of these models. What tax is chosen in equilibrium? You must know the objective, if it's to maximize the welfare then no tax is optimal. Figure 77: Subsidy The same goes for a subsidy as can be seen above. A subsidy is a negative tax. If you give the subsidy S, it decreases the price from rÌ„ to rÌ„ minus S. You have capital inow from K* to K'. You have an increase in the surplus of your other factors so the new surplus is A and B but there is a cost for the government which is the subsidy per unit times the number of units which is A + B + C. You have a net loss for the economy which is the triangle C. So the optimal subsidy policy is also 0. Those models say something that is quite profound. 70 Under the strong assumption that you have the factor that is taxed being mobile, the market will undo your policy. This is because there will be a reaction of your base, the factor that you want to tax which is the mobile factor will leave the jurisdiction where it is taxed. It is of course dierent if you did an analysis in a closed economy. In a closed economy, your labour and capital is xed so there is no way it can leave and there you can actually tax. It means that if your factor is immobile, it would be possible to tax. There is no way for the factor to leave the jurisdiction. It's exactly the same idea when we had the discussion on elasticities: the incidence of the tax. The incidence of the tax will always depend on the elasticities of supply and demand. The more inelastic you are, the higher the incidence will be on you. The ones that will always bear the burden of the tax are the inelastic/immobile ones. The question of mobility is an interesting one. In those standard models, we tend to consider labour as being the immobile factor and capital to be the mobile factor. in reality, if you see capital as money. The distinction is more complex Obviously you can transfer money from one account to the other with just a click on the computer. That's something that's arguably, in the absence of capital control, extremely mobile. If you see capital as a plant, it's dierent. A nuclear plant, it's quite a substantial amount of capital that is located somewhere, it's not mobile. Even nancial ows, the salaries of workers that are working on a coal mine, the coal mine is immobile and because the money that you need to use to pay those workers that are in an immobile sector becomes de facto also immobile: it needs to be attached to the workers. Mobility of labour is the same story. Yes, labour is not so mobile usually except at the extremes. Very poor people are relatively mobile, think about the expansion of the US, the ow of migrants to the US when the US was created before the independence was quite substantial. A lot of people who were poor in Europe went to the US. We see that today with the migration from Africa, unfortunately. We see mobility at the top as well. All the students in the class are well educated people and most of us are Erasmus or studying and will be working in a country that will be dierent than the country in which you were born. How mobile are you the more you specialize? A doctor that is specialised in treating extremely rare forms of cancer, maybe there is only two or three research centers in the world where this person can work. It can also create policies that are dedicated to some parts. There is literature at the moment about special regimes for the super rich: the Beckham law in Spain. You had a tax system that was designed for David Beckham to go and play at Real Madrid and would have less taxation than what others would normally have. What's the implication for scal federalism? There is an article by Stigler: he gave a short contri- bution in front of a US Senate Subcommittee. He makes the point that you need to assign the taxing rights at a level where the market cannot undo your policy. In this case, what you want to have is to give the policy at the level where the mobility is decreased. For example, if you have a tax on capital and capital is extremely mobile within the EU, less so between the EU and other countries, it would make sense from this perspective to have the taxing rights not at the level of the countries but at the level of EU because it doesn't matter if the capital would ow within the EU, it would still be the same policy that is applied in all these jurisdictions. The scal federalism consequences of this is indeed put your taxing rights at the level where the market cannot undo your policy. 6.1 Standard Model Of Tax Competition: Zodrow - Mieszkowski - Wilson Now we will talk about the Zodrow - Mieszkowski - Wilson tax competition model which is called the standard model of tax competition. We are going through the pdf le of the model and come back to the slides after nishing it. Professor quote: I will never ask you to replicate this (the model) at the exam but I want you to understand the results of the model: the impact of taxes on capital, the impact that it has on wages, the impact it has on policy making. George Zodrow and Peter Mieszkowski have published a paper in 1986 in a volume of some journal of economics and Jay Wilson also had in the same volume published a paper that was related/complementary. It's really the rst time that the models of tax competitions were formulated. 71 Oates 1972 : The result of tax competition may well be a tendency towards less than ecient levels of output of local services. In attempting to keep taxes low to attract business investment, local ocial may hold spending below those levels for which marginal benets equal marginal costs (. . . ) (ZodrowMieskowski 1986, Wilson 1986) Oates was an economist in the 1970s who looked at tax competition but at the time it was not formalized. There were non-formal discussions of tax competition and what he said was that the race to the bottom will lead to a moment where it will become inecient because it will decrease the tax revenues to a level where it's becoming not enough to nance the socially desired level of public goods. There was a fear if we continue this race to the bottom, we would actually decrease tax collection and that we will arrive at a tax collection that becomes insucient to nance the public goods at the level that is socially desired. The idea here is to try and formularise this in a model. There are quite a number of hypotheses in the model and the professor insists on this because there is a wealth of economic literature that is built on this model by changing some of the assumptions and to see how the results have been aected. In some cases, changing the hypothesis of the model may dramatically change the results. We need to be careful and always keep in mind the setup/hypothesis/limits of the model. The Hypothesis of The Model The following slide summarizes the assumptions of the basic model. Figure 78: Assumptions We have many jurisdictions. The idea is that even if we don't have those jurisdictions appearing in the model, what we mean by that is that we have an economy that is small, that is open (needed to have mobile factors that can leave the country), non-strategic (in the sense that the setting of the government does not depend on what the others are doing: there is no interdependence in the tax setting), and it's a price-taker. We have one tax instrument which is a rate t on capital and a simple production function that depends on capital K and labour L. That is constant return to scale: the sum of the products between the level of each of the factor times their marginal revenue equals the total production. Then you do some normalization: you x notably L as the numerator and so you can dene k as K/L which is the capital-labour ratio. We do it to simplify the computations. Very important: it is a well-behaved production function - it is a concave function that is increasing in k: the rst derivative is positive but it's concave in the sense that the second derivative on k is negative, it's increasing but the increase in the production reaches some limit so you have a concavity. We have some local residents: they have a utility function that depends on the consumption of x (the private good) and G (the public good) and those two goods which are the only two goods that are in the market are produced with a marginal rate of transformation between x and G of 1. With the transformation you have the choice and you need to trade one unit of the public good if you want one additional unit of the private good and vice versa. Competitive factor markets: you have competition on all markets - many rms, workers, and capital owners which means as a consequence that the production function 72 is set with productivity factor pricing which means that the price factors equals the return. We have the price for capital which is r: the interest rate and the price for labor which is w: wage. Now he spent some time walking through the math of the model in the pdf. Just review the pdf/math while reading the notes. How does t aect k? The rst question we may ask is how does tax aect the level of capital in your economy? We are assuming that K (the capital) is mobile across jurisdictions. If it's taxed, it may go elsewhere. L for labor is not mobile (original model was actually L for land, same outcome though). Since capital is completely mobile across jurisdiction, the gross return to capital will equalise. There is an arbitrage condition that says that the rate, since capital can move freely in all dierent markets, is the international rate of return. If you want to tax it you have no choice but having the marginal return to capital equal to the set exogenous net rate of return of rÌ„ plus the tax you impose on it. We know that the value of K that will satisfy the arbitrage condition will depend on the tax rate that you set, so on how you actually tax this capital. What we want to do is to show how this changes so we dierentiate with respect to the tax rate t. The second term is exogenous so it doesn't vary with t so it becomes 0. We know that k varies with the taxation. From the well behaved concave function that we have, we know that the second derivative is negative. We know that k varies negatively with the tax rate. If you tax your k (capital-labor ratio, in this case), it will decrease. There will be less capital in your economy, it will move away. This is the rst step in the model. First result: k varies negatively with the tax rate. How do wages react with variations of k? The second question we may ask is how are the wages changing if your capital-labor ratio in your economy is changing? We take back the equation that we had previously. We derive everything with respect to k. We are left with an expression: the derivative of the wage with respect to the capital-labor ratio on the economy equals 00 −kfkk . What do we know about this? We know 00 fkk is negative (well behaved function - concave) so the expression is actually positive. We know that if your capital-labor ratio in the economy increases, then the wage will increase. That's a measure of the productivity of your economy, your capital-labor ratio. We know in the long-run, the wages evolves with the productivity of your economy. If you increase taxation, the capital-labor ratio will decrease because capital will move away. If your capital-labor ratio in your economy is decreasing, your wages will decrease as well because there is a positive relationship between the two. Second result: A profound rst element of the model is that if you tax capital, capital will y away out of your economy, and it will decrease your wages. The economic incidence of the corporate income tax falls at least partly on the workers. If you decrease corporate taxation, it is increasing capital inows in your economy that will increase the productivity/capital-labor ratio in your economy, that will increase the wages. If you tax corporations, capital ies away, this will decrease the capital-labor ratio of your economy, the productivity of your economy, and in the long run the wages will decrease. Now we need to check what would be an optimal policy and for that we need to look at the budget constraint. We have two budget constraints: GBC and HBC. One for the government (GBC) that 73 nances the public goods with a tax on capital, K. We take the small notations (g, t, and k) by dividing everything by L. We also have a budget constraint for the households (HBC) which is that the consumption of good x (here again everything is divided by L) depends on the labor income plus some capital endowment that they may have (here in the model it doesn't really matter). How does t aect x and G? Now the next question becomes how does the tax aect x and G? What you do is dierentiate the two budget constraints with respect to the tax. The expression that we obtained by deriving the government budget constraint with respect to t will be used later. For the household budget constraint, the second term is zero because the endowment is a xed endowment: it does not vary with t. For the rst term, we know that the wage varies with k and we know that k varies with t so we know that indirectly, the wage will vary with t. That's exactly what we see here. The derivative of the wage with respect to t equals the derivative of the wage with respect to k times the derivative of k with respect to t. We end up with a simple expression: the derivative of x with respect to t is equal to -k. We have our expressions that we can use now in our optimal policy. What value of t maximizes u(x,G)? We are a benevolent government and we want to nd the tax that maximises the welfare of the citizens. The citizens have a utility function that depends on x and G and we want to nd the rate that maximises it. What we do is a rst order condition, we try to nd what is the maximum by deriving the utility function with respect to t and setting it equal to zero. What we need to remember also is that we have by assumption a marginal rate of transformation of 1. We know that we have an equilibrium in the economy on those competitive markets when the marginal rate of transformation equals the marginal rate of substitution between the two. The marginal rate of substitution is the ratio of the marginal utility of the public good to the marginal utility of the private good. We know that 00 fkk is strictly negative so my marginal rate of substitution is above 1. How do we succeed in having an equalization between the marginal rate of substitution and the marginal rate of transformation and that are all equal to one? In the model here, you must set the tax t equal to zero. We have the same result as in the Tibaut model. The optimal tax rate in this model to avoid the distortion in the economy is zero. Third result: The optimal tax rate in this model to avoid the distortion in the economy is zero. However, we can also do a little transformation. Recall that the elasticity of demand for capital: that's the proportional change of capital to the proportional change of its price. We also know that the derivative of k with respect to its price r is the same as the derivative of k with respect to t (the tax is a component of the price). That's 1 on 00 fkk . So if we change this into the model and express it in terms of elasticity we obtain the following expression: 74 This says that the marginal rate of substitution is the ratio of 1 on 1 plus the elasticity of capital with respect to tax multiplied by t divided by r. What are the ways to go back to where the marginal rate of substitution equals to a marginal rate of transformation that equals to 1? Again, one policy option is to set t to zero: the race to the bottom. What is the alternative? The alternative is having an elasticity that becomes zero. If your elasticity becomes zero, you can actually tax at any rate. It means trying to make your capital less mobile. Fourth result: A profound result of this is from a political economy perspective: if you have a factor that is mobile, in this case the capital, you are faced with two possibilities. Either you don't tax it so t = 0 or you try to make it immobile. Making it immobile doesn't necessarily mean having restrictions on the movement. It may mean, in terms of the scal federalism that we've seen, setting the assignment of the tax policy at the level where the elasticity becomes lower. In this case, if you have for example a world government that would decide on the tax rate that would apply on the capital, it would be taxed at the same rate everywhere and the capital de-facto becomes immobile because it's facing the same tax everywhere. You can use this model also to think about the tax assignment in a federation. This was not the purpose of the model but was something that people were able to interpret from the model. We are done with the pdf le and now are going back to the last slide of lecture 6. An important element that the professor wanted to tell us is that what we've done in the model is we tried to maximise the welfare of citizens. We tried to maximise the utility of the representative citizen that depends on the consumption of x (the private good) and G (the public good). You can rework exactly the same model with just changing this maximization and to instead maximize tax revenues. so-called You will have a very dierent model. If you want to maximize tax revenues, there is the Leviathan model. The Zodrow - Mieszkowski traditional model states that tax competition is a bad thing because it leads to underprovision of public goods. The Leviathan model says that by just changing what you maximise, they say tax competition is a good thing because it's restraining the size of the government who otherwise would continue to overspend and tax, tax, tax. . . There is no reconciliation between the two. In a model, you need to maximize something. Depending on what this something, it leads to dierent results/dierent rules. 75 7 Lecture 7: Aggressive Tax Planning Structures We will be talking about aggressive tax planning structures and it's pretty technical. What we will see now is to try to remember some of the eects of taxation and how those are basically used by companies to try to decrease global prot. It's a technical one because we will also be talking about anti-avoidance rules: the way countries try to react to aggressive tax planning by trying to impose some of the rules within their law to try to avoid prot shifting. There is a large distinction that needs to be made between what we call specic anti-avoidance provisions and general anti-avoidance rules. Specic anti-avoidance provisions are provisions that you put in the law with the intention to defeat a certain type of prot shifting activity or certain types of behaviours. For example, you say: if you do that... this is the remedy I will apply. Then countries, they can not foresee all the possibilities in their law, they cannot foresee all the possibilities that companies may have, and you know it's always a race with people being more and more inventive and trying to nd new ways of avoiding taxes and the tax authorities running after them and trying to impose new specic anti-avoidance rules. They also try to have in their jurisdictions/laws gen- eral anti-avoidance rules that more or less says: if this is something I've missed, that I don't have a specic rule, but that your intention was to avoid taxes while it was not the intent of the law/the rule was not designed for that, then I can actually apply a remedy, I can try to deny the benets of that. The way we've seen taxation in the 80's, 90's and even the rst decade of the 2000s was a barrier to the single market/integration. There were a number of Member States, we were 15 and up to 28 Member States, with various tax systems in place and companies have to face 28 tax systems with 28 set of rules. This is seen as a problem for companies that are trying to do Pan-European activities because they have to learn about all these tax systems. They may have some operations that's considered and will be classied as a transfer of debt in one country and another country may say that it is not a transfer of debt but a transfer of equity. This may have consequences and countries usually say they tax their activities and the other country may as well tax their activities: you end up with disputes and have to nd ways for a double tax relief. The concern was really about double taxation. This still exists today: there is still a lot of focus on trying to avoid double taxation. Indeed, it is one of the four principles of the treaty: it's the free movement of capital, alongside goods and people. Today, the idea is also to avoid that we nd ourselves in a situation where we have double non-taxation; that, in the end, economic actors do not pay taxes anywhere. Taken from the slide: ATP consists in taking advantage of the technicalities of a tax system or of mismatches between two or more tax systems for the purpose of reducing tax liability. It may result in double deductions (e.g. the same cost is deducted both in the state of source and residence) and double non-taxation (e.g. income which is not taxed in the source state is exempt in the state of residence) The concept of aggressive tax planning is something that they are currently pursuing. advantage of the tax system and its technicalities. It is taking You know that in legislation it's always dicult to qualify everything. For example if you have a hybrid product, it is dicult to say what is equity or what is debt. You have a number of criteria that will say if this looks more like this, it will be debt, if it's more like this, then it will be equity, etc. . . but you have some in-betweens/grey-zones where it's for the tax authorities to decide what it is but companies or economic actors will try to argue for one or the other and try and ll in their tax returns in a specic way. Same thing for the classication of VAT, for example. We know we have reduced rates and some companies may want to classify some of the products into specic categories. For example, imagine you have some marmalade, and you qualify for import duties if it's marmalade or not depending on the percentage of fruit, so you try and switch into one category or the other and play with the classication and always try to argue that your product falls in the lower category. It's also taking advantage of mismatches in the tax system for the purpose of reducing the tax liability and what you can end up with is situations where you have double deduction or double non-taxation or even a combination of non-taxation in one 76 country and deductibility in the other one. This is simply because the denitions between the two countries in the tax legislations are not the same. That relates to the example the professor gave between the denition of what is a resident of a company between Ireland and the Bermuda that allowed some companies to avoid taxation anywhere. This was because one was saying that they do not tax because they consider the residence of the company is where it's registered and it's not registered in their country. The other country said it's registered in their country but their denition of residence is where the decisions are taken and the decisions are not taken in their country, the board is having the meetings in another country so that country should tax the company. Basically no one was taxing them. It's a topic that was known by a few specialists. What the Commission decided to do a few years ago is to launch a series of studies year after year to try to dig into the issue. You have three studies that you can nd as taxation papers. The rst one was the rst study they launched: Ramboll - Corit advisory 2015 study (Taxation Paper #61). They asked an external consultant, a think tank/tax advisory company, to do a study on the most common forms of aggressive tax planning structures. In the end, it boils down to six or seven models of tax avoidance and most tax avoidance done by companies are a combination of these. Sometimes these were extremely complex with ramications, companies that own another one and a web of ownership that's quite complex to nd out but these are those seven. When the Commission is doing a matching for what is necessary for those tax optimization structures/aggressive tax planning structures to work (basically something missing in the legislation or something permitting it in the legislation) and compare this/match it with what's happening in the Member States. Then, the year after they did a second study by ZEW which is a German-based institute (ZEW on ETR of ATP Structures - Taxation Paper #64): imagine that you have some economic operations that goes directly from one company to the other one, what would be the eective tax rate on this operation and then what is the eective tax rate in the company that takes the aggressive tax planning route via, for example, a conduit company? By how much can it decrease the eective tax rate? Then, they started to work that out and notably started discussing this in the context of the European Semester. The European Semester is the cycle of economic recommendations to the Member States. Last year, for the rst time, they had country specic recommendations on aggressive tax planning asking a number of Member States to act/look at the phenomenon and propose legislation to address what they saw with a set of economic indicators (I'm guessing this part is the third study - didn't specically mention it though, just started talking about this: IHS, Dondena and CPB on Indicators of ATP - Taxation Paper #71). The economic indicators are, by nature, an activity that is dicult to measure because tax authorities do not necessarily know what's happening, you need very detailed rm-level data, you need tax record data, and so it's quite complex. But what you can check is whether or not the macroeconomic indicators are telling you something. For example, if you have a high level of nancial ows in the form of royalties in a specic country that is combined with an absence of withholding taxes or absence of specic taxation, you can suspect that there is something going on. It's never a denitive proof that you have aggressive tax planning between one country/towards one country/from one country but it's what we call a body of evidence. They have a set of indicators where, for some indicators, some countries are actually o the charts. You have, for example, a level of FDI that are multiples of GDP and most of it being special-purpose vehicles and then you say it doesn't match with economic activities. What they do in those cases is that they use a gravity model, models build notably in trade, where you try to explain trades or nancial ows or economic activities between the two countries by a set of indicators that includes the characteristics of the countries, the dierence between the two countries, and also, for example, the distance. You try to use this model to predict what the economic indicator of FDI, for example, would be and if you see that it's completely far o in terms of what you actually measure, the economic structures/activities, the underlying fundamentals cannot explain this level of FDI. So you go back to the country and say these are the gures, now do YOU have an explanation for this? Maybe there is a good one, we don't know but. . . What about the set of anti-avoidance measures? Remember, there are three main ways of do- ing prot shifting: the shifting of debt so that you can deduct your interest in the high tax country and tax the interest instead in the low tax country; there is transfer pricing, basically a mispricing of 77 operations between two related companies; and there is the location of IPs, notably the fact that you have patent boxes. You have major reforms that may actually try to address this. For example in debt-shifting, the advantage of using debt may disappear if you do a cash ow tax, a notional interest, or if you do a comprehensive business income tax where you disallow interest deductibility so you may actually have ways to completely change the system. It's more dicult politically to do it, you may be the only one and so most countries rely on anti-avoidance rules to try and address the excessive behaviour of companies/the excessive ways of playing with the tax system. Remember the idea is to try to deduct your interest (refer to Figure 79). You have a high tax country taxing at 50% and you have a low tax country at 10%. You have a prot of 100 and both a parent and a subsidiary. The idea is that you want the parent to pay interest that is tax deductible to the subsidiary because the interest that is transferred will be taxed at 10%. What you do is that you transfer debt, you create a debt of 1000 and you have an interest payment of 50. The interest payment will come as a deduction of the prot (-50) of the parent and will be added to the prot of the subsidiary. In the parent, there the 50 will be taxed at 50% so 25 and the prot will be 25. In the subsidiary, your prot goes up to 150, you are taxed at 10% so a tax of 15 which results in after tax prots of 135. You save on taxes simply by shifting debt from the low tax to the high tax country and having the company located in the high tax country to pay an interest that is tax deductible in a high tax country and taxable in the low tax country. Figure 79: Example They can do this ad libitum. As a country, notably the tax authorities of these countries that are losing tax revenues may want to limit the interest deductibility. They may accept that for X/Y/Z reason there may be a need to borrow from your subsidiary but they want to put a limit to this. There are two main anti-avoidance measures that a country can implement. The rst one is called the interest limitation rules (ILR)/earnings stripping rules. It says that they want to limit the deductibility of your net interest (net interest is what you paid minus what you received). They say if your net interest is superior or equal to a certain percentage of your EBITDA (typically in most legislations that's 30%), everything that is in excess is NOT deductible. You can actually deduct your net interest only up to X% (typically 30%) of EBITDA because as the tax authorities they consider that X% (typically 30%) is a normal level of interest payment considering your earnings. Everything that is in excess is not tax deductible so you don't gain by shifting prots. The other possibility is the thin capitalization rules and pay attention to the name. It's a rule against thin capitalization. Remember when you have a company, it has assets and liabilities. Assets can be divided into tangible assets and intangible assets etc. . . but in terms of the liabilities you have mainly equity and debt. That's the way you are capitalized. It's your level of equity in the total of your balance sheet. We say a company is thinly capitalized when it has a very low share of equity, it means it has a very high share of debt. It's a rule against companies that are thin capitalized. It says we look at a ratio and there is a wide range of possibilities that are applied in countries. Each of them has its own denition. It can be the debt-to-equity (D/E), it can be debt-to-total assets (D/D+E), it can be internal debt to equity (so you can either borrow from the outside like a bank or from other companies that are not related but you can also borrow within the group and the one that's usually used to do prot shift is within the group) or internal debt to equity owned by the one you borrowed from. There is a wide range of possibilities and if it's beyond a certain level, for example 1.5 for D/E, then either non-deductibility of the excess happens, there is one or two countries that says there is 78 non-deductibility of everything so nothing can be deductible as soon as you pass the threshold, and some other countries go even further and say there is a possible requalication: meaning what is in excess, they do not consider this to be debt anymore and consider this to be equity. It's a disguised equity that you've received from the other company that carries additional taxation. You have a wide range of possibilities: non-deductibility of the excess, non-deductibility of the total, requalication of your interest into equity. This remedy can be either automatic, the tax authorities says that as soon as you're over the rule you have to apply it, and some others will look at the situation and if the company can prove that it's a benign economic activity that is not for tax planning, they may allow the deductibility. So the tax authorities have some discretion in that case. Those two rules have the same philosophy, interest limitation rules and the thin capitalization rules: as soon as you have an amount of interest that you want to deduct that is above a certain threshold, dened either as a percentage of EBITDA or as a percentage of a ratio of some debt to equity or assets, then there is non-deductibility on the excess, on the total, or re-qualication. It's a measure that tries to capture the extreme cases. That's for the ones on interest. Then, there is another measure which is called the CFC rule. CFC stands for controlled for- eign corporations. The situation is the following and it's a situation that was quite frequent under the previous US tax system. Figure 80: Example You have a parent in the US where the tax used to be 35% that has a subsidiary in Bermuda that has a 0% tax rate. A lot of the assets are in the US company, most of the workers are in the US company, but strangely enough most of the prot is recorded in Bermuda. There is a level of prot to economic activity that is unexplained. Say there is a prot of 100 in the US and a prot in Bermuda of 10,000. Normally, what happened is that under the US tax system, you had a worldwide tax system. It means that they tax the prot of all its entities in the world as soon as it's distributed and they will provide a tax credit for what has already been paid abroad. Normally, the prot in Bermuda should not be kept in Bermuda but should be paid as a dividend at some stage to the US parent because the shareholders invested in the company, see the prots over there, but want to see the benet of this prot that has been made as dividends. Normally what the US would say is you have a dividend of 10,000 that is transferred, it has not been taxed in Bermuda, it's taxed in the US so 3500 and you apply a tax credit for what has been paid abroad. What has been paid abroad? Nothing, so the level of tax credit is 0. That's normally the situation. The company may have a trick to avoid taxation when it goes back to the US. It's a tax that's applied when it is repatriated so what they can do is just not repatriate. Keep everything in Bermuda for ages and accumulate all your prot there. Every year you accumulate it there and never distribute it to the US. You never pay taxes in the US, it stays there and hope that, as it happens every 10-15 years, all of a sudden, someone nice in the country that says tax amnesty, we allow you to repatriate everything at a rate of 5% or 10% and then you are good. That's what a lot of countries were doing and if you see the situation of many US multinationals, they have a lot of cash piled up in many foreign countries. The tax authorities in the US may not be very happy about that and say that at some stage, this prot here (in Bermuda) should come back to me and should be taxed at least. They don't care if you actually bring it back but they want to put some taxation on it so they can apply a CFC rule. The CFC rule says that if the tax rate applicable in the country of the subsidiary is below a certain percentage of mine, usually the statutory rate, then if it's below that (for example, in the US it was 50%, so the US said if the tax rate in an other country is below 50% of my own rate which is 35%, 79 so if it's below 17.5%), they consider that the prot that has been generated in this country, it is as if it has been generated in the US. So the tax authorities will go to the US parent and they apply the CFC rule, you had 10,000 that has been generated in Bermuda, the rate is below half of their rate, so they apply a 35% tax rate on this prot and you have to pay it. That's the way the system goes. Of course if there is a 10% tax rate in Bermuda, it would still be below the 17.5% but the US would say they tax everything and give you a tax credit for what has been paid already in Bermuda. CFC rules can also be combined with blacklists, so the tax authorities have a list of countries against which they automatically apply the rule. They can also have whitelists, they never apply those rules because maybe they are treaty partners or other members of the EU. That was the situation on paper in the US. In practice, there was in the US legislation (there is always a possibility in the US to always include amendments in legislation that is unrelated) a rule called Check the Box. The check the box rules said if you are under certain conditions (professor doesn't remember what they were but they were very easily fullled), the CFC does not apply to you. For that, you just had to check a box in your tax declaration saying they fullled the conditions. It's a loophole in the US tax legislation. Someone asked for a clarication of the CFC rule under the condition that there is a tax in Bermuda that is lower but not zero. Figure 81: Example Imagine the rate in Bermuda is 10%, so you have a tax of 1000 now. The US will say you have a prot of 1000 that has been taxed at 10% but the US CFC rule says that as soon as your rate is lower than 50% of mine, which is the case here, the US applies the CFC rule. So what do they do? They consider that the dividend has been distributed, in this case it will be 9000. Remember though that when the tax authorities look at the dividend, they gross it up. They look at the original amount so the US tax authorities will say it's 10,000, they apply the 35% tax rate, so 3500, and give a tax credit of 1000 and the tax liability in the US is 2500. In the end, what the company pays in total is the US rate, 1000 in Bermuda and 2500 in the US. The US tax system today has changed: there was a reform of the US tax system. There was a rst proposal to have a destination based cash ow tax which was heavily debated in academia. It was the cherished system by a lot of economists. It failed. The Republicans in the congress and the senate passed a tax reform - Trump Tax Reform - that decreased the rate from 35% to 21%, that allowed for two years of immediate expensing of investment so instead of depreciating you can immediately expense all your stu, and a move from a worldwide to a territorial system. Now the US says everything that is produced abroad we exempt and for the prot that has been parked, we also have a low repatriation tax. What happens for the companies? If they have the prots in Bermuda, they have absolutely no incentive NOT to repatriate the dividend to the US because it will be exempted anyway. The idea of the US is to say given the fact that any way this prot stays there and that we cannot tax it, because of the check the box or other things, instead we will exempt all of this but at least we have the repatriation of the prots into the economy that will actually spur investment and bring back money. That was the US tax system and they have a new rule that like a CFC rule that is called the GILTI rule (pronounced guilty). Sort of the same rule with very complex details. The other one that we will talk about is the Benecial Owner Test. It's actually a standard for EU countries between themselves and a standard for all the countries that follow the OECD model. You know that for tax treaties, usually you follow the OECD model. It's like a template and can choose 80 which article you follow and which interpretation that you give, there is a sort of model by the OECD with also a reading by the OECD of what each article means and then you can choose the rate that you apply etc. . . The idea is to have the same tax treaties that are compatible with each other. Quite often, to avoid double taxation, it is foreseen in those treaties that for passive income (all the income in the form of interest, royalties, or dividends; as opposed to active income which is the prot that you make) we avoid having a withholding tax or oer a reduced withholding tax. Here normally if you have repatriation of the dividend between the two countries, you can have a withholding tax, the country can say they apply a withholding tax of 15%. That would be in addition to the tax that you would have as the CIT in the two countries. To avoid that there is double taxation, usually with treaty parties, you have a reduced withholding tax or no withholding tax. You can understand that as soon as you have this rule that oers lower withholding tax, it's also a way for companies to change the way they will do their activities, they will try to go through places where the withholding tax is the lowest. What countries can do is to say that they agree to give a low withholding tax but on the condition that the company receiving this withholding tax is the real one that will receive it. What they mean by that is that they are not just an intermediate, it's not a system from which you want to benet from a reduced withholding tax because there is a transfer of interest, but then this interest will shift again to someone else. There is a benecial owner test. We want to make sure that the benecial owner is actually the one to which we apply the low withholding tax. Then there is the GAAR - general anti-avoidance rules. Countries say well beyond this there may be other ways for companies to try to play with the system, play with the mismatch, play with the uncertainties, play with the gray-zones. If it's something for which we don't have a specic rule, we have a safety net which is the general rule that says if there is something where we think it's not something that is done in accordance with the intent of the law, we reserve the right to act, to correct, to refuse the operation, to refuse the benets of some tax advantages. Oshore Loan ATP Now we will discuss a rst example which is an oshore loan ATP. The structure is the following. You have a parent company, the MNE, that is located in a high tax country and that has a subsidiary, Holding B, in another high tax country. They are linked by some ownership relationship, the MNE owns Holding B via equity. If you disregard the oshore part and only had the MNE and the Holding B, what can they do? If there is a transfer of equity and a dividend that is paid, the transfer of equity has no eect on the tax liability. There is no tax consequences of owning another company and the dividends that are paid to the parent are not taxed, usually it's exempted. Imagine it's within the EU and you have the parent-sub directive that says we exempt the taxation of dividends. They are not deductible in the holding company and not taxable in the multinational. Then, the chief tax ocer has an idea, they will use a conduit company located in an oshore jurisdiction (by oshore we mean a low tax country) and we will do the following completely articially: you will ask the MNE to inject some equity into the oshore company that will lead to a dividend payment to the MNE and with this money, the oshore company will lend money to the holding company, that will create the payment of interest from Holding B. Thanks to this, we can decrease articially our tax liability. The dividend that is paid is exempted at the level of the MNE, it's not deductible at the level of the oshore company, but what's interesting for the tax planner is that the interest that you can pay thanks to this system is deductible in the Holding country where it's a high tax and it will be taxed instead in the oshore country where there is a low or no tax. If you had done this operation between the two via interest and 81 debt, the interest would have been paid from the holding to the MNE, it will be deducted against the high tax in Holding B but re-taxed at the MNE against a high tax as well. The tax advantage would probably be zero or small because these are two high tax countries. What you want when you do this is to do this between a high tax and a low tax country. We change equity into a loan and thanks to this we have a tax advantage. What makes this possible are a series of conditions. It is the dividend exemption in the MNE country so they exempt dividends received. It's the low or no taxation in the oshore country. It's also the fact that the country of the holding company does not levy withholding taxes on outgoing interest. If they were levying a withholding tax on the interest, the tax advantage is decreased. These are the conditions for this to work. What are the things that the tax authorities in each of the countries can do? The country of the oshore: we consider that it's a lost cause, we cannot ask the country to raise its corporate tax rate. For the two countries remaining, to defeat the ATP scheme, what they can do is the following. At the level of the country of Holding B - they can have a thin cap rule or an earnings stripping rule that limits the deductibility of interest. They can deny the absence of withholding tax via a benecial owner test. They can say wait a minute, we allow you to benet from no withholding tax because we thought that the benecial owner was located in the oshore country with which they have a tax treaty (even if it's a low tax country), but actually we realized when we looked at the operation that this is not the benecial owner, the benecial owner is the MNE over there, so we deny the benet of the low withholding tax. The last thing they can do is the GAAR. They say all this is ne but we looked at your scheme and the intention of your scheme is to decrease your tax burden. There is nothing that is economically rational to it other than decreasing your tax burden so they disallow the benet of it. If we look at the country of the MNE: the one thing they can do is to apply a CFC rule vis-a-vis the oshore company. The oshore company says they receive 100 of interest, that's the prot that they don't tax and they send it to you as a dividend that is exempted. The country of the MNE may say that this comes from a prot of 100 that has been taxed at 0, if they apply their CFC rule (in this case as an example, your rate is 20% below mine), they do as if the prot generated in the oshore country would have been generated in the MNE country and tax it at the MNE rate. They would tax the dividend received at the MNE tax rate, that would defeat the ATP scheme. You see that it's relatively simple as a structure. The problem is, of course, for the tax authori- ties to know the scheme because each of them see only a part of it in the absence of exchange of information. That's why exchange of information becomes extremely important. The gure above on the right shows the rates that you have. The professor would like us to con- centrate not on the cost of capital which is the inverse way of looking at the eective tax rate but on the average eective tax rate. If we would do a direct cross-border nancing (that is, directly borrowing between the two companies), the eective average tax rate in the EU would be 20.9%. If we would do the same via another EU country, there is a slight increase in the eective average tax rate to 21.6%. 82 This is because not all countries exempt the dividends at 100%, some exempt it only at 95%. You have a slight increase in the rate. If we use an oshore jurisdiction, you have the following: if you use one with which you have a treaty, it goes down to 16.2% and if you use one with which you don't have a treaty it goes up to 36.4%. Why is that? What makes an oshore country dierent if it's a treaty partner or a non-treaty partner? The benet of the treaty is avoiding double-taxation via the absence or low withholding taxes. So, if you do it via a low tax jurisdiction, then the eective average tax rate is 16.2% but if you have a non-treaty country, immediately you would have a withholding tax on the interest that is paid and possibly a withholding tax on the dividend from the oshore company to the MNE. You see that it's not necessarily via the completely exotic tax havens that prot shifting occurs but it's sometimes with very low tax but with a treaty partner jurisdictions and one of the ways to do prot shifting is to do treaty shopping. You actually have models that try to compute this, it's linear programming. You take advantage of the benets of the tax treaties to avoid all withholding taxes/to avoid double taxation. Hybrid Loan ATP The second example is a hybrid loan ATP. You have a multinational that provides equity to Holding B and receives dividends in exchange. The Holding B gives a hybrid loan to another holding, Holding C, that pays interest in exchange. They are all high tax countries. The problem however is that there is a dierence between the country of Holding B and the country of Holding C on the denition of the hybrid loan. Is it debt or equity? The country of Holding B says it's an equity injection so the payment that you are receiving is a dividend and the country of Holding C says they have received debt and what they pay is an interest. They disagree on the denition. The MNE pays equity to Holding B, receives a dividend that is exempted, no tax aect there. There is the provision of some capital, some say equity and some say debt, and there is a payment from C to B and the country of Holding C says it's an interest so it's tax deductible in their country and should be taxed in the country of Holding B as an interest received. Country B says it's a dividend so it's exempt. In the end, to the group it's a situation where they have a deductibility in in country C and no equivalent taxation in the country of Holding B. We have a deduction that is oered on this operation, simply because there is a dierence in the tax legislation on whether this operation is an equity or debt. What are the conditions for this to take place? Dividend exemption in the MNE as before. No withholding tax on the interest paid in the country of Holding C. For the country of Holding B, there should be no withholding tax on the dividend that is paid and a dividend exemption on what they consider is a dividend that has been paid. As soon as this happens, the scheme is possible. 83 Referring to the gures below. What could the countries do to defeat this? The country of Hold- ing C can limit the deductibility via thin cap or an earnings stripping rule. They can also apply the benecial owner test and say the real benecial owner is not Holding B but the MNE and you only do this to basically avoid taxation so the benet is denied. They can also apply a GAAR where they say they looked at the way you planned your operations and we think that the intent is to avoid taxes and we disagree with this. The Holding B country could suppress the dividend exemption but it's normally a rule that is there in the EU to avoid double taxation so probably what they can do is a GAAR. Once again, they say you planned this here because you know there is a dierence of denition that you tried to exploit. The country of the MNE can apply a CFC rule. They say that you've received what you consider a dividend that has been exempted so it has not been taxed and the eective rate on which you've been taxed is below what the tax authorities consider a normal one. So the tax authorities consider as if this dividend is the prot and has been generated in their country, they disregard the fact that it comes from a dividend and they apply their own rate. That would defeat the aggressive tax planning scheme. What about the gures? It's 20.9% for the direct cross-border nancing: if you would go directly from the MNE to the Holding C. A very high eective average tax rate if we go via a hybrid loan via an oshore non-treaty country: 36.4% because in this case you would most likely have withholding taxes in both places. If you go via a hybrid loan in an oshore country with a treaty, it's 16.2%. And if you go via another EU member state, it's 14.3%. In this case, it's even more advantageous to go via another EU country than to go via an oshore treaty partner. Aggressive tax planning is, once again, not always the paradise islands, it's quite often within the EU. Patent Box ATP The third example is patent box aggressive tax planning. 84 You have a multinational that owns a holding company (Holding B) and the MNE has a patent on the design of a wonderful phone. They decide to give this IP right to Holding B. The holding company B is now the owner of the rights and licenses the use of this right to another holding company - Holding C: we allow you to produce this smartphone with my design and in exchange you pay a royalty. The royalty is paid by Holding C to Holding B. It's tax deductible in the Holding C, and taxable in the Holding B, but Holding B has a system of patent boxes. It says when you receive revenues from IPs, you don't tax them under the normal rate but under a reduced rate. This is equivalent to saying they exempt part of the base. There is tax planning that is done and the holding company makes a prot thanks to this. They receive royalties and they pay back a dividend to the MNE. The MNE would not do this directly because the MNE is not located in a patent box country. What they want to do is give the IP rights to a holding company that is set in the patent box country so that they can benet from this low rate taxation. The planning is basically to transform the payment of royalties that will be taxed at a low rate into dividends that will be exempted when they come back to the parent. What are the conditions for this to work? No withholding tax on royalties paid in the country of Holding C otherwise it's decreasing the benet. A patent box regime in the country of Holding B and no withholding tax on the dividends that are paid to the MNE. The MNE country needs to have a dividend exemption on the dividends received and also low or no taxation on capital gains that they will have when they transfer the property rights to the Holding B. Otherwise, if there is a gain that is made and taxed, it's decreasing the benet for the group. What can these countries do to counteract this? We ignore the patent box country. Imagine that this is within the EU, you could have a benecial owners test in the country of holding C. The benecial owner is not Holding B but actually the MNE so you disallow the absence of withholding taxes on royalties. You can also have a GAAR where you say you tried to fool us, all of this is just made to benet from the patent box and repatriate afterwards the benets as dividends that will be exempted back to the MNE. The MNE country can have the CFC rule: they can say that you have some prot here that was taxed at 5% (for example) under the patent box and they consider that this is below a certain threshold of their own rate, so they consider the dividend that is repatriated as prot, that is: as if it has been made in their country so it is taxed at the rate of the country of the MNE and will provide you with a tax credit. What are the benets? Direct cross-border nancing is at 20.9%. The eective average tax rate is at 40.7% if you go through an oshore country with no treaty. It is at 18.2% if you go via an average EU country that possibly has no patent box. The eective average tax rate is at 2% if you go via an oshore treaty country and -1% if you go via the most benecial EU patent boxes. It means that the tax system actually gives you a subsidy. You actually have a tax advantage in the real sense if you do this. This was due to some circumstances of patent boxes where the deductibility was even higher. You see 21% to 2% and close to 0%, it's not via the oshore non-treaty countries where you imagine it's only sunny and palm trees. It's happening within the EU. 85 8 Lecture 8: Policy Responses This lecture is divided into three sessions due to group presentations. 8.1 Session of 27/11/2019 We've seen in the previous sessions a lot of phenomenons. We've seen that CIT created domestic distortions that we may want to correct. We've seen that there were international distortions and those international distortions: it's both a mix of size of reactions of companies when there is taxation in terms of location, investment, and prot shifting activities and the fact that we have three ways of prot shifting (debt, location of IP, and transfer pricing, and a fourth one may be added: the use of tax treaties) and we've seen an additional element to this debate which is tax competition. We've seen that there has been a constant decrease of corporate tax rates in the EU, every area in the world, but specically in the EU that leads to some tensions. That's also something that policy makers may want to consider. Corporate Tax Harmonisation The policy makers have come up with the idea of having some sort of corporate tax harmonisation. In July 1962, there was the Neumark. Neumark Committee which was chaired by a tax law professor named What this committee had proposed already in 1962, that's not even ve years after the European Committee, is to gradually harmonize the tax systems in Europe, starting with turnover taxes (VAT) and extending it to direct taxation eventually. The proposal of this Committee was never followed by any policy action and you know that in terms of policy action, we are still under the rule of unanimity for taxation. That means that this can be a proposal from the European Commission to the Council, the parliament doesn't have a formal right to decide, they are consulted so they will issue an opinion on the proposal. It's the Council that decides at unanimity, and then if the proposal is accepted, it is adopted and there the Commission has a second role of guardian of the treaty to make sure that it's translated into national legislation. In 1970, there was also another report called the Werner report on the economic and monetary union in Europe with the idea that if you want to have a successful monetary policy, you also need a way to harmonize taxes. It's a debate that's still here today at least in some circles. You have this tension between those who say they have lost the monetary instrument so they want to keep the scal instrument to be able to absorb economic shocks, otherwise the only thing left is the regulatory/structure reform instruments and it's not popular and those who say well actually you can have monetary policy if you also have a form of harmonisation of scal policy and the tax instrument. There is an increasing number of papers that started looking at the eect of taxation on the measurement on GDP, the measurement on productivity, and also on the way we measure balance of payments. Balance of payment should equal 0 in total (you have the capital accounts and the current account). It is clear that if you have tax policies that create movement of capital, for example you decrease your corporate tax rates or your personal income tax rate, you become much more attractive for investors, you'll have an inow of capital, while this inow of capital will create a disequilibrium on the capital account that needs to be matched in the current account. You have this tension in the balance of payments, the composition of the balance of payments will be disturbed by these ows. For example, royalties are not registered under the same items/chapeau/category as, for example, dividends or interest. If you have a country that attracts a lot of interest and that are transformed into a lot of royalties or vice versa, for example, it may be that the netting out of the level of country, that the composition of the balance of payments will be aected. These are all elements that some people started to look at. Two resolutions by the council in 1971 and 1972 backing tax harmonisation, so you see that at the time, the council was in favour of harmonisation. In 1975, a proposal by the commission to harmonise corporate tax rates between 45% and 55% which gives you an idea of the rates that companies were facing at the time. Today, we're in the range of 20% to 25% on average. 86 The Nyborg Report in 1979 from the European Parliament trying to back sort of a tax base harmonisation prior to the harmonisation of the rates. In 1988, there was a Commission proposition also on the harmonisation of the base but it was never sent to the Council. In 1990, there was a shift in the policy of the European Commission. There was a new French commissioner and there was a shift from issues of harmonisation to issues on ghting double taxation. It was really the time when what people cared about what the single market and to make sure we wouldn't get any barriers, whether they were of a technical level or taris, but also on taxation, notably double taxation issues. You wanted to avoid the following: you would be double taxed when you do a cross border transaction whereas when you do domestic transaction you are only taxed once because this would be detrimental to the single market and create a distortion. In 1992, the Ruding committee: Ruding was a Dutch professor who tried to propose some min- imum standards in corporate tax bases. Then he called for a band for tax rates that everyone should be between 30% and 40% but no political actions followed this either. In 2001, a big communication on company taxation in a Single Market. There were several elements, there was rst of all a study on eective tax rates that followed the Devereux-Grith methodology that we've talked about. There was marginal eective tax rates and average eective tax rates and they tried to look at the sources of the dierences in eective tax rates across Member States. Then, it contained a long list of tax obstacles to the single market. A long list of issues where there should be policy actions by amending or proposing directives. There was also, besides those targeted elements, already an attempt to try to nd a comprehensive solution and to say what could we do if we were to harmonise corporate taxation in the EU? There were four solutions that were put on the table. The rst one was a full harmonisation and having a EUCIT: an EU corporate income tax that would be the same base and the same rate for everyone in the EU. That was then considered as politically unfeasible because there were too many diversions. Another option was a compulsory harmonisation of bases: all companies in the EU would have to apply the same way of computing the tax base for corporations and then they would be able to tax at the rate that they want. So each country retains its right to use the tax rate that it wants so Belgians was 33.99%, the Irish at 12.5%, but at least there would be a harmonisation between the countries on the way we compute the base that would be compulsory. There was then optional harmonisation of bases (CCCTB): so we would create an additional system along the existing one so where companies could opt for this EU way of computing the rates because people thought that it could be benecial to large groups that operate in various countries. The idea was to say that they should not force small companies, local plumber, to adopt a CCCTB, it's burdensome etc. . . but for large groups that operates cross-border why not oer them this EU system where they could decide to opt in? That was the idea to facilitate business. Then, there was a last proposal that at the time was also discussed and popular which was called Home State Taxation. The idea is to say that instead of going through the diculty of harmonising the base and agreeing on what should be this harmonized tax base where we need to decide what is the right depreciation rate? What is a tangible vs intangible asset? If we can deduct pensions or not? What is a pension payment? All the elements that constitute a corporate tax base. Why don't we simply apply the system of the parent company? If there is a group that is ultimately owned by a Dutch parent that has subsidiaries in, for example, the Netherlands, France, Germany, and Belgium: all the companies that belong to the group would simply apply the tax base computation of the parent. That was a system that was at rst sight quite simple because you don't have to create a new one, you only have the one of the parent that is already existing. The problem is the administrative burden. All the companies of the group would need to know how to recompute the base based on the system of the parent. That could be okay but the problem is that the tax authorities of each country would need to know about the tax system of all the others because if you are the Belgian tax authority and you have a subsidiary in Belgium of a Dutch parent, you need to know the Dutch system. If you have another subsidiary that is a subsidiary of a German parent, you need to know the German system. In the end, all countries end up having to know about the details of all the corporate tax system. That was not pursued for this reason. 87 Based on these and on what was considered in the communication as the preferred option which was the optional harmonisation of the tax base, there were discussions that started at the level of the Council in technical groups. ECOFIN in 2004 asked the Commission to lead a working group to explore the possibility of having a CCCTB. CCCTB stands for common consolidated corporate tax base. Common and consolidated are two important elements. Common means a harmonisation of the way we quantify/compute the tax base and consolidated means that at each group, we will consolidate the prots/results/benets so that it doesn't make sense for a group to shift prot from one side to the other because at the end of the day all this prot is consolidated and will be redistributed to the dierent entities according to a formula. That's the idea behind the CCCTB. The rst proposal came under Commissioner Semeta for the CCCTB in 2011 that was put on the table. It was not successful in terms of the political negotiations. There was some opposition and in the end there were no agreements at unanimity in the Council but the technical work continued. That's an important element because at the end of the day it also shows that there is some hope because if you don't want something you don't bother looking at the work on how to implement it. The technical work was on nitty-gritty details but they were important for tax certainty for taxpayers. We should not have dierent interpretations on dierent articles across countries otherwise this is defeating the purpose of having a common base. A new commissioner from France came and there a package was put on the table in June 2015 on Fairer Corporation Tax System in the EU and the Future of MNEs' Taxation. The idea there is that we were in a dierent environment. It was an environment where we had the rst scandals, notably the Lux Leaks. There were also some other scandals like the UBS scandal implying a lot of US taxpayers, the Lichtenstein Scandal. UBS basically was accused by the US government to having helped wealthy US citizens evade taxes abroad. There was a whistleblower who all of a sudden revealed the case. There were stories of people transporting information in toothpaste. The tax authorities asked UBS to provide the names of the people that were involved. UBS rst refused and then the US authorities basically said ok but if you don't we remove your license to operate in the US and that's when they agreed to negotiate and agreed to give a partial list to the US authorities of the names of people involved. In the Lichtenstein case, it was the case of individuals who had accounts there that were not declared to the US authorities. So it was a dierent environment where the issue of fairness was much more acute. After the scandals, there was for the rst time at the G20 a conclusion that was extremely strong on the issue of tax evasion and tax avoidance by saying that this is becoming unacceptable for policy makers. We were just after the nancial crisis, still in the aftermath of it, the public opinion started saying that they have suered from the nancial crisis, they paid for saving the banks, they don't understand why they were the only ones paying and feeling like they're the only ones paying for it because several governments had to increase taxes to deal with the consequences of the nancial crisis, and nd it unfair that some stakeholders, notably large MNEs, that they do not consider are paying their fair share. That was sort of the political landscape occurring at the time. They put a communication on the table along with a sta working document in June 2015 and in this communication they were basically proposing several elements of a package among which was a relaunch of the CCCTB. CCCTB This package was accompanied by an impact assessment: it's an analysis that is quite standardized at the level of the Commission. The process now for any legislative proposal that is being put on the table needs to be accompanied by an impact assessment, it's compulsory. We need to dene the problem, usually you draw a problem tree: the links between the dierent elements - where is the problem? Where is it created from? What are the consequences? You need to dene the objectives: it can be broad or specic. You need to dene all the options that you consider: there is always a default option of doing nothing and then all the proposals. Then you will look at all these proposals against the objectives that you've set in your assessment/that you want to achieve. There were many types of analyses: usually you look at the eects on the macroeconomic indicators. It very much depends on the quality and availability of the data. You have impact assessments that are more of the qualitative nature if there is no data available but there is also impact assessments that can go quite deep in the analysis with economic regressions, a review of the economic literature, etc.. Then this impact assess- 88 ment when it's nished, it goes to an independent body that is attached to the Secretary General but it's independent. It's composed of high ranking ocials in the Commissions and external academics who will give an opinion and the opinion is positive, positive with comments or reject and the idea is that you don't want to have a reject. If you have a reject, you have to rework on your proposal, adapt it, and usually once it's cleared by this board, it goes on the table of the college, and once it's adopted everything is published: the proposal, the opinion of the board and the impact assessment. You must also explain in your impact assessment how you've taken into account the comments of this independent body. It's quite a process and it's something that the professor's unit at the Commission is doing. Think of the independent body like a referee/peer review. The priority of the Commission with the proposal is fair and ecient taxation. There are issues of transparency. It's also the beginning of the BEPS (Base Erosion and Prot Shifting which is the name of the phenomenon but also the name of the program) process at the level of the OECD. When you talk about BEPS, sometimes your referring to the problem base erosion and prot shifting and sometimes the initiate at the OECD where there were 15 actions to try to tackle base erosion and prot shifting. One of the elements that was already there on the table is that there was an agreement on the BEPS that all the countries at the level of the OECD would apply a set of anti-avoidance measures. There are ve anti-avoidance measures that countries agreed to and they were translated in the EU into a directive so that Member States would apply those and it would be under EU legislation. It's called the ATAD - the Anti-Tax Avoidance Directive. The idea that we should tax activities where it takes place is another context of the proposal. There is another mantra today: we should tax activities where the value is created. Everybody agrees on that except nobody agrees on what it means. What is value? Where is it created? Is it where you sell? Is it where you've done the research? Where you've done the physical process of the product? Is it where you have your nancing? Is it where you created the IP? Is it the place where you put all the risk to create the product? That's the big question with a divergence of opinion on what exactly value means. How would the CCCTB work in principle? You have a group that's located in several Member States. You have a rst step which is the harmonisation of the way you compute the tax base (1. Tax Base Computation). So you have a common base, you dene how you allow depreciation, allowance for R&D, how do you agree on the same way of deducting some pension payments? how do you agree on what is a tangible and intangible? etc.. Once you've done this, of course, each country would still have its own rate. Even if you harmonise the base, which is the rst step. It could still be, in the example of slide 14, the group tries to report all the prot in Lithuania which has the lowest CIT rate and still make some prot shifting. The idea is to say let's try to avoid this, let's consolidate at the level of the group (2. Consolidation). We basically look at the group prot in a consolidated manner so that wherever you report it, at the end it's consolidated centrally. Then, we should redistribute this consolidated prot according to a formula (3. Apportionment based on 3 factors). It's formula apportionment. In the case of the CCCTB, it's three factors with equal weight: 1 Ai 1 Sales 1 P + P + 3 Ai 3 Salesi 3 1 W ages 1 L P + P 2 W agesi 2 Li A third on the share of assets to the total assets of the group, a third on where the sales are located over all the sales of the group, and another third that is based on labor which is a mix on: one half the share of wages and another half which is the share of labor. The idea is to have factors that are more dicult for companies to play with. You consolidate the prot and then you try to gure out for example what you give to Denmark. What's the share of Denmark to the total assets? What's the share of the total sales? And the share of labor/wage in Denmark? That will give you a proportion of the prot that will be reallocated to Denmark and that can be taxed at the Danish rate. The CCCTB doesn't touch on the freedom of Member States to apply the rate that they want. The idea is that they get a share of the consolidated prot depending on a series of factors that are dicult to manipulate by companies. It's much more dicult for a company to remove your labor somewhere, remove your assets somewhere, the sales: it's very dicult to manipulate where your nal consumer will be. It's 89 much more complex to do that then to articially shift prot as the way they do today. It's not without problems, some critics say ok companies cannot manipulate this anymore but it creates a tax competition on the factors themselves. Countries, instead of trying to attract prot, will try to attract assets and will try to attract people in order to benet from a larger share of the corporate prots. Some critics say you're actually shifting the tax competition from a tax competition on rates to tax competition on the factors. Again, they are much less elastic than the prot so it's already an improvement. There was a re-launch in October 2016 where the 2011 proposal was withdrawn and put on the table two new proposals: one for a common base and one for consolidation. Why the two stage approach? It's more for helping the negotiations since everything at once might be too hard to implement. The idea is to have the two but try to get an agreement on one before the other. One thing that has changed as well: the 2011 one was optional, it was to help companies who wanted to have a single system to operate across countries so they were free to do it. Here, the landscape has changed in the sense that they would make it mandatory for groups that have a worldwide turnover that is above EUR 750 million globally. This threshold was made to match with other thresholds they have in other OECD proposals. Why compulsory? It's because compared to 2011, they think there is also a problem of prot shifting and aggressive tax planning activities and that if they let companies choose between the two systems, they will always choose the most favorable one so it must be compulsory for them. They didn't want to make it compulsory for everyone because it needs to have some proportionality. It's a criteria for the impact assessment: proportionality of the measure. Let's not try to kill a y with a cannon. And for smaller companies it's optional, didn't feel like it must be compulsory. All the anti-tax avoidance elements have been put into the common base. Otherwise, companies could not apply them so it needs to be taken from the anti-tax avoidance directive into the CCCTB proposal. They had some technical modications to reect the evolution of the discussions since 2011. They had two additional elements that were not present in 2011. They have national legislation today on corporate taxation that for many countries foresee that we have R&D tax incentives. Why do they want to do this as policy makers? It's because the social return on R&D is larger than the private return of the companies so it makes sense to try to incentive companies, to try to invest in research & development and innovation because basically if they only look at the prot that they could make, these products are replicable by others afterwards, you may get a patent for a while but afterwards your product is available for others to replicate. We want to incentivize and need to keep at least the same degree of research in the EU as it is today otherwise we are basically lagging behind the other parts of the world and in terms of economic growth, it's also detrimental. They wanted to have an element which was a super deduction for R&D expenses where companies could deduct more than 100% of their R&D expenses. The other one is an element called the allowance for growth and investment on equity increases. It's an ACE - a notional interest system. It's like the system that we had in several countries where we thought there is a debt bias, we need to correct it, so we need to have in the CCCTB an element to correct it. They allow a notional interest on equity to be deductible alongside the interest payment that companies would pay on debt. However, they looked at all the systems and derived a system that would be proof against manipulation. You had a system, for example, in Belgium that was based on the stock of equity and where you could cascade the benet in the group. Basically the parent borrows money, injects equity in a subsidiary, the parents benet from its own equity notional interest, with the money the subsidiary benets also from the allowance for corporate equity, injects it into another one that also benets from the same ACE, etc.. There was a cascading of the benets. What they've done is a system that is rst incremental: they compare your equity level with your equity level ten years ago and the allowance is on the dierence. If you increase your equity base, you get this additional allowance. If you decrease your equity level, you have a negative allowance it means they increase your tax base. There was also a system that was anti-cascading because from the benets of the parent company, you were deducting the benets of the subsidiary if it has any. There was a system to make sure that basically you get one deduction/allowance but only 90 once on the original amount of capital that is put in the group. Some words on the fact that it's compulsory. Groups that are over EUR 750 million: it's about 1.6% of the total groups, so it's not a lot in terms of numbers but they represent 64.2% of unconsolidated turnover. They are few in numbers but matter a lot. These groups are responsible for two-thirds of all the turnover of all the groups. Figure 82: Preferred Option For R&D Incentives This is the R&D scheme. What you have is that for companies, you are able to have an immediate expensing of your R&D expenses: you can deduct 100%. You can deduct an additional 50% of the R&D cost on the rst EUR 20 million and an additional 25% above 20 million. For start-ups (companies that are younger than ve years with less than fty employees), you basically have an addition 100% up to EUR 20 million instead of the 50% so you have some good incentives to invest. Figure 83: Allowance for Growth and Investment This is the description of the Allowance for Growth and Investment. What is noticeable is that the yield is based on a 10 year treasury bond with a risk premium of 2 percentage points. Figure 84: CCCTB Re-Launch 91 8.2 Session of 04/12/2019 I skipped this session. No notes for the following slides Lecture 8 22-31: 8.3 Session of 18/12/2019 Anti-Tax Avoidance The rst directive we will talk about today is the ATAD - the Anti-Tax Avoidance Directive. It's actually two directives that have been there to implement what had been decided at the level of the OECD according to the BEPS rules. It's an anti-avoidance package and in it you have ATAD 1 which is a set of anti-avoidance measures that we've already explained in the past. The rst one is a CFC rule: now, countries have to implement in their legislation a CFC rule. Another is the GAAR: it's like a safety net, everything that you could not anticipate could now be captured by more general rules saying well if you're goal was to try to escape taxation then I retain the right to act. The third one is exit taxation: which is linked to the transfer of IP and the fact that sometimes this transfer of IP goes without taxation. There is some sort of capital gains that is made thanks to the transfer 92 of IP to another jurisdiction, What we want to avoid is that this is done in a way that allows them to escape taxation and that this would go to other jurisdictions without any taxes being paid on this. There is also the interest limitation rule: you cap the deductibility of interest, in this case, at 30% of EBITDA of the companies. Then you have a fth element of the ATAD which concerns the hybrids and the way countries need to deal with them. Hybrid: A product that would be considered on the one hand as debt and on the other as equity. The Anti-Tax Avoidance Directive 2 (ATAD 2) is a complement to these hybrid mismatches between tax jurisdictions. There is some protocol on how Member States should deal with this. Is it the country receiving the payment that decides or the country from where it issued? This ATAD directive is now being implemented in the Member States so there is a transposition period and the Commission role is now to check that there is indeed this transposition is taking place and if not to start possible infringement procedures. That is: rst letter of notication to the Member States, second letter with exchange of information with explanations from the part of the Member States and possibly, eventually, bringing the Member States to the open court of justice and try to get a sentence. Administrative Cooperation Another set of directives is the administrative cooperation between tax jurisdictions and the DAC - the Directive for Administrative Cooperation. There's been several elaborations in the directive that occurred with time. It's a directive that dates from 1977 and in the beginning the directive was not so eective. This is because the idea was that the exchange of information between tax administrations should be done on request mostly: that is, a tax authority needs to go to another tax authority and say that they have a suspicion that there is something wrong with tax payer X so do you have information on this and on specic elements that are already identied so that they can get this information to try and deal with the case at home. The problem with this is that it's very dicult for the tax authority making the request to already ex ante identify the exact persons that are involved and the type of documents that they need. This was foreseen also from the side of the tax authorities receiving the request as a way to avoid the following: a country would say give me all the information on everything and that it becomes unmanageable for the tax authorities having to provide the information. But most of the things then was based on the exchange of information on request and not on an automated way. What DAC 2 did as from 2017 is to extend the article 8 to elements of nancial accounts. It's was pushed from the US and it was after UBS and Lichentstein scandals where the US tax authorities also wanted to get all the information of their own citizens and ask the banks that they have to comply with this otherwise they would be taxed rst of all and they would not be allowed to operate in the US. There was a rst extension of DAC where the bank secrecy in the EU was basically coming to an end. In the past you had the Savings Directive and if you had an account in France, the tax authorities of Belgium wouldn't know or get any information from France on the fact you had a bank account or how much you had on it. Then the Savings Directive was implemented and there was a choice by the tax authorities of the country where the bank account was located to either exchange the information with the tax authorities of the country of origin or if they wanted secrecy to could apply a minimum taxation on the amount that was earned on the savings account. For example, if you had a savings account in France that earned 1000 euros of interest, the French authorities instead of providing this information to the Belgian tax authorities could have chosen to apply a 15% tax rate on the amount that you earned and to give 75% of this amount to Belgium. They would keep 25% for collection costs and give the rest to Belgium. There were three countries that continued to apply this banking secrecy until the very last stage of the Savings Directive: Belgium, Austria, and Luxembourg. Then, there was the Lux Leaks scandal and DAC 3. A ruling is an agreement between a taxpayer and a tax authority where the tax authority has to say in the ruling: if we do a certain type of operation, this is how the tax authorities will tax you and in most cases compulsory for the tax authorities to respect what they said. As a taxpayer, you can go to the tax authorities and tell them: look this is what I intend to do, I want to buy this company, sell those shares there, make a swap here, etc.. could you tell me how I would be taxed if I do this? The tax authorities basically will look at it, do a report and say they will tax you X% on this, this is exempted, this is not exempted, etc. . . Then they have to respect this. It's completely 93 legal and creates tax certainty. The problem that appears in the Lux leaks les is that people would see that a lot of the deals were actually providing specic companies with specic advantages. What is the problem with this? It could be considered scal state aid. The commission opened a number of scal state aid cases because it's competition issues: some companies have a specic tax advantage that other companies do not have. This may create problems of competition. The idea there is that tax authorities should exchange their rulings on an automatic basis every six months where the tax authorities would between themselves provide them in an automatic way (the Commission built the IT system for this exchange), they would provide the other interested tax authorities (those that are related to the deal) the information on what they've been agreeing to. What's interesting is that this information is exchanged between the tax authorities, the European Commission is setting up and running the IT system, but Member States never gave the right to the Commission to know what's exchanged so they don't have access to the data, they only run it. There is a depository of all these rulings, the aim is to say tax authorities may actually see what others are doing and may actually be a deterrent for the tax authorities to do such specic rulings. Then DAC 4 that implements BEPS 13. The idea here is that for specic companies, those that are active in several jurisdictions and with a turnover that is beyond EUR 750 million, they would be an access to the country by country reporting (CBCR) for the tax authorities. There, the idea is that the tax authorities usually receive either the information for their own countries on an unconsolidated level of the nancial accounts or access to the consolidated accounts where they have the global numbers but have no clue on the split between countries. One way to detect whether those companies are doing aggressive tax planning is to have this split between countries because usually aggressive tax planning means that you may have a lot of prot that is parked in specic low tax jurisdictions but there is no real activity there. The idea there is that tax authorities would have access to the CBCR information and that would be a way for tax authorities to spot things. Same principle, no access for the European Commission: they are running the IT system, it's an exchange between the tax authorities. In parallel, there are two initiatives for having the CBCR public. There is one that is already implemented: that is for the extraction companies and the nancial sector: now the nancial sector has to provide a country by country reporting. There is also a proposal to try to have public country by country reporting for whole (might have misheard?) companies, the proposal was sitting in the Council and it was outvoted last week where a number of Member States opposed the fact that there would be public country by country reporting. You need to hear the arguments. The fear of some countries and companies is to say if I give this information for my company, this is sort of strategic information, the way I organize my business is a strategic information for me and I don't want my competitors to know about it especially if those competitors are not subject to country by country reporting as this is the case outside of the EU. The argument may have some validity but still the EU Commission believes that public country by country reporting is a transparency obligation for companies. Then you have DAC 5 that was also a reaction to the Panama Papers. There the idea is to have access to who the benecial owner is for anti-money laundering reasons. There is a legal obligation to indicate who is the benecial owner for the tax authorities. You may see, for example, when you go to the bank to get credit, a loan, or even if you put a certain amount of money on your account, there is now an obligation for the banks for anti-money laundering reasons to ask you for information about the origin of the funds. For big bank transfers now, there is a legal obligation, otherwise they can have a huge problem, to nd out and ensure that they know where the money is coming from. Then, the latest that will now be implemented is the DAC 6. In DAC 6, they now ask all intermediaries (people who as tax advisors, nancial institutions, etc. . . anyone who is building a tax scheme) has to report to the tax authorities what they do before the entry into force of the scheme: I'm a tax advisor, I advise a client, I now have a legal obligation when I advise this client to rst go to the tax authorities and say this is the scheme I will provide to my client and you tax authorities should know about it. These are things that have been voted on last year in May 2018 that is now being implemented. Transfer Pricing Transfer pricing is the topic we will now talk about. There is of course an organization at the level of the EU in terms of a forum where tax authorities and experts can actually discuss those issues. It's quite a dicult process and it's very tedious so there is an expert group that works on those cases and 94 there is also an arbitration convention. So when two tax authorities disagree about transfer pricing rules to be applied and the reference being chosen, then there is an EU instrument that they can refer to where there are also procedures on how to deal with those cases. Common EU List of 3rd Countries for Tax Purposes Now we will discuss is the EU list of 3rd countries for tax purposes, which the professor considers a huge success for the EU. The idea is that for many years, there is a group at the level of the Council that is called the code of conduct for business taxation. This code is gathering the Member States and what they have been doing over the last decade or so is a review of all the tax rules for corporate taxation that are in place in the EU. They have a number of criteria to determine whether or not the schemes is so-called harmful or not. When is a scheme harmful? It needs to full two conditions. The rst condition: it's a harmful tax scheme if it provides low or no taxation. This alone is not sucient, there needs to be a second condition. The second condition: there is a list of elements that you can nd, a second condition, for example, could be that it's not transparent, it departs from the usual rules of taxing, if tomorrow for example you say all companies active in this specic area should be taxed not on their prot but at a rate of 1% on their cost, that could be considered as low or no taxation and that departs from the usual rules of taxing. Another example of a second condition could be that it is ring-fenced. Ring-fenced means it's only available for a certain number of companies. So the code of conduct had a review of all these schemes and those that were considered harmful by the code, working at unanimity minus one. For example, imagine they reviewed the tax code of Belgium, the other 27 countries should at unanimity decide this specic rule in Belgium is harmful based on the two criteria, then Belgium has the responsibility to roll-back (replace it by something that is not harmful). The code of conduct has been doing this within the EU where there was a large number of measures that were considered harmful and that have now been removed. The code now is looking at third countries, there was a selection based on economic criteria of which countries are the most economically relevant for us and there was a review by the EU of the legislation of all those countries and the creation of a black and a grey list. So it considers the issues of or not there is indeed lection (1. transparency, fair tax competition and whether real economic activity behind the schemes that are oered. There was a se- Selecting) of the countries based on the economic indicators. A screening (2. Screening) was done by the code of conduct and the Commission working as a team. A listing (3. a monitoring (4. Listing) and Monitoring) of the eorts done by those countries to remove those harmful measures. When they issued the rst list in December 2017, they had 17 blacklisted jurisdictions and 47 that were on the grey-list. What's the dierence? All those jurisdictions had been contacted because they had an issue. Those that did not reply or did not commit to change were put on the blacklist, those who said they agreed to review and to change the rules with a specic calendar were put on the grey-list. There were nine jurisdictions that were postponed: there was a hurricane that went through those countries at the end of 2017 so were not able to commit or anything. From all those jurisdictions, we now have from October 2019, only nine jurisdictions are left on the blacklist and 32 that are now on the grey-list. All the others have changed their tax legislation to comply with the EU requirements. One question that is linked to this. Of course there is a reputational issue with being put on the blacklist, even some research showing there was an eect on the prices of companies located in those countries. There is also the possibility to reect on possible EU common sanctions against those jurisdictions. What is implemented now is two things. Individual Member States can refer to the blacklist for their own personal sanctions. For example, the Dutch have a blacklist of jurisdictions for the CFC rules and for the Benecial Ownership etc. . . and the blacklist specically says they include those that are on the EU blacklist. Also, what they've done is increasingly so they have cut nancial aid to those jurisdictions. The EU is progressively not giving any money anymore to those jurisdictions. 95 Communication on Qualied Majority Voting The last topic we will discuss is the communication on qualied majority voting. You know that one of the stumbling blocks of moving forward with the EU tax agenda is the rule of unanimity. You need 28 Member States agreeing on the details to have something that is implemented. It worked well for that, every time there was a scandal, after 6 months, there was a new DAC legislation that was implemented so when they want something, it's possible but on the big projects, it's becoming more complex. The current rule is that you need unanimity at the Council and the European Parliament has absolutely no say. They can pass resolutions explaining what's their opinion but they don't have a formal say. What they did in January 2019 is to try to open a debate and to propose a move to qualied majority voting in four steps. Why? Because there are a number of problems with unanimity. The rst one each country has a veto right, you can hijack the discussion and say if I don't have what I want on the agricultural policy I block the tax issue, etc. . . Usually, there is also a risk that the agreement will be at the on. It's also lowest common denominator: only a few things that the Member States would agree self-defeating because if you have something that is voted at unanimity, you also need unanimity to change it and so it's dicult for Member States to commit on something they know will also be extremely dicult to change. Member States want to keep unanimity because they want to remain sovereign. The question is if you want to have something but one country is blocking you, are you still sovereign? Can you still do what you want? Can you still implement it? Then, the European Parliament, which is an institution that is directly elected has no say on those issues. The proposal has 4 steps. The rst step (Step 1): to go to QMV with issues on cooperation and mutual assistance in ghting tax fraud, evasion and administrative initiatives. common interest on trying to improve tax collection. Then support other policy goals (e.g. ght climate change - green taxation). modernise existing harmonised EU rules (e.g. VAT) Then, Things that are of Step 2 would be all the tax measures to Step 3 are all the measures to Step 4 would involve major projects such as the CCCTB. How would they do it? Using article 48(7) TEU: the passerelle clause. The clause says that if there is the consent of the European Parliament, no objection from national parliaments, then the Council can, deciding at unanimity, move to QMV in certain areas. You need a unanimous vote to go to QMV, so to get rid of unanimity, but at least the treaty foresees this possibility. They are not calling on Member States to reect on this, to discuss, to open a debate on this possibility and try to be pragmatic with the tax agenda. That's everything for the course. 96