 Good morning everyone. Thank you for the introduction and thank you for the invitation to come talk with you. I understand that most of you are from the development community and or the inequality community. I am from the tax community so I'm I'm coming from another world but they're really not unrelated worlds. There's a lot of overlap between the the kinds of topics that I and people in my community think about and the kind of topics that you and people in your communities think about. I want to talk about some of those overlaps today. I have a half hour only so I'm going to just touch on three questions briefly and I hope that if any of the things I say are of interest you can and you want to talk more about it you can bring it up in the question period or be in touch with me after today. So the three questions I want to touch on are the role of taxes in inequality and poverty reduction and second what I call a tax systems approach to thinking about taxation and why that approach is especially important for questions that arise in developing countries and third I want to pick up on the recent interest in measuring inequality using tax return data. I'm talking about Toma Piketty's book and the work by Piketty and Sayez and talk about the advantages and disadvantages of using tax return data to measure inequality and trends in inequality. So first the role of taxes in inequality and poverty reduction so let me first say that I recognize the controversy surrounding whether countries ought to address not only reducing absolute poverty but also lowering inequality and how we should think about these two related but not identical objectives. I have to I will admit that my own view is the former that I think the object of policy should be to reduce absolute poverty and to think about inequality only as secondary to that objective but I'm not here today to try to convince you of that view. What I do want to say is that there's no question that the progressivity of the tax system that is how the burden of taxes is borne by different income groups. The progressivity of taxes is one of the important tools to address inequality but it is not in my view particularly effective for eliminating poverty at least not in a direct way. Unless you think about negative taxes which go to low income people but although in some countries like the United States the tax authority operates systems which of negative taxes and in the U.S. it's called the earned income credit. There's no need that the tax authority should operate such transfer programs. They can be delivered through the tax system but need not. And the reason that I believe that the tax system is not directly effective for eliminating poverty is if you rule out transfer programs and consider those separate from the tax system there is no way to increase transfers to low income people at all through the tax system. And so what we should think about as the role of taxation in addressing poverty reduction is the role of the tax system in generating funds for the better instruments for addressing absolute poverty and that is progressive government spending on programs such as education and health. That's these are the best instruments for addressing poverty and this requires raising revenue. So now the tax system has to become part of the conversation for addressing poverty reduction because it is the method for mobilizing the resources needed for providing these programs. How to raise revenue and how to raise revenue as good or as well as possible. This is the sort of thing that people in my community spend a lot of time on and modern tax theory I will date back to about 45 years and so we've been thinking rigorously about this question which we call optimal taxation for about 45 years and so I will try to now distill for you in one or two sentences what we've learned in 45 years. So a lesson of tax theory is that in order to raise money for for example for education and health programs in order to raise revenue with minimal disruption to the economy and that is certainly a goal you would country should avoid taxing mobile factors and tax bases for which the behavioral response is large. So in other words other things equal you want to tax bases for which when you levy tax the base shrinks so you don't want to tax bases which can easily move offshore or which can easily move out of the tax net for whatever reason and we summarize this behavioral response by talking about the elasticity of the tax base. So that's a statistic that tries to summarize the responsiveness of different tax bases. So summarizing 45 years of thinking other things equal a country should try to identify tax bases for which the elasticity of the tax base is low. If you're talking about income tax and that becomes the elasticity of taxable income that's the concept that many in my community have been trying to measure in the last 20 years and that becomes the central empirical parameter and under some assumptions it doesn't matter why in a particular tax base is very elastic. It doesn't matter whether it's because the base will move offshore or there will be more evasion or people will work less or there will be less investment. Any of those things that reduce the tax base are, they are measures of the inefficiency caused by raising taxes. So that's the first question. The second question, the second issue I want to talk about this morning is what I call a tax systems approach to thinking about taxation and I'll tell you what I mean by that. So although the modern study of taxation which I date about 45 years and which has made tremendous strides in the I think in the rigorous understanding of the effects of taxation and how taxes should be levied it has I think missed some of the important issues that are of particular relevance to developing countries and I'll be more specific about what I mean. The modern theory of taxation focuses almost exclusively on tax rates and tax bases. So for example how what rates should we apply to income? Should we tax income or consumption? So tax rates or tax bases and as the modern theory has spent much less attention on two other aspects of tax systems and that is what I call remittance rules that is who or what entity should actually remit the tax to government and third, enforcement rules. Now you all know that no government can announce tax rates and a tax base and just sit back and let the money come in. You need a tax administration. You need rules about the consequences of not remitting the taxes that are owed including penalties, audit rates, audit rules, information reporting and modern tax theory in my view has ignored these latter two issues. So let me say that a tax system has three parts. The first is tax rates and tax bases. We in my community have spent a lot of time thinking about that but second, remittance rules. Who should remit the taxes? Is it businesses? Is it multinational companies? Is it individuals? Is it employees? That's second and third enforcement rules. The other thing tax a tax system approach recognizes is that there are multiple sources of cost that need to be considered in designing a tax system. The modern approach to taxation has focused almost exclusively on one kind of cost and that is the cost that arises when people change and businesses change their decisions because of the tax system. So for example, to what extent do people work less if income taxes go up? To what extent do businesses invest less if a country tries to tax the return to investment? But there are other sources of cost. There are administrative costs, the kind of budget of the tax authority which is often non-trivial and constrained and there are compliance costs. Those are the costs borne directly by the taxpayers or by the third parties who help with assist with tax collection. Second, there are multiple types of behavioral response. The modern theory has looked at these real responses such as a labor supply or investment but has relatively ignored two other types of response, avoidance and evasion, the distinction between the two being that the latter is illegal and the first refers to legal ways to change your affairs to reduce your taxation. In a developing country context, I think you would agree it is silly to develop and take seriously a tax theory framework that doesn't consider avoidance and evasion. And finally, the third element, the third thing a theory of tax systems must consider is that there are multiple instruments in a tax system. A tax authority and a government has to decide not only on what basis to tax, what rates to apply to those bases, but on a whole vector, a whole set of other things. For example, the enforcement system, how much resources to give the tax authority to audit tax returns. What rules should be used to pick out the returns to audit? What should be the system of information reporting? Do firms have to provide information on the wages they pay to employees? Do banks have to report information on what interest they pay, et cetera? What should the cutoffs be for taxation? Should there be, for example, exemptions for very small business? So below a certain size, the tax authority will explicitly or implicitly not be concerned with those small businesses. This will save administrative costs, but of course, will also provide an incentive for businesses to stay small, which doesn't sound like something other what you would want to have. So let me try to summarize in a few sentences some lessons from thinking about taxation in this kind of framework with three aspects of a tax system, multiple sources of cost, multiple types of behavioral response, and multiple instruments. So first, a higher, if you're considering trying to raise revenue with higher rates, that should be accompanied by having a broad, effective tax base, including enforcement. Another way to put that is this central concept, the elasticity of the tax base, or the elasticity of taxable income when we're talking about income tax, is not a fixed parameter. If a government is worried or wary of raising tax rates because it's afraid that the tax base will evaporate, there are other tax instruments which can reduce that response. And so this elasticity of taxable income is a policy instrument, the broader the tax base and the better the enforcement of the tax system. The lower is the elasticity of the tax base, and the lower is the efficiency cost of trying to raise more revenue. But of course, increasing enforcement, for example, I want to be clear, that's not a free lunch. It costs more resources to have more effective enforcement, and it also increases the effective tax rate. Obviously, if you're a business, and suddenly whatever tax rates are on the books are now being enforced, that increases the effective tax rate in the same rate as actually increasing tax rates would be, but often that's a more effective, more efficient way of raising more revenue than just raising the rates with a narrow base and an inefficiently enforced base. Okay, so that third question, third and final question I want to talk about is related to the measurement of inequality using tax returns. And I'm sure all of you are aware of the fact that this is a very hot topic these days. In part or mostly, I guess I would say because of the work of Piketty and Syas and the recent book by Tomah Piketty that uses mostly, not only, but mostly data from tax returns, income tax returns and estate tax returns to measure inequality and trends in inequality. So I want to talk a little bit about, and I noticed looking at the agenda of this conference, there are some papers that do exactly that. They use tax return data to get a sense of the nature of inequality and how it's changing. So I want to talk about the advantages and disadvantages of this approach of using tax return data to measure inequality compared to, say, surveys that are of income and wealth. So, first of all, advantages of using tax return data in many countries, an advantage is you can get very big sample size. So this data is collected for operational reasons and in many countries, individual tax returns are meant to cover the whole population. So in principle, you have a universe of tax data compared to surveys where, of course, it's costly to collect data and surveys tend to have a much lower sample size. So that's an advantage and especially if you're interested in inequality at the top for higher income people, then you have a 100% sample of high income people, which is very difficult to get any other way. You probably know that surveys of high income people tend to have very low response rate and therefore very difficult to draw conclusions from. So one advantage is using tax return data is that you have a big sample. The second advantage of using tax return data compared to survey data is that the numbers that are provided have real consequences. And this is the time that normally I would tell a joke about how the way you say a sentence can change its meaning. So rather than tell the joke, I'm just going to explain what I mean in the tax world. So, conveying data to the tax authorities has real consequences. So that, you can make that out like an advantage. So unlike a survey where if you lie, make up that you're rich or make up that you're poor, there's no real consequence for lying to a surveyor on the telephone or with a clipboard. But lying to on your tax return has real consequences. So let's think about how I could play that either way. It sounds like a good thing because they're real consequences, especially if they're real consequences to understating your income. You would think twice about doing that because you might be caught accused of tax evasion and penalized. So people won't do that lightly. So you might say, well, I'm going to, this data is better than survey data because they're consequences to understating income or wealth. Okay, now I'm going to take the same thing and turn it around. There are consequences to understating your income and wealth, but there are real rewards to understating your income on a tax return and understating your wealth and a tax return if there's a wealth tax, and that is the tax saving you get if you're not audited. And so we know about tax evasion in this room. So to the extent that there is tax evasion, even in the face of enforcement and penalties, one might suspect that income figures on a tax return are understated and maybe one might suspect that the income figures put by high income people on a tax return are understated. And so I'm taking the same point about tax return data that there are real consequences to tell you that that's an advantage and that's a disadvantage. And in fact, these advantages and disadvantages are the classic trade-off in the modern theory, positive theory of tax evasion, where it's a Gary Becker model, where you take the chance that you won't be caught and then you save money because you have a lower tax liability and on the downside, there's some chance you are going to be caught and have to pay some penalties. So we understand these trade-offs. These trade-offs have to be kept in mind if you're going to use tax return data to measure inequality. You have to think about whether to what extent the numbers that people write on their income tax return are understated or not. There is in the last 20 years a very big literature, empirical literature on tax evasion, its determinants and its magnitude and its nature. I will try this joke though, which I'm part of this literature and it's a colleague of mine once said at a conference like this, he made the following statement about empirical research on tax evasion and he said, empirical work on tax evasion is easy and straightforward except for two things. You can't measure the right hand side variables and you can't measure the left hand side variable. Well, I won't talk about the right hand side variable, but the left hand side variable tax evasion, of course, is not easy to measure. If it were, there would be a lot less of it. Some people take this as a reason not to get involved in this literature. I'm the type of person that takes this as a challenge and I would say in the last 10 years there have been a lot of interesting work using administrative data, using randomized field experiments, et cetera, to understand the nature and determinants of tax evasion. One last thing about using tax return data to measure inequality, one phenomenon that's observed in taxation, we call income shifting, so imagine that a given kind of income can be considered, depending on the details, individuals' income or corporate income. So imagine I run a business and depending on the choices I make about incorporation, say, this income can show up on my individual tax return or, for the most part, it can show up on a corporation's income tax return. Well, what choice I make depends on the relative tax rate of individual income, personal income, and corporate income. So another caution for people using individual tax return data to measure inequality, be careful when the tax system changes in a way that changes the incentives for people to report income on their individual return and a corporation's return, because you might be observing much less income after a tax change of high-income people, and you might say, well, inequality has gone down, but no. In fact, what's happened is that these people have found it in their interest to move the income into the corporate, into their corporation, and you could be very misled, and there are many examples in developed countries where this exact phenomenon has occurred. So I'm going to conclude now and just tell you about, summarize my messages from the tax community to the development and inequality community here in this room, and so I've touched on three issues. The first, I will summarize by saying taxation is important for mobilizing resources, that is, raising revenue to provide for the expenditure programs that I think are critical to poverty alleviation. The tax system can also be used directly for reducing inequality, but I think that is not the right way to think about the relationship between tax systems and inequality. In particular, or in part because the more progressive, the more progressively one tries to raise revenue, the bigger the efficiency cost tends to be, and again, other things equal, you would like to raise revenue in a way that doesn't repel investment, that doesn't deter labor supply, et cetera. So that's my first summary. My second summary is that especially in developing countries, but not only in developing countries, you should be thinking of taxation in a tax systems approach that is to think of that there are multiple sources of cost, there are multiple tax instruments, and there are multiple kinds of behavioral response. And that a tax system consists of much, much more than just thinking about what basis to tax and what rates to put on those bases, but there's a whole array of other policy instruments that can be very important such as withholding, information reporting, auditing, and the same tools we use to study tax rates and tax bases and their effect. The modern empirical and theoretical tools that we've developed over the last 45 years can be used to study these other aspects of tax systems. And finally, I certainly recommend the use of tax returns to help measure inequality, but we need to be aware and maybe even wary of the disadvantages as well as the advantages of using tax return data. And we should go into that with clear eyes and with wary eyes that things like income shifting and evasion can substantially alter the picture of inequality that you would like to get. Thank you very much.