 Cool. Thank you for having me. Good afternoon. So, most of what I'm going to present is a joint work with my great colleague Javier Garcia Bernando. And the title of my presentation is Profit Chifting of Multinational Corporations Worldwide. You know, it's, in a way, it's great to be speaking as the last speaker, so I'll be skipping much of the stuff. But there are a number of reasons why we care about this specific type of tax avoidance. And one of them that I will discuss in more detail is the lower government revenues, countries worldwide. But it's also the uneven level playing field of companies that do engage in this tax avoidance at the dawn. It might also lead to people perceiving globalization that might have some other beneficial effects as inequitable. And, of course, allow me to mention, as was already mentioned today, that sustainable development calls, illicit financial flows, is one of the targets. And for those interested why Profit Chifting is illicit financial flow and how it can be measured, we have a book out with Oxford University Press that you can learn even more than in today's presentation. But to the research that I'm going to present, what our paper, where the most value at it is, is that we provide evidence on what specific countries are the origin and destinations of Profit Chifting. So, for example, there are detailed studies of Profit Chifting, such as Nadine's paper for one country. There are other cross country studies, which maybe have 50 or 60 countries, usually not more. So here we wanted to look at, you know, basically all the countries in Africa and other continents and see how prevalent Profit Chifting is in those countries. And also specifically to which countries the profits these multinationals shift to. So that's kind of, you know, our focus. We have also two methodological contributions. So country by country reporting was mentioned already as one transparency regulation introduced over the past 10 years. For us researchers, it means we have a great design for research discontinuity design, but that's another paper. It also created data, which is the case of this paper that we can employ. And I was, you know, glad to see that IMF uses in their evaluation of PIL 2. OECD does the same, so that seems to be agreement, pending agreement that country by country reporting data provides some of the best information that we can have for multinationals. And around the time this paper came out first, we were some of the first to use this data, but kind of every presentation that I have of this paper, it gets outdated, of course, because there are many more papers coming out with this kind of data. And we're only, as I will discuss, using aggregate data. We don't have access for this paper for the firm level data. The methodological contribution is about the relationship between reported profits and tax rate. So most papers in the profit shifting literature have modeled this using only linear function. Then the question came, again, about 10 years ago, is it really only linear? So we have the quadratic function and the question is how nonlinear this relationship is. We argued that it's extremely nonlinear and that you should use extremely nonlinear functions, such as a logarithmic one to model the relationship. And we structured a paper and four questions. So how big is the profit shifting? What are the tax havens to which the profits are shifted? Are the differences in profit shifting according to where the multinational is at quarter? And how are various countries impacted? And are the low-income countries hit more strongly than other countries? To preview the results, it seems that low-income countries tend to be more hardly hit, although the results are not as conclusive as they might be. There's many papers including people in this room that we build on, but for the sake of time I'll skip the discussion of the literature. So let me start with a bit more detailed discussion of the country-by-country reporting data. So there's some information in the aggregated country-by-country reporting data, I will say CBCR, for around 190 countries. What's good is that there's one data set for profit-making and another data set for profit-and-loss-making. So that allows us to use one of them for estimation of the backward-looking effective tax rates and the other data set for the rest of the analysis. In this paper we're really interested in the wide range of countries. So if you go to the appendix of our paper, we end up with estimates for around 200 countries. But it has its cost. So we do some data imputations using OrbIS and other data sources in addition to the CBCR to be able to have meaningful information about so many countries worldwide, while only some of them report the CBCR reports. That's kind of our headline results. We kind of for the methodology purposes, but also to have more credible data source. We also in one section of the paper, we present only the results using the data for U.S. at quarter multinationals. That has the benefit that we have more confidence in the data. And still we cover around 30% of multinationals profits worldwide. So we have these two parts of the paper. And the CBCR data. It's a phrase that I like and has been used at this conference a lot. So is the glass half full or half empty? So which take the one from me on the CBCR data? So I'll be the optimistic one. So it's glass half full, although of course there are imperfections. But still we seem to know much more about the geographical coverage of multinationals than we knew before. And now with the higher quality data forthcoming and more data becoming public and shared with researchers, every year we know more and more. One of the technical details around the country by country reporting data that I would like to discuss in detail is the double counting. That is unfortunately present there. So relatively recently after the, I think even before the data was released publicly for the first time, we heard that there's going to be likely some double counting, because there hasn't been clear guidance on reporting for holding companies of profits in the data. So we were aware of it. I personally understood it was there. I didn't expect there would be so large as we found out in another paper with Kavir and Gabriel Sukhman, where we may use of the US data having the availability of data for US multinationals from various data sources. We especially rely on CompuStat. And we find that around half of the profits of US multinationals is double counting. So I changed my priors, I thought it would be less. So we feel it's important to correct for double counting in the data. And we do so using several sources in the worldwide analysis too. So we use corrections for specific countries, for example, so Netherlands and Sweden and other countries issued kind of corrections for double counting. So we use that, we get rid of the stateless income. We use our correction from the other paper for the United States. And so we want to make the paper kind of more conservative, kind of, you know, hopefully realistic. But we can't be certain with the corrections if we are not overdoing it or whether we are getting rid of the double counting in the right places. And there's just one table from the other paper. And basically we compare CompuStat and CBCR when we try to arrive at the same sample of companies or adjusting the CompuStat to arrive at a similar set of firms as should be reporting in the CBCR. And then, for example, most of the profits double counted seem to be in the United States for the US at quarter companies. It's here the numbers above 50%. But, you know, it's half full as I say. So I want to inform you about the double counting because I think, you know, this is a real issue that needs to be addressed. But it's also good news that the kind of the reporting guidelines were not clear. We knew that five years ago. And so the guidelines have been fixed. And companies have been reporting according to the corrected guidelines already for 2020. So when data are published for 2020, which should be any months now from the OECD, we should, in theory, it should at least not suffer from this specific problem that has been identified. There might be other problems outstanding. But let's see. It's also, there are other initiatives companies are sharing their country by country reports. European Union has mandated that many companies will need to publish parts of their reports publicly. So let's see. There will be other improvements in the CBCR data. So I remain upbeat about the usefulness of the data. But let me move on to the methodology, which is relatively standard in this literature. And I presented a couple of weeks ago with Jim Hines present in the audience and, you know, still referring to the Heinz and Rice after 30 years. So I don't know if the paper was so good or we were kind of partly stuck in the literature. I'll keep that as an open question. So the standard model use so-called text seminalistic model where we have profits on the left-hand side and we try to estimate what the profit should be using capital and labor and then whatever it's pick up by the tax rate is likely caused by profit shifting. So it's about the profit book and tax rate in simplicity. And as I said, the linear specification has been used in most studies that have used this specification. But already Heinz and Rice and it was highlighted again powerfully by Tim Doud and his co-authors in a general public economics paper in 2017. The quadratic makes a lot of sense. So that's an improvement that most papers now do account for the non-linearity in one way or another. What we're saying in this paper, the quadratic is really improvement on the linear, but we feel it's not enough. And one of the ways to reason this is to look at this graph where we have the CVCR data, effective tax rate on the horizontal axis and profit per employee on the vertical axis. And this is logarithmic scale on the vertical axis. So companies reporting profits in Bermuda, Cayman Islands, Isle of Man, Jersey are extremely profitable and they all have extremely low effective tax rates. So there are no countries that will be extremely profitable and have higher tax rates or vice versa. So there seems to be something extreme in the non-linearity between the profits and effective tax rate. And one way to go about this is to use instead of the quadratic function to use the logarithmic function, which we do in this paper, but we also present all the results for all other functional forms, including linear, quadratic and logarithmic. If we take the case of Jersey, which in the data has the ETR of almost zero, it makes a lot of difference which of these functions use. The difference wouldn't be so big for countries that have higher effective tax rates where the lines, estimated lines are much closer. What this means for the distribution of the profit shifting? So let's look at the first line, the Jersey with the lowest effective tax rate. So we have then logarithmic quadratic linear at the last three columns. So the more you take into account the possibility of the non-linearity, the higher amount of profits reported in the tax even are labeled as profit shifted there. So it's 99% logarithmic, 89% quadratic and 55% linear. So that's what makes it different that if we use linear, still most profits in Jersey are labeled as profit shifting, but it's much higher share if we use quadratic or logarithmic. So unfortunately, as far as I know, we can't say for certain what's the real profit shifting. So it's hard to, we do make a case that the non-linearity is extreme and that we should use logarithmic or other similar function, but I don't see that there will be a good something that we could strongly compare it with. So this was based on the US data. Now I'm moving to conclude using the OECD data where we sum up the results according to the ETR and the results are similar as with the Jersey example before. This is the way to present all the results. So on the left-hand side is the countries from which profits are shifted and on the right-hand side you have the tax havens, Cameron Islands, Netherlands being some of the biggest ones. And then this is a graph showing that lower income countries tend to lose more as a share of their corporate income tax revenue and that's the same for African countries. My last slide is about putting this in the perspective of other studies that exist here. So in this paper we put the total of profit shifting at around 900 billion US dollars annually and the revenue loss is around 200 billion US dollars. Right now there are studies, so when I presented it for the first time, about two years ago it was the highest estimate among the list. But here now you have some studies that are having higher numbers. So it's kind of broadly comparable, it depends on what kind of take you want to take if there are differences across the studies or if they point to this being a big problem. Thank you and looking forward to the discussion.