 Hello, everybody. I'm Yukkapirtila. I'm presenting here joint work with the chair, Maria Joste and Tina Kaidu, Uganda Revenue Authority, Joseph Kello also from the URA and then our colleague Pia Rattenhooper from UNU-wide. And our title is Taxpayer Response to Greater Progressivity Evidence from Personal Income Tax Reform in Uganda. So I want to shift the focus a little bit. I mean, we are not bypassing revenue here, but we are also worrying a little bit about the distribution of the tax burden across the taxpayers. So by way of motivation, one notices when one looks at the data on the income tax rates and also on the revenues that the personal income tax rates in the sub-Saharan African countries, SSA for short, are low in international comparison. So here's information from a study by McNap and Granger. What you can see at the bottom line, the yellowish line, is the average highest marginal tax rate on personal incomes in African countries and it's close to 30 percent, which is significantly below to what it is in the OECD countries. So that would be the line here. Is that maroon? I'm always very bad with colors. But anyway, so there's a big gap and the rates are low. On the other hand, we know that income inequality in African countries is substantial. So here's data, apologies for the small numbers, but here's data from a set of countries. Uganda is not included, but Uganda wouldn't be very different, I would guess. Which shows, first of all, how high the income inequality genie coefficient is for some African countries and how little it's reduced by the tax system. So the disposable income inequality is almost equally as high as the marketing of inequalities. And the tax system doesn't even do much for poverty reduction. On the contrary, it can even increase poverty numbers because also some of those who are close to the poverty line need to pay the tax. So, of course, this begs the question whether it would make sense to raise the tax rates for the high income earners in the African countries. There's also some optimal tax background that we can look into when thinking of taxis and setting the tax rates, also including for the top income earners. And this is work stemming from the Nobel Prize winning economist Jim Merleys and much of subsequent work. It shows that the socially desirable top tax rate is relatively high if the inequality aversion in society is at the high level or if the pre-tax incomes are unequally distributed. On the other hand, it would have to be a lower one if there are major efficiency losses from those high tax rates. So if high tax rates then lead people to report lower incomes to the tax authority. So it would be key to know some of these, to quantify these things, especially how reactive the tax base is to change these to tax rates. And that's what we try to do in this paper. There's little prior evidence on how the tax base reacts when the top tax rates is changed from developing countries. So that's, we would argue, our contribution to the academic literature is then to provide that sort of evidence. So what do we do in the paper? We use the Uganda Revenue Authority administrative tax data to examine the so-called elasticity of taxable earned income. And we use the, for the empirical analysis, we zoom into the tax policy reform in 2012-13. When, among other things, the Ugandan tax system was changed so that the top marginal tax rate was raised by 10% its points, namely from 30% to 40%, which is a substantial increase in the tax system. So we want to gauge the revenue and inequality consequences of this policy change. Taking into account those potential behavioural reactions that some of the taxpayers will be reporting lower incomes now after the reform because the tax rate went up. Why did Uganda carry out this reform? In our understanding, that was a response to the fact that the tax system hadn't been changed for years. And when inflation had been going on, that had led to an increase in the tax burden for the lower-medium income earners. The government wanted to alleviate this burden, but recoup some of the revenues by increasing this tax rate at the top. So it was a progressivity-increasing tax reform. And a new top tax rate was introduced on persons earning more than 10 million Uganda shillings a month, which is at the ballpark of 3,000 USD. And given that the average Ugandan taxpayer has relatively low incomes, this tax was actually a tax on the top 1% of the income earners, quite closely to the top 1%. So here's information about how the tax rate schedule looks like in Uganda. This is the, I have the pointer here, yes. So these are the marginal tax rate schedules, and this is the average tax rate that increases more smoothly. So what we do in this paper, we really focus on the top group of income earners. And we are looking at those income earners who are about the threshold value, whose marginal tax rate jumped by 10 percentage points. And we are comparing their income developments to those who are just below who continue to pay the 30% marginal tax rates on any additional income they will be earning. In the paper, we also look at the sum of the other groups affected by this reform. But in this talk, I'm now going to focus on the most salient aspect of the reform, which is the top 10%, a top 1% increase. As I said, we use data from the administrative records of the Uganda Revenue Authority. So the data includes monthly pay as you earn returns filed by the employers on behalf of the employees. We are covering five fiscal years in this analysis from 2010, 11 to 2014, 15. And a particular point that I need to raise is that while all the employers at that time had a unique tax identification number, which means that we can follow the same employer across time, not all employees had one. So that means that we are actually using the data in a cross-sectional manner when it comes to the employees. And it could be a disadvantage, but on the other hand we know that the so-called panel economic techniques using the individual level data could actually be problematic because of the so-called mid-reversion issue, which we are going to bypass here when we are using the so-called cross-sectional approach. So the empirical approach is a variation of a different difference in differences design, where we are focusing to the tax reform of 2012, where we have two pre-reform years. Unfortunately, we don't have more of those pre-reform years because that's when the digital system started. And we look at the three post reform years as is quite customary in the literature. And we are comparing the so-called treated group, which is a top 1% of taxpayers. And then we have a comparison or control group, which consists of taxpayers, who are then the next 4% of income earners. And we express then the results as an elasticity of taxable income. And this measures then the change in the reported earnings relative to the change in the one minus the marginal tax rate, namely the net of the tax rate. And when we do the analysis, when we actually use this narrow control group, we do find that the treated group's income declined more than the comparison group incomes, but the decline was not statistically significant from zero. And on the right, you can see this event study type of plot, which compares the income developments of the treated group relative to the incomes of the control group. And the incomes decline, but then the confidence bands include zero. So this means that this is not statistically significant. However, if we were to use a broader control group, we would be detecting a significant reduction in the incomes. But we feel that the narrow control group is more comparable to the treated group, and that's why these are our favorite results. However, the macro level results, if you wish, hide some heterogeneity in the results. And if we zoom in to the taxpayers or to the employees not employed by firms, what treated by the large or the medium taxpayer office, we actually find a significant negative impact on their incomes. However, we also find that in those firms where the top incomes decrease the most, there's also a sizeable increase in dividend income, suggesting that there's something called income shifting going on in the firms that some of the employees, rather than having salaries, get perhaps now more dividends out of this. This is partly speculation, but this is indicative in the results. So often tax researchers worry about what are the mechanisms that give rise to results. We are saying that at least part of the mechanisms here seems to suggest that some income shifting is taking place. This is, of course, very different from people actually reducing the actual effort, because this can happen even if the labor supply remains the same. We also, as I said, we also look at the revenue and inequality implications, and we come to the conclusion that the reform as a whole contributed to revenue, so that increased revenues in the Yukandan case. We also find that even if the elasticity that we detect 0.5 was significant, it would mean that the new top tax rate wouldn't be about the so-called laffer curve maximum, so that would mean that the Yukanda government would increase revenue by lowering the tax rate, that wouldn't be the case. And then we also find that the reform led to a mild reduction in the inequality among the salary employees that we are looking at. So before the reform, the Gini coefficient amounted to 0.635, and now after the reform it went down to 0.606, so it's still high, but it's a little bit less than it used to be. Although if there's income shifting going on, some of this may not be real reduction in inequality, so that with that caveat this result can be seen. So I summarized before Maria shows me the zero sign. All right, so what we did, we investigated the 2012 reform in the Yukandan case. In this presentation we focused on the top group, the top 1% of income earners, but in the paper you can find results for the other groups as well but for our preferred empirical approach we don't find a statistically significant reduction in top incomes, but we do find a significant impact among the smaller firms, maybe then suggesting that some income shifting may have taken place and then we find that the reform I think was a positive one because it led to a reduction perhaps in inequality and it contributed to the revenues. So we would argue that the results are informative also for similar lower income African countries when they consider reforming their personal income tax systems. Thank you, that's all.