 We are going to see the tax system design and tax-loving collection in the context of South Africa. We are going to have three presentations and then after that we are going to have the discussant who is going to bring up the new common idea about all those three presentations. Maybe a couple of things before we just have the first presentation. My name is Imabre Salimana. I am a research fellow at UNU Ida. I am working at the SATI program based in South Africa. Just a little bit introduction about this session where we are going to see the tax system design and tax-loving collection in the context of South Africa. Much is known about the South African economy where we have issue of unemployment and inequality. This is most common critical issue in South Africa. This program has a big implications in terms of the development, in terms of the job creation, in terms of investment, and there are some other issues like roadshedding and others. All this has massive negative impact for the development. The police makers in South Africa, they try to bring in some different programs so that they can make sure that they can have different lovin services, including internal domestic mobilization. This is why different programs, research development have been putting in place to be able to help the police makers to have a concrete and efficient ways to correct tax revenues and help to minimize the economic loss in the country. In this session, we are going to have the first presentation by Professor Nadine, who is the director of the Institute of Public and Regional Economics at the University of Monster. This is also one of the workstream leads at ASEA Tired Program. This is the program we have, a tripartite program between National Treasury, South Africa Lovin Services, and UNU Ida, where the tax data lab was established at National Treasury in collaboration with SAS to bring in those big data sets to help the researchers to provide relevant evidence-based research. So Nadine plays a key role in the development of one of the workstream on domestic revenue mobilization. Welcome, Nadine, for the presentation. Okay, so thank you very much. It's great to be here and present you a new paper that we currently develop in the context of this ASEA Tired Program that Imabla just mentioned, and it asks the question, what happens when you tax the rich and its evidence from South Africa? So background and motivation for the study is that in general, when you look at less developed countries, there tend to be in many of those rising levels of income inequality that has sparked in these societies discussions about potentially increasing the tax levies on high-income earners in these countries. And if you think about the implications of that type of policy reform, then the promise that this holds is that you may increase revenue collection in these countries, and many countries struggle with low tax-to-GDP ratios, and simultaneously may reduce after-tax inequality. The downside, of course, is that there may be efficiency costs, and these efficiency costs may be high, especially in lesser developed countries. Why? Because high-income earners may have many options to evade and avoid taxes, and commonly these countries are characterized by lower tax enforcement capacity than more developed economies. And if you think about real responses, so here you tax potentially the very high income, very high-skilled individuals in the society if they reduce their labor supply or their effort provision that may have real economic consequences that may potentially be more detrimental than what you see in developed countries. So if you think about the responses of taxpayers to this type of policy reform, it's really first order in terms of importance for policy discussions, whether you want to do this type of reform to understand how taxpayers react, and this is the context of the study. So we look at South Africa and South Africa in 2017, so in February 2017 the government announced to increase the top marginal tax rate in the personal income tax scheme from 41 to 45%. And this treated the very upper end of the income distribution. So the 0.5 top income earners in the country were affected by this reform. So income earners or incomes above 1.5 million rounds were affected. And the government said what we want to achieve with that reform is exactly what I just gave you and the motivation they wanted to raise revenue collection from the personal income tax and they wanted to decrease after tax inequality in the country. And the question is just did that work out? So to give you a bit of context, what you see here is a graph that plots you on the vertical axis, the top marginal tax rate of countries. So each of these acronyms is a country and on the horizontal axis, the gross domestic product per capita. And what you see is there's a positive correlation. So better developed countries tend to levy higher marginal tax rates on income earners presumably because they have more tax enforcement power or better systems like in terms of third party reporting that just helps the tax revenue authority to really enforce these high top marginal tax rates. And South Africa is the red dot and what you see is that even before but more clearly after the reform the top marginal tax rate was quite high or is quite high for a country of that development stage. So in terms of which individuals were affected, so the reform as I just said affected the very top end of the income distribution in total a bit more than 80,000 individuals were affected by this reform. This is a small fraction of the total taxpayer population in the PIT system but this population contributes a massive fraction of the total revenue collection, 22%, so it's 0.6% or 0.58% of the individuals that are treated but they in total contribute 22% of the personal income tax revenue collection in the country. So this is arguably relevant from a revenue collection perspective. So what we aim to do in this research paper is in the first step to understand in a broad way what these individuals did. So they were treated, they faced a higher top marginal tax rate and the question is how did their tax income reporting change because of that reform? And this is very difficult to identify empirically so if you think about it then one quick approach might be to say okay I just look at changes in income reporting of these treated taxpayers relative to some controlled taxpayers that have lower levels of income and are therefore not treated. And the problem and this is what you see here in this graph is that the income trends are very different at the upper end of the income distribution and at the more lower in the middle of the income distribution. So what we see in South Africa and other countries as well is actually income trends that are reflected here in this graph. So you need to understand the graph so on the horizontal axis it's just the income level and on the vertical axis it's the income trend. So how does the income trend over time correlate with your income level at a certain point in time? And the correlation is actually negative so if you focus on the blue graph and it's also negative in South Africa what we observe is that at the upper end of the income distribution the income trends are smaller than at the lower end of the income distribution and the reason is mean reversion. So we have a lot of mean reversion in the data so people may in one year have a very high income and then they revert back to the mean in the next year. And so what you see here is these underlying income trends are very different at the upper and lower end of the income distribution. So our empirical identification strategy now what we do is that we say we take a pre-reformed period to identify the blue line so there was arguably no change in the PIT system and that period we use to model differences in income trends at the upper and lower end of the income distribution and then we look at the reform period and this is what you see here in red and in the reform period the reform effect kicks in and what we would expect is that individuals at the upper end of the income distribution may actually reduce their taxable income reporting because of this tax shock. Okay so this is the strategy and what we use in terms of data it's the universe of personal income tax returns from South Africa for the years 2011 to 2020 and this information combines the actual PIT returns so the tax returns that the individuals submit and pay information submitted by employers so third party reported income information and as dependent variables we use two income concepts a broad income concept and then actual taxable income after deductions that the personal income tax system allows. So what you see here is just pure descriptive revenue collection in the PIT system in South Africa across tax years starting in 2011 until 2020 so the blue line gives you overall PIT revenue collection and the red line gives you revenue collection from the treated taxpayers so taxpayers above 1.5 million round income and the reform so the first treated year was the year 2017-2018 which is indicated as 2018 in my graph and what you see is looks pretty flat so what we had is we experienced this increase in the top marginal tax rate but if you just descriptively look at the data this does not show up like in a take up and tax revenue collection from this group of taxpayers so here in this table in the first line you see the number of individuals that report income above 1.5 million round over time and if you look at this line and the first this number then you see it goes up until 2017 and then the number of individuals with more than 1.5 million round of income tends to drop and the second line gives you how much taxable income do these individuals report and you equally see a drop in 2018 which already descriptively points to a taxpayer reaction that they might reduce their taxable income in response to the reform so here again you see the graph with my empirical identification design and here you see the estimates from South Africa based on the real data so the blue lines again to understand the graph so on the horizontal axis you see the taxable income in a pre-period and taxpayers with more than 1.5 million round of income they are treated and the blue line gives you a pre-period so in the base this is the period between 2012 and 2015 so it's always a three-year gap and taxable income change of taxpayers in different percentiles of the income distribution and what you see here is this declining trend that I also theoretically show you driven by mean reversion that we see in the blue pre-period and red is the reform period so this would be the period between 2017 so pre-reform until 2020 and the tax year 2020 is not affected by COVID so this is prior to the COVID outbreak and what you see is that in the we call that validation region so taxpayers below 1.5 million round of income these curves overlap so you see that the income trends are very similar in the pre-reform and the reform period and in the treated regions so taxpayers that were treated by the reform you see that these curves diverge which is suggestive of a response of these taxpayers to the reform and this is broad income and you see the same type of pattern if you look at taxable income and account for deductions in the tax system so this translates and this slide is more for the academics in the room this translates into a substantive elasticity of taxable income in the South African economy this elasticity of taxable income tells you if you increase the net of marginal tax rate by 1% by how many percent does the taxable income reporting increase so here we need to think in reverse it was like a reduction in the net of tax rate by how much does the taxable income in your economy is reduced and the first this is again talking to the academics in the room the first column is a reduced form estimation the second column is an instrumental variable estimation where we instrument the actual change and the marginal tax rate that taxpayers face by the policy induced change by this policy reform the increase in the top marginal tax rate and this elasticity is very high so that means that these taxpayers reduce their income by 10% around 10% in the treated region so now you may have concerns about this approach and may say ok perhaps something else happened perhaps there there are just changes in these underlying trends over time at the top end and the lower end of the income distribution and in the paper like one strategy to avoid this type of bias is just to look at so-called placebo tests and say we look in the pre-reform period and just in the pre-reform period ask ourselves over time do we see that this red and the blue graph over time does it change and we do not see any systematic changes over time in the pre-reform period which kind of substantiates that what we see here is really a reform driven effect driven by the increase in the top marginal tax rate so what the graph that I showed you initially already suggest is that this response increases across the treated taxpayers so what you see here is this is only the treated group and this tells you what is the elasticity of taxable income so how elastic do they react to the reform across the treated regions in the upper the 10 would be individuals in the very upper-decile of the treated region and what you see is they respond more strongly the higher up their income so high income individuals within the treated region respond more strong and then one thing we are really excited about in this project is that the richness of the data really allows us a bit to dig into how do these taxpayers respond and how do we see a response and over all taxable income and of course kind of the question is how do they respond is this in which income component is this evasion is this avoidance is this a real response and of course for the country it's very relevant to get an answer to their question and how to think about this policy reform so what you see here is just the distribution and the composition and not the distribution sorry the distribution of income taxpayers so 1,500 this is the 1.5 million individuals and what you see is just what type of income do they earn and normal income would just be IRP5 income so pay income reported by employers to net of in yellow allowances and then orange fringe benefits so this is also reported by the employers and what I found interesting is that these income earners they report relatively little business income this is different in other countries so at the very upper end of the income distribution you commonly see higher fractions of income earned by business income so we still want to dig into that so we haven't understood that very well but and this is like interesting descriptively and investment income plays some role at the level of the relative to normal income it's like not huge huge okay so here this might be one of the most interesting tables at least to me from this paper is so here we look at responses in different income components and the first line gives you responses and pay income so IRP5 income net of fringe benefits and allowances so the elasticity is 0.26 or 0.41 which is a response but not like massive it's not the one that we saw in overall taxable income and then in the next line you see gross income of these taxpayers minus this third party reported employer income and there you see like a massive response so the response is driven by income that is not covered by third party reporting or even remitting of the employers business income does not show a response so business income does not play a major role and we have not I think we need to look deeper but if we look like in a plane where we do not see like a massive response of business income we see a quite substantial response of investment income we see a bit of a response in deductions this is actually quite interesting in the study per se of taxable income and this may relate to the design of the South African personal income scheme where deduction is relatively simple with relatively few deductions and so deductions are commonly perceived to be one and how taxpayers respond to these types of reforms and what we see is even in a system which is very lean with relatively few deductions we see a strong taxpayer response here in this context and fringe benefits and allowances also respond quite a bit this allowance response is driven by different types of allowances there's like one category other allowances this drives this a bit and yeah so this yeah so this is like the main takeaway here so this is the top of the income distribution and shows you like the last interesting finding I can convey for today and shows you in terms of if I only look at the income that these individuals report how it is actually composed between monthly income so that the income that individuals get every month and annual income this is in red and this would be bonus payments or incentive payments so this is payments that people get like once a year and then commission and this graph is in to show you that this annual income and so bonus and incentive income is quite important at the upper end of the income distribution and if we check how these income components react we see essentially nothing in monthly income and a quite substantial response in annual income and this is consistent with prior findings in the literature that suggest if these very top individuals face higher tax rates they may reduce the effort provision and may not be as successful in achieving their targets within the firm as before and this means that might be a real response that you see in the economy so asking is this like avoidance evasion or is this a real response I guess this is the result that most looks like a real response in our study and then we link this to the firm side so essentially if they really show this type of behavior then it might end up showing up for the firms as well so we basically match our personal income tax returns to the corporate income tax returns and look at the fraction of tax payers within these firms that are affected by the reform and then ask ourselves do we see like a reduction in output and activity of these firms and the good news is there are not many firms treated in the economy so it's one tax payer that is treated by the reform it's only like 1.5% of the firms that are treated and we see a bit of a reaction which is quantitatively, relatively modest however so this would be sales of firms of different size in the economy that are treated and we have different treatment indicators and the fraction of tax payers that are treated or whether a firm is treated at all and we see like a 3% reduction in output in response to the reform so this looks a bit like a real response it's arguably quantitatively not massive but I think so what we think is interesting here is that you see the richness of the data that allows you to track these types of reforms and also control for confounders of course you may think these firms are very different that are treated from other firms but we can absorb like differences in industry trends differences in firm size trends here and still the effect shows up okay then you can think about what happened to revenue collection and inequality in the country so this would be revenue collection so what you see in the upper line is if the tax payers would not respond to the reform so static just the marginal tax rate increases what is the additional revenue collection that you have in the economy and then you would collect around $8 billion more in taxable income but we see this massive response by the tax payers which essentially suggests that South Africa collected less revenue because of the reform because of the strong reaction what about inequality inequality may have dropped so here you see the Gini coefficient in 2017 if you just account for the tax change the drop in the Gini would be from 0.626 to 0.618 and if you again account for the behavior response you get an even stronger reduction in the Gini coefficient you need to keep in mind that not all of that is plausibly a real response much of that is avoidance or evasion so this might just be reported income inequality not actual income inequality however if you think there are evasions or avoidance costs if you think about consumption inequality that may have dropped so summing up so what we find is the tax payers responded quite substantially to that reform we see that the response is centered around income that is not subject to third party reporting with the exception of this incentive and bonus pay it may have some had some mild repercussions on the real economy and likely it did not increase revenue collection in the country thanks for listening