 So this is work, I would say, very preliminary work and maybe Javier set the bar very high and now we go to South Africa and it's some similarities but he's stolen some of the things we're going to talk about already so we should chat afterwards also to see how the comparison works. So this is work on top income inequality in South Africa and this is evidence from this administrative tax data, it's individual level tax data in South Africa and this work is with a PhD student, Chandra Jacobs at the University of Cape Town, Marie Lebrun at UCT and then Yocca and Marlis Pekus with us at UNU wider. So if you attended some of the presentations about South Africa yesterday, you'll already know income inequality is persistently high in South Africa and a big part of the discussion yesterday about income inequality was about the data and how to get reliable estimates which data sources give us what we want to find and so a lot of the empirical work in South Africa is focused on survey data and household incomes and then as Javier said the challenge here is always under reporting or non-response at the top end of the income or the distribution. Now tax data provides us with better coverage at the top income or the top end of the distribution but then of course it's got this little coverage at the bottom and that's always going to be your shortcoming. Base case scenario, you merge survey data with the tax data and there's at least one paper where they try and use both survey data and tax data. They don't actually merge it and it's not yet possible to do this in South Africa but we hope in the future that this is something that you know we can maybe another project for a PhD student to start thinking about. So what's our contribution here? Here we're thinking about measuring income inequality. This is for South Africa so in the formal sector for income earners and then now we've got this period so pre-pandemic 2011 to 2018 and using this sort of new individual level tax data. I also want to add here, so I'll tell you a little bit about the data. The data set has two main sources, this is information from the payroll and so for South Africans in the room this is IRP5 data so this is information that firms report to the tax authorities. This is then merged with personal assessed data so self-reported data or self-reported tax assessments that you send to the tax authority or ITR12 records and so here you get the income earners and well those who also are self-employed. I think I mean one of our advantages here is actually coverage of the top incomes and so our focus is a little bit in that direction and we've only very recently sort of moved in that direction as recently as last week but the coverage of the top incomes we think is sort of what we want to focus on in this work and then if there are any ideas I touch on this sort of panel element and so I'd like to hear at least from the crowd hopefully if there are any thoughts on what we can do with the panel element of the data. Unfortunately with tax data especially at the top end there's problems of tax avoidance and tax evasion and we don't touch too much on this but this is something just to keep in mind as we do the analysis. Okay so this is our overall sort of income distribution graph or K-density for the for the panel period that we have and this is real gross income for each tax year and it looks like this sort of the slight shift in the distribution to the right over time and then these little peaks over here that we need to still a little bit investigate. I know it's not driven by business income we tried to remove that last week but it might be some either round number bunching or there might be some kinks in the text schedule that's kind of driving this so we you know this is at the stage that we are thinking at the moment. Here are the real income real income in percentiles so we've got the 10 the 50 or the 25 the 50 the 75 your 90 and your 99 and then all except the 99 are on this axis and so for the period that we're looking at we're not seeing a lot of income growth it's quite sort of flat when you hit the 71st percentiles this kind of starts to go up and you can see this year the scales of course different for the 99th percentile but you are seeing this sort of income growth right at the top and this is not a new story there are you know research previous research but from earlier periods it kind of shows that this is sort of what's going on in the South African context in terms of the Gini coefficient so I'll before tax Gini see if I can swap hands so before tax Gini over here is around point six four and then after tax it's about point six one and at least the squares up quite nicely with the survey data at the sort of point six four this is sort of what we saw yesterday and some of the presentations and then it was nice to see obvious presentation of the sort of you know what you can start to think about in terms of tax reforms in terms of changing this kind of distribution we haven't got to that stage yet this is just sort of way where we want to point to in the different estimations yeah in terms of the top income she is over time so there's a paper by Marie LeBrandt and co-authors they use survey data and they show that I think this is for the period 1993 to 2008 so kind of just before this before this period and they show that income has become increasingly concentrated at the top and this is the story kind of continues like we this isn't really changed much in the South African context so what do I want to say about this graph so for your top 10% we're looking at 48% around 48% closer to 50% of income and for your top 0.1% so 0.1% yeah so 0.1% it's looking at about 4% what's capturing about 4% of the total income now in the tax data the tax data in South Africa we don't have a ton of demographic characteristics we know where people or you know the age of individuals and we know the agenda and so this is the one one where we've sort about slicing the data in terms of sort of looking at within the males and then within females and then top income groups so I think the one thing to point out here is that in the beginning of the period in 2011 they're far more men in the data set than they are women but the numbers actually equalize quite a bit later on for these top income groups so the genie for men genie coefficients for men are the blue lines for the top 10% the top 1% and the top 0.1% and you can see they are a bit higher than those for women except here at the bottom and then I think the one other thing that we might want to sort of tease out a little bit here is that the the there's a slight increase in the income share there's a slight increase in the income share for women I think at some point I've said genie but I meant income share okay so this is a comp so I guess some of the other interesting things that you can think about in the tax data is that you actually have different sources or source codes for each of the incomes and this is quite well reported in the South African tax data maybe the one thing that's missing from this data and we sort of have some of it is around dividend income and so that's something we still need to think about but in terms of earnings and interest business income capital gains this is sort of what we expected to see where the business income and capital gains kind of becomes more important in the higher day sales we break this down a little bit further we look this at this top 10 10% this becomes a lot more prominent so we see this business income becoming a lot more important and then of course the capital gains particular for this 0.1% group yeah so this is the part that I think is a little bit under explored and we want to put a little bit of thinking into this yet and so we literally started looking at this this week and I think the main so this is our transition matrix so what we where people are at in 2011 versus 2018 and then this is the sort of your 0 to 90% and then your top 10% that's broken down and the majority stay here at the bottom between these two periods but I think the I mean the overall message we're getting from this was that mobility for the formal sector top income earners is actually quite low almost 65% of individuals in the 99% I'll actually stay sort of in the in the top 1% so if you look here yeah this is sort of getting to this they kind of remain in this after I mean eight or nine years of the tax data so somebody this is so we I mean inequality remains high this is not we're not you know presenting some kind of new radical story about this but we do think it's consistent driven by high income high and maybe growing incomes at the top end of the distribution there's not much inequality between gender relax not something that we we see I think one important thing to mention in the South African context is that race plays a big factor here this is not something that we can actually do in the tax data you don't need to report your tech your race to the tax authorities that's something they collect as such but for us we want to at least look at do some comparisons on some of the earlier estimates we have a little bit of information about regions so where people are located we might want to think about how that how that changes from place to place but we definitely open to hearing more ideas thank you