 Thank you. I'm going not to talk. I'm not going to talk about tax havens I'm not going to talk about anything to do with that, but I'm actually going to talk about my learning experience of getting into The tax fairies and your domain some fairly new to this Luckily, I had co-authors Sam Corbina and Eva Risto who's not new to this But the reason why I started and looking at this is actually I can start with an anecdote It's a because my father he owned a company and he had a lot of employees in this company and at that time when he Actually negotiated wages with with his employees. He actually did what you call an ad hoc type of negotiation because there were different tax brackets in the Danish tax system and He wanted to say it's also beneficial for you to stay under This tax bracket because then you're taxed a little bit lower and I will benefit as well from this and Because of this I will give you some of the services that my company can provide for example a car Free of use we don't have to record and then so forth in that sense It gives you an idea of why I have looked at this The differential bunching impacts across the income distribution because the idea basically comes from this Of course a lot of people have done this and I have to say it's not as innovative this paper It's more my learning experience. I would say than the two previous presentations The unique part here is that we are using I think more or less for the first time the sampe and tax admin data For more or less any analysis of this sort. So that is the new part of the paper so Given that I'm new I could also read the the articles with a little bit more skepticism than Just citing office that most of you will know that I didn't know and I could also be maybe a little bit more critical For because the first question I actually asked was related to domestic resource mobilization There was all of these cool methods econometrics very hardcore in the end a lot of identification assumptions Causality claims and so forth in the papers, but in the end they didn't talk a lot about the magnitude of the effect and Then I came to think about my anecdote again How much did it actually matter that my dad paid just below the tax bracket? How much is the aggregate effect of this and in my mind it cannot be that large Even though the papers are really looking into this This was my main reason for going into this so the domestic resource mobilization So how big distortions does it actually create to tax collection in that sense? It could Change Yeah, I'll trigger real economic responses Effect labor supply it could also give rise to illicit behavior in form of reporting responses And my claim is that the first one would be huge effects. Whereas the second one May be more limited in terms of the impact There's a lot of literature on this using taxpayer data to analyze this and The most important papers on contribution is Emmanuel size and Henry Cleveland I think some of the papers and of course jetty and co-authors Important to know is that the pay as you earn income tax in Sambia is a graduated system where tax liability increases progressively In and each Bracket is associated with a fixed marginal tax rate So we have this discontinuity and that is what we are going to exploit in the analysis If we look method in terms of the methodology We are basically borrowing a lot of the methods from Cleveland and Basin and the recent paper by Bell in South Africa So we analyze Behavioral responses to discontinuous jumps in the personal income tax rate accounting for issues related to reference point problems and Some of these round number problems that we are also facing we will basically also correct for that according to the literature What we know is that this may drive a wedge between the structural elasticities and the observed elasticity That is estimated and that is due and the first one is to show this was Cleveland and was seen in the case of Pakistan We also know that reference point effects amplify bonching and It may overstate the true elasticity. Okay, so this is our point of departure. I Will not have time for the literature as all the other office as well But there is a lot of literature on this even in developing countries We have papers in South Africa your right China Mexico and so forth and the important lesson here is that they are finding finding more or less complete opposite results of bunching and very few of them are actually Telling us anything about the combined effect or the magnitude the aggregate So what I learned is that in order to be taken seriously in this literature I needed to combine all of these methods. So I actually have learned quite a lot during the last year or so I've worked in development many years, but not in this domain so we are basically Taking into account the distinction between kings and notches. We are basically taking into account reference points and number preferences we also Distinguishing between wage-earners and self-employed that turned out to be quite a substantial difference in the literature Whether you consider one of the other with different results in different countries again, I have to say and Then we also need to distinguish between what is behavior and real responses because that is going to affect the magnitude and the reason why I think a lot of papers are not Discussing this much. It's because I will show you that it's not impressive at least what we find in terms of magnitude So we apply these bunching approaches to tax admin data over the period 2014 to 21 and I will give you the results that we find significant evidence of excess bunching at the first kink of this tax schedule in Sambia and we also find a limited evidence in the second kink of the tax schedule, but not a lot or Not significant evidence at the third kink. We cannot find bunching We find a lot. It's important in the Sambian case at least to correct for Round number bunching so that is actually also reducing these elasticities quite heavily This is just to give you a picture about how the tax schedule looks in Sambia and how they changed over time just to keep it Very simple and these are basically the bunching estimates and the counterfactual distributions Where we have the distance from the kink at the zero level Okay, so very standard according to the literature. There's no new things to report in terms of methodology The round number bunching I have to say appears to be very time dependent and I cannot explain that I don't have an explanation for this today. I don't know enough about Sambia I'm talking to my Sambian colleagues about this, but we are finding finding much more or the largest round bunching Round number bunching in 2021 as compared to 2016 So what is the reason for this result that we actually seeing changing responses to round number bunching over time? According to the literature that should be constant So what is going on? Is it a data mistake? I cannot tell you today. We're working on that So what are the results that we get in Sambia? We actually get results that are in line with Papers in South Africa where observed bunching Acts sharply and immediately to changes in the location of the kink points. You saw we had changes in the kink points immediate changes that could suggest that there is a behavioral response Real responses would result in a more scattered response around the kink point than we observe generally We do observe it, but we would have expected to have larger effects more over real responses require Adjustments along different dimensions, which may require more time So we also would see maybe a more gradual response over time Adjustment of for example working hours will take time and will not have an immediate response It will depend on the contract given that we are observing formal sector workers okay reporting responses may Are less detrimental to welfare as is shown in some papers compared to economic responses? So we should take into account these aspects Results differ heavily from the bell paper that is also from South Africa They find that the largest bunch of bunching responses are at the high kinks. We find it at the low kinks Is it something to do with the difference in? behavior in South Africa and Sambia may be but it could also be that the What do you call it the real income that you have in Sambia? Where your taxed is actually at a higher level that is comparable to the high kinks in the South African case That we are looking into as well They also find evidence for bunching among the self-employed workers and not for wage workers. We find it significantly for wage workers Luckily our results are consistent with recent evidence taking into account reference point problems as we do and Therefore we actually believe that our results are a little bit robust to this consistent with the Mexican paper or the Badges and social paper we find that the behavioral response to the kinks is driven by reporting responses And that's the important point here That means that I would expect Low impact on the rest domestic resource mobilization because it's not a real response good Now we are working on the counterfactual distributions trying to estimate the laws and in Taxes taxes collected given the counterfactual distribution comparing it to the individuals around the bunching Moving them to where they should have been in the counterfactual distribution and Then just trying to calculate the overall impact on collected taxes We're working on this working on different methods on doing this because there are at least eight different ways of doing this We have only done two until now So how big is the mix miss tax revenue arising for this excess sponging? How big or important is it in the Sambian case at least where we have reporting biases? You can say and not so much real effects So we do these simple calculations and we actually find that The missed revenue can be estimated to a roughly 26 million The Cushia and that is only about 0.25 percent of the total Taxes collected from the pay-as-you-earn schedule. So limited effect what we are doing now is trying to redo all of the other papers and Calculating the impact that they get because they do not report it But I bet you it will be low The domestic resource mobilization effects are these kinks So I will be a little bit strict now to all the other office as well When we are talking about domestic resource mobilization, maybe these kinks are not that important in the tax Thank you