 Good afternoon everyone. It's very nice to be here and thank you to the organizers for inviting me. I'm Christos Kotsjiannis. I'm based in the UK, University of Exeter, and I'm also leading the Tax Administration Center, which is a center that is doing research on tax administration primarily, but taxes more broadly. Now this project is, as you can see, the title is Audits and Compliance, Evidence from Ugandan Administrative Data, and I'm delighted that I've teamed up with an excellent team of researchers with different expertise, David Hennig from UCLA, Yuga Petila, who's also here, and he's going to answer all difficult questions if you have at the end, and Inuluka Salvadori, who's based in Barcelona. Now this paper is about audits. It is about understanding whether tax audits have an impact, a compliance impact. Oh, before I continue, this is part of a project, a long-term cooperation between UNWider and the Uganda Revenue Authority, and also, I need to say something else, which is a visit disclaimer that the views are our own, and they don't reflect those of the UNWider and the ERA and its management. Okay, so what's the motivation again? We would like to understand whether tax audits, operational audits, no tax audits, no work in the sense of providing compliance in the future. We know that there are two things with audits, no tax audits. One is they have a contemporaneous impact. There's a verification of states, businesses, tax payers get audited, and some tax liability is verified if it has been underreported. But there's also, more importantly, that audits change behavior in the future. And this is the, what to try to tease out with this research. And this is what we call the dynamic impact, the inter-temporal, how tax pays behave in the future once they get audited at some point by the tax authority. Now, the type of tax basis we are looking at in this study, at this point in time, we're going to be extending that as I said in a minute, is to evaluate the corporate income tax audits, and also the VAT tax audits. There is a plan to look at PAYE tax audits, but this is something that we haven't done as yet. And by the way, what I'm going to say today is kind of tentative. We're still working on this project. It's a long project. Quite data crunching is not easy, but we're getting there. So the result I'm going to show you, we think are interesting, and it's unlikely that they would change as we go through some robustness exercises that we're planning to do shortly. Now, so what I'm going to do is I'm going to go through quickly what we know in terms of literature, and also give you the gist of the findings, and then tell you how we do it very quickly. It's not going to be technical. It's going to be kind of a very high level presentation, but I'll be happy to talk to you if you find that interesting and go through the technicalities and more details of this study. Now, over the last years, there has been, and I guess because of tax authorities becoming more open, opening the doors to scientists, to play with the data, look at the data, examine the data. There has been work on evaluating tax audits. It started with a very nice paper early in 2000, 2010, using Danish data, and then from then on, more papers in the UK came out, mostly from developed countries. The intensity is kind of not continuous. Again, as I said, more tax authorities, kind of opening doors to us, and mostly from the developed world studies that we have, mostly tax audits perform what we should be expecting to be performing, which is kind of they have a positive compliance impact, mostly, not always, but mostly, that's true. Now, in terms of the developing countries, the evidence is more limited. There are some nice papers, and I have the references here, in the lookie at Pakistan, BET, tax audits in Pakistan fighting no impact of those audits. Myself with some others, we have looked at Rwanda, trying to assess again the impact of audits, but more importantly, different types of audits, how they impact compliance. We find some interesting findings there, which is kind of counterintuitive, possibly. And then there's another paper in the lookie on network effects in South Africa, and more recently, another study on Cameroon, which is again ongoing. Now, what this study is going to do is kind of contributes to this evidence. External validity is difficult with these studies, because countries are different, they have different institutional frameworks underlying them, so it's difficult to take a study and say, if we apply these rules, the same is going to happen to the country we're looking at, so it is important that we kind of continue this effort and try to understand the mechanism underlying any impact, positive or negative that we find. Right, so what is the, let me go through the snapshot, just as I said, the results are very tentative. Now, what we find when it comes to the corporate income taxes, there is some positive future compliance going on in terms of no CIT, no liability, so once businesses get audited, then they comply in the future. I have a caveat here, which is the setup is not yet perfect, as you've seen in a minute, but we believe that this positive impact is going to stay even if we find here the analysis a little bit more as we should. So, let's keep that, there's a positive future compliance going on. Now, the evidence also seems to be suggesting that actually what is driving these compliance is not all businesses across the spectrum, but it's those businesses that are kind of declaring nothing. So these are what we call the nil filers, so the nil filers who have been audited, they seem to be driving these compliance, this future compliance, positive future compliance after the audit. Now, also we have been trying to tease out whether, because no tax authorities are doing two types of audits, they do comprehensive, which is kind of very intensive, so-called line by line, and also they do narrow scope, focusing on the single issues, very specific audits. And what we find is that if we exclude those narrow type issues, then there's no major change in the results, so as you've seen in a minute. And importantly, something that hasn't been identified as getting the literature, we find no difference between those who have been found compliant or non-compliant at the time of the audit. Typically, the literature differentiates, finds differential impacts in terms of compliance for those who have been found compliant and non-compliant, with those non-compliant driving the future compliance. So here, that's not coming through, and I think this is a very interesting finding that is worth exploring and understanding more. Okay. Now, a little bit on the specifics now. On the aggregate, we find not a consistent dynamic response. We find in significant, so this is for the VAT, for the VAT. On the aggregate, we find no consistently dynamic response, but we find in a significant impact view the first and the second year. And positive impact started from the third year. Focusing on the impact of audits on the probability of reporting positive ad liabilities, those business who report positive ad liabilities, we find a significant negative impact, suggesting that the impact of audits depends on the sign of the VAT liability reported. I have only five minutes. I'm going to speed up a little bit. Thank you for that. And as I said previously, even for the VAT, it doesn't really matter. The result doesn't depend on the audit outcome. I'm going to skip the institutional details a little bit. Nothing strange here. I mean, this is a conference on developing countries. So Uganda, the tax GDP is low. Kosamu tax contributes a lot and things like that. So let me skip that. Now for the audit process, they do audit. Audits are risk-based. They used to scores. Risk and financial importance scores. The selection is not deterministic. There's some discussion going on, which we need to be careful in the empirics. For that, and as I said, there are two types of audits, comprehensive and issues, issue-oriented. And the penalty is not 2% and an interest 20% on the income of the reported. Now for the data, we have the universe, I think, of the 2014-2021 for CIT, VAT, and currently, as I said, we're working on the PAYE. We have the universe of the data on the records and information selection process. So we take all this data and then what we do is, and also we have the audit waves because we have audits every year. So we have all these audit waves to use. We use the matching to identify the counter-faxual and then we're using this constructed control group within the different diff to estimate the impact of the audits and either matching or we use various variables to do that. So this is for the CIT. Now graphically, I'm going to show you what I said at the beginning. This is for the CIT. And as you can see, the line, the vertical doting line is where the audit has happened. So as you can see for the CIT, we have following the audit, CIT declarations increase. This is significant because the confidence intervals are above the zero. And we're using a very, as I said, matching and the impact is the same. Yeah, so you see a positive trend. So in the CIT or the aggregate, the average treatment that treated is significant over the years. Now for the VAT, things change and we try to understand why this is the case. We're looking at a little bit more detail in the VAT rules to see how why business behave like that. So for that, we find significance two years after the audit. So there's sluggishness. There's a late reaction. We don't really understand that, but we'll try to kind of look into the data and see why this might be the case. I think this is interesting. As I said previously for the CIT, whether business have been identified as compliant or non-compliant, it doesn't really matter. They comply better in the more in the future. So again, this is something which is, we find interesting, something that hasn't been sort of observed in the empirical literature before. For VAT, things are the same. So there's a sluggishness and it doesn't seem to be, there's some significance years after the audit. And that's independent of the type of behavior. Businesses were compliant or non-compliant. And let me conclude. I have one minute left. So this is a conference on revenue mobilization. So evaluate tax audits is very important. We need to understand whether these, for revenue mobilization, we need to understand whether the audits work and if they don't, we need to find ways of fixing them. So evaluating is important. What this project is about is to evaluate these tax audits. We find some positive compliance going on, new filers are driving the response. And the VAT audits do not have a compliance effect to the SOTRA, but only from the third year on. Now, what the research doesn't do is it doesn't evaluate the quality of audits, including the processes, the allegation to audits, et cetera. This is something that is important. Very little work is going on on that. I think that needs to be more carefully looked at. It's important because they affect the outcome. And it's an issue that deserves further attention. Thank you very much indeed for your time. Thank you.