SA-TIED Seminar Series: What tax data reveals about profit shifting in South Africa

Good afternoon everyone and welcome to today's session of the essay tide seminar series. My name is Abna Labi Odum and I'm going to be facilitating today's session. We're joined by Matthew Abbugna from the tax justice network and he's going to be presenting some new and exciting essay tide research which at a very high level looks at how tax returns as well as customs data. Sorry about that. Can help us identify uh possible profit shifting by multinationals in South Africa. A few housekeeping rules before we kick off. We would ask that you please keep your microphones muted for the duration of the session. Matthew's going to present for 40 minutes and then we'll go into a Q&A discussion. Matthew, over to you. >> All right. Thanks very much, AA. and the rest of the team for organizing this. Um so as you mentioned so I'm Matthew and I work um for tax justice network and it is my pleasure to present this joint work with Ron Davies who is a professor investor at um CH University. So in this work as abena mentioned so we are basically asking a question as to whether the data that is available to SAS can they use that data to be able to um identify multinational enterprises that are high risk in terms of profit safety so that they can narrow down um the number so that they can focus their auditing resources on those firms that um exhibit features that are very similar to firms that engage in profit sifting. And that's what we did basically on this um research project. Um so to begin with so we we we all know that profit safety is is is a big issue globally and that has to do with the fact that um it involves multinational enterprises moving profit from one country to another usually from countries that have relatively higher corporate income tax rates to countries that have um lower corporate income tax rate. And the idea is to basically um move the profit into those countries so that at the end of the day um the global profit the global tax that this national enterprise will be paid will be lower as a result of this profit sifting. So recent estimates um actually suggest um that in in annual basis around 1 trillion where um multal enterprises profit accepted to countries that are considered as tax savings and tax savings are basically countries that have corporate income tax um effective corporate income tax rate which is um often below 10%. And why would that matter for South Africa? So if you look at South Africa, the best estimates available suggest that on annually South Africa loses between 1.3 to 1.8 8 billion every year as as a result of profit safety and this is a huge number to a country like South Africa which is also a developing country and probably would require this amount of money to um assist in the development financing process. So to even put it in more perspective so the dollar amount might sound more abstract to many of us. So if you look at that in terms of the run um we're talking about 24 to 20 um to 33 billion worth of run of multinational enterprises um profits are sifted out of South Africa annually and this is almost equivalent to um the water and sanitation department um annual budget as well as the department of human settlement annual budget. And this clearly shows us the fiscal stakes that um profit sifting has on on the economics development of of South Africa. Because if we are basically losing a department back to profit sifting then that raises a serious question as to how we can tackle that. Now so profit sifting is not a new thing. The existing research has clearly indicated mapped out the various strategies that multinational enterprises use to move um profit from these high corporate income tax jurisdiction to these lower um corporate income jurisdictions that we called tax haven. And one of such channels is the transfer pricing where we have this group of companies that um do transactions among themselves which is um absolutely legal but then they do in such a way that they underpric or overpric this um internal transaction in such a way that the firms that are located the affiliates that are located in a high um corporate income tax country like South Africa would end up um taking more of the cost in such a way that at the end of the day um it has a lot of internal payments to that group in a manner that most of the profit are actually being moved out of a country like South Africa into the other fleets that are located in the lower um corporate income tax jurisdiction. That also applies to the depth sifting just in a similar manner where the group arrange the depth in such a way that um those countries those groups that are in the higher corporate income tax jurisiction takes more of the debt. So that at the end of the day then they have more deductible debt um servicevicing payment which eventually reduces the amount of corporate income tax that they would pay in those high income jurisdiction and that is also similar to the strategic allocation of intangible assets usually in these tax saving countries where these multinationals do not actually even have like real economic activities but they locate these um strategic entangleable assets there and then in in turn sell those assets to the affiliates in the higher corporate income tax jurisiction so that at the end of the day um they reduce their corporate burden there. >> Yes. >> And what is the problem? So this is well documented but then the challenge still exists because we know for most developing countries like South Africa is extremely difficult to be able to detect this anomalies and this practice by multinational enterprises largely due to the fact that most of the most developing countries tax revenue um authorities operates in a really hard um resource constraint environment and most of these small town enterprises are actually more advanced when it comes to some of these activities. So it becomes extremely difficult for these resource constraints um countries to be able to detect that. And if you look at the estimates actually suggest that even though these developing countries uh are not able to detect some of these things and also to prevent um these leakages they they are the mostly that are affected in terms of the the amount of tax revenue that has been moved out of their country relative to their country tax basis. And this is probably has to do with the fact that most of these countries really are not able to assess the global the universe of the multinational enterprises that operate in their country. So what they basically see based on the information that is submitted to them is activities within their country. So they're not able to um see what are the other activities in the affiliates of those mult enterprises in other jurisdictions. So that constitute um a big issue for them to be able to detect this professor sifting. So what we did is to try to address these challenges by basically developing a simple practical toolkit that would allow tax authorities to use the data that's available to them to able able to flag firms that behave like profit sar audit resources to those firms to actually establish whether if indeed they are engaging profit sifting and to provide a basic um a brief institutional context. So we know that South Africa is one of the countries in Africa that is mostly industrialized and also a middle income countries and at the moment has one of is one of the countries with the highest tax to GDP ratio in Africa which is around 25% relative to the African average of 16%. However, that is still below the OECD average of 34%. In other words, if you're comparing South Africa to APS in Africa, then we would say that South Africa is doing well in terms of its um tax to GDP ratio. But that is actually also below the OECD level. And in terms of the corporate income tax contribution to um South African total revenue, we're looking at about around 19%. The corporate income tax contributes around 19% of the country um tax revenue basket. um given at its um initial 28% corporate income tax rate which has eventually reduced the 27% corporate income tax rate and South Africa is one of the countries in Africa that has this robust rules to prevent um profit sifting by multinational enterprises which includes the transfer pricing rules as as well as um a host of other rules And recently they started um implementing one of the OECD um rules which is a country byc country reporting to be able to see the activities of multinational enterprises beyond um South Africa as well as the implementation of the global um minimum tax. So what remains is that yes in terms of the legal point of um thing South Africa actually has this robust legal system to be able to combat um profit sifting but notwithstanding that the South African Riven authority still faces this um capacity constraints which is common to developing countries as I mentioned in terms of even having access to the right information when it comes let's say transfer pricing to be able to make the necessary comparison to know whether this transaction is actually within the accepted um am pricing mechanism or otherwise. So in that case what is needed is to be able to use the resources that is available to the authority like the data and the human resources to be able to at least um flag those firms that behave like profit saf so that they can um focus the audit resources on those ones. So in terms of contribution so the generally there's a lot of studies that actually look at profit sifting in terms of how much M& are um sifting profit out of their um one country to other and if you look at studies that are closely related to us is is Blinka who actually says that multinational enterprises in the UK are often reporting profits which are very close to zero as compared to local firms And if you look at um other states, Rasmuseno also did similar studies looking at the firm level in South Africa to say that most of um the multinational enterprises in South Africa um engaging in transfer prices which are dis um not in accordance with the accepted um pricing mechanism. So what we did is to basically extend these studies to include a multi-dimensional approach not just measuring how much profit is being sifted out of South Africa but to also look at which of these firms are actually engaged in this um profit sifting so that we can be able to um identify them for the tax authorities to actually conduct um an actionable audit on them to determine whether these firms are actually engaging profit sifting. And this methodology is something that is applicable to other developing countries. And in fact, we are actually um extending this to a couple of other African countries as well as um developed countries just to ensure that our methodology is robust across um all countries. So this is a basic conceptual framework of what we construed our profit safety. So assuming we have two multinational enterprises that have one affiliates and say South Africa which has a corporate income tax rate of around 27%. And let's say has another affiliate in Arilan which has a um corporate income tax rate of 15. So basically the affiliates in Ireland um would would relatively pay lower corporate income tax rate as as compared to the affiliates in South Africa. So in that case as a group then the group has the incentives to sift cost towards South Africa and their revenue towards Ireland in the sense that the cost that is being sifted to South Africa would reduce the amount of corporate income that would be will reduce the amount of profit that will be reported by the athletes in South Africa and then subsequently reducing the amount of corporate income tax that will be paid. um and then most of the profit that is shifted to our land will now be reported there and then will be taxed at the tax rate which is 15% which is um relatively less than what is being taxed in South Africa. So if that is that happen then we expect the affiliates in South Africa to report relatively lower profitability as compared to it industri PS because it's it's actually moving cost it's taking a lot of cost from um it athletes in other countries and then also moving profit to the jurisdictions that have lower corporate income tax rates and this is done through the um the channels of profit system that I mentioned earlier on which is basically um the in intra group transactions. So we expect that at least if you're engaged in profit sifting you should report lower profitability. You should import a lot from tax savings especially if you uh your affiliate in tax saving and then you should also have a higher intragroup debt as well as higher management or royal management payment. And this basically would leave trails in the data that these firms report to the tax authorities so that we can use that to be able to construct the framework that flags these firms as um behaving as as as if they're making profit. They're sifting profit out of these countries. So basically our concept our conceptual framework points to the fact that if if you're a multinational enterprises and you engage in profit sifting then we expect you to report lower profitability and that should also accompaniate by other channels of profit sifting. So how do we establish this um that the company is is behaving normal and otherwise it's not shifting profit against companies that are sifting profit. So remember I mentioned that we're looking at profitability and the other condition of profit shifting relative to the firm's industry. So in other words, if you continue to report u relatively lower profit as compared to your industry average, then would assume that you're behaving abnormally. So if you look at this hypothetical um profit distribution, so what we're looking at is firms that continue to report profits within here. So if you're multinational enterprise that operate in South Africa and for all the years you continue to report profits that are um two standard devas below your industry average then we would consider that to be abnormal behavior because you cannot continue to report um profit below your industry average for all the years you operate. Especially if this is accompanied by relatively higher imports from tax savings as well as the other condies of profit sifting then we would consider that to be an abnormal behavior which probably warrant attention. So in other words that uh you'll be get you'll be flat by our methodology which then allows the tax authorities to be able to follow up with the actual audit to understand what exactly is um contributing to a lower profitability with regards to your industry peers. So in terms of the methodology we use two um two measures or two proxies for firm profitability. First we use return on assets as well as uh return on wages and then this is how we benchmark our we benchmark our profitability. So first what we did is to to regress the return on asset on our firm industry as well as the age of the firm and then we have a dummy which indicate whether the firm is an is a multinational enterprises or is a local firm so that we can be able to isolate the residuals over the period in which the firm appears in the data. So if your residuals is let's say two two standard devas below your industry average then in that case we consider that you're reporting lower profitability and we apply the same to the other countries of profit sifting as well and we know that economic activities are not um smooth over over the period. So maybe one time profit lower profitability is not enough to be flagged as a firm that is potentially engaged in profit sifting. So because of that we we look at it over the period. So if a firm has just one year lower profitability that alone is not considered as as a red flag but rather something that may be a genuine economic conditions that results in the lower profitability within that period. So we that's why we we ensure which I'll get into uh in the later how we ensure that firms that have one year lower profitability are not flagged. So these are our red flags. So first for you to be flagged you must be a multinational enterprises and that is so because profit sifting by it definition is is a crossber activity. So local firms that will not have that opportunity because most of their economic activities are restricted within the country. So even if they engage in profit sifting that money is lost to a different firm which doesn't help them. So but multinational enterprises that have affiliates in other countries are able to do that. So first for you to be flagged one you must be a multinational enterprise and two you should have abnormally lower profitability relative to your industry PS and then you should also have higher imports from tax havens. The same applies to intra firm depth as well as um the intellectual property management fees as well. So first, so you must just trigger the first um red flag which is abnormal lower profitability and then if you have any of the other two red flags then we would consider you as potentially engaged in profit um system. Otherwise uh in other ways our red flags um you would trigger a red flags which means that uh the tax authorities would have to look at that and so we use three different options. So first we look at our standard deviation which is the two um 2.5 standard deviation in terms of the profitability. If you report profit which is 2.5 below the industry average then we use that as as our main outcome and then we also look at um the threshold. So if you are a firm that continuously report 10% below your industry um profit average then we also use that as an outcome variable and also because we have a lot of data limitations so we use a dummy if you have ever engaged in trade weight tax savings then we use that also as an alternative outcome for the regression. So in terms of the data, so first we use the corporate income tax data which allows us to extract our profitability measures which is basically the return on asset as well as the return on wages and then the intrafirm transactions as well as interest payment and royalt royalties and management fees. And then we also use custom data to um extract the trade flows like transactions crossber transactions we're able to determine whether a firm is trading with the tax um havens and then we also use the ownership class as if if you have an affleate which is outside South Africa then we considered you as a multinational enterprises. So firms that have ownership or that have affiliates outside South Africa are considered as multinational enterprises. So in terms of our sample so we implemented a couple of risk fixings. So first for you to be included in the data you must um report all the required variables um more than two years and that's what I was explaining earlier on that because economic activities um changes over time maybe one year a firm might be had a bad year so because of that it can report relatively lower profit. So for you to be included in the sample you must have at least two years of all the variable data so that we don't just pick one year per year as as as under economic performance and then we also restrict it to firms that has turnover um firms that have turned over below 2.5 million South African run also excluded uh in in in the data as well as um assets [clears throat] that are also equal. So if you have asset that is below 500 K as well that is also excluded from the sample. So at the end of the day we have about 26,500 and 65 M and here observation as well as um 156,000 domestic firms. And what we also realized was that for most of the firms we see that the distribution was not so smooth. So some of the firms have this um really huge numbers because we're comparing domestic firms with mult. So what we did is to winize those um extremes outliers in such amount and so that we can have at least reasonable comparison so that we don't um our estimates are not drive are not driven by those um outliers. And this is a basic summary statistics. We're looking at the various red flags or indicators and we are comparing domestic firms with multinational enterprises. So generally what we did see is that um the two groups are not significantly different when you look at the varial um the various indicators that we're using. If you look at it in terms of the mean. So we're seeing that basically the domestic and the local multinational enterprises are generally the same in terms of profitability heaven exports as well as the other indicators but I think what is more important is when you look at the dispersion that's the standard deviation then we realize that there are a little bit of differences between the multinational enterprises and the local firms. So most of the multinational enterprises appears to have um higher heaven imports as well as interest payments and that clearly means that maybe even though we have a very smooth comparison between the two groups but we realize that there are few multile enterprises that are actually engaged in higher heaven imports as well as um interest payments as well as management fees and that itself raise a preliminary um issue. with regards to profit sifting as to why maybe there a group of few firms should um actually engage in these high level activities with regards to these indicators and then when we look at it even graphically so you look at the profits we're at this graph we are comparing profit um using a return on asset for both domestic and mult enterprises so we're seeing that so if you look at that graph the gray is for domestic firms as well as the red is for multinational enterprises. Now what we did see is that generally we see a very uniform distribution with with regards to uh profitability of these two groups. But if you look at it, we see that most of the local firms as well as the multinational enterprises are all classed around the zero. If you look at it, it appears more multinational enterprises are reporting lower profit as compared to the domestic firms. And that should not be the case if um we if we're holding everything constant because we know that with multinational enterprises, they're usually big and make a lot of profit as compared to local firms. But in this case we are seeing that most of it they clustered to run the zero which um earlier studies has contributed to uh they moving profit from one from this high countries with higher corporate income taxes to those um countries that have relatively lower effective corporate income tax rates. So this is um our main result. So in here we're looking at the number of firms that trigger each of the six flags that I have indicated. And if you look at the button so basically here we are trying the various method that we use combining that with our primary methods and what we did see is that the numbers are consistently not different um for the various indicators that we looked at. So if you look for this zero these are the domestic firms so they cannot trigger flax because they don't have that and if you look at lower profitability for the first one we're looking at just five fms then the numbers keep increasing as we change the methodology and then up to the the safe flag. So if you look at this table, so the take away from this table is that yes, for most firms they do appear to engage in look like they're engaging profit sifting, but this is something that is relatively um concentrated in the small group of of fans because if you look at the firms that go beyond three that are um triggering the third flag, the really the number keep dropping as well as when you move to the C FE and then they say flat. This basically tells us that even if indeed profit shifting is ongoing, this is something that is um very concentrated in the very few firms. In other words, if auditing [snorts] were going to be conducted then most of the attention should be around for this for firms that are triggering more maybe probably firms that are triggering flags that are beyond three. So maybe the four, five, six or even the fifth. In other words, it becomes easy for the tax authority to just isolate these few firms that uh engage in profit sifting. And the numbers are almost the same if you try the alternative um measures that the alternative um cut points that we use as as as outcome variables for our trend on for profitability as well as the other indicators um of profit sifting. So now the natural question is so what happened to those numbers? So here what we try to do is to look at so for all the firms that are flagged that are consistently flagged if auditing were to be conducted today and indeed they engage in profit sifting how much revenue would the will sain from that and what we did here is to basically say is that for every firm the average the it profitability should be it industry average in other words we're assuming that you should have a profit levels, which is equivalent to your industry average. And we basically multiply that by the current 25% corporate income tax rate to drive our various revenue potential revenue gains. Um and if you look at the distribution from the decal groups from 1 to 10, we see that for the 458 firms that are consistently flagged, the revenue distribution is concentrated on just these 44 firms that are contributing more than 63% of um the total revenue gains. In other words, if you look at all the firms that are flagged, we're looking at just 44 firms that um if auditioning were to conducted to be conducted today and indeed they engage in potent uh they actually engage in profit sifting, then we expect that this um 44 firms would give us around 63% of the potential revenue losses that um is ongoing in the country. And this is very important for us with regards to this methodology because this is what we we're seeing that profit sifting is not a rampant thing among multinational enterprises but is concentrated on few firms and if those few firms are properly flagged and then audit attention is is is focused on those few that has the highest revenue um gains then that actually gives the taxes authorities the space to prioritize that auditing process so that um the relatively few resources that is available can be concentrated to those can be focused on those firms that have this highest um revenue st. So in other words, these 44 firms probably would be the primary um firms that should be audited if um this was supposed to be audited at the moment given the amount of revenue um gains that is expected from that and then here we look at the sectors. So we look at so for these firms which sectors do they belong to and this is also important because as tax authorities you should also know which sectors in your economy are most likely to engage in profit sifting so that even your routine auditing process that can also help so they know that okay if it is the manufacturing sector then you know that firms in the manufacturing sector are most likely to engage in profit safety. So that comes to the back of your head when you're conducting your routine auditing. So in the case of South Africa, what we're seeing is that for most of the profit system that is ongoing is actually concentrated in the wholesale and retail sector as well as the service sector. These are the sectors that are contributing most to um the profit shifting that is ongoing. And for us we think that this basically has to do with the fact that if you look at the the wholesale and retail sector this this is mostly crossber transaction because most of the things are either imported and and then uh resale. So because of their greater participation the global valid chains then they do this um it does a lot of intragroup transactions and all that and that um probably allows them to maybe um engage in profit sifting and that's why we're seeing these huge numbers for um especially the wholesale and the retail sector because of especially the crosswater each of their activities. So these are few limitations to our methodology. So first we realize that we do not have uh in the case of South Africa we do not have the full information with regards to the me and ownership. So we do expect that we should have um know where each of the memes and that operate in South Africa, where they are affiliates are across the globe so that we can be able to determine um whether these emmes uh have affiliates that are located in tax havens, how much transactions have been done with this uh ammon. So we don't have this clear information. So that is one of the caveat of our of our methodology and this also has to do with the fact that uh probably maybe going into the future most of these datas can also be made available where um multinational enterprises are encouraged to provide all information with regards to where their affiliates are so that we can be able to construct that to determine where their affiliates are and then trace those transactions and then that also applies to the bilateral um country trade data. because we for most of the countries we do not have that uh for we we were not able to observe that uh with regards to M&Es by country so we just know that okay it's an international transaction but which country does that go that is sort of a limitations that we're not able to take into consideration or we or the current studies um does not include and then we also realize that another important information that is missing is especially economic zones which um we also call them the mini tax havens within countries because they have special tax regimes. They pay lower corporate income tax rate. So most firms that have branches in these special economic zones may also sift much of their profit into those locations. However, the currency is we do not have that u information. So we're not able to include that in our red flag methodology. And then so for our red flags, this is not a proof that these firms are actually engaged in profit sifting, but gives us the indication that they're doing something that is potentially wrong, which um would require a full audit. So this is a full audit that can determine whether these firms are actually engaging profit sifting or otherwise. And that also applies to our revenue estimates because these are indicative because we're using the industry average. So which I mean in in in practice may not hold because we know that even though the industry average some firms may earn more below. So our estimates are also just indicative that if these um firms were to be having profit levels which are the same as their industry average then we expect those um expected reven as well. And then I think one of the things we struggle a lot was also royalties and management phase because we could not um determine whether these were crossber transactions or these were just within um a group within South Africa. So probably this is also a limitations that is worth highlighting. So how can we make this better? So I think at the moment what we're doing is using the corporate income tax returns financial statement as well as um the custom datas to be able to establish the profitability as well as the various channels of profit sifting as well as the multense status and then their tax saving affiliates. But if we were to have access to like a full country by country reporting and then a very granual related party transaction that would allow us to even do more with this methodology so that we can even go further to see whether these frames that we flagged how much profit is reported in South Africa and how much profit is is being reported in the other countries and we can actually calculate the misalignment using um either employment numbers to to determine that okay so this is how much profit they reported in South Africa but this is what they supposed to report based on the level of economic activities so we think that if we have access to this data our methodology would actually be more robust because we can do more activities that are more uh with regards to this exercise so to conclude so in this project what we're doing is to develop a very practical, scalable, lowc cost methodology that allow tax authorities to use their existing data um to be able to flag firms that are behaving like those engage in profit sifting. Now in the case of South Africa we are seeing that is profit sifting as documented in the literature is not something that is rampant but rather something that is um concentrated among few firms and that also applies with the revenue um at risks. So we're seeing that nearly 63% of the revenue that is at risk are just 44 firms are contributing to that high number and this is something that is directly actionable. So because we're using the data, so once the data you plug it on your data, it's able to flag this from FMS for you to be able to now determine which of them you should audit based on the audit selections. And that can also even help when it comes to maybe agreement with third parties or third party countries because it allows you to know um which countries have the firms that are behaving like profit shifters. So that when you're signing those agreement then you can tell that okay because multinational enterprises from your country have the potent likelihood to engage in profit sifting you would um tighten whatever agreement you're signing. So for the next test we're looking at um trying to incorporate the ownership data right with the M& so that we can have like a full universe. So we can do all day pretty good things that we can do to be able to like calculate the misalignment against economic activities as well as reported um profits and to also look at mult enterprises that have affiliations in special economic zones and then uh very general country level trade deal. um disagregation would is what we're looking forward to including in the methodology and this is something that is also um replicable across other countries and we're actually having a a bigger project that have currently I think we doing for about 10 countries which is part of uh our initiatives called the administrative data for tax justice where we are implementing the methodology and I think currently about eight countries um as well. Yeah. So that is it. So I think uh I don't know what I'm on time but then I'll pause here and then take questions. >> Thank you Matthew. You are on time. Just before we go into the Q&A session, I noted that a few people joined us a bit later because of the confusion between uh Teams and Zoom. So, I just want to assure you that you will be getting a recording to this session and it'll be accompanied by Matthew's slides. We'll now go into the Q&A session. I would ask that if you have a question for Matthew, you raise your virtual hand or you're welcome to type your question in the chat. Matthew, I'll just scan to see if there any questions. Okay, that is an appreciation to you. We have a question from Kurbus who asks, "What would be the most practical way for SARS to test this approach? Should it begin with the 44 highest ranked firms or audit a broader sample that includes both flagged and unfl flagged firms so that the models accuracy can be assessed? >> All right. Thank you. So for for now so what we we're doing is so if indeed they were going to audit I think for me the advice would be to focus on those 44 firms because that is where you you expect most of the the revenue gains. So the first auditing uh starting point should be those 44 firms but in terms of the practical implementation so what we are now going to do I forgot to add that in the next step. So we're going to run the same methodology on the audited data so that we can look at um how many firms are also flagged based on the audited data and then don't know maybe if it is legally allowable then we can we can try to compare that with firms that SAS has probably audited then so that we compare which one which of them is uh more robust in um identifying firms that are risks in profit sifting. So I think for the next stages that is what we're doing and for any implementation like practical implementation yeah that could be the best thing to do is to um run it on the audit data and then you compare the two and then probably look at what SAS has also audited to see whether there's any um agreement with firms that they have cons internally considered high rates as well as what we have also our methodology has also flagged Thanks. Okay, we have a message from Anuk. I I hope I'm pronouncing your name correctly. Firstly, he'd like to encourage you to consider the tourism industry as one of the key sectors for the future analysis given its significant potential to drive economic growth and employment creation um as well as foreign exchange earnings. And then he notes, I believe the presentation would be even more impactful if it focused on a specific sector and offered clear practical and evidence based policy recommendations. Would you have any comments on that Matthew? >> Yeah, thanks. Yeah, so um so it's it's still a work in progress and for us this is not like the usual academic exercise we do. So this is more like uh a policy relevant research we're taking. So for most of the countries that are engaged that are part of these studies, we're engaging the tax authorities in every stage of the methodology. So and as we engage them we look at what is feasible what can be incorporated and what cannot be incorporated and it basically has to do with the data because uh some countries has good relatively um rich data for if you take country like Norway which is part of the study they have almost all the relevant data we need. So in that case we can look at the industry specific and the sector specific. For South Africa we can also do that but um some of the results we cannot present them like in in in the open for other countries because of uh confidential issues because I think in Kenya we have results that are like pointing to two or three fans that people can actually detect but so we would we don't want to do that. So but if it is within the in country presentation we can show details uh to that but for the sector specific analysis that's something uh we'll consider going into the future and to look at that especially somebody says the tourism sector take a look at that in the in the case of South Africa. Thanks Matthew. I hope you are answered. And then there's one on so there's one that asks how sensitive are the 5.29 billion estimates and the firm rankings to alternative profitability benchmarks such as mash domestic firms or me specific benchmarks. Uh so in terms of the sensitivity so at this moment what we basically the two measures we use are the return on asset and return on profitability and if you look at the two in terms of return on wages and profitability the the firms remains the same. If if you look at with the firms that are contributing to a larger share of Raven losses, the same number of firms for both the return on wages and the return on profitability. And I think at the moment we are revising the methodology for some countries that we have data challenges but not South Africa because South Africa is probably one of the countries in the global south that we have this rich data but for the two for South Africa the current results are robust if we whether we're using the return on asset or return on wages is the same number of firms that are contributing to the greater share of of the potential or profit losses. >> Thanks, Matthew. I'm trying to scan to see if we have any questions. Are there any questions from the floor? >> Okay, we have two. Are you happy to take two, Matthew? >> Yeah. Yeah, sure. Yeah. Yeah, sure. I think we still have time. >> Okay. Quas asks, "How often would the model need to be updated as tax rules change, particularly following the global minimum tax and as firms adjust their behavior in response to stronger enforcement?" And then Bluntler asks um how does the methodology distinguish between genuine operational loss so high operating costs versus artificial profit extraction? >> Okay. So thanks. So I'll take the first question about uh the global minimum tax and whether how our methodology would have been. So so for the set of our methodology. So if a firm really wants to avoid being flagged then it probably would have to begin to do the good things. In other words would you would have to produce like report higher profit relative to your industry average so that you don't get flat and then you should also have lower um interest payment as compared to your industry PS as well as all the indicators of profit sifting. So basically what he's doing is that if you don't want to be flagged then you probably would have to be doing what is right and otherwise you must be within the accepted rules not to be flagged and for the global minimum tax I think for the developed countries if that comes to effect maybe this might not be something that will be relevant to them but I still believe that even if today all African countries are implementing the global minimum tax we still need this because the the bank match I think is the 750 million cut off point that's too big for most of the firms here and probably most not so many firms would be obliged to fulfill the minimum global uh the global minimum tax because of that cut off point and if you look at that not so many firms in Africa have assets would have profit annual profit up to that point so for the firms that are still below that uh that you will still need this methodology to be able to um determine whether they they're reporting the right profit or not. And then to the last question um >> I think we have two more actually. Oh, are you still responding to Tantla? >> Yeah, I think the the other question the last question I forgotten. Sorry. >> Yes, it is from Tanla. >> Could you please repeat it? He asked, "How does the methodology distinguish between genuine operational loss?" So, high operating costs versus artificial profit extraction. And then he has a follow-up question, but we'll get to that, but maybe you can address that first. >> Yeah. So, for the general economic losses, that's why we required that the firm at least have a minimal two years presence in the data. So that if if it is one time off event then it should not the system should not just um should not just flag you because you reported one time profitability which is lower than your industry average. But if it is reported consistently over the years you appeared on the data then that's where the system is flagged. So if it is generally one year profit lower profit then it definitely won't be flat as as part of uh firms that are potentially engaged in profit sifting. So if it generally you you have a bad year then the the methodology will not flag you. But if consistently you appear to have a bad years every year then that I think itself required attention because u you cannot continue to have bad year all the years. >> Great. Thank you Matthew. And I see our colleagues from SARS are here. Lillian. Hi. Um, Lillian says, "Earlier you raised the issue of GMT. How do you foresee this to play into expectations of model results going forward?" Then Dantler had a follow-up question where he notes that previous essay tide research showed that 10% of multinational firms account for 98% of estimated shifted profits. Does this new administrative administrative detection framework confirm that profit shifting remains concentrated among a few mega me and all right thanks so I'll take the first question first so I think I've already explained the GMT that for developed countries yes this may not be important to them because for most of their most most enterprises have that um stressole which is the 750 million um profit every year but for African countries we most of these will not apply because um most of our firms would report profit which is below the 750 million but the 750 million below 750 million is still a huge number so we should be even if it is 100 million it should we should be interested in that so for develop developing countries this will still be applicable even if we have the global minimum tax in place today because um we do not really have those big firms across all the countries. So we still need to look at the smaller firms what they're doing as whether they're reporting the required profit or not and then the last question talks about the numbers. So what we are seeing is is is the same. So we're basically confirming that what the earlier states did said that profit sifting is concentrated among few firms and that is the same for South Africa. I think we're talking about 1% of the firms multinational enterprises and for other countries is also the same. I think Uganda is is about 1%. Kenya, Kenya is also around 1.7% and where I think Rwanda is also around 2.3%. So basically what we are seeing is that for all the countries we get where we're implementing the methodology the numbers are relatively small. there just few firms that appears to engage in this u practice but it's not something that is rampant among all multinational enterprises thanks Matthew Pla and Lillian I hope you're covered please give us a thumbs up if you are um then we have about 3 minutes remaining Matthew so we'll just take one more question from who asks once audits have been complete completed. How could SARS feed the results back into the model so that the indicators and rankings improve over time? >> Yeah. So if if audit is done and then there's a feedback so we know which firms are actually engaged in the the profit sifting and which are not uh engaged then in that case we we would be able to calibrate the model um based on what the audit fun and in this case so what we're doing is basic we look at all the indicators involved and which of the indicators are these firms um noted to be using. So if it is something that has to do with maybe imports or exports or if it has to do with intellectual property then we can calibrate that. But for now what we do is if a country wants to implement this at the moment we leave we leave it to them to do the auditing. Once they find out which firms are engaged in the are actually engaged in profit sifting and we know the number of firms um then we can calibrate our model to be able to reflect the new findings that the audit did so that if if we don't end up with false positives all the time and and that's what we are currently trying to do for some countries where they would actually conduct audit to see what are the funds we flag are actually uh engaging profit safety once that is done then we move to the next stage of um feeding that back into the model to be able to reflect the real um situation on the ground. So as I mentioned this is a work on progress. So we we're looking at both angles trying to take feedback from the tax authorities as well as uh the research end of things from ourselves. I hope that answer your question. >> Thank you Matthew and Kiss. I hope you are covered. That brings us to the end of this essay tide seminar series session. A huge thank you to everyone for joining us today. I have shared the link to the essay tide website. If you're interested in any of our other research, I would invite you to please take a look at the website. And if you'd like to be made aware of other upcoming essay tide seminars, please click on the register for the newsletter link and you should be added to our database. And then Matthew, thank you once again for joining us. We really appreciate your time. We will be sending a link to the recording as well as Matthew's slides. Thank you. >> All right. Thanks you and thanks everyone for