 So good morning, ladies and gentlemen, as I've been introduced, my name is Ezekiel Sweber, and I'm from Tanzania Revenue Authority, and today I will make a presentation on the paper that we did, of course, in collaboration between Tanzania Revenue Authority, UNWIDER, and the local university in Tanzania, the University of Dar es Salaam, and the title for this paper is the Effect of Risk-Based Approach to Tax Examination, and we are drawing evidences from Tanzania Revenue Authority, and we will come across the following contents. We will see part of the introduction and directed, see the empirical strategy and the data that was used, and a bit of the graphical inspections, and then we will see the implication for the study. So the main focus and functions of many tax administrations has been based on the following areas. One is to increase the tax, expanding the tax base, enhancing the tax compliance, minimizing tax evasion using various, of course, enforcement strategies, and increase tax revenue at the major goal and expectation for many tax administrations. We can try to check on the sustainable development goals, and here we are drawing a reference from the meeting that was held in Addis Ababa in 2015 that was addressing the taxi initiatives in Africa, and they say in order to meet the goals for the 17 sustainable development goals, we really needed to support our tax systems, and we really, really see the commitment from developing partners to supporting developing countries into these areas. And of course, various enforcement strategies, including random audits and examination used by these tax admin authorities. However, it has been confirmed that risk-based approaches and machine learning in these practices are limited, and there are methods which are tested and promising, improving detection and probabilities for non-compliance. So literature has been confirmed that, but limited and existing systematic evidence on the impact of these risk-based strategies is observed in literature, and for our understanding, no studies has been conducted before to assess how risk-based compliances, interventions are working in developing countries, and this paper now draws and going to text the examination revenue impact of this risk-based tax examination, which was piloted in Tanzania Revenue Authority. And the pilot, of course, was blind jointly by TRA, and a Finnish tax administration, and we received funds from the government of Finland. And the intervention specifically intended to flagging taxpayers from the examination on the basis of data-driven risk assessment, and it was designed in two ways. One to improve the existing practices of choosing taxpayers to be examined in which the practice which was aligned on the staff description. This intervention was limited, of course, to the personal income tax or corporate income tax session, and the following five objectives was intended during the planning of this intervention. One is to increase the revenue collection, but the two is to improve the skills of these tax officers who are into these procedures or conducting this examination process. And of course, the third objective, it is to focusing on the efforts to the risky taxpayers instead of just selecting taxpayers and then ending up of not receiving the intended objective. And the objective number four, of course, is to treat equally and all taxpayers by upholding the principles of fairness and justice in taxation. And the last objective which was planned for this intervention, of course, was to reduce the time spent by examining taxpayers with limited risks. And directly we go to the approach or the empirical strategy that was used during testing our examination. So we use the methodology, the different methodology, and the outcome variable with the taxable income, and we provided the data salam, the treatment variables, and we given the value of one if the tax offices is locating data salam where the intervention was taking place and the data salam comprises of about five tax legions and the rest, which is zero, we provided it to the other tax legions which are outside data salam. And the post was the variable receiving the value of one if that outcome, it is from the intervention year which was from July 2019, which is the year in which we tested up to June 2020. So we taken this as the financial years and we did for zeroes if that information is of the past years where the intervention was not there. So we tested or draw the same level panel data, which of course we extracted from Tanzania revenue authority system, and each year contained about 25,000 observations, which we believed it was enough information to test our intervention, and it covered the period of four years from 2015 up to 2020 to the last year of the drawing our information, and of course it was the year where the intervention was taking place. And we are checking on the data, and table number one on your left-hand side, it entails a number of firms which were adjusted taxable income, which was observed to include from one financial year in both the treatment and the control group. If you can see starting from 2016 for the control group, and if you can see starting from the same year for the treatment group, and you compare to the 2019-20, all these taxable income were increasing based on the information that we have drawn. And table number two summarizes the number of adjustment that have been taking place during the pilot period for the treatment group. And table number two there, we had drawn some observation where we observed one of the tax region in the piloted area, which is Ilaala, has relative more adjusted income depicting like 70% if you compare to the reaction from other tax regions in which this intervention was piloted. And we are asking ourselves maybe why this observation now is coming, and we heard some results in hand, and we say probably there are relative low knowledge of recording and reporting financial transactions for a taxpayer, or perhaps there is a high responses of taxpayer to adhere with the TRA regulations. But again the question was why this now is happening during this time and not at any other time, so we said maybe it was the impact of the piloted intervention. But again, still we need to test if it was really observed or not observed. But again we checked critically on the information by describing and we observed some of the characteristics of the firms in the controlled and uncontrolled firms before and after the piloted interventions are compared. And the firm in the treatment areas were observed to be large and more profitable. But again the largest number of firms that we observed during this analysis were from the service sector, wholesale and retail. And again we went further now to test the assumptions which now will come up and see if our assumptions and the methodology was critically in place. So we applied the different methodologies as I said before. And of course the crucial assumption in this different diff is to test the parallel trend assumption before if it is holding. And in the absence of the treatment, the average difference in the outcome between the treatment and control group observed to be remaining similar. As you see on your left hand side that table, this assumption was holding properly well. And after the intervention during the 2019, it was observed that there is a change. So we want now to measure the impact during that change that was observed. And quickly now we go to the results which were observed during our analysis. And our result we confirmed that it shows a 10 to 15 increase in the adjustable taxable income where the risk based tax examination were implemented. So this was tested by our methodology and we confirmed that there is an increase in the adjusted taxable income. And this would suggest the use of the intervention to flagging taxpayer could improve the efficiency of tax examination and lead to an increase in tax revenue. And of course the impact was found in the corporate income taxes as I said before. It was limited to that area and they arise predominantly from the services sector. And again we went further to show the pilot growth in tax revenue in the piloted area. Main streams from the increase in the amount of the tax paid. And we observed that the number of firm filing income adjustments remained constant and not affected the tax revenue which was collected. And this highlights the fact that focusing on improving the efficiency of the tax examination is more beneficial for the TRA than conducting more tax examination cases. And again we tested the cost of implementing this intervention if it was cost wise. And we tested this and we measured that the cost of the intervention was costing about 60 pounds and estimated revenue which was increased at least during the intervention period was about 18 million pounds during that period of intervention. So we confirmed that our intervention was cost effective because it was at least giving us a substantial revenue collection during that period. And now we conclude by providing the implication for our study. And we have drawn some four implication one. It is possible for the risk-based tax examination to increase tax revenue collection in developing countries. So we cement and build on that. And again there is a tax revenue potential when firms are undergoing tax examination. And again improving the efficiency of tax examination can be more beneficial than conducting additional tax examination. So it's better to rely on these risk-based intervention so that we can see we are examining risky taxpayers who will be beneficial in our areas and not investing more of the efforts to the taxpayers who does not have risk or any risk sensitivity. And again these interventions hold the great potential for being further expanded because the results have been confirmed to yield more results. So we are expecting to expand it to other areas because this it was beneficial and of course to see if it is viable or not viable again by automating the risky-based system through our system so that we can be able to get the potential and a more revenue. Thank you very much for your attention.