 Thank you and good afternoon everybody. My name is Gemma Wright and I'm presenting a paper that has been prepared jointly by three of us from SASBRI, myself and Michael Noble and Helen Barnes, and two colleagues at the University of Essex and the EuroMod team, Catherine, Gassio, and Crisola Venti. It's been a collaborative endeavour to one of the first to put together some of the SouthMod models together and look across the countries at some of the arising findings. As Remy has made the case very well, the taxation and social protection systems are crucial policy instruments for governments to pursue distributional goals of reducing inequality and even going so far as eliminating poverty along the lines of agenda 2030. But informed policy decisions require the ability to be able to assess the distributional impact of public policies and the effects of measures on inequality and poverty rather than relying on who shouts loudest or what was done in the previous year. Evidence is required on these things and also the ability to be able to undertake ex-anti-evaluation of reform ideas. It's also important to be able to estimate the fiscal impact of public policies and potential reforms. Whilst researchers and policy makers in developed countries make heavy use of tax benefit micro-simulation models, these have been used to a much lesser extent thus far in developing countries and people have much less access to such research tools. There's extensive literature on the distributional impact of taxes and benefits internationally, but again very few studies that focus on lower and middle income countries with some important exceptions cited there as examples. Our contribution and intended contribution in this paper is to look not just at consumption but also and more explicitly at different income concepts and to pull together six state-of-the-arts micro-sim models that have been developed or updated as part of the southmod project. We assess the distribution and composition of incomes and the effects of taxes and benefits on poverty and inequality in six African countries for a common time point of June 2015 for five other countries and one day later for Tanzania which has a tax year that starts on the 1st of July. These six models draw on local national household surveys. They've been developed by and as part of a collaboration with UNU wider Southern African Social Policy Research Insights, the URMOD team at the University of Essex and local country teams, many of whom are present in the room. All of the models are underpinned by the URMOD software which has been developed over a 21-year period for countries in the EU and provides a common platform and very well-tested methodological approaches. It's extremely flexible and freely available and provides a shortcut or a head start in the process of building tax benefit models. Our analysis is based on three low-income countries, Ethiopia, Mozambique and Tanzania, two lower middle-income countries, Ghana and Zambia, and one upper middle-income countries, South Africa. The surveys we use for the Ethiopia Socioeconomic Survey 2013-14, the Mozambique MozMOD model uses the Household Budget Survey 2014-15 for two quarters. The Tanzania model TASMOD uses the Household Budget Survey 2011-12. The Ghana model uses the Ghana Living Standards Survey, GLSS 6 for 2012-13. MicroZAMOD uses the Living Conditions Monitoring Survey 2015. SA-MOD uses the Living Conditions Survey 2014-15. Of the six models SA-MOD is their eldest, 11 years old. All the other models are one or two years old. We're still getting to know them and the results that they're yielding. We simulate the cash benefits in each of these six countries, a certain number of the in-kind benefits, social insurance contributions, direct taxes in the form of employment taxes, personal income tax and simplified taxes or turnover taxes. For indirect taxes, we simulate value-added tax and excise duty on a selected number of items relating to alcohol, tobacco and fuel. In the process of pulling the six African countries together to look across them, we've encountered a range of data and simulation challenges in preparing a technical note with our observations and also recommendations for future steps for harmonising the models. It's inevitably an iterative process with such young models, but we've identified a lack of comparative subpopulation variables and consistent category definitions for available variables. We've been able to generate a small subset of harmonised variables, but there's more that can be done on that. Consumption data is not included in SA-MOD underpinnings dataset, and though it is available in the Ethiopian input data, we're unsure about its quality, so it's excluded for those two countries. In the implementation of when we simulate the benefits, we assume full take-up of benefits, whereas in practice, of course, there's not necessarily full take-up in the countries, and there's also restricted roll-outs where there may be many more people who would, in principle, be eligible for the rules, but they're solely rolled out in certain parts of the country. We've taken that into account to a certain extent in Mozambique, but not in the other countries. The countries each have countries' specific up-rating indices. As a general rule, income-related monetary data is up-rated using the consumer price index to common 2015 time point, but in some countries that subsets of income variables were up-rated using components of the CPI or earnings inflation indices. Equivalent scales vary across the countries, and poverty lines vary across the countries, so we had to make decisions as to how we were going to look at these six countries in a common way. There's also a paucity of external statistics for validation, but on the SouthMod website page, there are links to the country reports for each of these six countries where that validation data that we were able to obtain is compared against the simulated data. We considered poverty and inequality with using a range of different applied income concepts. We use original income, which takes into account employment income, self-employment income, including farming and other market incomes. We look at disposable income, which in addition adds simulated benefits in cash and kind of deducts the direct taxes and social insurance contributions, and then post-fiscal income, which additionally deducts the indirect taxes, so the expenditure on fat and excise duty. We also look at consumption. The six countries across the board of very young average age is ranging from Mozambique with the lowest to South Africa with the highest average household size of four to five people, and very young populations with Mozambique having almost half of its population aged 0 to 14. We ran these models having undertaken a fair degree of harmonising, and were able to look at these different income concepts across the six different countries. This first slide of results shows quintile shares and mean and median incomes using the disposable income concept. Again, I should stress this assumes a full take up of the benefits and also full compliance with the tax arrangements. South Africa has the highest mean and median disposable income, and Mozambique has the lowest mean and median income. I can see it's incredibly skewed towards the fifth quintile, all extremely unequal countries in terms of the distribution of disposable income. This next slide provides poverty rates using different income thresholds for the six countries. We use the $1.9 a day that we heard about discussions about this morning and the prospects of being able to reduce that to zero, but also the lower middle income and upper middle income New World Bank poverty thresholds of $3.2 and $5.5 per day. We've harmonised them all by implementing per capita equivalent scales, but down at the bottom the NES, the penultimate row, shows the national equivalent scale and the national poverty lines for the different countries. At the bottom is the World Development Indicators published consumption poverty rates, which are reassuringly similar for the four countries shown. If we break down the income concepts just using the $1.9 per day poverty threshold, we have a striking finding that in all countries apart from South Africa, if you move from original income to disposable income or from original income to post-fiscal income, poverty stays roughly the same or in fact increases. In contrast in South Africa poverty plummets as a result of the tax and benefit system. I think this is a very interesting and important finding for us to reflect on. Looking at inequality, it's slightly more positive for all countries inequality reduces when you look from original income to disposable income. There's a strange exception in Mozambique where inequality increases from original income through to post-fiscal income using the genicoefficent. Thanks to the role of that, it seems. I'm assuming I have very small amount of time remaining, just two concluding slides. We've seen that with the exception of South Africa poverty rates using the $1.9 per capita per day threshold appear to be largely unaffected by the tax and benefit arrangements. In some respects, this is unsurprising. As we know, for example, overall in Africa, the ILO has identified that only 18% of people in Africa are covered by one or more social protection benefit. We know that the benefit amounts are very narrow in their coverage and they're usually very small in amount. But it's a sobering discovery and one that will change because of innovations such as Remy has been talking about, of the progressive rollout and increase of these benefit amounts. In contrast, income inequality is reduced by the tax and benefits arrangements in each country if you compare original income with disposable income. But most interestingly, one of the, I think, very interesting points also is that income inequality is higher than in South Africa in all five comparative countries, whether one uses original income, disposable income or post-fiscal income. We're used to hearing about South Africa as one of the most unequal countries in the world. Here, these preliminary findings are suggesting that if you use these income concepts, then income is extremely unequal in these other countries. The use of the URML software as a common platform has enabled us to apply common concepts and terminology across countries. It is a very powerful resource for us all as researchers to have access to. There is much more to be done to hone the comfortability of the country models and to take into account issues such as compliance levels and to take up and roll out of benefits. There's much more to be done as well to scrutinise the quality of the underpinning data, especially the income data, because we are at the mercy of the quality of the income data when we're looking at these poverty and inequality measures. Overall, South Mods models do provide a very good basis for exploring and potentially improving the tax benefit systems in these six African countries. There's a great deal more to be done. Thank you.