 I will be presenting this paper to the school the mitigating role of tax and benefit packages for poverty inequality in African countries amid the COVID-19 pandemic. So this is a joint work with a lot of people including Javier and I'm not sure if there is anyone else here yeah including Javier and a lot of colleagues from Unwider colleagues from Essex research assistant etc. So briefly about the self-mod project so this is a ongoing project at Unwider. The idea of the project is basically to develop tax and benefit micro simulation models in developing countries using the aeromod software and use models for applied policy research. So I'm not sure if you know the aeromod models so basically the aeromod models exist at least for 20 years for European countries and they use to do tax and benefit micro simulations so they can for example simulate just one very simple example if you take a child benefit in Finland what would be the impact of this child benefit in Portugal for example so you can simulate the characteristics of that policy in Portugal. So basically what we do in the project is develop these models for developing countries so mainly in Africa so we have in the project around nine countries we are developing the model for more two countries at the moment for Rwanda for Zanzibar and we also have the Latin American models our colleagues here represent the papers using the Latin American models and there is a huge initiative that Havier and other people are leading to create these models for more countries so okay so this is an active collaboration with national teams so this is a very important part of the project so for every country we have at least two citizens that are based in that country working with the models and those people are very important for the project because they are working in the university or in the government and they bring to us the daily life expertise so if we need for example information about some policy that exists or some new policy office if there is new data set that we can update the model so they are part of the project they help us and do this kind of updates in the model maintaining the model and also delivering training so every year we deliver at least one training per country in the models okay using the models so the models are freely available so you can go to the Unwider web page you can download the models and you can use you can take the data sets and do the simulations that you want and another important part is in this capacity building initiative is that we are developing our online training tool so people from all over the world will be able to have a training to start using the model because we cannot train everyone that want to use the model we have limited resources of course so what is this COVID-19 study so it was a large research project but the distribution effect of COVID-19 and also the role of the existing tax and benefit system and the new description policies that were created amid the pandemic so in this in this project specifically we work with Zambia Mozambique Uganda Tanzania Ghana and recently we had Ethiopia but they will not show any results for Ethiopia in this presentation and we also have the models for all the countries and there are other papers that I think now we are working papers only right but developed by our collaborators for example Javier working with his same Ecuador paper for Vietnam a paper for South Africa and the main objectives of this study is to estimate the effects of the pandemic on poverty inequality but this is that is not the main goal because estimated impacts of the pandemic on poverty inequality you can do in different ways even using more advanced methods that we use here so what we I think what is the main contribution of this using the micro simulation is the possibility of assessing the impact of the existing tax and benefit system for example if you have the the the right to unemployment and unemployment benefits or some additional cash transfer that were created during the the pandemic in some specific country to see how those policies mitigate or not the adverse impacts of the pandemic so in this in this project specifically was a very joint effort with the national teams because the information about the policies created in African countries were not easy to assess so for example in the case of Mozambique specifically they had an expansion of the cash transfer and this was announced in around June July of 2020 but if you we simulated this policy for example for Mozambique based on this document we would be doing something completely wrong because then because of the national team in Mozambique we received extra information that this policy had problems to start because basically it was international organization money that it needed to be transferred to the Ministry of Finance and then they have all the bureaucracy to transfer this money to the people the policy only started in 2002 so this is one example of how important is the having those national teams as a partner in this project so this this project has some steps and we are writing at the moment literature review about the role of tax and benefit the role of taxation and social protection during crisis in developing countries and one of the things that these literature review shows is that one of the there is various cars evidence that are only a few number of papers that deal with this issue and the main problem that we cannot observe is the lack of up-to-date data so for example if you go to African countries the situation is completely different from Brazil for example in Brazil we are doing a national survey conducted by the government every month to assess the impact of the pandemic on incomes labor or etc and we don't have this in the in African countries the World Bank didn't I fought to do that with the foreign service but for a lot of the countries that we have in the data set the World Bank if data have two limitations for us the first one is that they don't ask for incomes and the second one is that most of the data are not available so for example they did the survey in Mozambique they did in Tanzania but the Minister of Finance in those countries decided to not make the data available so in this study specifically we use the World Bank from service for one country for Uganda as a kind of sensibility analysis for informality try to say a bit about that let me see my time so first thing is we don't have date a data set so we need to develop these crisis data set so developing the crisis data set is basically assuming that there was some shock in the country and we do this in a very simple way so we take the GDP trained in the country and we observe that we observe the GDP country the GDP trained in the country and we see we can see that in 2020 there is a drop in the GDP okay so the analysis pretty simple is just to assume that the GDP in 2020 would be the GDP that would follow that trend and then we take this drop that is because of the pandemic in 2020 and we extend this drop to the micro data so we assume for example if we observe in the in the country data that 10% of the people in the agricultural sector lost their jobs we go to the micro data and we random select 10% of the people in the agricultural sector and we move them to unemployment okay so this is unfortunately what the maximum what we can do we can test different ways to to simulate those shocks but it's a limitation because of the lack of available data okay and we also do we also use the World Bank Foreign Service because we have micro data in the World Bank Foreign Service to do a different way to assess to assess who lost income because of the pandemic so basically we estimate the probability of the person to lose income in the World Bank Foreign Service and we recover this coefficient and we apply to the model micro data so instead of it but instead of random selecting people to lose their job and to lose the income we calculate we basically select people with more or less probability of losing their income based on the World Bank Foreign Service for Uganda okay so we develop this crisis data then we model the tax and benefit policies during the coronavirus this study is only for 2020 okay and and we also do a lot of the compositions techniques so first the crisis data set as I mentioned let me show sorry I thought the graph was here so we do this random allocation method that I just explained it and then we do what we call the importation method using the World Bank Foreign Service to Uganda to show that doesn't matter which which method you use the result is quite similar the biggest difference is that in Uganda we can see a higher impact for informal workers because basically we have information if the worker is formal or informal in the World Bank Foreign Service okay so then we model the tax and benefit policies and what we can observe is that in Africa tax and benefit measures and the household level in response to COVID-19 have been very limited and the biggest explanation to that is that they don't have budget the government don't have money to do policies just as an example in Mozambique they have a cash transfer program that's called PSSB and this cash transfer program Mozambique transfers basically eight dollars per month for each family benefited for the program and this money is the discussion transfer is paid for by international organizations and some countries for I think Sweden and the UK and they only cover one third of the country so one third of the eligible households for the cash transfer in Mozambique received the cash transfer okay so basically they don't have money to do social protection policies okay so in Mozambique they basically reduce utility fees like water electricity in Zambia they create an additional cash transfer program in Ghana they pause a large social feeding program during the lockdown so basically in Ghana they did the kind of a negative social protection policy so instead of providing more support they took one of the main supports they had in the country and in Uganda and in Tanzania we didn't have any relevant policy in 2020 so we don't model any policy for Uganda in Tanzania okay so as you see as you can see here in Mozambique in this example specifically we are not modeling the additional cash transfer because it started basically in 2021 so and then okay we thought we we apply the method and we are comparing headline results basically impacts of the pandemic on the mean disposable income income basic poverty inequality and then we do the some of the compositions basically showing how much the shock was mitigated by the automatic stabilizers so that is basically if you are in the formal sector and you lost a job you have unemployment insurance this is a one kind of automatic stabilizers and new policies adopted by the the COVID-19 because of the COVID-19 that are those policies here okay okay so basically what we do this is a very is a image that explained basically so we have a data set the data sets are basically how service conducted by the government so we have micro data for the data that are produced from the government of those countries and we we will apply this way of shock the the incomes of some individuals in those micro data and we will create this new data set that we call the counterfactual data set and then we have another data set that is the the the shock data set after simulating the policies okay so basically the idea is you need to have a micro data you need to have the policy systems and you create different scenarios basically so jumping to the results first we show that that is decrease in the main disposable income for all the countries in this study then we apply the the composition techniques to see if the drop in this main disposable income was mitigated by the new discretionary policies created during corona virus if it's explained by the automatic stabilizers or it's explained by the just the drop incomes so what what we see is basically that the automatic stabilizers do not play much role here because it's not statistically significant and that is a very small effect of the new policies implemented in mitigating the effect of the pandemic when we look to poverty and inequality the results are quite similar so when we look to the poverty rates the poverty rates increases and when we look to the composition exercise we show that the covid policies had very limited effect to mitigate the pandemic effects and the automatic stabilizers in fact they had affected in a different direction in increase the inequality and mainly because the automatic stabilizers affect many people in the in with higher incomes so if you if you think about african countries basically people that are in the formal labor market are people that are only a small percentage of individuals and only those individuals are affected by automatic stabilizers so if you have a shock that affect everyone only the top earners will be benefited by some kind of automatic policies so the poor in those countries will be the most affected by any kind of shock because they are not protected by anything so this is the the intuition behind this positive effect here then the effect on poverty gap quite similar and the effect on gene coefficient also quite similar okay so I mean increase in the inequality and automatic stabilizers also not helping too much and also very limited effect of the covid policies so this is another exercise that we do we basically decompose the effect of the pandemic on mean household disposable income across quartiles and we can split what is variation in earnings variation in covid related policies variation in automatic stabilizers and what is the disposable income and as I mentioned before you can see that only the top earners they have the their income are mitigated by the automatic stabilizers you cannot observe here the effect of automatic stabilizers while you can observe some effect on covid related policies it's the existing tax and benefit policy so can be fiscal policy can be also as I mentioned unemployment benefits for example so yeah okay so this is this for Zambia so this is the effect of the energy emergency emergency social cash transfer when we go to and as I mentioned very limited automatic stabilizer effects when we go to Uganda that didn't have any policy so we can observe also that there's so much automatic stabilizers only working for the top earners and and I mentioned as I mentioned we we we had access to the word bank from survey data only for Uganda so we could compare the two methods of allocating people to unemployment because this is the main the main way that we act we simulate the impact of the pandemic is moving people to the unemployment so this is the random what we call the random allocates allocation method that we observe like 10% of the people in agriculture lost the income so we go to the micro data we random select 10% we take the income then we we also did the imputation method and the notation method is the one that I mentioned that's different so we use a regression to estimate the probability of people losing income based on the word bank from survey and we take the coefficient estimated but based on a lot of observed characteristics and we input this in the input data set of the model and the biggest difference here you see that the the shape of the the figure is quite similar but the main difference is that when we use this imputation method because in the word bank from survey we have information who is formal and who is informal we can basically we observe a higher effect of the the the on informal than on formal employees thank you