 This is the session which is termed design matters, the potential of social protection systems to deliver during crisis and I would say all three papers were inspired of course also by the most recent crisis but also more generally by the question how well suited are our systems to deliver when people need them urgently. So we have today with us Adnan Shahir and Yooka Pettile who will present the first paper. Adnan is working with Union Wider and has a PhD from the University of Insubria in Italy and then we have Yooka Pettile who is the forefather, the father of South Moscow so to say, who is with the University of Helsinki and with what. Now they're going to be presenting jointly and then we have Miguel Lino Sarasua who will present a paper followed by Annalena Opel. Miguel is with SOAS in London and Annalena is with LSE so we have a lot of people from London here today actually. So with that I'm going to take more of the time and I will hand it right over to Yooka. Good morning everybody. So this is a joint work with Adnan who's here and in fact also with Ravi Kanpur of Cornell University and the chair herself, Pia. And what we do in the paper is that we compare the poverty reduction impact of targeted versus universal transfers during crisis times. So we at Wider we think that I mean because one of the key goals of tax is not only to raise revenues but also to contribute to equitable distribution of income that's why we often analyze taxes and transfers or benefits at the same goal and so this is one example of such a study. Let me start with a little bit of background so of course we know that there's vast analytical literature on social protection policies in developing countries and there are various studies out there which try to investigate the extent of social protection offered by the current systems including by the World Bank and the colleagues at the so-called commitment to equity project CEQ for short. But in fact I mean quite a bit of this works looks at the sort of if you like steady state impact so in normal conditions the poverty reduction impacts of social policies. There's very little formal analysis on how tax and benefit systems fare during crisis times when incomes change. There's informal discussion by the World Bank there's a book out in 2020 about so-called adaptive social protection policies but it's it's pretty informal so what we try to do in this paper is that we analytically examine how different targeting approaches perform when people's income change and we do it using a micro simulation approach and that's why I'm lucky to have Adnan as a co-author because he's an expert and we have a little bit of analytical thinking before we embark on the micro sim part. So as I already alluded to of course the key objective is tax and benefit and transfer systems is to alleviate the long-term or chronic poverty. But also these systems should provide social insurance and in the developed world this would be programs like unemployment insurance or health insurance so people who were not poor but who are not able to work anymore then get some cushioning provided by the state to their to their consumption because they begin to be eligible for these benefits. So this can also be examined via the lens of so-called automatic stabilization and with this we mean the automatic increase in benefits and then a reduction in taxes when people's incomes decline and a colleague at Wideren University of Helsinki, Kvabenardua Babio has a paper on the extent of automatic stabilization for developing countries and he finds that the extent is very, very small in most low and low middle income countries and it's much smaller than in the case of developed economies. So Norway and Finland for example cushion something like 50% of the income loss to people who lose incomes whereas the system in Ghana provides an order of magnitude less of cushioning. And we believe there are three key reasons for this. One is the sheer size of the government which is way smaller in developing economies. The second is that there's a large share of informal workers who don't pay the tax so if they didn't pay the taxing to begin with when their incomes decline their tax burden can't go down. And the third reason is that the many of the benefits that are available are so-called of the proxy means test type and these don't react to changes in incomes at all in the short term. So what are these proxy means test type benefit BMT for short? So these are benefits which are where the eligibility is calculated on the basis of household level indicators which measure the assets and the household composition and a score is created and then if the score falls below a certain threshold value then the household is deemed poor and the household is eligible for a benefit. Now if these conditions change the eligibility can only be tested whenever new information about the household characteristics is gathered and this happens every third year, every fifth year. So in the meantime if somebody loses incomes the benefit system doesn't adapt. So that's the backdrop for this study where we try to analyze targeting or not targeting the benefits when crisis occur. So we ask whether the system should be targeted as it is now in many developing economies including in Africa using these proxy means test transfers or should it be more universal? And of course the other end of the spectrum is a completely universal benefit, universal basic income for example. But of course there are some middle ground and the benefits could be so-called categorical where they go to a certain category of households like households with older members or household with young children. And we think that there could be a trade-off, economists love trade-offs and here's one. So if the targeting works in quote unquote normal circumstances then almost by definition it's the most cost-effective way of reaching the poor. But when shocks occur and if the shocks of the type where the profile of the poor chains then the initial targeting may not be the correct one anymore. And this is what we try to then examine first with a small analytical exercise and then with simulations for the case of Ethiopia. So we have a couple of theoretical points in the paper. So the first one is more or less what I already said that if the benefits are appropriately targeted before the crisis they may or may not be appropriately targeted after the crisis has occurred depending on how incomes change. In particular if the if the crisis like the coronavirus pandemic mostly hit people who were not originally recipients of these transfers then the targeting efficiency is worsened. We also show we think for the first time that there's a link between social protection budget and poverty reduction. The poverty reduction increase is so we get greater poverty increase in universal shock in targeted systems in comparison to untargeted systems. And let me highlight that a little bit more here. So here's a thank you here's a chart where we have R on the horizontal axis shows the resources available for the poverty reduction programs in society and in the vertical axis you can think that as the poverty rate. And we are comparing two systems U and T. T is a targeted and U is an untargeted system. Now of course if resources are zero then there's no difference between the two systems. I mean obviously because if you have zero shilling to spend then there's zero poverty reduction as well. Now the targeted system performs better when then resources start to increase. So that's why then the curve is steeper so the poverty reduction is more for any resources spent. So this is during normal times. But a mirror image of that is that if resources then decline and you are somewhere do I have a pointer here I don't know anyway let's forget it. If resources decline and you are somewhere in the middle and that can be seen I mean that's one way of thinking of that is that society becomes poorer. You can see what happens to poverty. The poverty increase in the targeted system is actually greater than in the universal system. So that would be the trade-off. So the predictions of the theoretical exercise if you wish is that if budget does more relative to the overall poverty gap like they are in lower income African countries targeting matters relatively little. You get very small poverty reduction in any case. Poverty levels could well be higher in more uniform systems. That's because targeting by by definition is more effective. But the increase when shocks happen in poverty may be smaller in more uniform system. And now I hand over to Adnan who who then shows how we try to tackle some of these issues in the case of Ethiopia. Thank you Yuka. So my presentation focused on describing the social benefit system in Ethiopia and to display results on the three types of benefit arrangements. So PSNAP or the Productive Safety Net Program is the major social protection system in Ethiopia. The program launched in 2005 in collaboration with the development partner mainly in the rural area. Then following 2016 the program expanded to cover major urban area including capital at this. So this PSNAP which is a type of permitting but most of the identification is done by communities at local level. The background assumption is local community or in neighborhood better knows who is poor than the government. So this PSNAP program has two wings. So it offers support for conditional and unconditional types. For the conditional benefit the household must have at least a single member who could provide labor service. So we use ET mode the Ethiopian tax benefit micro simulation model which is a family of models developed by Southwood project in Africa. So these models simulate income tax direct taxes and this benefit program. The ET mode is a static model. It doesn't show the impact of policy change at macro level but the output from ET mode can be used to show the effect of the policy in using CG models through integration based on the need of the researcher. Currently ET mode cover up to from 2014 up to 2020 policy system. It uses input file compiled using the social economic survey. So we consider two shocks the COVID shock actual COVID shock which we computed based on the deviation of 2020 GDP from it is counterfactual number which compiled using the change or through the extrapolating the linear growth from 2017 up to 2019 which is COVID period and then the agricultural shock which literally 10% reduction in agricultural income. Then having this shock from macro we imputed in the micro simulation model by transmitting the randomly individuals in paid employment to unemployment with zero income. So we have three system scenarios the existing committee which is a productive safety net program. In this case the beneficiaries remain the same both in baseline in crisis cases. We have categorical benefit we use simple category like the number of kids if the household kids in the household more than four or if the individual is more than 65 gets the benefit and then simple UBI universal basic income distributing equal amount of money for everyone. Then we also simulated a scenario with the expanded budget for the individual receiving in the baseline case just by multiplying the benefit amount by tenfold. Okay so we see here the main results when we see the poverty head count for the three scenario in the first the second row there we don't see any change this is due to the tiny budgets are located for the benefit program in Ethiopia. There is huge poverty on average more than 43% but the budget are located is very smaller than what is needed to pull out all from the poverty line. So in order to I mean eradicate in Ethiopia poverty government has to allocate more than 140 billion per annually but what currently expanding is around three billion this is why we see small changes across the benefit scenario which is support the theoretical framework you can discuss in the lower use based on the lower budget amount given the higher poverty rate in Ethiopia all the benefit arrangements reduces poverty but we see a change when a benefit amount increased by 10% so in this case committee perform well better than the remaining benefit arrangements then we tried to see what would happen if the profile of the poor changes then we implemented five times higher covid shock in this case most of the individual lost a job due to the magnified shock covid shock is where none poor in the baseline case so now we see the paradox because permitting with higher benefit lead to lower poverty line both in normal as well as in crisis time but a system which was ineffective in normal time become more effective in reducing poverty during the crisis so permitting which is lower in reducing during the shock time but fair better than the other at normal time so the concluding statement is social assistant budget is lower in Ethiopia and then the poverty during a crisis those who were not covered better by the benefit system suffer a lot and and permitting system also works better with a higher budget thank you