 The presentation today is very fitting. As some of you may recall, we started in early May with Andy Sumner reflecting on the damage of COVID-19 for global poverty. And today we sort of come full circle and we're going to listen about the effects of poverty on lockdown compliance and the associated challenges, which have been many and very complex across most of the world. And to discuss this important topic, we have with us Professor Olivier Bargain from the University of Bordeaux. Olivier has a very distinguished career with very important work on public and on labor economics. He's also a member of the Council of Economic Advisers to the French President. And he has been producing some of the very earliest and very exciting research on the links between COVID-19 and social behavior, both in Europe and in developing countries. And he's going to focus on developing countries in our talk today. We also have with us Amina Ibrahim, who will act as a discussant to the presentation. Amina is a research associate at UNIWIDA based in South Africa. And Amina is going to discuss Olivier's paper and also give us a perspective about the evolution of COVID-19 in South Africa and its experience. So look forward to hearing from both. Before I pass on the word to Olivier and to Amina, just a bit of housekeeping. I wanted to remind everyone that webinar is being recorded. All participants are going to remain muted. And it would be really great if everyone could use the Q&A function to ask questions during and after the presentation. You will see the Q&A function right at the bottom of your screen. If you scroll down to the screen. And if you have any questions during the presentation, please post. I will be keeping a track of all the questions and I will address them to the presenters at the end. But please feel free to start whenever you have a question. So without further ado, thank you to everyone and thank you to the two presenters. And I will pass now over to Olivier. He'll speak for 20 minutes and Amina will come afterwards. Olivier, very welcome. I know the team. Thank you very much. Thanks a lot to Patricia, to Ana, Amina and all the team at wider for this nice invitation and this next occasion to present latest work on the topic of poverty and COVID in developing countries. This is a paper and it's extension that's been co-authored with Ulbeck Aminjanov, who's also at the University of Bordeaux in our development team. And I will start with a quote from a UN World Food Program Executive Director that said at the same time while dealing with the COVID-19 pandemic, we also are on the brink of a hunger pandemic. And I think everybody has gained consciousness over the past weeks and months that there was very high risk in many countries in the world. About this and there was a lot of actually literature on India, but we are going to focus here on Africa and Latin America. I'm mentioning a few of the recent papers on the fact that strict social distancing, strict lockdown policies is not always fitting in the context of poor countries because indeed it puts these people at risk, especially these people who live on daily informal jobs in poor countries, poor regions of the world and for whom staying at home is not an option. Teleworking is not an option. And for this strict enforcement, enforcement of lockdown policy might be extremely dangerous if that reduces their access to earnings and to means to live. This is as simple as that. So I mentioned here a few papers, one of Patricia and the team at wider on the extremely interesting connection between this and the risk of social unrest through poverty and a violation also in social capital in poor regions, a paper of the self, Tico Ferreira and World Bank colleagues on poverty and COVID. And other contributions, the one from Martin Robertian was, I think, an article in VoxDev focusing exactly on this issue here. I'm mentioning also the work of Robertino stating that indeed a large share of the population in poor countries does work informally and depend on daily income and cannot afford to stay home. So far there was not that much evidence on the effect of poverty and on compliance to a lockdown on health policies during pandemic. And this is what we try to address here today. So the attempt here is to examine whether poor areas, poor regions in developing countries indeed comply less with social distancing rule. And for that we use data that have been available for some time now, which are the Google Mobility Reports. Google has provided indicators of mobility based on mobile phone information at the regional level for many countries in the world. So we are going to use information about nine countries for Latin America and Africa. But I mentioned at the end that we actually have many more nowadays as throughout the month of June, Google has released information for now more than 40 countries. And we're going to illustrate what is the main result of the paper here is that people in high poverty regions actually move significantly more after a during lockdown compared to low poverty region. The region with less poverty can afford slightly more to stay home and be less exposed to the danger of the pandemic. We will connect that to the actual outcome in terms of health, in terms of the diffusion of the virus by looking at the effect of mobility on the diffusion of COVID-19 and then connect the two elements of the equation and then look at how poverty rates increase or exposure under the rate of diffusion of the virus in the sample of countries we are considering here. So again, the aim is to explore the pre-pandemic variation across countries and across regions of four countries in terms of poverty level. So for that, we need the first ingredient which is the pre-existing level of poverty at the regional level, measure as a share of population of each region living below a national or international poverty line. So we are going to use different types of poverty lines. The standard ones being the World Bank $1.9 per capita per day as the extreme poverty line or the 3.2 or the 5.5. We will use what is the official recommendation by each government in each country in terms of poverty line. And sometimes we will use alternative poverty line compared to the World Bank ones which are based on official poverty statistics or on the recommendation by local government but not necessarily available and we will have to recompute this poverty line using household survey data for these countries. So the nine countries under study are listed here are Argentina, Brazil, Colombia, Mexico and Peru for Latin America and Egypt, Kenya, Nigeria and South Africa for Africa. We will use poverty in three different forms. A simple one and I'll show you the results in graphical terms which is really straightforward then it's a simple ways to have binary poverty that is regents below the national poverty rate or above the national poverty rate. We can have a little more of a variation by using their size would be the regents below the 25th percentile of national poverty which we will call low poverty regents and the regents in between the 25th and the 75th percentile and the regents above the 75th percentile high poverty regime. Or we can directly use the continuous measure of poverty and in order to make it more comparable across country you can also take the Z score which is a form of standardized poverty deflated by the mean poverty level and divided by the standard deviation. The Google mobility report which is the other main ingredients of the empirical exercise here is a measure of how much people move and during the COVID period and of course before the COVID period as a benchmark. These are aggregated and analyzed data from users of mobile device using the location history data in their mobile phone and it is measured as the number of visits or duration of stay at different location compared to this benchmark period which is pre-COVID for the regional area that we are looking at. Nothing happened at that time. That is the month of January until February 6th. The type of mobilities are very interesting because you have things like retail and recreation, grocery and pharmacy, parks, transit station which is how much time you spend at a transit station or at a bus stop. But the main one we are looking at of course is workplaces. We want to know people go to work as simple as that. And the residential area is basically the time you spend at home and that means not doing all this other activity and complying with lockdown police. We are going to look at the month of February and April 26th. A lot has gone on until in the more recent period and I can say brief proverbium but this period is the most interesting one and the period for which we had the Google information when we actually started to work on that project. What you see here is a timeline. 0 is March 1. And that's the mobility intensity and you start with a level of 100 here when we are in the pre-COVID period and you see this big drop in mobility. You see some countries that really do decrease their mobility a lot here like Colombia for instance. You see some countries from Latin America like Brazil with these strange policies. So there's a reduction but it's not as much as for other countries. And for some African countries here like Nigeria you see the decline which is kind of slow and plateau which remains quite high. We also indicate that poor countries in Africa actually on average cannot really afford to decrease their poverty or their mobility in terms of work mobility, work movement as much as other parts of the world. If you put all this information now in graphs and I hope it's not too small and you can see here we have the tercile version of the poverty within pink, the high poverty region, the region with a high percentage of poor people. The purple one which is a median rate of poverty at the regional level and the blue which are the lower poverty regions. These are the variation across all the region of in our data which are nine countries and more than 250 regions across these countries. Showing that again that's the mobility in terms of work mobility going to the workplace and compared to the January period here in early March or late February you were somewhere close to the initial level around 100. So it was very, very similar and you see no difference and that's very important to say that. You see no difference between the high poverty region and the low poverty region. They tend to behave in the same way. We are going to use what econometrician and economists call a difference in difference approach looking at before and after the introduction of lockdown policy and the impact of COVID on these countries. And what's very important in these methods is that the different groups that we are looking at tend to not necessarily behave in the same way but at least have parallel trends have trends that are common or parallel to each other in the period before the things start. And what we've got here is that these three groups of poverty, three groups of region actually not only behave in a parallel way but even in the same way as we see these three line actually do overlap almost completely which makes the point even stronger that there were really no difference between them before. What you see and that's the main result of this research is very simple. You see a monotonic pattern here. You see that the regions with low poverty rates or at the counseling level are those who reduce mobility in terms of workplace more than the medium poverty region which themselves reduce our mobility even more than the high poverty region. So the poorer you are, the less you reduce mobility meaning the more you still have to move out of your home to go to work. And therefore to expose yourself really to COVID-19. This was the results using the variation and you see this, it's not only a line. You see the surfaces in blue which are the confidence intervals at the 95% level for each group. And you see that this confidence interval do not overlap meaning that the difference between mobility reduction across these three groups is statistically significant. You see that for Latin America where the same kind of results may be not so strong and for Africa you see a lot of variation and you also see that overall Africa reduce mobility less than Latin America showing that at the continental level or the country level then the poor country also cannot reduce our mobility as much as slightly richer countries. But you also see an interesting variation across the different groups here. Another piece of information is how these things evolve compared to other type of mobility. So we are in again in blue and red we've got that's gonna be low poor, low percentage of poor and high percentage of poor so the high poverty region and low poverty region for mobility and we compare to the light pink and the light blue here for another type of mobility which is recreational activities. And we see that because it's less essential than a mobility could reduce more but most importantly the difference between the two type of region it's much smaller in this case. Same here going to the grocery that or the pharmacy that reduce less because these are vital activities very essential activities. But again there's no reason for the poor and the less poor to behave differently in that respect. And you see that the difference again between the light pink and the light blue is small compared to the difference between the two type of region in terms of work mobility. Same here for transportation. I will just skip all the technicality and just focus on the essential result. If you look at this estimates here in pink it is the actual difference between the pink and the blue between the high poverty region and low poverty region compared to the pre COVID and pre lockdown announcement era. I forgot to say one important element here is that the dash line here March 20 is the data that we are using as to define the before and after situations and that's actually what we use in this estimation. But again the result don't really change too much compared to the you could choose some date that is more in line with for instance the World F organization date of announcement that COVID was a pandemic which is March 11 and which is more in line also with the lockdown announcement in Latin America or slightly later for Africa. But this choice of the time cut off is not extremely important. The effect is so strong that it is not sensitive to it. So what do you see here is this number three or four that's out of a state of 100 the impact of being in a high poverty region compared to a low poverty region in terms of your mobility. This is in terms of, I can mention later in terms of elasticity what it means but you see it's around three or four point of percentage difference between this type of region. And we see here that the effect is much larger in Africa on part of Latin America and you restore a bit of the effect for Latin America. Once you exclude Brazil which is an outlier for the reason that we know. I will, time is running and I don't want to push too far but what we see here is that the estimate that I was looking about around four for what mobility is not found when it comes to retail and recreation or going to the grocery or transit station, the other type of mobility. So that's what the point I was making when I was showing the graph with light blue and light pink curves. We don't see a difference between the poor region and the rich region, not rich but the poor region and the less poor region in terms of this other type of mobility and the zeros here are just p values for the difference between this coefficient and the coefficient for work mobility meaning that we have an absolute rejection of the quality of this numbers here. That means that there is extremely significantly a larger drop in mobility, of course, poverty regions here for work compared to the other type of mobility. So it's really a work story, it's really a going to work type of story. The last element I would like to convey to you today is how does it translate in terms of decision of the virus? We have here an estimation of what we've just seen today here was an effect of poverty on mobility. The effect is, if I take for instance a one star deviation in poverty, so the difference between two regions that are one star deviation difference from each other leads to for instance a 10% difference in mobility. So the number we've seen before, three or four point of percentage looks small but a standard deviation is a standard or a very regular type of metric to actually assess how much variation there is across regions and it's true also within countries. And this type of normal difference across region in terms of poverty leads to a variation in terms of mobility of 10%, which now start to, in terms of magnitude, start to become relatively important. But we want to go beyond that, we want to go beyond the causality between from poverty to mobility. We also want to show how much an increased mobility or a smaller reduction in mobility leads to an increased diffusion of COVID, which we measure in terms of upcoming growth rates of COVID-19 cases. We find that a 10% increase in mobility which was about the effect of our one star deviation difference in poverty. A 10% increase in mobility or difference in mobility between two regions leads to around four, five percent increase in the epidemic growth rate. Again, it might seem kind of a small variation, but as we had exponential diffusion of the various, same somehow to some extent by the policy in place or by the lockdown behavior at different degrees, this upcoming growth rate of COVID can actually over a weeks materialize into a quite large number of additional cases. If you take the average in this nine countries in Latin America and Africa, we had around 200 cases, cumulative cases recorded by March 20th, which is, if you remember the cutoff points that I used to say there was an after in this kind of difference approach. And by May 3rd, which is the end points if we look at the two week forward COVID diffusion after the last data that we've got in some of Google mobility report, then we ended with more than 22,000 cases on average at the country level across the same country. So what we see here is that the elasticity that we have calculated this difference of around four, five percentage points in the growth rate of COVID would lead to an additional 2,500 cases by May 3rd. So it's 10% and it's quite a lot in the end, even for the initial percentage as well. I would conclude that I'm already past the time, I'm sorry for the chauvin, for me to interrupt me as she was very kind, but I really have to wrap up. Poor people whose livelihood depends on casual labor are less likely to comply with social distancing requirement even though they were strict or less stringent across countries. What we calculate is an average effect. And we actually illustrate that here by showing that indeed countries with a high rate of poor people in these regions are in a way compliant less by reducing their mobility less. One drawback might be that we use Google mobility report that are based on the fact that people actually have mobile phone. We have checked and despite the high poverty in some region, the penetration rate of in terms of smartphone and mobile phone is extremely high. In the worst case, what would happen is that we would underestimate the true effects of poverty on compliance with this health measure, simply like the fact that in the poverty region, in the high poverty region that we are looking at, only the less poor actually have a mobile phone. So the less compliance, the lower rate of compliance in this high poverty region would be even stronger if we could actually measure the poverty of the extreme poor or the poorest in the high poverty region. Maybe we'll discuss more, some extant extantion of this result in what comes next, but I just wanted to end by saying that we now have this similar result for more than 40 countries in the world and more than 500 regions using the latest release of Google mobility report. And we also find a strong effect of the policy response and the income support that have been put in place in many countries, an effect on mobility and exposure to COVID, but also a reduction of this gap between the high poverty and low poverty region that we have shown in the result today. I should stop. I'm sorry for going beyond my time. Thank you very much. Thank you very much, Olivier. Fascinating presentation and really interesting results. I'll ask now, Amina, to discuss the results and speak about South Africa. So Amina, 10 minutes. You will need to unmute. Thank you, Patricia. I just want to share my screen. You can see my presentation. Okay. So thank you, Olivier, for a very interesting and thought-provoking presentation and paper. And so I've been considering what this rock and hard place is in South Africa while we remain in lockdown and maybe I can share some light on the scenario. Okay, so the paper's interesting as it considers the challenges to the lockdown where large parts of the population are forced to work to escape extreme poverty or hunger. And it considers the impact of the responses to COVID and not the virus itself. And so I guess as to the literature in terms of the economic impact versus the health impact and there's a large discussion going on about that in South Africa at the moment. And then we also, you know, the paper also moves on into this discussion on estimation of how high poverty rates translate into faster spread. And through this channel of increased work mobility. So your findings are suggesting that lockdown is challenging, if not almost impossible for poor areas and this is now proxied by your mobility, Google mobility data. And this is interpreted as the necessity for people to actually go out and work and earn in this really difficult trade-off between poverty or getting the virus. And I think this combination of the Google mobility data, then your survey data on poverty rates and the COVID data is actually a pretty interesting nexus. So despite lockdown measures, the virus has spread considerably. And in South Africa epidemiologists suggest that we are only now entering the peak phase which could last perhaps even until September. And so these are just the nine countries that are under examination in your paper. And I just wanted to see where South Africa kind of lies and they sort of, South Africa is kind of in the middle. But it's nice sources to see where the, that these countries actually, you know, had their first cases in a very similar period. But the one thing I guess I missed maybe from your paper and I mean, we haven't spoken about it in your presentation so far is, I know a lot about the lockdown and what's been going on in South Africa, but I know less about lockdowns in other countries and kind of some similarities and differences would have helped me to contextualize some of this information. So in South Africa, the lockdown is initially viewed as a necessity to flatten the curve. And what the lockdown has done really is put the South African government time to prepare the health sector. And lockdown is just one of the collection of strategies for the government where it's including social distancing and preparing or assisting with businesses and looking at mitigation strategies at the workplace. But we also know that lockdown cannot continue forever. It logically has to finish at some point. And the virus is still going to exist once the lockdown ends. We just simply cannot wait until there's a vaccine rather because we have no end date or we have no idea when that'll happen. Some say 18 months, but there's a lot of uncertainty around that. And then of course, a complete lockdown is not possible in South Africa but anyway, right? So in South Africa in particular, people don't have the kind of savings where they can hoard or collect food supplies for long periods of times. And we also require essential services. So we need our hospital staff and we need our municipal workers. And the government was very specific and published a list of essential workers that actually needed to go out and work. But then we also need transport workers and our grocery stores workers to keep moving about and returning to their workplace each day. So in South Africa, our lockdown actually started on the 27th of March, which is quite close to your 20th of March in your estimation. So we've been in lockdown for 102 days already. We have a staged lockdown set up in South Africa where the first five weeks we were in level five lockdown and this only permitted essential services to move around. So you weren't permitted to leave your home. There was a curfew in the evening even for those on the essential services you needed a permit to go out and you could only leave your home to get groceries. We moved to level four on the 1st of May and so more movement was permitted. So the curfews were extended but people only permitted to actually leave their homes in the morning for exercise purposes between the hours of 6 a.m. and 9 a.m. They were very specific about this. And then we're currently under level three of lockdown and that started on the 1st of June. And this includes some business travel and then some phase return to work and schools but those also have their own set of challenges. So the rock in the hard place in South Africa is what I sort of want to touch on. South Africa has a set of challenges related to lockdown some of which are sort of broadly touched on in the paper. The density as I want to show you in this picture, this photograph is actually much higher in Poirier area. So the top part of the photograph has a poor community which is I guess more dusty and very tightly packed and then you've got a less poor or rich community on the other side where everything's very spaced out and green. And so your Poirier areas, so the density in Poirier areas are two-fold. One is actually you're closer to, you know, households are actually closer together but within households there are more people living together. Then the second thing that I fixed, but both low poverty areas are high poverty areas, perhaps more so the high poverty areas is that schools and daycares have been closed. And what this means I think for people who do have to go out and work both as essential workers or informal workers is that they need to then, if there's nobody else in their household to look after their children, they actually need to hand their children over to somebody else to look after them whether it's a neighbor or it's a grandparent. And so that's actually increasing your interaction with other people. And the last point that I've been thinking about is this South Africa's got these high levels or still has high levels of HIV and TB. And so we've got a population with lots of co-mobilities and in some cases these co-mobilities are located in poor areas. So the Western Cape or the areas in and around Cape Town actually have clustering of TB in certain areas and this would also exacerbate the spread of the virus. But overall I think we can buy into the argument that mobility, work mobility has a severe consequence in poor areas. So what has the South African government done to reduce the need for poor people to go out and earn their wage? The first thing that they did very early on was there was a cap on prices of essential items easing the burden of the poor and there were several complaints lodged and the government's taking action against businesses who were price-couraging. The second is social relief of distressed food parcels and vouchers and so this is part of the government strategy. This came in a little bit later but civil society has also been pretty active in supporting these efforts with large-scale food distribution and this is a photograph of this food distribution in a poor neighborhood in Cape Town. And then the last point I want to mention about the support is that there has been an increase in social security. Again, this may have come a little bit late but there were two parts to this. One was a top-up of existing grants. The child support grant which reaches a number of families and has been in place for quite some time but also a new grant for those falling outside of the social security system, informal workers in particular. The reality, though, is that this didn't all go smoothly. So, broadly, the measures were, they garnered support and they were perceived as a necessary response from the government to the looming hunger crisis. The photo on the right of a residence from several informal settlements in Caligia who gathered and they were demanding water during the lockdown and this is people sitting on the steps of the municipality building but they have been violent food protests and looting of businesses all over the country as well. And this pandemic has certainly highlighted problems in South Africa's social security system and now that more people are more heavily reliant on this system, these issues have been exacerbated and they don't simply disappear. In some ways, these measures also assume that the state is capable of scaling up the social security system. And so there are lots of problems in terms of the implementation of actually these social security, such as we have a system now where those who need a new grant, the special grant have to go out and apply for it and a number of people haven't actually been approved for their grant further delaying the access to the money or maybe they won't access it again if they don't try and actually fight the system. So it becomes quite difficult for those who are in desperate need to do so to actually access the money that the government actually set out for them. And then there's being corruption and food parcel corruption in particular and this is in two forms. One is where the food parcel just doesn't actually reach the people at all. Somebody else's pocketed the money somewhere along the way and the other way is where leaders have provided food parcels only to those in their constituencies. And so this makes for a fairly challenging and tense situation in South Africa and it's not clear where we go from here because these challenges are not new and it seems difficult that we'd be able to fix them during a pandemic. Thank you. Thank you very much Amina. I will start by throwing in two questions which link very directly to Amina's discussion and perhaps Olivier could address them a bit further. First is on levels of formality and informality. Is there a way where your analysis could be correlated? I know you're doing it at regional level so it's very difficult to kind of pinpoint areas where you might have slum areas and formal settlements and so forth but is there a way of correlated with levels of working formality? And the second one is you mentioned right at the end very interesting point about social assistance programs dampening the effect of poverty on the compliance which links with what Amina was saying about the social protection program in South Africa and many places in the world. The governments have tried to implement some kind of social safety nets. Could you maybe talk a bit more about these and how and whether you have some insights about the levels or the type of programs that may be working? I know this is new research but perhaps those points could be maybe you could talk a little bit more about that. Olivier, if you. Thank you very much. I will shut my screen again with you. Can you see? Yeah, okay. I've just run in real time in life an estimation for South Africa which I had but I had for the binary poverty and here we have actually that's from the three groups that I was showing before like the low poverty, medium poverty and high poverty. If you take the two extremes as you can see here the result that we had applies to South Africa as well with the high poverty region reducing the mobility less than the lowest tercile of poverty region. If we have just binary group, we cannot overlap and there's not so much difference in South Africa across a region that we had on average but if we take really systemic cases then it does work. Another point that I'd like to make is that we had to use a lot of countries and a lot of regions because if you use only country variation or if you focus on one country, of course, you can, it's very interesting as you just did and it helps to really tell a story but if you want to have an overall like a global view of the effect of poverty on higher exposure to COVID then you need to use regional variation across many countries. Even country variation itself is complicated because we know that country would indeed design maybe the policy response to the overall poverty level of the country itself, et cetera. So there are many confounding factors that come into play. Well, if you use regional variation and in your estimation you have region fixed effect as we had then you escaped from some of these issues. So for what we did, we really had to use the variation across countries and many regions in the world. But that said, the point we've made on average is true in general and as you see on this graph for several countries, Colombia, Dominican Republic, Ecuador, at the bottom, Chile is, we don't see something extremely strong. It actually goes the other way around for Chile and Brazil has this strange policy as we know, so these are two outliers. But otherwise, for many countries in New York, you see here El Salvador, Honduras, Mexico is also a bit of a outlier, but not that much. Paraguay, Peru, Uruguay, et cetera. Lebanon, Jordan, Egypt, Kazakhstan, Philippine, our story also work at the level of each country. The last point I made is related to policies and it was extremely interesting to see more closely what's going on within the country in South Africa. If we look at the global picture again, then you've got here, for example, the graph on the right, these are the variation in mobility and reduction in mobility during COVID time across now three groups of countries or regions which are the variation in terms of income support. You mentioned several policy designs. This is another picture we will dig into the nature of the policy schemes that are put in places. But here it's the overall type of income support, the outstanding income support beyond normal social assistance that has been put in place during COVID time. And you see that region with no income support reduced mobility, much less. And you see the mobility rates still extremely high compared to those with low income support in purple and in blue, those with a high income support. And the point we were making about difference across regions in terms of poverty, you see that the income support actually does reduce this gap. Here on this graph here, you've got the light pink and the red, which are the low poverty region and high poverty region, which don't receive any income support. And you see they don't reduce mobility that much. And we've got this gap, the one that I was pointing at during my presentation. And in blue, you see the light blue is the low poverty region and the dark blue, the high poverty region in countries we've receiving any form of income support. And you see that first of all, their overall mobility has reduced much more than the pink curves here. But also the difference between high poverty region and low poverty region has shrinked, meaning that this policy responses have managed to reduce the difference in exposure to COVID across the regions in this country. I hope it answers some of the concern or question and I'm very happy to answer some more. Great. And that was also the question on informality versus formality, Olivia, whether you'd have some insights on that? Yes, it's a very, very good point. We thought about that at first. We thought about actually the trust issue that you're addressing in the paper I mentioned in my introduction as we work on that for Europe. And on the labor market side, we thought about informality. In the end, we focused on poverty as it's in the most pressing question. But that said, I also motivated our empirical work to get the argument of the fact that the poor people have to go to work and these jobs are an informal job. The thing is that there's not probably not in a variational trust region in terms of the nature of the labor market. You have informal worker everywhere. We could try, we could attempt to look at the variational trust region in terms of the right of informal work and to see if it has an effect and be a complement to the story we were telling. And it's definitely extremely important. Great. I have a couple of questions here related to the data. So the Google mobility data, I think it's been quite widely used now amongst researchers looking at various effects of COVID-19. Can you say a bit more about how the data is generated and also about the reliability? You mentioned the difficulty of capturing some of these effects in developing countries where the use of smartphones might not be very well used, et cetera, but you maybe talk a bit more about how the data is generated and how reliable it is and whether there are any sort of sources, any alternative sources for which robustness testing could be done. Yes, that's a critical point, obviously. The Google mobility reports give some scores in terms of, indices in terms of mobility intensity, which is based on the time spent at near location that correspond, that are identified by Google as being helped for transport or pharmacy, grocery, parts, et cetera, rather than the time you spend in your, identify as your normal residence location or identify as your daytime job location. So based on the mobility itself, the movement itself or the duration, combine this information, leads to yield this course in terms of mobility in different type of activities. The second point is, so the critical one is it representative. We've got here on, let me show this graph here. It's a rate of penetration of mobile phone, so it's not necessarily a smartphone, but as you see, there are different type of indicators in the third column, but that's usually the number of access per 190 times or the number of cellular phone per 190 times. And of course, some people can have more than one phone or more than one SIM card or one more subscription because they share the number in the family and so on. So this rate can be above 100, but they are very high even in countries like Kenya, or Nigeria, the poorest country in the least here, and South Africa as well. So people and even poor people have mobile phone. I think it has become extremely important for everyone around the globe to be able to stay in touch, to communicate. And there's a lot of literature and economics on how mobile phone are used for access to credit, for instance, of access to different type of credit operator through mobile phone in poor regions. So it is not exactly the idea, the preconceived idea that poor regions don't have a smartphone or mobile phone, they do have. But indeed, as I explained earlier, maybe we're wrong, but our intuition is that the bias that can operate here and the fact that the extreme poor may not have as much as the less poor, then that would tend to underestimate the effect of poverty that we are highlighting in our work. And the things are probably even more serious than the gap between low poverty region and high poverty region that we obtain. Simply because of the fact that indeed the poorest people may have less rate of smartphone holding than others. Even though there are those who probably still have to commute or to work more. So we probably have a lower bound of the true effects of poverty. Right. Question now for Amina. So you've discussed, one interesting or worrying consequence of some of these issues is rise in social unrest. And we have a paper at you when you are addressing some of those issues. And Amina, you refer this to the case of South Africa. And that seems from what you said, it seems like this has happened even while cash transfers in a variety of also other social programs were being implemented. So does that then indicate that these programs perhaps are not having the effect that they should have or where would these sort of protests be coming from? What is your understanding of what's going on? So I think it's a, I mean, part of it said it came a little bit too late. And so when you've got a large community that is very hungry, I mean they have video footage of trucks driving into poor communities and people just kind of storming the truck. And you grab as many bags as you can instead of standing up and queuing and waiting for your bag of food. And so there's that kind of response that's been happening. And then it's also, I guess, part of the corruption story, because if you, or some people are getting the food parcels and others are not getting the food parcels, that creates difficulties within a community as well. And what goes hand in hand with this kind of social unrest in particular in poor communities is police enforcement and in the case in South Africa is police brutality. So in the early days of the lockdown, there were actually more cases of people who had died from at the hands of police than people who had died from the virus itself. And so this is, I mean, then this ties into the whole discussion about work mobility going out. You can't physically be in a small space with a lot of people all the time. So very, very challenging situation in terms of compliance in South Africa. Okay, well, we're getting to an end. We have about a couple of minutes left. So I would take the opportunity now at this point to just ask any final words from both of you. And I know we have an audience that is particularly sort of interested in the policy implications of some of these results. So I guess the question here coming up is what can we do? You know, social protection is coming too late, unrest is increasing, poverty makes it impossible to many, many people to stay at home. What can we do? So, Olivia would like to go first or Amina? There you are. Well, I think this is a terrible time. And at the same time, it's an outstanding time because as we thought that we couldn't increase debts and help people also in rich countries, we realized that we are able to actually unblock situations. So there's this huge question about the debt of poor countries that's on the table. But there's also a researcher and for policy making an extraordinary time as we can really know see how this outstanding targeting or universal transfer or whatever the form it took, what effect it has not only on, of course, on helping people in our time, but also try to see how much it has reduced. It's able to reduce inequality, how much it's able to help people in the longer run, maybe to infer from, for instance, from the mobility figure I showed, but for all the type of indicators to invert it to find how much the reduction in poverty gaps across regions of the world happened during that time. And of course, it's only bound to be a temporary time, but if some of this support, some of this cash transfer, some of this help and help could be sustained and because we have seen the effects and the effectiveness and if we could monitor and measure that, there are several World Bank reports that actually provide a lot of interesting information like John Keeloney and Etal and so on on the nature of policy response across the countries, all the countries of the world. And if you can monitor that over the next month and the next years and see which of this program will remain and how much you can help people, then it is a fantastic opportunity to seize in that respect. I think that's the main message here. So I think I have two points. One, I think, I mean, in theory, it's nice to have a new social grant and have this value and tell everybody about it, but I think the implementation is important because you can't actually get that money to people. You know, you're going to continue with the social maze, you're going to continue with the looting and just very tense situation. And then the other thing that I think this other thing governments may stop with a little bit is that initially they had announced that they would have a top up of the child support grant specifically. And so the child support grant goes to each child under a certain age, but it later transpired that actually it wasn't going to work out that way, it was going to be just a top up for the household. So whether there was one child or the household or five children in the household, you were all just getting X amount of money. And so that's not necessarily helping. So you're going to need the top up in terms of the cash transfer, but you also still need the food parcel. And so you're going to need this combination of policies to get us to where we need to be. Thank you. Very true. Thank you very much. We are completely out of time now. So it's left me to thank you both for fascinating presentations. And it's really a great opportunity to see these early results about what helps COVID is affecting so many people across the world, everyone. And I'll take the opportunity to wish everyone a good break, a good summer break wherever you are. And I hope to see you all back after it. So thank you very much, everyone. And thank you for all the participants that have stayed with us over the next last couple of months. Thank you. Thank you very much. Thank you. Bye-bye.