 And my paper is part of a larger project that we had on looking at the impact of refugee migration on host communities. And in this case, we look at different factors, labor markets. And it was also in conjunction with Maastricht University, so Melissa and Craig were also involved at some point in other projects related to this. But what this paper does is basically we're trying to tell a gender story. We're looking at the impact of hosting refugees on the hosting economy, and particularly looking at the case of Tanzania, and looking at the household level. So we're looking at household changes. In this study, basically, we'll talk about the context of the refugees in Tanzania. We had a very good background this morning with Maria who presented a paper on the same area. And I'll go over it a little bit too, but then we had some background on that. Then I'll talk about the gender impact. What are the mechanisms that we think might be at play here? Then I'll talk to you about the data and the results and some of the conclusions. So in this study, what we do then is we focus on the household dynamics, on the consequences. So basically the idea was thinking, well, when we look at the impacts of hosting refugees, are there any gender impacts? And we thought, well, no, the impact may not be gender neutral, so we should look more into this, and we try to explain the channels of how this may have been going on. And maybe most of you, probably most of you are very familiar with the setting. We look at the refugees in Tanzania. We call it the refugee shock because there was a big inflow of refugees into Tanzania in the 1990s, 1993, 1994. There was a major ethnic conflict in Burundi and Rwanda, and over a million refugees abandoned these two countries and moved into Tanzania. Specifically to this part of Tanzania, which is Kagera. One of the things about the conflict is that a lot of people had to move very quickly and basically distance was a big determinant of moving into Tanzania. Just to give you an idea of the magnitude of the refugee inflows into Tanzania You have here the refugees from Rwanda and Burundi into Tanzania. And you see how in the early 1992, 1993, there's a big spike of the number of refugees coming into Tanzania, and that decreases towards the end of the 1990s. And what we do here, what we do in this paper is we have information on the hosting population before the refugees got in, and when they were less refugees into everything. Just to give you a geographical perspective of what's happening here, as you see what we're going to look at is this region of Kagera, which borders Rwanda and Tanzania. Basically, our identification, or the idea, the way in which we're trying to identify these possible effects is because a lot of the refugees had to be very close to the border. There were a lot of logistical and geographical reasons for this to happen, but if you look at the dark points, those are the refugee camps. So you see how close they were to the border. So when we started thinking about the gender impacts, we tried to sort of like what would be channels to which there would be gender impacts of hosting refugees, and one of the ones we thought about was the competition for resources. Refugees in some of these environmental literatures, refugees have been called resource degraders, because when they got into the camps and year by the camps, they had to cut trees in order to use the wood for shelter and cooking and to clear space for cultivating crops. There were some soil erosion, depletion, and a pollution of water resources. There was some evidence that refugees in Tanzania used more firewood per person than the locals, and there were reasons for these. First, they were less likely to put out fires between meals because of the lack of matches, and also they depended more on dried food, so they had to cook longer than what the locals had to do. So there are some UNHCR estimates that at the peak of the refugee crisis in Kagena, the camps consume about 1,200 tons of firewood each day, and by 1996, there had been a substantial phase of land being partially deforested. We got this from where it was, but UNHCR shows the green areas and the parts that are less green, which are the ones in dark, and you see how by 1996, you see that there's been some environmental degradation. So in rural Tanzania, it's common for households to collect firewood and to collect water on a frequent basis, so additional times spent on these activities can restrict involvement in other activities, and women were more likely to be doing these activities anyway. Just to make sure that this population really depended on natural resources and we look at the dry season and rainy season, and if you see in 1991, they were highly dependent on natural sources of water. So the competition for resources was one of the channels we thought of. There is this anecdotal evidence of people saying that they had to work longer, they had to spend more time doing certain household chores because it took them just longer to be able to achieve those. We also thought about the role of casual labor, and this comes from a previous work that we had done on looking at the labor market impacts, and the way we thought about it was there's a big literature on OECD countries, mostly in the U.S., looking at the impact of low scale migration on the labor supply of women. For example, Patricia Cortez was looking at how the impact of low scale immigration had resulted in more labor supply of high-scale native. Cortez and Cesare basically found that individuals with high enough productivity outside the household for them was optimal to sort of outsource the household course and increase the time dedicated to employment. After this, there's been a couple other papers having the same findings. Women that are in a better position to be able to use these low scale migration were working more, so their labor supply had increased, but there were also having more kids, and they were also spending less time on household chores, more time with the kids in educational activities. There's a lot of these findings in the high income country, but we didn't find anything that sort of the low income country. In this case, it could be a channel that we could explore. In the low income country perspective or the refugee context, there is a surplus of casual labor, and reports, for example, suggested that in some areas close to the camp, the wage rate for casual work had decreased by 50%, and there was evidence that refugees substituted casual work, casual local workers. So there was some substitution of refugees, of locals for refugees in the casual sector. Some of the local women then could have employed these refugees willing to work for low pay and to help their household chores, dedicated more time to other activities. And we thought, well, maybe more likely it's going to be women with higher productivity, and we tried to think about ways in which we could measure this. And one way is sort of looking at literacy and math skills, since this is a very brutal area, obviously we cannot differentiate between high and low scale in the same way you would do in a high income country, but then in this context, just basic literacy and math skills could make a difference. So literate women would be less likely to compete with refugees in the labor market. They could maybe probably get more likely a position with an NGO than other people. They could take advantage of new work opportunities or use cheaper labor supply represented by refugees to help with the household chores. An illiterate woman would be less likely to take advantage of the presence of cheap refugee cash for labor, and they would still need make adjustment to increasing competition for natural resources that was represented by refugees. And sort of the last channel we thought of was in terms of the changing demand. So the working in food crops versus cash crops, women typically are responsible for crops that are meant for household consumption and men are responsible for crops that intend to generate income, so the cash crops. So with this changing demand, a consequence of the refugee chocking Tanzania was that there was an increase in the demand for specific agricultural products. And there was evidence of international agency increasing the demand for wood and the price of free farms. So qualitative evidence has suggested that male members of the household started dedicating more time to these crops that were typically women's crops because they were generating cash. So what we're doing here is we look at this case in Kagera. We use the Kagera Health Development Survey, and our identification is basically based on a semi-natural experiment. Forest migrants, so we rely on the fact that forest migrants were not even distributed across Kagera. There were natural topographic barriers, logistical decisions, and distance for the countries that have had them be closer to the border as we saw this morning. Mariam talked about this. So it's possible to use distance to the camps and to the border as an identification strategy. We obviously have not been the only ones who have done this. There's been previous papers who have used different sort of similar, in the similar vein, with little differences, the same identification strategy. So we used two rounds of the Kagera Health Development Survey. We used the 1991 wave, which was sort of before the refugees came in, and then we used the 2004, which is after, not after, but then once the refugees had come into the region. So this data was initially conducted in 51 communities, but individuals were tracked over time even if they had moved out of the community. Over 90% of the households were tracked, so the attrition rate is very low. It was a good recontact rate in the 2004 round of the survey. So to measure the impact of refugees, basically we used GPA's data for the distance to the refugee camps, and then we used the inverse as a measure of intensity, how intense was the presence of refugees. We also sort of weighed this by the population of each camp, and we interact this with a time domain, so that we can see sort of what happened after. So we focus on the impact of the shock on three different activities. We look at farming, we look at outside employment, and fetching water and collection of firewood as a measure of household food. In the 1991, they had other measures, but then this was the only one that was sort of tracked in 2004. So we focus on the likelihood of engaging in these activities and the time dedicated to these activities, just to give you an idea. Just to sort of think about whether these trends had to do with some sort of pre-shop data, we look at the impact of refugee shock on the likelihood of engaging in time spent on these tasks in 1981, and we see that there is no significance. So this was not correlated with task activities. So if you look at the chair of people engaged in different tasks, you see that in general, when you look at fire and water, the likelihood of women and men to be engaged in these activities was not so different. But then when you look at the 2004, you start seeing bigger differences, so there's a 25 percentage point difference between male and women in the 2004 wave, but then this is driven mostly by people that had an above-medium shock. The similar results we find with our spend per week in different tasks, and this is what we do. We have sort of a model in which we have a household dummy indicating whether the individual had engaged in a given activity, the ones that I mentioned. We have a timed dummy, we have the refugee shock, a female dummy, and we have distance to Burundi, Uganda, and Rwanda's control variables, and then we have many other controls. And then what we're interested in is this last coefficient, beta-coefficient, which is the interaction of being a female, the impact of refugees, and time. Because we want to look at the differences within household, what happened with women after the refugee shock. So what we find is that women are not engaged in these activities, so this is sort of the coefficient we're most interested in. We find that women are more likely to be doing farming activities, less likely to be involved in outside employment, and more likely to be doing household chores. So there was definitely a gender impact of the inflow of refugees with women spending more time in farming and household chores. But then, yeah, this is just an interpretation. Just to give you an idea of the magnitude using the median value of the shock, the results indicate that the presence of refugee led women to be 9 percentage points more likely to engage in farming and fetching water or collecting firewood, and 18 percentage points less likely to engage in outside employment than men. We find similar results when you look at time. At time allocation just basically very similar results. I have five minutes, but I should be... And again, when we look at... So to put this into context, using the median value of the shock, an increase of 1.4 and 1.8 hours per week, we find that there's an increase of 1.4 and 1.8 hours per week in time dedicated to farming and fetching water and collecting firewood. And the equivalent relative decrease in outside employment for women is close to 8 hours. So once we had the results, we thought, well, but there certainly must be differences when you look at differences across women. More productive women or certain differences across women could make a difference in the results. So we look at the results for different skill levels. So we had a division of gender and pre-shock literacy level. So we look at literate women, basically basic things as reading or doing simple math, because we thought maybe they could be benefiting from the additional supply of sheep labor and it would be a channel you would think might be at play here, but then it might be compounded on the general results. And when we look at the results by literacy, we find evidence that women, literate women, were more likely to engage in outside work, outside employment, whereas illiterate women were more likely to be doing farming and collecting fire and water. So the results were being driven basically, so if you separate by skill levels, the results were a bit different, a bit different. Very similar results when you look at math skills, because we look at whether they can sort of read and then we look at the different math skills, same results. We look at different results for age groups and basically we found that the people that were 30 or less in 2004 were the ones driving these results. Again, we had more time spent on farming activities and household chores and less time spent on outside employment. And we looked at the results for children as well, and in here we found that the time dedicated to outside employment was higher for girls, but then they also had more time dedicated to outside employment and household chores. So in summary, we find that hosting refugees had different impacts on the time allocation and activity choice for women and men, with women less likely to engage in outside employment and more likely to engage in household chores. But the results differed by skill level, which we thought was a very interesting result because we find that literate women are more likely to engage in outside employment in response to the shockwaves. Literate women are more likely to be engaging in farming and household chores, basically firewood and fetching water, collecting firewood and fetching water. And as I mentioned, this is part of a larger project that we had and any questions, I'm happy to take them at the end.