 Okay, so switching gears a little bit from ethnicity to gender, but still on human capital investment. Belinda Archibong, this is co-authored work with my co-author Frances Anana, Georgia State University. So essentially, we're going to look at the impact of climate-induced disease on gender gaps in education. And so we started out this project by saying there's been a lot of investment in trying to close this primary enrollment gap, right, gender gap in primary school enrollment. So the OECD released a report, I think, last year, a few years ago, basically stating that, you know, it's great. We've closed the gender gap in primary school enrollment as a result of all of these free primary education programs in developing countries, but we still have this very persistent gap in completion rates, right? So education attainments, this is the kind of a million years of education on the X-axis, female average years of education on the Y-axis, this is a 45-degree line. And you see that a lot of the poorer countries are the blue diamonds over down here or to the right of this 45-degree line, right? So you still have a situation where men, boys still have more years of education completed than women in poor developing countries. So we said, okay, why is the reason more? What is the reason for this? You have all of this investment in education. You have all of this investment in closing the enrollment gap. Why do you still have this persistent gap in educational attainment? So I'm from Nigeria, my co-author Frances is from Ghana, and we've all heard stories growing up about, for instance, when people were sick in the household who got to stay home from school, it was the girls, it was the women, right? So my mother tells me these stories all the time. So we said, okay, you have a kind of double burden here, where you have a situation where you have low educational attainment, you also have this gender gap in educational attainment, and you have a disease burden in many of these poorer countries, especially in Sub-Saharan Africa, right, in the tropics. So here is these countries in yellow, these are these 23 countries in Africa that are part of this meningitis belt, right? So they are countries that are frequently exposed to meningitis epidemics. The cycle differs per country, three to 14 years, depending on the country. And we said, okay, what happens if you are in one of these countries in this belt and you're exposed to disease and you have this kind of education problem as well? What happens? What is the effect of the disease? What is the effect of the epidemic on the gender gap in educational attainment? So that's what we do. So we said, okay, let's look at evidence from one country, which is Niger, so one of the poorest countries in the world. Very low rates of education attainment, this is from 20 times about two years of education on average, and you see this gender gap between men and women, 2.8 years completed for males versus 1.5 years for women, lower than the Sub-Saharan African average, and of course also lower, Niger is also lower than the world average as well. So we said, okay, what is the contribution of this, I'll talk about the climate aspect, right? So this is a disease that is driven by the dry season, the hamatan season for people who know Africa, I guess this is, lots of people here know Africa, but the hamatan season in Africa drives this meningitis epidemic. So what is the impact of this climate-induced disease on this persistent gender gap in educational attainment? So just a preview of our results, all I'm going to do is look at evidence from a quasi-experiment. So this is, you know, you have this sudden health shock, right? Sudden epidemic that happens in 1986 in this year. We're going to show you that the hamatan season, the dry season characterized by high wind speeds, high dust concentrations, lower temperatures, really very strongly predicts this meningitis epidemic. So this is evidence that comes from the environmental health literature, and we're going to show this econometrically with regressions as well. And what we find is that higher meningitis exposure during the epidemic really reduces the years of education for girls who were school-going-aged, right? So primary school-going-aged, secondary school-going-aged as of the time of the epidemic. So these are girls between about the ages of 6 and 20 when the epidemic hits. So we're going to show that in a cohort study. And the magnitude of the effect is something like 3% to 4% reduction in years of education for girls relative to the mean. So we're going to show with some evidence that we think that one primary mechanism here is that there are income effects, right? So this is a health shock, a sudden meningitis exposure for the household. I'll show you some evidence that it's quite costly to treat, not just in terms of the direct cost from treating with vaccines, but also in terms of the opportunity costs from lost days of work, lost days of school, et cetera, for gone income for the household. I'm also going to show you that because of these income effects, it seems that one thing that's happening here is that in this context where you have this bright-priced tradition, right? So when women are married, there's a transfer of income from the groom's family to the bride's family. What happens is that parents who are then liquidity constrained who experience this health shock that is an income shock to the household are selling off their daughter, marrying off their daughters at earlier ages, right? So they're in exchange for this bright price. So that's why we're going to show you that the age at first marriage falls pretty significantly by about a year from 15 to 14 in these epidemic-affected areas. I will skip over the conceptual framework, but again, income effects, income shock, the health shock is this income shock to the household. And then we have this marriage response. So tell you a little bit about the meningitis belt and this meningitis epidemic. We look at Niger, which is this country right here, 95% of Niger's population rests in this meningitis belt. So this is this area in dark orange. So 95% of the population is here. So as I mentioned, three to 14 years is about the cycle for Niger. It's a sudden event. You don't know exactly when it'll hit, but when it hits, it's pretty devastating. What is meningitis? So infection of the lining of the brain. So very nasty. You get stiffness, fevers. In the worst cases, neurological damage and death caused by this bacterium. The epidemiology is quite complex. So only recently are the environmental health people now trying to understand what actually drives these epidemics in this region. And they have argued that the hamatans season, the dry season, I mentioned high wind speeds, high dust concentrations, explains something like 25% to 30% of the variation in meningitis in Niger during epidemic years. Another thing to note is that while there are vaccines, there have been limited effectiveness of the vaccines over the years because this bacterium, neserium and agititis, tends to mutate with every epidemic. So it's really a terrible shock that happens, sudden episode, because you can have the vaccine, but still get the illness. In terms of the epidemics, there have been six epidemics. So this is data from the WHO World Health Organization. Six epidemics from 1986 to 2008. We look at the earliest recorded one, which is 1986, which had about 15,823 cases per 100,000 population with a mortality rate of about 4%. So who's affected? Young children and teenagers are particularly vulnerable to infection, which is a really big deal for a country like Niger, where the median age has remained about 15 years for the last decade. So it's a very young population that's vulnerable to this epidemic. Another thing that's going to be important for our identification of the effect of meningitis epidemic on the gender gap in education is the fact that you have limited inter-district migration in Niger. So you have migration going down to Nigeria or going up here to, I think it's in Libya, up here, during, for instance, the droughts, et cetera. But within the country, you have limited inter-district migration. So what we're going to do is identify the impact of individual exposure to the 1986 epidemic based on where you live, your district exposure to the meningitis epidemic. So we only have data for meningitis cases at the district level. So that's what we're going to use to identify exposure. So why do I think this is primarily an income effect story? So I mentioned that the health costs of treating meningitis during the epidemic is quite high. So we don't have data from Niger, but we do have some data from neighboring Burkina Faso, which is also in the meningitis belt. And here households are spending something like 34% of per capita GDP on treating meningitis and follow-up costs associated with meningitis during an epidemic year. So 34% of per capita GDP is pretty significant. So we think that this is a pretty significant income effect for households. In terms of the vaccines, right? So the vaccines are supposed to be free. But oftentimes people pay for the vaccines because there isn't asymmetric information between healthcare workers and individuals about whether it's free or not. And then you have that. So let me tell you a little bit about the data that we use. We are going to use meningitis cases from the WHO and we are going to use DHS data on years of education completed from 1992 and 1998 as our outcome variable. So we have individuals across the country, across all 36 districts of the country. And essentially what we're going to do is a cohort study. So say, let's look at the people who would have been at the age to be in school, right? Not necessarily in school, but school going aged as of the time the epidemic hit. So these are the people born between 1960 and 1992. We'll look at the primary school age population that is between six and 12, the secondary school age population between 13 and 20. And as kind of a control cohort, not a control exactly, but a control cohort, we'll look at the non-school green age population which is the zero to fives when the epidemic hit. So just back on Niger, about 20 million people is quite homogenous in terms of religion and ethnicity, over 50% Hausa, I think something like, depending on the source 98% Muslim and also quite poor. Okay, so this is what the epidemic looks like in 1986 or the distribution of meningitis cases from the epidemic looks like in 1986. So these are the districts in Niger. This is 1986, mean weekly cases. So we have something like 15 cases on average per week per 100,000 population down here in these southern districts in Niger. So just to give you an idea of the difference, you look at a non-epidemic year like 1990 and this is the distribution. The cases are much lower on average, something like three cases per week. So just to give you a graphical kind of visual of just the difference in meningitis variation in the country during an epidemic year versus a non-epidemic year, this is Niger epidemic year 1986. These are the districts, the 36 districts in the country. And this is a non-epidemic year in blue. So very kind of very different outcomes there. So I mentioned how we construct the cohorts. We're going to look at school-going age populations, primary school-going age in six to 12, 13 to 20, and then the non-school-going age population zero to fives during this 1986 epidemic year. So Niger has free primary education. The mandatory school-going start age is age seven. So we'll show, or I might not get time to show it, but we do a bunch of robustness checks. We change the kind of age category cutoffs, marginal changes to the age categories, no difference in the results. And so what we're expecting to see is that we should see the effect of meningitis exposure for this school-going age population but not for the non-school-going age population. And our framework is kind of pretty straightforward. People who do regressions, difference in difference, you have district fixed effects, year fixed effects, year of birth fixed effects, you cluster your standard alerts by district. We try to be as robust as possible in looking at basically trying to estimate this gamma, which is this gender gap in educational attainment. So first set of results, focus on 1B. So the outcome is this case cohort variable or case cohort outcome. And this is the three to four percent, so ages six to 12, ages 13 to 20 that I was talking about earlier, right? This reduction in years of education for girls and no effect for this non-school-going age population. So robustness to different specifications there. Let me skip over that. We also do it as a robustness kind of, we look at a non-epidemic year, 1990 as a test. And here again, you don't see an effect for this primary school-going age category. You do see kind of a positive, we haven't got a chance to explain this, but it's positive effect on this non-school-going age category in a non-epidemic year. And then you see an effect for the 13 to 20s, but this is kind of some of the category from the six to 12s in 1986, spilling over to 1990. So just to show you, we did an instrumental variable analysis. I won't go into all of that, just to show you the first stage that Hamilton, the kind of previous year, dust and winds, very strongly predicts our meningitis case. It's during the epidemic year, right? So this is IM statistics, very high there. So I won't show you the second stage, but the second stage results in the IV also, they go along with the OLS results and the diff in diff. Okay, let me talk a little bit about the mechanism. So I mentioned that we think that a major mechanism is this early marriage story, right? So income effects 34% of the capital GDP. One of the reasons that we think that this is an early marriage story is that Niger has the highest rate of early marriage in the world, right? So 75% of girls in Niger are married before the age of 18. So one of the things that you can think of is, well, this is a disease, a heart shock, a covariate shock, because the whole country has experienced this meningitis epidemic. So who's marrying who, right? The man's family has also experienced this shock, right? So how are they getting married to these girls? So one of the things that we said is like, one response is that you are marrying across and we'll show some evidence on the wealth category. So slightly, it seems like older, richer men are marrying poorer, younger girls. And also, Niger is a polygamous country, right? So we thought this was interesting. We had this in the paper. We had to move this to the appendix because of measurement error. And I'll tell you what I mean in a minute. So this is the DHS data for men and women, the women's sample and the men's sample. So this is the DHS women's sample for school going age women in 1986 and in 1990. So what I want you to take away here is that when you ask the women, right? So we said, okay, this is not one-to-one matchings in terms of marriage. It's potentially one-to-many matching. So you can imagine where a richer guy is able to marry more than one wife during the epidemic year, right? So when you ask, so we wanted to see if there was an effect on the number of wives. So when you ask the women, how many wives does your husband have? They say up to seven. When you ask the men, how many wives do you have? It's always top-coded at four, right? Because in Niger, it's a Muslim country, you can only marry by law up to four wives, right? So anyway, measurement error, but when you look at the women's sample, there is a positive significant increase in the number of wives during the epidemic year. Look at the men's sample, there's no effect. Anyway, so we thought this was important because of course there's a large literature showing this very strong correlation between the age of first marriage and educational level. This is just a correlation. We confirmed this in our data set, specifically for, especially for this school-going age women in both the epidemic year 1986 and 1990, the magnitude of this effect is pretty stable and significant. Men's sample, it loses the effect between an epidemic year and an unpaid year. These are just purely correlations, just showing the correlation between years of education and age at first marriage. And this is the kind of marriage reduction story, right? So this up here is the hazard rates for women during the epidemic year, 1986. So who were school-going aged? So basically what this is showing, so this is the high meningitis affected districts above the national mean and the kind of low meningitis affected districts in yellow. All this is showing is that women are twice as likely to marry younger during the epidemic year than an unepidemic year. So these women who are school-going aged, right? So econometrically, I mentioned it's something like a year reduction in the age at first marriage from about 15 years to 14 years on average. So we do a bunch of robustness checks sometimes. So it's not rainfall. People say it's a rainfall. There are no, it's not like there's a concurrent rainfall shock going on when the epidemic hits. And we also find that the effect of this meningitis on the gender gap in, sorry, the kind of age at first marriage results is concentrated at the more asset constrained households. So the lower wealth quintiles as well. So I'm concluding. In conclusion, what we find is that the gender gap widens during the meningitis epidemic year. So again, really, I think from a policy point of view, pushes us to think about the kind of joint relationship between education and health in trying to determine policy, especially in the presence of future climate change, which is supposed to worsen disease environments in the tropics. So I mentioned that we're doing future work on trying to understand the marriage market implications as well. Thank you.