 Okay. Well, thank you very much. Thank you, Gian Rango, for the presentations. It is great to be in this classroom. Again, I presented this paper, Not only this paper, a different paper, in this exact same classroom About five years ago. It was a different topic. But anyways, today I'll be presenting some work with Jose Martinez. The title of the paper is Community Price Shocks on Child Mortality, Evidence from Anti-Coca Policies in Peru. So the objective of this paper is twofold. The first objective is to identify a causal relationship Between agricultural commodity price shocks and child mortality. And the second objective is to be able to set some light on The pathways, the mechanisms that are behind this. The big motivation for doing this is that income, aggregate Income, is more volatile in agricultural settings. And it's also the case that most of the poor live in agricultural Settings. And given the prevalence of Cranes constraints and market imperfections, we worry that The poor will not be able to weather an aggregate income Shock. And therefore, some of the Welfare dimensions such as child health may be at risk. Now, we started this paper because there is a literature On aggregate income shocks and child mortality. However, the evidence for the developing countries is quite Mixed. Sometimes we find that mortality Increases, sometimes we find that it does not. So what we wanted to do with this paper was to get a clear Natural experiment and be able to identify what are the Channels at play and then set some light on potentially Why are we getting these mixed results in the literature. So to fix ideas, let's think a little bit about a Theoretical framework. So let's assume that the Production of child health depends on two goods. We can have health promoting goods, think about medicines. And time. Time intensive health investments. Think about going to fetch clean water, hygiene, and many Other things that do matter for the health of a child, but Do not necessarily represent a monetary cost. And now let's assume that wages drop, say, because of some Commodity price drop. And the question is, well, what Happens with child health? And the problem is that the Effect, in fact, is theoretically ambiguous. Expenditures in the presence of marketing imperfections may Go down, so households may not be able to wear the shock. Therefore, they may reduce their consumption and they may Reduce their consumption of some nutritious foods, also the Health cut from promoting goods. But then what is not very clear Is what happens with the supply of labor of the primary Caregiver, right? So if we think about maternal labor Supply, when wages are lower, what should be happening with A regular supply curve is that you should be working less If they are paying you less. But in developing country Settings, what we sometimes see is that, in fact, people supply More labor to compensate for the income shock. So we have these two opposite directions and then we do not Really get a clear prediction there. We also face a challenge, which is an identification Challenge. We're really trying to get to the Gospel effects and we may be seeing selective migration or We may be seeing selective fertility in response or in Anticipation to the shocks and therefore it gets a bit tricky. So what we will try to do in this paper is try to get this Right, the identification challenge right and be able to put Signs, what is positive and negative here and then Conclude what is happening here in this particular setting. So what we do is we exploit a natural experiment, the Peruvian coca industry. As you know, probably coca is The main input for the production of cocaine. And what we do is we're exploiting an abrupt price Change, a price drop that came because of some foreign policy That i will explain later. It represented about a 50% Drop in the price for the producers of the coca leaf. So from one year to the next there were 50% less on farm Prices. And as i will explain later we Will use variation across the space on the cultivation of Coca and the variation over time of this abrupt decline in the Price and several data sets to be able to pin down what is Happening with child mortality and behavioral responses of the household. I will give you a preview of the results now and then i Will go into some of the highlights of the paper. So what we find is that the 50% price drop is equivalent to A 5 to 9% increase in child mortality. That is an important effect. And we're also to pin down where Are these life losses happening. We see there are losses in Inutero, then we see that the babies that survive in Inutero are smaller at birth and then we also see increased Mortality during the first years of life. When we look at the mechanisms we see the households do Reduce their total expenditures so they are not able to Totally cope with the shock and they reduce health Expenditures in about 30%. In addition we see that mothers Increase their labor supply when prices drop. They try to get additional paid income. And we have evidence that they are reducing their time at home And additional evidence of investments in health. So we see for instance there are fewer prenatal checks being Done, there are higher risk deliverers being done at Home instead of at a hospital or a health post. So we find evidence that there are investments in the production Function of health that are diminishing and then we are Consistent with the increase in mortality. We're also able to look at some of these effects and see The differences of what is happening with some districts In Peru, districts are the smallest administrative unit. So if we look at the wealthier districts, these are still Poor but wealthier districts, then we see that there is no Increase in mortality. We find they are able to Smooth their consumption and mothers react differently. They work less, in fact, which is consistent with what we See in the literature in developed country settings. So what we find here also tries to put the science to the Mechanisms i was mentioning but then we are also able to predict Differently the impact in mortality based on the baseline Characteristics of wealth of the districts with this shock. Okay. Now let me have a few words on the Setting and then i will go into the identification and the Main results. Coca as you may know is the main input for the production Of cocaine is only produced in three countries in the world Bolivia, peru and colombia and at the time of this, the Period of analysis for this study peru was the most Important producer in the world. 60% of the production in The world came from peru and bolivia and colombia were Producing the remaining 20% percent. Most of the production goes to the legal market. There's a legal market, too, but most of the production goes Into the legal market. You know, the way this market is arranged is not very Vertically integrated, especially back then. We will have a small scale farmers, usually producing About an hectare of coca, which will then Sundry the coca leaves, sell them to some middleman. This middleman will either sell it to someone else or By himself, he had the technology, will transform Big amounts, big volumes of coca leaves into coca paste, Which is an intermediate product. About an hectare of coca leaves will make one kilo of coca Paste, so that volume is definitely shrinked. And then it will be smuggled through different roads in The amazon jungles and at the time small airplanes will Ferry coca paste from the jungles in peru into colombia And then in colombia, the process will be refined from Coca paste into cocaine and then finally exported to North american markets. What we see here is the map of peru and as you can see There's almost a perfect diagonal line in the country. The districts, the smallest of units that were saying red Are the producers of the coca leaf. And the reason why they are grouped here is because of The agricultural suitability needed to produce the coca Leaf. So what we have here are the andeans mountain and to the Right, the amazons. And it's exactly in those slopes and sunlight and all that That these leaves are able to grow. If you look at the maps of bolivia and colombia, you Will also see a special concentration there. So what we exploit is the shift of the policy from Central american countries to the focus of the Interdictions to the so-called source countries, bolivia, peru And colombia. And in particular a policy called the air bridge denial Which was a policy promoted by foreign governments and That colombia and peru did by forcing the landing of these Small airplanes that were ferrying coca from the Amazon jungles into the colombian jungle. The policy was enacted in the end of 1994 and started in 1995. What we see here is the real price of the coca leaf. And what happened right after 1994 when it's enacted in 1995 is a 50% drop in the price of coca. The anecdotal evidence of the implementation and the Strangency of the policy follows quite nicely in fact the Pattern that we see in the prices. From the point of view of the producer of coca, from the Farmers, this is a negative demand shock. They don't have anyone else to sell their coca to. Their coca leaves to. And in fact this period here is known as the crisis of Coca, the crisis of coca. There are some evidence of farmers leaving their lands And their crops to try to find temporarily better jobs Because the prices were so low, we don't know really, But the prices were so low that they were not covering The costs of producing coca. So what we do for identifying the impacts of this price Shock on mortality is we will be comparing cohorts of Children born in the very first year of the Implementation of the policy, this is in 1995. These are children that would have been or were born in 1995, conceived the year before. We are comparing them to the cohorts immediately before Them, the 1994 and the 1993 cohorts. And we are going to do this comparison continuously across The map of Peru. So basically we are building a difference in Difference strategy comparing cohorts of children over time And across the space depending on how much coca there was Being grown in each district at baseline. So before the implementation of the policy. Now i have to say a few things about how we measure Mortality. But first let me tell you the Main exploratory variable is going to be a measure of the Potential coca revenues in each district. So for each district we will have a measure of how much Coca there was being planted before the policy. And then we will interact that with the price. The price at the national level for each year. And to measure mortality things get a bit trickier. Because as you may imagine there are no really good Admin records for these areas. We shouldn't be expecting to have very good admin records In most developing countries i would say. But then we could use for instance something like the Moravic and health service. The DHS data and try to relocate children back in Time exactly to where they were born. Unfortunately DHS data does not show us where people Were born they just show us the year of birth. So what we do is we look at census data and we are going To use a missing children approach as our preferred Estimation and we are also going to be using the DHS data As i will show you later. But what we do is we use a We will be using a population census that took place 13 Years after the policy in 2007. And with that census we are going to relocate people To the place where they were born at the district level and the Year that they were born. And with that we are going to Create counts of people. We are going to create Cohort sizes at that district year cell. So in 1995 in district x ten people were born there. In this other place nine people were born there. And we are going to infer how much mortality or excess Mortality there was in relationship to this shock by Knowing how many people were missing. So the idea is that if we see that there is a systematic Relationship between the price drop and lower counts of People, lower counts of survivors, then we can Classify that as excess mortality. That also has a couple of advantages and is that Cohort sizes will also give us an idea of Losses which is something we wouldn't see just by Comparing children that were actually born and if they Died later. And it also picks up on Mortality taking place later in life. So dhs data, for instance, will ask for mortality of Children under five. Here we will get mortality Of children even at further point in time if that were To happen. Okay. Well, we include a bunch of controls. As i was saying at the beginning, a key on our Innovation is we're going to be looking at this very narrow Window trying to get away of any variation that happens later. In 1996, a year after the policy, we would think, Okay, families are already doing something. They are averting more pregnancies and so on. Now let me walk through the main tables. I'm going to show you a bunch of tables like this. They look quite similar. We have the main dependent Variable here. I'll be telling you which Are the samples that we're using here. And here we have the coefficient of that main Explanatory variable. And we're going to be focusing On the implied effects. The implied effects are the Effects for the median, sorry, for the mean coca district. And given the 50% price drop that took place in 1995. The first two columns here are focusing on the Cohorts 1993 to 1995. So this is the first cohort Sposed in the two previous ones. And this is a robust check Only at the very narrow window of 1994, 1995. What we see is that cohorts are reduced with the price Drops. So there is excess mortality increasing. The cohorts are 0.4 to 0.5% smaller. This translates, in fact, to a lot of additional mortality If we were to attribute it, for instance, to mortality After birth. So this is about 5 to 9% For the baseline level of child mortality in this setting. We also use the demographic and health surveys. Again, this is not our preferred estimation method because Of the district of birth. It's not available there. But we're able to show some evidence that there are Inuter losses taking place. We do that by adapting This cohort strategy to the probability of seeing a Live birth. We are also able to see that Children after being born alive are also more likely to die in The first couple years of life. We take identification very Seriously. So we try to do a bunch of robust Checks. We ask questions, for instance, Okay, who will be more likely to avert mortality? Do we see any characteristics of mothers, for instance? Are welfare mothers better to avert births? So it's not an impact of the shock. It's an impact of the Behavioral anticipation. We do not find anything like that. We find, for instance, that it's poorer women who are Experienced in most of the losses, which is what one would Predict. We find evidence of the Survival of the advantage of males are more likely to die Both in uterine and alive. We are able to predict that. We do not pick up any pre-trends and the results are Also robust to a bunch of controls. Now let me walk you through the mechanisms. What we have here in the dependent variable is log real Expenditure per capita. Again, we have the Implied effect here. What this is showing us is That total expenditures are falling about 7%, but then Expenditures in food, which are important, and health Expenditures are decreasing quite a bit in about 20% So we know that the first mechanism we were thinking About, the health-promoting goods, are probably decreasing Here. Then we use additional data from Before and after the shock, different household surveys. We look at what is happening on the labor market. We see that first, if we look at only females, Females are more likely to start working for, sorry, To have a paid job in three percentage points Based on, sorry, 54% They're working more hours and they're also more Likely to spend less time on household chores. Of course, we don't see in the survey how much time Are you actually putting to your children, but we do Have a question on household chores. For males, we do not see changes in the extensive margin. We do not see they're more likely to work, but most Of them in this age group were already working. They are not more likely to be working more hours or to change Their household chores. There are also papers on this Shock on child labor and we see those margins also being Moved. And it's that people who were Not working as much for paid work before are usually the Margins that work as insurance here. They're going to be starting to work more to make some More income. If we look at mothers, Mothers of children 0 to 5, relative to other women, it is The case that mothers are also working more. So it is not the childless women. It is mothers in general. They're working more hours And they are having less time on household chores. Finally, we try to look at the health investments. And this is the result probably of the combination of these Two mechanisms, right? Households have less money to Work and they are spending less in health. And household primary caregivers also have less time now Because they are working more for a pay. What we see is that children 0 to 5 years or age are not Necessarily more likely to be sick, but they are much less Likely to be taken to a doctor when they are sick. We also see that pregnancies, when women are pregnant, They are less likely to have at least one prenatal care Appointment. The number of prenatal care Appointments decreases. There is a higher Likeliness of observing a delivery at home, which is Also a higher risk type of delivery for this type of Settings and therefore lower probability of the birth Being assisted by a medical professional. Now to finish, let me show you the tergeneus effect. What i mentioned at the beginning of different predictions Depending on the baseline income of these districts. So here we will classify districts into very high Poverty and very low poverty. First i'm showing you the labor market outcomes of women. Then the real expenditure per capita, so the two main Mechanisms here and then what is happening to the cohort Size, our measure of excess mortality. We see that for the high poverty districts, women are Working more, more hours, less time on household chores. For lower poverty districts, women are working less. That's the prediction of a regular slope of the supply Of labor. We don't find a statistical Effect here, but it is telling that the design is Positive on the household chores. For expenditure, higher poverty districts, we see the Decrease in expenditures overall and also in health. We do not see that for the lower poverty districts, the Sign is negative, they are probably not able to smooth Consumption completely but much better than the high Poverty ones. And then overall, we see the Decreasing cohort size, so the increase in mortality for The high poverty districts and we do not see an Increasing mortality in the low poverty districts, if Anything, the rates of mortality may be decreasing, which Is what we find in countries like the u.s., for instance, when with the fluctuation of unemployment. Okay, so thank you very much. I will conclude there just By saying that what we try to do in this paper is Exploit a careful natural experiment in Peru and we Try to pin down what are the mechanisms at play That explain why our cultural commodity price shock may Result in the increase of child mortality. Thank you very much.