 So thank you very much for, can you hear me? Yeah, okay. Yeah, so thanks Mohamed and Victor for inviting me and for letting me present this paper. Actually, this is a project that I have with two colleagues here in Bocconi, Adwid Alfani, who's an economic historian as I am, and Marco Bonetti, who's a bias statistician here in my department. So we're working on this project since I would say one year or so and we are about to finish the first draft. So any comment is really very well welcomed. So, wait, okay. So basically this project, this particular paper that I'm presenting today is part of a bigger project that I have since a couple of years, I would say. And I'm trying to collect for Italian city-state and more generally for Italian north and central Italian cities, data on mortality and especially data on mortality related with the waves, dramatic wave of plagues that started in Western Europe with the black fat in 3047-52 and ended in the mid 17th century. And what I'm trying to work on with this big project and what I'm trying to show you also today is an analysis of the factors that affected the individual risk of death during this major pre-industrial plague. So prevented that we all think that these waves of plagues were to some extent exogenous but the mortality that the chance of surviving of the individuals was related with their social demographic or socioeconomic status. And that's what I'm trying to reconstruct. And what I'm showing to you today is the first step of this analysis that we're doing focusing on a particular case which is the city of Carmaniola. You see it located here in Piedmont for my kingdom, the subaudian state. It was a city of 7,600 inhabitants close to Turin, Turin is here and Turin was 25,000 inhabitants. So it is about one third of Turin, capital of the state in the 1630. What we investigate is a micro demographic database on the city of Carmaniola during the 1630 plague. And what we do is to estimate and reconstruct, sorry, reconstruct theoretically and estimate the risk of contagion and the probability of the individuals to be interned in the plague ward, which I will call Lazareto from now on. That was the name of the ad hoc hospital for the internment and the cure of the, those who got, were infected by the way. And also we will try to reconstruct what was the impact of the health status of hospitalization and the interaction with the individual characteristics on the individuals on each patient on their chance of survival. So more roughly speaking, whether those Lazareto were effective in curing the sick or not. And yeah, I'm gonna go very fast through the literature slide because the literature slide is always very, very boring. What I want you to keep in mind is that we are aware that there is a literature on that. Many, I mean, Guido and I, we have to different extent contributed to this literature. The literature analyzes many factors or many aspects of the impact of pre-industrial plagues. When this literature goes to analyze the micro, the granular level, the micro level of the impact of those plagues, the literature has either tried to reconstruct the patterns of the transmission of the disease or when they have tried to reconstruct how individual factors affected the impact of the disease, they mainly focus on age and gender. Now, what we do here is we try to explore two dimensions that have been neglected also because of data problem in the past. One is how the socioeconomic status affect the chance of survival. And I will try to, I will explain you later what is socioeconomic status in this case and also the impact of the hospitalization. Now, we focus on one plague in particular, which is the last big plague in northern central Italy is the plague of the 1629, 1630. The plague actually originated in 1623 in northern Europe, low countries, northern France and England. And then also thanks to the 30 years war, it's 1618, 1648, he spread in continental Europe, first in Bavaria and Switzerland in 1628, 1929 and then he arrived in northern central Italy. And the overall mortality rate in northern central Italy has been estimated around 30, 35% of population with two millions estimated victims. And we study, as I said before, in particular, how the plague hit the city of Carmaniola. As I mentioned before, the city had about 7,600 inhabitants in 1630, just right before the plague. The city was a wallet city under the subodian state, but with its own government. And the government had a lot of autonomy in dealing with daily organization of the city. Among those prerogatives, the government was governing the cities in times of plague. And in fact, in January, we have seen in the proceedings of the government that the government was already alerted of the possible arrival of the plague from close community who had been infected. And immediately in that month, the government appointed two officials and then the officials became more because this became a semi-permanent health board in the city who were just appointed to try to prevent the contagion. And what did they do? They basically implemented measures that were very well known in the Italian cities at that time. They imposed travel bans to people traveling from non-infected regions or they tried to set up sanitary cordons to avoid people from infected regions or goods from commodities from infected regions to enter into the cities. However, in April, we know that first cases were reported in the city probably also because of the arrival of troops, military troops from Central Europe in the Piedmont region. And in the next month in May, a Lazareto was set up beyond the city walls. This Lazareto will be first a temporary structure made of huts far from the city. And then it will become so big that the city will expropriate a monastery and will put the patients permanently for one year and a half into this former monastery. Now, just to give you an idea of the diffusion of the contagion during the year and a half in which the two years actually which the plague spread through the city. I show you here what is the best measure but still it is a proxy measure of the spread of the plague to the city. So this is the number of admission in the Lazareto. You will understand in a second why this number is a proxy of the plague spread. We don't have better measures. So we don't have any number of that or infected through time. We only know that starting from the first week in which the Lazareto was functioning which is the first week of May 1630. You have a first wave of high number of admission in the Lazareto which is basically dated around August 1630. And during this period, the plague is reported to be a bubonic plague. So a boobos and antisemic death. And then you have a second wave which is around 1630. And here the plague is spreading through fall and it is reported to be a pneumonic plague. Then the plague and the admissions, let's say to the Lazareto declines and you have a last recrudescence in May 1631. Now this, as I said, is the best measure we can reconstruct of the plague spread in the city over time. So just for you to understand better why this is the best measure, I wanna explain with some detail what is the data set that we have. So the data set was firstly put together by Mario Abrate, an economic historian of Piedmont in 1972 who published this paper, this book in which he provided an extensive analysis of the 1630 plague in Carmaniola. We have recovered all the archival sources that Abrate has used and we have checked all the archival sources reconstructing the work of Abrate. And which are those archival sources? First of all, there are two documents that were compiled by the government of the city during and after the plague outbreak. One was the list of all the patients that were admitted to the Lazareto in the year that I've shown you before. So the name of the list is the individuo estatic condotti al Lazareto. So the individuals who were brought to the Lazareto. And in this list, you have the individual level information group at the household level. So if Mattia Fogazzato had been interned in the Lazareto, you know that Mattia Fogazzato belonged to a certain household. And of those individuals, we know the date of admission, the length of stay, the cost of hospitalization and the health status at discharge whether the individuals was alive or he died in the Lazareto. Now, let me anticipate that the cost of hospitalization is basically redundant information. We have discovered that basically there was a fixed amount of money that was spent for each patient every day. This amount of money was unconditional on any status, any characteristics of the individual. And so it was just correlated with the number of days that the individuals were staying in the Lazareto. Another list that we use is the list of all the plague victims in the city. This was compiled after the plague ended and it provides the full list of name and surnames of deaths at home. Now, unfortunately, and this is the first big limitation that we have from this data set, there is no date of death. So we cannot reconstruct in the city any time to event analysis. We can only know that these individuals given the list of name and surnames belonged to probably eventually some households that was interned from the list of admission to the Lazareto. Finally, we have also reconstructed the last part of the work made by Abrade which is to connect or match the individuals in these two lists from the last complete tax sensors for the city of Carmaniola. This was compiled in 6021 and it reported at the household level, the neighborhood of residents of the individuals, the age, the gender of each household members, the occupation and the fiscal group of the household head. Now, this information for us is crucial because what we want to know is how the social economic status affects the probability of entering into the Lazareto and your chance of survival once you are in the Lazareto. Fiscal groups are going from one to five and the richest individuals are in fiscal group one, the poorest are in fiscal group five, the fiscal group six is all the miserable in the city, the very poor that were exempted from taxation. And just to give you a sense of the city, this is the city of Carmaniola, which is the proper city. It's not a rural village, right? You see the wallet boundaries of the city and these are the neighborhoods of the city. Now, what we've done, we have double checked the Abrade's work. Here is just one page or I would say the first page of the list of those admitted to the Lazareto names, surnames, the number of days they stayed in the Lazareto. This is the date of admission. This is another date of admission. And here you have the status at exit. This cross means that they died. This line means that they survived. What are the limitations of the data set? These are important because these dictate basically our statistical work. First of all, we are able to have age, gender and socioeconomic status information only for the individuals that come from the intersection between the three lists. So what we're missing are the immigrants and the children below nine years old because they're not in the 60-21 census. There is no date of that at home. So no time to event analysis is possible. And we have no information of those who caught the disease, remain at home and survive. So they recover it from the disease. We only have information of those who died at home and those who enter into the Lazareto either died or not. Any measure of mortality case fatality is made on a measure of projected total population which is based on the natural marriage and fertility rate that we can reconstruct from the surviving parishes register from one neighborhood of Carmanhola. So we have estimated the structural population at the 1630 Carmanhola just right before the plague by using, by projecting the 1621 population using the natural fertility in marriage and mortality rates from that parish. Overall, and just to give you a sense of what is the situation, what is the outcome of the plague in general, we have about 25 mortality rates in the city, including the victims at home and the victim in the Lazareto. And that compares pretty well with respect to rest of Northern Central Italy, I would say. And we also have a mortality rate within the Lazareto. So this is only among those who were admitted, which also compare pretty well. 28% versus the other measures that have been estimated for other plague words in other city 30 years before. Now, don't compare this measure with this measure concluding that mortality rate was higher in the Lazareto because of course the populations are different between the two groups. Now, recap our research question. We want to understand the socioeconomic factors and the impact of socioeconomic factors on the individual's chance of survival. But given the limitations of the data, what we do is dictated by those problem and gaps in the data, the absence of the date of death at home and the absence of the complete picture of the population in 1630. So what we do and what I will show you in the next 15 minutes, if I can, is the two-fold analysis that we're doing. First, the theoretical model, we construct a theoretical model of the process of admission to the Lazareto and we compare prediction versus realization. And we infer conclusions from the comparison. And then we just, within the population in the Lazareto instead we are able to do a logistic analysis of probability of being discharged alive given the individual demographic and socioeconomic factors. Now, regarding the process of admission to the Lazareto, what we know from the historical accounts and this comes from the government provisions that we have intensively checked in the last year. The city officials were appointed, a group of city officials, usually there were two for each neighborhood and they were regularly visiting the citizen's house and checking their health status. How regularly, every two days, they report conditions of each household. And when a member was found dead by plague, ill or was suspected of being killed, the whole household had to be sent to the Lazareto. So what we conclude is that in principle, what we assume, if the whole household has to be interned when at least one member is infected, the smaller the size of the household, the lower should be the probability of an individual of being interned. To this assumption, which to us seems to be pretty reasonable also because it's historically grounded, we add three other assumptions that to some extent we admit are not that reasonable. One is that the size of the household during the plague outbreak remain constant. The plague outbreak lasted one year, not that unreasonable. The event of the first infection of each individual in the same house was independent, meaning that if I get infected, my wife get infected with the same probability doesn't depend on my probability of being infected. And each individual has a fixed probability P greater than zero of becoming infected in a household with no infected individual. Now, this probability being fixed, this is the most unreasonable assumption that we have to make, means that this probability is independent on individual characteristics of the individual. This member of the household, but not on the characteristics of the household, you will see. Then we have assumed that the size of the members of the household being N equal to the household head plus all the other members, follow up was on distribution being discount data with the expected value being lambda greater than zero. And if P greater than zero is the probability of an individual to be infected, then the probability that all the individuals in a household of size N greater than one are spared by the plague is just one minus P power N. Now, what is the probability that a randomly selected household is spared? It is the probability that it is that the household is spared conditional on the size, the expected size of the household. And these, which is not exactly my cup of tea, but I can guarantee to you that this return, this formula, which just show you that the probability of a household to be spared decreases has the expected size of the household grows. The larger is the household size, the lower is the probability that that household will be spared by the plague. And so the higher is the probability that the household will enter into the lazareto. Now, okay, this is nice, but what can we do with this model? First of all, we can compute theoretically the probability of being admitted as function of any P and the expected size of the household. Or considering any subpopulation, this is important, like for example, a neighborhood or a fiscal group, we can observe that the subpopulation size is equal to the expected size of the household times the number of households. Now, if the population is fixed, then the number of households is decreasing in the size of the households. So the larger the households are, the smaller is the number of households in that subpopulation. What does it mean? It means that the number of households that will enter the lazareto in the subpopulation will decrease in the size of the household. So the larger the households, the less households enter into the lazareto in that subpopulation, but being that household small number in the subpopulation, the proportion of households in that subpopulation will increase in the size of the households. So basically speaking, what you can do is you can estimate the proportion of households admitted in a subpopulation as a function of the size of the households. And why is that important? This is interesting because we can take the number from the 60, 20 census and estimate the actual average household size by neighborhood or by fiscal group. And then we can use our model to estimate the expected proportion of admitted households in subpopulation and compare with the proportions that we observe in our data sets. And this is the main result. So fiscal groups are shown here, six, five, four, one, two, three. So the larger the fiscal group, the larger is the size of the household. And I think Greg Clark will be happy here. And then we have clustered those fiscal groups after extensive selection model across a combination of neighborhoods. I don't have time to explain this, but believe me. So basically the red points here are the predicted proportion of households per group of fiscal group and neighborhoods, all right? Given the mean size computed on the real data. And the blue point are what we observe in the data set. And the first conclusion we have from these analysis is that all the sensitive group entered into the Lazzaretto, according to our model predicts very well how they entered into the Lazzaretto, except the poorest group, which was admitted less than what we would expect. The main conclusion we draw from that is that since the low means size of the households in the poorest group, it was less worthwhile to intern all the members of the Lazzaretto once one members were found healed or died. So the poorest were admitted less than what we would expect. Another thing we can do is to estimate the expected average size of the admitted household. This also comes from the model. And we can compare the predicted household size admitted in the Lazzaretto versus the one we observe. And we can compute the difference or the percentage difference and see that first, our model overestimate the size of the households that were admitted. Obviously our model predicts that all the members were admitted. But the difference between those admitted and those who were left at home is different across fiscal groups. And for instance, the poorest one were interned more or more completely than the richest one. The richest one were resisting to the internment in the Lazzaretto. We have another piece of evidence which showed the average number of days between the first and the last entrance by household members. And you see that while the poor households were actually entering almost altogether into the Lazzaretto once the first member was entering, in the richest household members waited more to enter. So this piece of evidence shows the second part of the story. If we reproduce the process of admission, we see that first the richest entered given to what we expect, the poor were usually left behind. But within the households, the poor when they were entering into the Lazzaretto were entering altogether. While the richest were trying to avoid to enter into the Lazzaretto. Now, I think I only have three minutes, right? Yeah, I would say yes, three to, maybe like up to six minutes or so. Yeah. But so I have to go fast in the second part, which is, well, so what happened when they entered, right? So what we know from the history is that once they were in, the individuals actually received a very decent treatment. Medicines, heating, vitals, their clothes and mattresses were legually clean. They were medicated if ill and they actually received enough space also to isolate the infected from those who were just interned to protect them from infection. There is a part of our work which is concerning with the fact that we want to, so since we basically, we want to estimate a logistic model regression of the probability of being discharged conditionally on the observed demo and socioeconomic individual characteristic, but also the health condition at entrance. I hope you got from my description before the fact that once one individual was entering into the Lazzaretto, all the other members with some distinction across groups were admitted to the Lazzaretto. So we actually not able to observe what is the individual's condition at entrance. And here we have adopted another method and I don't have time to explain it, but basically what we have done is to observe the distribution of the length of stay of the two groups of individuals. So the only things that we observe, the outcome, those who died in the Lazzaretto and those who were discharged, and we were able to make some hypothesis, also given our knowledge of the history progression of the disease per individual on their health at entrance. So basically what we were able to do is to distinguish whether they entered sick or not, whether they caught the disease within the Lazzaretto or whether they caught the disease outside the Lazzaretto. So for example, you have here a huge number of individuals who were discharged on the 30th day, which is the last day of quarantine. For all of those individuals, you can assume that they have been entered healthy, they made their whole quarantine within the Lazzaretto and then they were discharged. Now, I only have time to show you the results of our logistic regression. Remember, we test the probability of being discharged alive against the condition at entrance and another certain, a long set of individual characteristics. Gender, the physical group, geographic area, age, and some interaction, those were more, which were more significant. Now, the main results here are, I would say, three. First of all, the chance of survival decreases with age. So the older you were, the higher was, the lower was the probability of being discharged alive. The second result is that physical groups also matter. And the poorest one, who are the reference category here, are those who have the highest probability of being discharged alive. And finally, gender doesn't matter per se, but it matters when it is interacted with the condition at entrance. And the healthy male have a highest probability of surviving than the healthy females. What is the interpretation here is that females were entering healthy. If healthy, they were asked to take care of the unhealthy and they eventually caught the disease. So, and then the probability of being discharged alive declined. So, conclusion, sorry, from the model of process of admission to the lateral, we have shown that the larger and richest households because they're, so the richer households, because they were larger, faced a higher theoretical risk of contagion and internment. And in fact, in proportion with respect to the group, they were interned more. But the richer households, the individuals within the richer households were likely to oppose resistance to the internment. Once we're in, the individuals from very poor households had higher probability of survival. Why? Well, first because the richer household entered in worse condition, exactly precisely because of the fact that they oppose resistance when they were interned, they were sick. And the other reason is that probably, we are pretty confident in that the poor individuals enjoyed better living conditions than at home. And then they could actually increase their chance of survival or maybe isolate themselves from contagion. And that's it. Thank you very much.