 Moj tudi količe je Flore Gubert in Sorana Toma, tudi v Franciju. Tako, če smo prezentovali, je to več preliminarne stave. V komentari so nekaj, ker smo počutili naši argument in analizirati. Tako, če smo izvistimati v tem paperu? Tukaj, smo počutili, da smo izvistimati vsečenje vsečenje vsečenje vsečenje in sošiljej kapitali. Načo, ki je zvednji sečenje informacije, zvokaj smo pošli na Senegalizm vsečenje in izgledaj se način o data, ki smo vsečenje vsečenje. Nasložili smo, da se pri vsečenje Senegalizm vsečenje nekaj sečenje vsečenje in sečenje, in, drug observations of both network access and network IUs on labor market outcomes. So what determines the quality of their job, both upon arrival and later on in the migration trajectory. The peculiarity of our data sets is that it's a multi cited survey. So we have within our data, observation of Senegalese migrants, both in 중요한 kontekst and in African context. And we are speaking about migrants in France, Italy and Mauritania. Unfortunately, we still haven't exploited the fourth sub-sample that is composed by Senegalese migrants in Ivory coast that is one of the way forward part of the story. One of the questions that we will try to explore is how does the context of the reception shape the role of networks in labour market attainment. Why we think that this is relevant? The mayor of the del, before, towards the seasonal introvert, is the veryw scaffibo. He spoke at theize, which is important, as it speak for the individual everywhere, whether there is a direct current in their購oval. So, first because migrants' labor market attainment and trajectories are a major concern in the policy debate. On one hand, they proved to be ta z vsočiko tegoje zvaljuj Dra chemije, in poten bi bom vsoči, da zvoučalimi informacijne netvore na laborstvom prijovali, ker je tudi na neko jezpravnjo nječvom iz Technico. Po druhovu, zelo toh, da in je zvečenja v Afrikanih način neUSTW doživljenj, da je se in rukovil, in pretekl poštati, da se zelo, da drugi motiv, v tega nasčenosti, je da izvrduzila v tega včas. Oželjno, in včasno bilo zelo, da so tem počke, tega je vse vse zelo v del, lahko kako je počka nekaj vsega vsošanja vzelo in tudi nekaj vsega nega nekaj vsega, nekaj nekaj vsega. Daj vsega počka se je zelo vsega vsega vsega vsega, ig nuevo in stvoでき sove in Excellence, the main focus of this literature is to identify, I would say it may be not over summarizing, but it is to identify which kind of social capital help and which kind of social capital do not help. So the distinction into bridging and bonding is the main point, I would say of this literature. This will not turn out as a very relevant point in our data. We will try to develop others, not saying that we is not relevant at all. But this will not turn out as main, as a crucial point up to now in our analysis. The data that we use, is the database, that is collected for the mid-dust survey that was conducted in 2009 among Senegalese migrants in France, Italy, Mauritania and Côte d'Ivoire. da ovo se razli, zdaj smo najbiha tudi predse data seti, zato sem zelo se včetnimo na 4. Se je da način včetnimo nekaj nekaj 900 obzervacija. Taj odse svet je zelo dozeni v modulji po vsega post-migrače in na kredne neselje in vsi frane in familije. To je neselje in vsega in vsega in vsega in in in in in. In je vsega in in in in in in in in in. Poj developed even in Fieri in in in Italy There are quite like say old data I mean we are used to work with this data That dates like ten year spec But a lot of thing occur in between So this is something that can be described So how far our emphasize can be useful to day Meaning that, in between there has been the economic crisis and there has been the refugee crisis And the political spotlight on migration Pozvo, da sem tukaj oznatil, da se je načo način. Tako pozvo, četko. Prosti, če se prihvozago na vnikosti skupci. Tukaj nekaj, da se prihvozago, je to izbog moritani in svojšeljšeljšeljšelj. Da to ni pozvodnji vstav, da se izbog moritani tukaj to izbog moritani, da način svojšeljšeljšeljšeljšeljšeljšeljšeljšeljšeljšelj. in zelo je však jezda v Italiji in Franske, as we will see more deeply in our analysis. But overall we can say that the migration to Mauritania with respect to European countries displays a higher share of women, younger, people migrated earlier, and on average less educated, just to have a snapshot of what we are, because of course we are speaking of very heterogeneous people moving. So what do we investigate and how? So as I announced we are doing an analysis in two steps, where the first one is to identify who are the people who rely on networks to find a job, so where the dependent variable is network use, so the job search channel, and the second step is how the different networks and job search processes affect job characteristics, so where the dependent variable is labor market attainment. For both steps we have measures of both first and last jobs. Methodological problems are huge when trying to identify the relationship between social capital and labor market outcome because everything is intertwined. So we have two main problems. One is reverse causality. You can get social capital from your employment, and to solve this we use time. So we have measures of social networks before the first job and before the last job. So in terms of reverse causality we try to use time to identify the good direction. But the main problem is endogeneity, so the idea that there can be some other factors that affect both network access and labor market outcome and both network use and labor market outcome. We will use instrumental variable techniques to try to address this, and this is one of the main points of our conclusion, is exactly that we are not exiting the endogeneity problem, and this we will see later on. So, which are our dependent variables? I am going through the variables that we use. So in our first step the dependent variable is the variable derived from a direct question. So how did you find your first and how did you find your current job? And so that we know if this is a formal channel or an informal channel, and within informally if it's through family networks or friends network. Labor market outcome is mainly measured by the ESAE score that is derived mainly from the work of Gazenbaum, Trayman and quarters done in the 90s, that is the idea of associating a score to the ESAE score classification, the international classification of occupations that tries to capture the average income and average education level in each occupational sector. It has a lot of limitations as an index because it's based on weighted average of education and income among male population in 16 countries. So really we are quite aware of the limitation of this index and comments are more than when come on how to deal with it. We try to couple this with other indicators. One is the probability of being employed at survey time, and lastly, we use categorical variables. So unskilled manual, skilled manual, unskilled manual, unskilled non-manual skilled and self-employed to have a more categorical distinction of employment. Social capital variables. We have, of course, the use of social capital in this is first a dependent variable and then an independent variable. And in turn, but it's always the same one. The access to social capital is measured as a family network, both at arrival and before the current job, size of network known before migration and before the current job, and whether there are some natives in the network, meaning French in France, Italians in Italy, and Mauritaniians in Mauritania. Again, so this is a snapshot of the descriptive statistics of social capital variables. Again, as we see, so there is overall in Mauritania greater use and access of networks, but this doesn't capture all because, for example, the network use to find the, sorry, to find the first job. So the share of people using network to find the first job is the highest is in Italy. So, I mean, this is to go back to the fact that we have to explore more also this aspect. We control for a number of intuitive variables, so education, education in the host country, location of the family in back home in Senegal, characteristic of migrations and especially whether the person entered undocumented. And the sex dummy and the destination countries dummy. So, first block of analysis is who uses networks to find a job and we have both, the first table is upon, for the job obtained upon arrival and the second one is for the current job. So, first of all, what we see is that here is that both in Mauritania and in Italy, it seems that, so the model, sorry, is a multinomial logic model where the reference category is formal channel and so the coefficients tells us whether both the family and the friends. So, who uses the family and the friends channel more with respect to the formal channel. So, we have a higher use of both informal channels, both in Mauritania and in Italy. So, overall, what we can say is that who uses network more? Women, both family and friends network. In undocumented migrants, especially on what concerns the family network, and the people arrived young, age at arrival, especially on what concerns the family network. So, these are the major results of who uses network. It seems that network access has a positive impact on network use. If we look at the number of the probability of using the family channel, it has a positive correlation with the availability of family members at destination upon arrival. And there seems that there is a degree of substituability between family and friends network. So, having relatives in the destination countries lowers the probability of using friends network. But still, there is a correlation between network access and network use. Basically, at surveys for the last job, it's not exactly a survey time, but the determinants of network use for the last job are very similar. Undocumented migrants display a different, so before they had a higher probability of using a friends network, now this doesn't hold anymore, but they have a lower probability of using the family network. What is interesting here is that education starts to play a role. While for the first job, education didn't play a role in the probability of using networks, here it does negatively. So, more educated people tend to use less to rely less on informal channels. So, main findings, just to summarize what I just said, is that, initially, youth, women and undocumented migrants have higher probability to find job through informal channel. These results holds for the current job, but not for the undocumented, for people entered undocumented. Education lowers the probability of finding a job through informal channels, but not for the first employment, only subsequently. And there is a correlation between family network access and probability of finding a job through informal channels. So, social ties seems to play a role in job search method, and they kind of substitute each other, family and friends channel. Second block of analysis. So, how does all of this impact the employment status? So, this first table displays some OLS, and finally, an IV model to try to account for endogeneity of the occupational status upon arrival. So, the dependent variables is the EZ code that I presented just before. I don't show some controls, but in the end, I will try to comment a little bit on them. So, what we have is that, apparently, so in the first column, what we see is that the size of social network, so, one of our measures of network access, seem to play a negative role on labor market attainment. I don't show this, but this is also holds, if we look to bridging social capital. So, to the number, not the total number of people in the network, but to the number of natives. So, the French in France, Italians in Italy and Mauritania in Mauritania. Still, the coefficient seems to be negative. If we interact, so, this result is also true for network use. So, I don't know if I have a pointer. What we see is that also, both using friends and family channels seem to have a negative effect. Okay, try to speed up. From this table, there are two main things that are relevant to our analysis. One is that these results hide differences in the different countries. So, it seems that friends network has a negative impact in Italy and Mauritania, while the family network has a negative impact in France. So, there are differences of the role of networks as we, this was one of our questions, but probably, second main result is that this is not robust to instrumentation. So, what we do in the last column is to instrument the network use, so, the use of network of informal channels with its predicted probabilities after a multinomial logic. And this, the negative effect of network use disappears. And also, I don't show it, but the effect of network access. So, this is the main thing that I would say should be retained is also in the following slides. So, we tend to find an apparent negative effect of both network use and network access that is not robust to our instrumentation techniques. So, here what we, we display is the probability of being employed, so a probit model and an IV probit. In the probit model, what we find is the size of social network seems to play a negative role, especially driven by the Mauritanian and the Italian sub-samples, but this is not robust to instrumentation. Occupational status for the last job. Again, we find that, let's look to the second column. I skip something if not I will totally run out of time. We find the negative effect of having found the job through the network that is interesting because it's totally driven by the French observation, while in Italy the coefficient displays a positive sign. So, this is another thing to be explored more, but again, this is not robust to instrumentation. I can go back to the instrumentation, so what we use as an instrument in the different cases, but I don't stay on this now not to run out of time. The last slide here, what we use is the categorical distinction of occupation. So, try to see which is the probability to shift from the unskilled manual to the unskilled manual, to the unskilled non-manual, to the skilled and to the self-employed. There is no endogeneity control here, so we cannot claim any causality link. What it seems is that the network seems to channel people towards the self-employment. And this is, I wanted to report this because it seems consistent with other recent research. So, even if we cannot, it's not, we didn't do any robustness check. So, two slides of just of a conclusion. So, the main findings of this part of the analysis is that social networks seems to play different roles in different contexts, and there is not a sharp divide Europe-Africa, but also the European sample displays a great extent of heterogeneity. There is an apparent negative effect of both network access and network use, but it doesn't show to be robust to instrumentation. So, we cannot claim causality in this negative effect, in this negative coefficients. I didn't speak about controls. Just say some words. Control play in the expected way. Education and diploma destination have positive effect on labor market outcome, while being undocumented upon arrival has a negative and long-lasting effect. Not on the probability of being employed, but on the quality of employment. Ethnicity variables are often significant. We are wondering what does it mean? Does it capture, like a rural urban divider, does it capture network, some network that is not captured by the other variables? So, this is the question we are left in mind. So, overall, networks, what we would retain from this work is that networks are highly endogenous, and this should put into perspective a big bulk of pessimistic literature on network and labor market attainment. And it is really necessary to look at who uses networks. So, in our case, especially women, youth undocumented, less educated. That means that we could be interpreted possibly, and this is open to debate, if there are more vulnerable categories who resort to networks, but probably in the absence of networks that they would be worse off. And so, this is the main message that I think can be drawn from the analysis up to now. To be done a lot, I already anticipated something, explore more the differences across countries besides the network interaction, introduce the devorian sample, do an analysis on wages that is the only continuous variable that we have on labor market attainment, and to understand more what this ethnic dum is represent. Trenk ju, I hope not to be too long.