 The topic of this talk, as you can see on this slide, is about diaspora externalities, which is something I've been working on in the last three, four years. And Finn asked me to try to focus wherever possible on South-South migration. So you could have diasporas anywhere. It's a conference on developments we're interested in. All of this lecture is about migration and development, the effect of diasporas and migration on the development of developing countries. And where possible, I will try to emphasize what is specific, if anything, about South-South migration in this relationship. Unfortunately, as you know, the data constraints prevent from sometimes from having this specific loop. But wherever possible, I will try to have it. So it's an overview. It's not a research paper. It's an overview of recent research with an overrepresentation of my own work. I think would be the right qualification for this lecture. So let me first explain the title, the diaspora externality. Why do I call it this way? Because as we know, as economists, externalities are kind of unintended consequences, or at least consequences, on others' well-being. When someone makes a decision, a consumption decision, a production decision, a migration decision, this is based on cost and benefit analysis for that person on good and for the loved ones and the family and so on. But there may be also consequences of others which are not mediated through the market. In this case, economists call this an externality. And I think this really what diasporas are about for home countries. These people make decisions to move, not considering whether this is good or bad for the country. But once they do move and they form these diaspora networks, these migration networks, they can affect the home country in many ways, which are externalities. So I will review the different type of externalities. Migration externalities can be good or bad as any externalities. And for example, we have examples of negative migration externalities in the literature. For example, there is a big literature in political science emphasizing that diasporas often take part in fuel-saving civil conflict. That would be a negative externality. In economics, the Dutch disease effect of remittances has been studied, and it's a negative externality. And maybe the most famous one is the brain drain. Externality people live, but by doing so, the stock of human capital in the home country is depleted, and that is bad for anyone left. We also have examples of positive externalities. You can flip the brain drain story into a brain gain story. If you factor in the incentives to educate, then the brain drain becomes a brain gain. In that case, that's a positive externality. But I think the most relevant externalities are linked to the role of diaspora and migration networks. So what I will split the presentation into the contribution in my view of diasporas into favoring the economic integration of the home country into the global economy. So that will be part one on diasporas and economic integration to the global economy, which has been studied for quite some time, but I think there is new interesting pieces of evidence. But also, more recent dimension, I think is the cultural dimension, and I will try to argue that diaspora networks can also affect the cultural integration in a very broad sense, and I'll try to be more precise when I get here. So the focus on the talk, as I said, will be on development. I will try to focus wherever possible on South-South migration. It's not always possible, as I said, because of data constraints. And also, I'm an economist, so I will over-represent economics literature and my own work as part of it. Okay, so the roadmap for the talk is, as I said, two parts. The economic dimensions of globalization and the cultural dimensions of globalization. Okay, so for the economic dimension, quite traditionally, I will start with trade. I will move to FDI and other financial flows, and then move to knowledge and technology diffusion. I think diasporas play a very instrumental role here in favoring these other dimensions of globalization. So it's not just migration is part of globalization, it also works as something which is increasing the connection to the world through other dimensions of globalization, like trade, investment flows, and knowledge. And then I will move to culture, emphasizing mostly the political dimension of culture, what is called political remittances, preferences for democracy, voting, and these types of things, how are they affected by migration. I will then move to fertility and other dimensions, other aspects of what anthropologists have called social remittances. And then I will move to an exercise which is asking whether migration makes home and host country culturally closer or more different, okay? And then conclude. So let me start with economic integration, diasporas and the global economy, and quite naturally with trade. It has been known for quite some time that there is what economists call the trade creating effect of migration. It's not that obvious when you think in the standard neoclassical framework, migration and trade are seen as substitutes. Either you move goods or you move people, but if you won't move one of them, you need not to move the other ones. So they tend to be substitutes. In reality, we see this is not the case that when you have more migrants moving from country A to country B, you also have more trade between country A and country B. And the reason is two effects or two channels for this effect. One is information. Trade is not free, trade is costly. You have transaction cost and the fact that you have migrants between the two countries reduce this transaction cost and therefore favor trade. You also have a preference effect, but the economists are more interested in the information effect, okay? The networks in the sense of seeing migrants as vectors of the information allowing to reduce transaction cost across countries and therefore favoring trade. And the evidence for this information effect is quite large. This started by with work, again in economics. I know there's been earlier more qualitative studies outside of economics, but for economics it started with work by Jim Rauch studying ethnic Chinese networks and showing that when you have Chinese diasporas this is improving trade mostly between countries with a Chinese diaspora. So not between China and the countries where the Chinese are, but say between Thailand and the US because you have Chinese in both countries, okay? And the argument was that they reduce information asymmetry, they improve trust and this is evidenced by the fact that this beneficial effect of migration shows up only for trade in heterogeneous goods which are goods for which there is an information problem but not in trade for homogenous goods such as commodities, okay? So that was the initial work. Then there has been also work looking at different countries vis-a-vis the rest of the world in for the US, for Canada or cross-country studies showing consistently that when you have a bigger migration stock from country A to country B this is good for exports from country B to country A, okay? So this has been shown as I said in case studies bilateral between US or Canada on the one hand and the rest of the world or in a fully cross-country bilateral setting. However, to quote Gordon Hansen these studies are not fully satisfying for economists who are obsessed with endogeneity with the direction of causality, with omitted variable explaining the joint evolution of trade and migration and so to quote Gordon Hansen it's difficult to draw causal inferences from these results since immigration may be correlated with unobserved factors that affect trade such as trading partner cultural similarity or bilateral economic policies, okay? So what we are really after is something more experimental so we cannot artificially create experiments so we use experiments provided by history and that's called natural experiments. Ideally we would like an experiment where you take immigrants from one country unexpectedly you throw them into that country and you distribute them spatially in a way which is exogenous in a sense that it is independent from the future trade potential with their home country. You forbid trade for some time and 20 years later you open the relationship for trade and you see what happens and you check whether the places that happen to receive a bigger batch of people turn out trading more with their origin country, okay? So if you try to think of an experiment like that and we ask God to carefully design such an experiment he did that for us and that's a paper by Chris Parsons and Pierre-Louis Vesina which is first coming in Economic Journal and I'm a big fan of this paper so I'm advertising it and it's called Migrant Networks and Trade The Vietnamese Boat People as a Natural Experiment. This is exploiting the inflow of Vietnamese boat people to the US after the end of the Vietnam War in the late 70s on trade with Vietnam after the trade embargo was lifted in the mid 90s so really 20 years later. So this graph tells you all about the paper. Yeah, the pointer is working so you see here the inflow of Vietnamese immigrants to the US. You see the two first spikes are really the first refugee wave. Those who left right after the fall of Saigon in 75 and those who really took on the boats in 79 during the Vietnam-China War and these are the numbers that came to the US, okay? And afterwards in the late 80s and early 90s you also have another second wave of people and also the early wave could move after some time. So what they do is to take these guys as an instrument for the size of the migration networks of Vietnamese in different US states 20 years later, okay? And it's a very good predictor and these guys were allocated exogenously because there was a special government program to do this in this way and also because these people were sponsored by US families which are scattered in the US somehow exogenously independently of future trade potential with Vietnam, okay? And here you have trade with Vietnam. It's zero until 90, the early 90s. Actually, there was trade embargo on Vietnam until 94, I think and after 94 you see that the trade is speaking up. Okay, so what they do is to ask whether states, US states that received more refugees, Vietnamese refugees in the late 70s happened to trade more with Vietnam 20 years later, okay? And indeed, this is what they find and this is the result of the regression and you see a positive slope and the slope is the elasticity of trade to migration and it's about 15%, okay? Which is slightly higher than the non-experimental studies I mentioned before, which got an elasticity of around 10%. So if you have 10% more immigrants, you have 1% more exports, okay? So, and you see the table with this regression and this is the IV, so using the stock of migrants predicted by the early distribution of refugees, okay? And the result is interesting, you see at the extensive margin and the intensive margin, so, and they do also interesting placebo test and so on, so I cannot do, it's not a presentation about their paper, but just to say that I see it as the best paper in my view on this topic, okay? Let me move to another economic dimension, which is FDI and other financial flows. You've seen this picture maybe 20 or 50 times, it's about the rise of FDI from OECD to developing countries. The interesting thing is that it's not going everywhere, it's going mostly to, one of the two is Asia and the other one is Latin America, while Africa is the blue one here, not much picking up FDI and this is the Middle East not picking up FDI from the OECD countries either. Okay, so globalization is about the flow of investment. What are the connections with the flow of people? We can use the same information argument, migrants from country A to country B improve information, lower transaction costs, they tell investors about, they speak the language, so they could connect people, they have business connections, they have political connections, which as we know or can be useful for FDI, they can find partners and so on. So they reduce information and that should, if we believe the main take from this financial literature about the home bias, which is caused by lack of information, if you reduce this lack of information, you should see more diversification of investment and of portfolio, okay? And so this is what the literature is looking at. There is some micro literature, for example this paper by Folien Kerr in Management Science, showing that US firms that have foreign inventors signed on their patents, invest more in the home country of those inventors, okay? So that's micro firm level connection and it's nice to see it. There are also cross country studies, so I did a study 10 years ago with Maurice Cougler, published in a small thing in economic letters in GDE, Beata Javorsi and co-authors have about the same result that it's a bilateral exercise between the US and the rest of the world in both cases, showing that when you have a bigger stock of immigrants from a given country to the US, that increases FDI from US firms to the home country of migrants. It's not the case everywhere, it's the case mostly in the service sector and in sophisticated manufacturing sectors, okay? So there is this dynamic complementarity if you want between migration and FDI. And in the recent paper, we're not yet finished, although we've been working on it for quite some time with Aubry and Rechef. What we do is to try to connect trade and FDI using the Melitz framework where you have heterogeneous firms and these firms are choosing the mode of entry to the foreign market. So if you want to sell on the foreign market, you can do it either through trade by exporting or by investing through FDI in that foreign market, which means that there is a connection that you should look at trade and FDI not separately as the rest of the literature which are presented as done, but jointly. So what we do is to develop a theoretical model based on this Melitz framework, but I think the intuition is clear from this graph. This is a measure of productivity. So you have very productive firms here and this is the distribution of firms. So you have a lot of firms with low productivity and very few firms with high productivity because the fixed cost of doing FDI are very high. Only the most productive firms will do FDI. The fixed cost of doing exports are a bit lower. So firms with intermediate level of productivity will do exports and firms with relatively low levels of productivity will not sell on the foreign market, but will stay on the domestic market. Okay, so assume this is the threshold to do FDI. So only the firms in this, the density of firms will be here, okay? And to do exports, this is the threshold. So only the firms here, sorry, here, this is the threshold for export. So these firms will do exports, okay? Now assume you have something in our story migration which reduces the fixed cost of doing both FDI and trade. So this threshold will be lowered. The threshold for doing FDI will move from here to here. The threshold to do exports will go from here to here. Okay, and we ask, how will this affect FDI? You will have more firms doing FDI and not just the firms between this, sorry, here, but also the firms between the two thresholds. And you have more firms doing exports. But as you can see, those who start doing FDI do it at the expenses of exports. Okay, so our prediction from the model under some parameter restrictions is that when you have more migration, you should see an increase in the ratio of FDI to trade. Okay, which is the outcome we are looking at, the ratio of FDI to trade. So when you look at trade and FDI separately as the rest of the literature, you see these positive effects on trade for the first two columns and on FDI for the other ones. We've elasticity of about 15, 10% consistently with the rest of literature for trade and around 20% for FDI. But when you look at the ratio, which is also good to do because you're accounting for many things which you should account for, then you see something around 10%, meaning that when you have more migration, you increase FDI by 10% more than you do for exports. Okay, which is confirming our theoretical prediction. Now, FDI is not the only type of financial inflow to developing countries we want to look at. As you see from this graph, FDI is in orange. It's important, but it's not the only financial inflow to developing countries. You see that other financial flows such as the other financial flows are mostly bank loans, all right? And the portfolio flows, the purchase of equity and bonds are also not negligible, all right? So what is the evidence on migration and the flow of other financial flows to the home country? Okay, if you have people from Ghana going to France, will that increase the likelihood that the French bank will lend money to a Ghanaian company? Okay, that's the type of question we're asking, or that financial investors in France will buy more equities from Ghanaian firms as part of their portfolio, all right? And we find such an effect in a paper which is with Kugler and Leventhal like forthcoming at World Bank Economic Review. Okay, I will have to speed up a bit, I guess, so I will spare you the regressions. What we do is to look at the direct effect but mostly at second-order effects. We say if this is a problem of information, this positive effect that we find should be stronger when we look at skilled migrants versus unskilled migrants. Should be stronger when we look at long-term loans which have a strong risk in a certainty information component and not so much at short-term loans. When we look at countries which are culturally very distant because this is where migrants can help and not so much for countries which are linguistically or culturally closer, et cetera. And we consistently find all these heterogeneous effects which tell a plausible story every time in the same direction suggesting that information is indeed something behind the channel, behind this result. It's for now that the data I presented is only OECD to developing countries. So it's only North-South. Still when you look at, you interact the effect of migration with the fact that the home country of migrants is a developing country and not just any country. You see a stronger effect manifesting itself when we do the Poisson estimator which is indicative of the fact that the effect works mostly at the extensive margin, not the fact that French bank will start lending to Ghanaian companies and not that they will lend more. All right? So it's really the extensive margin that seems to matter for developing countries. And the last aspect of the economic integration I want to talk about is about knowledge and technology diffusion. And I think this is the new really topic. Trade, FDI have been studied for quite some time. There are improvements, refinements, it's nice. But I think that innovation and technology is really the most frontier, I would say, dimension of this connection. So let me start by saying that something well known for people who study knowledge diffusion, innovation, is the fact that knowledge diffuses in concentric circles. Okay? Very easily when you're close to an innovator and then when you're far from the place where the innovation has been invented, then it's difficult for knowledge to circulate, okay, even with internet and so on. Okay? And the reason which has been, first this has been very well documented, before being documented this was conjectured by people like Calpolani or Kinesarro who argued that knowledge can be separated into what is a codified knowledge and what is tacit knowledge. And tacit knowledge requires direct human interaction to be transmitted. Okay? So building on this insight that if a good chunk of knowledge is tacit, meaning it needs direct human interaction to be diffused, our hypothesis was that then the international diffusion of knowledge should follow the international movement of people, migration. Okay? And this is what we wanted to check mostly with Danny Bahar, co-author of mine. We have this paper where we ask, so first thing, that the main measure of knowledge diffusion which has been used so far is patent citations. Okay? So you follow how knowledge diffuses by following where the patent is cited. Okay? We chose another route. The hour measure of innovation in connection with the work by Ricardo Osman and others at Harvard CID is to say, well, if you look at the basket of exports of a country and that's basket is changing over time, so you start exporting goods that you did not export before, then your productivity must have reason. If you start exporting something and meaning that you must have learned something in the process. Okay? So what we do is to look how migration affects the evolution of the export basket of countries. So we have two papers, one which is a full cross-country exercise, forthcoming at the Economic Journal, where we ask, say, I'll take the example of Ghana if I see that the result will be that if Ghana wants to start exporting beer, I know that Ghana is producing beer because I had two yesterday, one and a half. I don't know if they are exporting beer. To export beer, maybe Ghana needs productivity and quality shift. Okay? And one way to get this would be in our story to have migrants coming from Germany or to have Ghanaians emigrating to Germany and after some time knowledge will diffuse, okay, through these human interactions. So in that fully cross-country paper, all we knew was whether Ghana has immigrants in Germany or whether the immigrants to Ghana come from Germany but we don't know where they work. Okay? In a second exercise that the second paper we zoomed on Germany because for Germany we have access to the labor survey which tells us where immigrants, in which industry, immigrants work. So now we can ask whether having Ghanaian immigrants in Germany actually helps Ghana start exporting cars or beers or goods for which Germany is good at. Okay? So these are the papers we are doing building on this inside that migrants are carriers of knowledge, of tacit knowledge that they can transmit. Danny likes this picture of Franschuk Valley in South Africa so he put it on the presentation for our paper one, two years ago when we started presenting it. So the reason it's here is that Franschuk Valley means the French corner in Africans. It's a valley near Cape Town where the French you know that were expelled from France at the end of the 17th century, established themselves and they were good at making wines coming from France and that's where the South African wine industry started. Okay? So because I'm a very dedicated researcher I decided to do field work and for this talk I went to Franschuk last week and the field work was to do wine tasting and I confirmed South African wine are very good. But I wanted another one product is not enough so I tasted beer also and I asked you know beer bar give me your best two beers. And South Africa is a producer of beer but you see these ones come from Namibia which was a German colony so they know how to do proper beer and like the Dutch and the British. Sorry for it was in a room and indeed Namibia is by its standards much bigger exporter size adjusted of beer than South Africa. So the story is about this slide having so now it's about US and Italy if you have Italians coming to the US the US will become good at making pizzas but also Italy will become good at making hamburgers. Okay so that's the story that once you check the data you consistently find this result will spare you the details of the empirical strategy but we exclude bilateral trade so that it's not a network effect as in the first part of the presentation that we capture we control for global demand for the goods so it's not just picking up possible connections between migration waves and demand for certain goods we are instrumenting properly doing a gravity model and so on. So we're reasonably confident that our result is not an artifact of something else. And it works at the extensive margin. This is for both immigrants and immigrants. Okay so again really the story is not just Germans coming to Ghana but also Ghanaians going to Germany at the export at the extensive margin and at the intensive margin. Okay so the growth of export is also positively affected if you already exported the goods the growth of exports will be significantly affected. So this is also you have much stronger effects for skilled migrants than for unskilled migrants. I think this is consistent with a knowledge diffusion story. And here we were able to focus on South-South migration because we have really global trade data and global migration data. And I think this is a very interesting and important message that this is not just a story about North-South migration but this knowledge circulation works for South-South migration. The point estimate is slightly lower but it's not that lower and it's robust. Mostly for immigrants but also for no it's yeah you have 10% significance for one of the regressions. Okay so I think it's an important message to convey that if you want to diversify your export basket and this is a concern for many developing countries migration both in and out including to other developing countries is one way to go. Okay that we'll come back to that when I wrap up with policy conclusions. And that works for South-South migration both at the extensive and intensive margin again. Okay we have another paper I will maybe not talk about just that zooming on Germany because now we know where migrants work in which industry and we find let's say the same results. This is for immigrants from anywhere in the world to Germany. Okay and the channel here could be just having migrants present could be return migration and we focus even more looking at Lugoslavlian refugees that came to Germany in the 90s and went back after a few years. Okay maybe the only yeah. Okay let me just for that Germany paper we can look not just at skilled and unskilled migrants but also because this German labor force survey is quite detailed. We also know whether people not just whether they are skilled or unskilled but also whether they are white collars or non-white collars whether they work in occupations with high analytical and cognitive content or in manual tasks and whether their jobs are classified as a high problem solving or low problem solving jobs. And you see that time and again consistently the results are stronger for type of occupations that are skilled. Okay so it's consistent again with knowledge diffusion story. Okay let me move now to the second part the cultural dimension now. Here I want to be tribute to Peggy Levitt. She's an anthropologist at I can't remember the name of the university near Boston Brandeis, Brandeis University and she is the one who coined the term social remittances and defined them as migrants transfers of behavioral and cultural norms to their origin communities. So many of the things that economists do are already known by other social scientists just that we're putting a number on it and we're maybe more precise about mechanism but that's at least the way we see it. So this is known for quite some time that you have cultural transmission of values, of preferences, of institutions through migration but can you show it when you look at big numbers can you show it in a case study where you have a careful identification strategy this is what the economics literature is about. So let me start with political remittances. The first paper here looking at cross-country evidence is a paper by Antonio Spillimbergo in American Economic Review. This paper was about foreign students and democracy showing that countries that have a lot of foreign students in one time period will have better democracy outcomes in the following period. Using comparative indices of democracy and UNESCO data on foreign students. He also had more anecdotal exercise on where leaders were educated whether they were educated at Harvard or at the Sorbonne for example and that also seems to matter. But the take from his paper was that foreign students are good for democracy only if they studied in a democratic country. If they went to the Soviet Union they didn't do any good to democracy in their home country. That was his paper. I think it's very nice because it was the first to tell the story. I think though the story is not specific to foreign students it could be generalized to any type of migrants and this is what we did with Doquiello DiGiani and Schiff in the paper published last year in the Journal of Development Economics just that migrants in general and we find to make it short the same result that when you have migrants openness to migration is good in terms of democracy. We also have IVs and we think our result is stronger in the sense that his result was based on the fact that his main variable of interest was what he called a democracy score at destination a way to the average of the democracy score of where your emigrants go. But the interaction with the number of emigrants didn't matter which is kind of difficult to interpret. In our paper we have both the result that where people go matters but if more people go that also matters. However, again, this is cross-country study even if you do the most sophisticated instrumental variable strategy you will only have convinced a sympathetic reviewer and so what you are really after is a nice natural experiment where you can really be more confident in terms of identification even though the external validity may not be certain. There are a few papers there are papers on Mali there are papers on Cab Verde looking at emigration from those countries or out of Mexico. I see some problems with these case studies in the sense that countries like Mali or Cab Verde had emigration and started very long time ago. And so they could have affected the evolution of democracy in the country and you don't know exactly what you are picking up. So what you'd like again is a natural experiment where the experimental conditions to test the effect, the causal effect of migration on the evolution of political preferences votes democracy can be trust. So we believe with these three co-authors in a paper which is on Moldova and published this year in American Economic Journal that Moldova provides such a natural experiment. So the paper is titled the effect of labor migration on the diffusion of democracy evidence from a former Soviet Republic that former Soviet Republic is Moldova and the reason why we are interested in Moldova is that it has the two features we think are essential to empirically look at this relationship between immigration and democracy. First, condition building on the insight from Spillinbergo you want to compare emigration to a non-democratic destination with emigration to a democratic destination. And this is exactly what happened with Moldova that has about 40% going west and 60% going east mostly to Russia. And second, a bit earlier and second the timing is important so the context is that Moldova had no emigration before the Russian crisis of 98. It was specialized in agriculture there were some minorities that left Moldova after independence but for the rest the country had certainly no emigration to the west before the fall of the Berlin Wall and even until the late 90s had virtually no migration to the west and very very little migration to the rest of the Soviet Union. So Moldova was specialized in agriculture. It was hit much harder by the Russian crisis than Russia because it was so much dependent on agricultural export to Russia so when the crisis hit they had to find alternative sources of income and started emigrating quite massively. So what you see here is that this migration to the west each dot on this scatter plot is a community we have something like 800 communities and the outcome variable is the votes for the Communist Party in the elections of 2009 the Communist Party was still ruling Moldova in 2009. It was the last ruling Communist Party in Europe and the 2009 election put an end to its rule. So you see here that migration to the west decreased the vote I mean these are simple correlations maybe this is completely spurious but there is a negative correlation between migration to the west for the Communist and a positive correlation between migration to the east and votes for the Communist. You would think this may be driven by self-selection on political preferences but if you think another minute it goes the other direction if people who want democracy emigrate to the west then the share of votes for the Communist in their community should go up. Because if they had not migrated and voted against the Communist and conversely if the people who want communism in sense of Putin style regime authoritarian regime go to Russia they would have voted for the Communist Party which is offering the same platform of nationalism and interventionism and so the share of votes should go down for the Communist. So self-selection into migration on political attitude is playing against us is going the other direction so what is the force there must be a counteracting force and we believe this is political remittances. So this graph is a summary of the paper you see that we had elections just before the Russian crisis and these were you see the share of votes for the Communist here so the black line is the average share of Communist votes the red one is for communities which had strong emigration to Russia and the blue one is for communities which had strong emigration to the EU. And interestingly you see that when you compare the two type of communities they don't diverge until 2005 so this is a differences in differences exercise so we have this common trend assumption which is validated and so on and they start diverging only after 2005 for the last round of election which is consistent with a story in terms of preference changing, building up transfer, it takes time to change and it takes time to make people around you change. So that's consistent but we have different sets of results looking at more precise things that back support really an interpretation in terms of preferences changing. And this is the same thing but in the table showing that the prevalence of emigration to the west is associated with negative effect on communist vote. The point of estimation is interesting, it's 0.7 0.6 meaning that if you have 1% more share of emigrants to the west, votes for the communists will decrease by 0.7 0.6% which is quite high. It's almost a one to one, it's really high if you consider that this should compensate self-selection into migration on political views. So we see it actually as a lower bound estimate. And we're controlling obviously for all the things you may want to think about. We control for pre-migration results so it's a different if it's a change in votes for the communists on differential migration east or west. We control for initial the intensity of the economic shock of those different communities. And we control for demographic characteristics, ethnic representation, geographic characteristics if you're close to the Romanian border, to the Ukrainian border and so on. Okay, so that's for political preferences. Let me move to the other domain of culture which I think has been most studied which is fertility. And this started by a paper by demographer, Philippe Farg, during Morocco and Egypt he noticed that communities in Morocco who have a lot of migrants tend to have lower fertility than otherwise similar communities, while in Egypt communities that have a lot of immigrants tend to have higher than otherwise similar communities. And he conjectured that this could explain by the destination. The low fertility region while Egyptians go to the Gulf countries, high fertility more traditional by religious standards for example compared to Egypt. And so his conjecture was that there was some social remittances, what I call Malthusian remittances that you would transfer a norm that small families are fine, that gender roles should be more balanced, etc. Which are part of culture of Western Europe that are transmitted by migrants, by returning by contacts, by phone calls letters, whatever and conversely for Egypt that tend to become more traditional if you're exposed to the cultural influence of the Gulf countries. So that was a conjecture but this conjecture has received support in different studies, a cross-country study by Ben Doki and Schiff or a careful examination of the case of Egypt by Bertoli and Marchetta. So here the case study I want to talk about is a case study in economic history with two co-authors who are economic historians Guillaume Daudin and Raphael Franck and we look at France. We look not at international migration, we look at internal migration in France. And France is an interesting country if you're interested in the fertility transition because it is the first country to have experienced the fertility transition one century before England. So the fertility transition started in France in the late 18th century while it started only around 1880 in England. And this is a puzzle, it's called the French fertility puzzle. How to explain this given that England was more advanced economically, was more urbanized, industrialized and so on. So our conjecture is that contrary to England and the rest of Europe France was not an emigration country, the French did not emigrate. Okay, they tried to in the 18th century but their colonies were taken by England relatively quickly. They had only one settlement colony in Algeria but for the rest they didn't move overseas. They didn't go to the US or to the new countries for example, contrary to England, to Germany, Italy and so on. But they moved internally a lot. And what we want to check is because initially France at the beginning of the 19th century was not culturally integrated, you had regions of very high fertility and regions of very low fertility, mostly Paris. And whether this migration to the low fertility regions had any social remittance effect, transmission same type of mechanism I described. So for this we do a big data work which is to build a matrix of bilateral panel migration data between the different French regions, the département and there are about 100 départements for this period of the late 19th century. So from mid 19th century to beginning of the 20th century you see here that France, if you compare France to England, Italy or Germany you see that France was clearly on a declining trend already by the mid 19th century while this started later everywhere else, most in England only after 1870 1880 and France was lower the point here is 0.3 while England was at close to 0.4 Italy was at 0.4, Germany was at 0.4 this is a type of fertility index you don't want to know about but take it as some measure of fertility. The new countries were high land to labour ratio and you needed to have big families if you wanted optimal technology you get land in the US you need big it's a family farm you want seven, eight kids to do everything properly. And so that in terms of selection it means that when you have international migration those who want big families are the ones who will live. Those who stay should have lower family that should decrease through self-selection the fertility of those who remain. For France it's internal mostly rural to urban migration it goes the opposite. Those who would leave the country to go to the city are those who don't want a family or want small families or so on and that should increase fertility in the regions of origin exactly the opposite. So there must be something else going on and this is what we are after. You see here for France the evolution of migration here the evolution of fertility over the period and we ask whether there is a causal relationship between these two. More migration less fertility. And we build our matrix and we use an instrument for migration which is based on the diffusion of the railroad network which is exogenous to demand conditions because this is the French system so it's not market forces it's the state who wanted to connect the different parts of France in those designed by engineers and not by entrepreneurs. And what we find is that migration explains internal migration explains two-third that sounds a lot of the decrease of fertility and the convergence in fertility across regions for the period which study and Paris itself explains two-thirds of the effect of migration what it's only one quarter of the internal migration. So there is a very strong effect of migration what we show is not just migration but this what matters is whether you go migrants go to low fertility regions including Paris and we see that the effect of Paris is totally disproportional and this is consistent with a cultural diffusion dimension because France was getting centralized politically they wanted to diffuse the central culture out of Paris and so on. So again this is I think going the same direction as the studies I presented before. Then there are other dimensions of societal change there is this interesting paper on the hajj on the pilgrimage to Mecca published in the quarterly Journal of Economics so this is clearly South-South migration showing for Pakistanis if you want to go to do the pilgrimage there are too much demand there is a quota so there is a lottery. So what they exploited is to compare Pakistanis which happened to have one lot of slots through the lottery to communities that didn't. But participated in the lottery. So what they show is that those who went to the pilgrimage again think of Pakistan as a very fundamentalist on a liberal to fundamentalist Islamic scale and there in Mecca you mix with Muslims coming from the rest of the world on average there will be more liberal than a Pakistani. And they show that attitudes to tolerance religious tolerance to women this type of thing to superstition are significantly affected and decreased become more moderate if you want for those who mixed with other Muslims during this experience. So this is a kind of illustration of the fact that migrating is a transformative experience and that this transformative experience can then spill over to the rest through social networks. Alright, so let me conclude with I have 5 minutes it's okay with the last dimension of culture which is really asking well what about migration this cultural aspect of globalization so what we ask here is essentially whether migration between two countries make them culturally more similar or more distant. This question has already been asked for trade theoretically in this paper by Olivier, Tonig and Verdie both theoretically and empirically again with Tonig and Verdie and other co-authors around here is the trade integration when you have more trade between two countries at least two cultural converges the countries become more similar maybe because you are consuming goods that you are eating hamburgers viewing American movies so you become culturally closer to the US. Okay, that's kind of the point of the paper. So we wanted to ask whether first this is true for migration but not just maybe migration is a stronger driver of cultural convergence than trade. So what we do with two co-authors, Solene Sardochot is a PhD student Arthur Silva from Laval University is to have both a theoretical model which is based on, for those who know this the Bizin Verdie framework of cultural transmission where you have the horizontal transmission by being exposed to people you pick a role model if you want but also vertical transmission for your parents which can invest in transmitting their own cultural trade. So we have this model and we have actually two models, a purely compositional model and a diffusion model, a cultural transmission model and we compare the predictions of these two models and we will just catch the forces that we capture in the theoretical model is the selection. So think that you're in a country where some people like football, some people like baseball and if you're in Ghana and you like baseball maybe you want to move to the US to be surrounded by other baseball lovers because then you want your kids to love baseball as well and it's more likely that they will pick a baseball lover as role model in the US than if you stay in Ghana. So that's the story of the selection and that should create cultural divergence because if baseball lovers in Ghana move to the US then the two countries become more dissimilar, less baseball lovers in Ghana and more in the US so the two populations become culturally more distant. The second effect is the diffusion. If Ghanaians move to the US they can diffuse their love for football or they can send to Ghana, sorry I speak with hands so be careful with the microphone they can send if they are in the US they may be more powerful at sending preferences for baseball to Ghana. That's the third one, the social remittances. So in the world we kind of put these different forces together and compare two models one based purely on composition through self-selection and one based on including the cultural transmission or the forces that lead to cultural transmission. We compare the predictions of these models one predicts divergence the other one predicts convergence that's clear but we have interesting more refined predictions what happens when you have more economic gains meaning more wage differences between the two countries a compositional model would predict convergence while the transmission model would predict divergence. If you increase now the initial cultural distance between the two countries the compositional model will predict divergence while the transmission model will predict convergence. So we are not able with the data I will explain in a second to decompose the effect we find only the overall effect but clearly if divergence obtains then it means that the self-selection and the compositional dimensions dominate while if convergence obtains then it means that the diffusion transmission channels dominate. And this is what we do we build different statistical measures of cultural similarity for robustness not just use one there are many in the literature so the data comes from the world value survey so a rich set of questions in a panel setting for many countries and the migration data is bilateral migration data by the World Bank and the specification we test is whether the cultural similarity CS or cultural similarity between country I and J at time T is affected by migration in the previous period bilateral migration in the previous period bilateral trade in the previous period and we have a very rich fixed effect structure with country time destination time and most importantly dyadic fixed effect country country fixed effect so let's look so this is about cultural similarity so if beta 1 for migration beta 2 for trade is positive this means migration or trade create more similarity convergence or if it's negative it's less similarity divergence and here what you see is that the effect of migration is almost always positive and very significant except here for one specific measure so but the ones we really believe is the last column which includes also the bilateral fixed effect so if I focus on the last column with the full set of fixed effect migration is creating more similarity for the three indicators and it's quite significant let me comment on trade here you can see this as a horse race, trade is no longer robust and significant once you include this bilateral fixed effect which was not included in the previous study I mentioned while migration survives this introduction so for the second order predictions the effect of income differences we add GDP gap between the countries here and you see that the bigger GDP gap decreases similarity not with the compositional model for all indicators and when you introduce initial cultural distance you see that the interaction with initial distance is negative as predicted by the transmission model but against the compositional model so in a nutshell what we believe is that the true forces that shape this cultural countries are forces of convergence through cultural diffusion and transmission and not from self-selection and composition here we were also able to focus on self-south migration and I think the result we get may be disappointing but it makes sense with the whole story if it's about cultural transmission that you would pick a role model that you're influenced by others so you see that north-north nothing happens south-north is exactly the result we get but south-north nothing happens so this story of cultural diffusion and convergence is really for south-north migration but not for north-north or south-north and again it's consistent with a role model where you want to emulate say a successful person and for migrants who emigrate to rich countries probably in poor countries they are seen as successful candidate role models while those who go to neighboring countries may not play such a role but I'm speculating but I think this is consistent with the theoretical argumentation so let me conclude with two slides one migration is contributing to the economic integration of the home country into the world markets two it's contributing to the cultural integration also to the global culture and the global economy in terms of what did we learn I think as I said that these things are not entirely new it's not something very surprising but I think it's good to have some experience about the direction of causality and to be able to quantify put a number how much is an elasticity and this quantification I think is especially important if you want to convince policy makers that they should consider diasporas and immigration seriously when thinking about development so what are the policy implications I think for home countries there are obvious I've been advocated for things that allow for people to move more freely to keep dual citizenship and facilitate cross border facilitate diasporic involvement in local politics and policies and so on there is also a policy implication for receiving countries which is that if by having migrants coming you can reduce fertility in their home country improve democracy and improve their integration into the global economy which is exactly what development policy is about maybe migration is the right way to go rather than something else and I will conclude here