 So welcome everybody to the afternoon session of this conference, of the conference on forecasting the future of global migration. For those of you who have connected more recently, let me introduce myself. My name is Julia Lendorfer. I'm the head of research and migration law at the International Organization for Migration in Austria. And we are also the national contact point of the European migration network. And in this framework, we organize a yearly conference. So this year, we're looking at whether and how global migration can be anticipated. In the morning session, we had 190 participants connected. We had a very engaging discussion around interesting topics. We explored forecast, scenario studies and early warning systems. We also looked at innovative approaches to forecasting by using online data searches and we discussed possible implications of migration forecasts on Europe. This afternoon, we will focus on migration predictions and proactive migration management with a particular focus on how migration forecasting has been used by policymakers and alternately how policy has influenced migration forecasting. In this first section session, we will look at the nexus between predictions and policymaking and I'm extremely honored to be able to introduce for brilliant speakers and experts on the topic. I'm very happy to be able to welcome Britta Behrendt, who is the head of the Division for General Migration Policy at the German Federal Ministry of the Interior. Britta will speak from the perspective of the German EU presidency, which is focusing on digital transformation to explain how monitoring of migration flows is done in Germany. I'm equally pleased to be able to present Teddy Wilkin, head of the data analysis and research sector at the European Asylum Support Office, EASL. Teddy leads several large projects aimed at understanding and predicting flows of asylum seekers to and within the EU so that the agency can provide targeted operational support to member states. Teddy will present EASL's focus on early warning and forecasting systems and how to bridge the gap between data science and policymakers at the national level. I would like to extend a special thank you to Michael Clements, who is connecting from Washington DC and this early morning for him. Michael Clements is the Director of Migration Displacement and Humanitarian Policy and a senior fellow at the Center for Global Development, where he studies the economic effects and causes of migration around the world. Today, he will present results from two recent papers, which look at the effects of economic development on migration. And to complete this outstanding panel, we are particularly happy to welcome Matthias Czajka, Professor of Migration and Integration and the head of the Department for Migration and Globalization at Daniel University in Austria. Matthias will conclude the set of presentations by providing a view on how to address and reduce uncertainties in migration forecasting, particularly for policymaking. Each presenter will speak for 10 to 12 minutes. And after each presentation, we will have a very brief Q&A session. This is mainly for clarification questions, because we would like to reserve 20 to 30 minutes at the end of the four presentations for a panel discussion. To ensure uninterrupted presentations, participants are muted, but we really do encourage you to engage. Please ask questions via the chat on the bottom right side of your screen. I will then channel your questions to the speakers, and please make sure to really send your questions to all participants so we can all read them and they also can come directly to me. Please don't use this channel for connection issues or IT issues, right either directly to the host or to the email address, which will also be posted in the chat. There's also an e-signature list, which will be shared in the chat, so please make us a favor and sign it. We'd really like to know who participated today. And finally, both the PowerPoint slides and the recording will be made available to you after the conference on our website. So without further ado, I have the pleasure to hand over the virtual podium to Pita. Pita, I think you have muted. Now it's better. Absolutely. Thank you. Yeah, actually, for me, it's the first time to do such a conference from my home because usually I'm at my office every day now. Maybe I'm one of the only ones who stays at the office like every day, but today I'm at home because there's a huge strike going on in Berlin. So I get started right now when we're looking at applying forecast in Germany. I would like to start with the general background information on the monitoring of migration in Germany. So I get to the first slide. One for the regular situation and monitoring of migration movement. We created, or my team created a cooperation platform, a network, so to say of all the relevant authority. It's a network of migration experts from the relevant ministries. My ministry of interior is the host of this network. And there are like other ministries involved like the foreign office, the ministry for economic cooperation and development, ministry of defense ministry of justice and finance. And I think that's nearly all and there are the relevant agencies involved as well. The network focuses on monitoring migration and the migration situation in the countries of origin and the country is right now focuses mainly on now casting, but also contains some prognostic elements of future migration dynamics, but mainly based on expert assessment. The institution involved in this process is the joint analysis center for it. Which is a kind of cooperation platform. It's not an agency. It's a platform or cooperation platform. And to this platform, a part of this platform are all the relevant agencies. And maybe I get to the next slide to. Get a better picture of this. The participating authorities are the federal office for migration and refugees. The foreign office, the federal police, the federal criminal police office, domestic intelligence, foreign intelligence. And central customs authority. They are based in but some at the federal office of federal police office, but they do cooperate in a kind of platform based approach. The focus is to exchange the main focus clearly is to exchange information between the different authorities. And there are different sources for the information used. You can see it like national sources from the sources or a national asylum statistics, and so on. But we do not use personal data. This is really important. No personal data being used in this. In the garden, as we call it. Here you get the picture. It's really nice old building and you see again all the participants, but I already told you who the participants are. It's not, it's not really easy, but it was, I think for Germany, it was a really important step to create this platform because it guarantees in time exchange of information. And as you might have realized, there are these two actors or players, this corporation platform, the garden and this network of migration experts and the network of the migration experts. It's more the ministerial level, like my colleagues from the other other ministries of participating in these meetings and they take place like every six weeks. And the garden, the corporation platform is, as you might call it, the operational heart of this process. So they produce reports and ad hoc reports. So they are really the operational heart of it. When it comes to migration forecasting. We are still, we are still at the beginning of a process because in this migration network. Of experts, we realize that we until now mainly focus on now casting and only sometimes. We look at the future, but we all realized that this has to be changed to be a little bit more strategic and to be prepared for strategic changes. So we realized that members of this network have already created kind of my great forecasting tools. Well, like for example, there is a tool of the foreign office called preview. There's another tool of the foreign intelligence service and another tool of the Ministry of Defense. But those tools, they were created like within the last one or two years, I would say, they mainly focus on crisis prediction and conflict prediction. So only the tool of the federal intelligence service focuses in part on migration as well. So we realized that we need a coherent approach to migration forecasting. And I would like to tell you how we did this. So I get to the next slide. Because instead of creating kind of theoretical discussions first, we took a kind of hands on approach, I would call it because we organized a meeting of experts. And we created a task, a common task for all of these tools. We said we want to have a kind of pilot prediction. So all the three tools, they had to analyze the same situation. They had to look at the region, which was like agreed upon that was North Africa, including Egypt and agreed timeline, like six months. And then they all had to work on that task and they had like four weeks time and then we had a meeting and all the different actors they presented their results. And we are still analyzing the next steps because it's clearly that they had like different views on the and different perspectives. And at the end of the day, there was a broad content that we would like to share the experiences and the different approaches to create a new tool focusing on migration forecasting. So it was really a kind of pragmatic way. So we want to start now with this process. And at the national level, we want to start taking two steps. Right now, we have the first step is this one. We agreed that all the actors, the three actors, they present their tools to the garden to this cooperation platform, and they start immediately to check if they can use the results for the reports right now. This is the first step using the existing tools, which mainly focus on crisis prevention for the garden and the second step will be that we get together and create a common tool for migration forecasting by cooperating with our partners from the other ministries and we would like to have this new tool at our federal office for migration management. But this is still, we are still at the very beginning, because it was just agreed upon like this week for last week that we will choose this road and it's really work in progress and we can get started. And our main focus is to cooperate and have this kind of migration forecasting tool as a to get a broader sensorium. And I would like to call it this is just for us. It's not like being able to get absolute clarity on the future is about having a different perspective perspective, which might be really fruitful to get to better preparedness on the national And then the really interesting point of the European level, because as you all know, the new migration pack was like presented this week and we are really happy what we read about the blueprint on migration management and crisis preparedness and I think Susanna will tell us a little bit about that later on, but we as the general presidency, you're really glad to be part of this process and I think putting kind of effort into this preparedness and being able to be with more strategic at an early age. It's really, really, really important for all of us and I think we all agree from that point. So what's that look. Yeah, and we, during the presence presidency as we started, you might know, on the 22nd of July with the migration for 0.0 event, and the aim of this was to support the digital transformation migration management. And one aspect we focused on was using the data for migration forecasting as well as you know. So we're really happy to continue with this process we have mapping process going on with all the member states and we will present the results. And maybe one of the next people meeting, we will have step out sessions conferences. And so we are really happy to be part of this and our MSP support and accelerate the implementation of the blueprint because I think it's really important for the whole migration issue as a whole. So, from getting from crisis to comprehensive management based on preparedness. This is a really valuable goal for all of us. And don't forget think it's digitally from the start, because we always have to keep in mind that it has to be digital at the end of the day having written reports is nice but I think it's kind of old fashioned and we and we have to get to platform integration at the end of the day that I think that's the way ahead. So maybe one last line. Yeah, and at the end of the German presidency we have on the 11th of December, we will have a follow up on the migration for 0.0 conference. And I hope we will be able to present to you some of the results of our national process and some result of the European process as well. So it's then at the end of the German presidency, but it's not the end of the transformation keep on going. And I'm really happy to hear maybe Katie as the next one. But first of all, I want to give the word back to you. Yeah, thanks. Thank you very much for it. I think it's always fascinating to learn how a member state and a nation state deals with these kind of questions and issues. Thank you so much for sharing really those insights from the ministry. There are no questions in the chat. Please do feel free to post questions. I do have a few questions for you. So I would be really interested to understand a little bit the time frame how you're what you think you said that you're doing now casting now so basically very short term but how far do you propose to look into the future with the models that you're developing. I've also heard in the morning a lot about uncertainty that there are a lot of elements of uncertainty in any predictive methods. And I would be really interested how you are thinking your team is thinking about uncertainties and how they can be built into into the German approach. And lastly, some policy aims, but maybe you can describe these in more detail. So is it really more policy aim to look at sort of asylum seekers or refugee flows or are you also looking at potentially understanding family migration, labor migration, sort of these mid to longer term trends better. So I think you still knew the bridge. Yeah. Thank you. So when it comes to the time frame, right now, what we're doing this kind of work we do right now is, I think we're looking what might happen in the wintertime so it's more kind of just expert knowledge but with our migration forecasting tool which you want to develop in the next month, maybe in the next year, going to be quite a long process at the end of the day, I think. But I think we aim at having forecast from like about six months to one year. So that would be our starting point. But I think we're going to this process will be a long journey. And I think we're really happy to have this close cooperation between the policymakers and the guys who have to create the tool. It's going to be kind of laboratory ongoing laboratory because we're having a kind of cooperation all the time and dialogue is key for this process. So I think we will have to keep on adjusting the thing all the time. Maybe it's, I think it's a process without an end because you always have to adjust the tools to your needs. And therefore, but right now, we would like to develop a tool for like six months or one year, but I think it's an ongoing process. Then your second question about uncertainty and certainty. I had an idea when I had a look at your program because he said this today our session is the nexus between prediction and policymaking. And from my point of view, it's more about the nexus between the fact finders and the decision takers. Because that's from my point of view, the key part of it you have to decide and you have to act. And it's not about just monitoring and knowing about the situation. You have to create the next to create action and to create help for people in need because at the end of the day. This tool is not just about early warning and preventing people from coming to our country. It's about helping people in wherever they are. So, I think it's a very, it can be a really positive thing at the end of the day, but you have to always keep in mind that we're talking about people even if our language sometimes is a little more. It's more how to call it, it's like when we're talking about early warning, we have in mind that we're talking about people at the end of the day. So, and that's maybe when I was explaining about the German process. It's really very prismatic one. It's not about analyzing migration flows on a kind of abstract scale about that's happening in this country and what can be done and who is able to finance the action. So it's really, it's really concrete and at the end of each meeting, we're identifying what has to be done who is responsible for the action and who is able to put the money into the action. And then it's really we have a list and we do controlling as well because from my experience otherwise. It just doesn't work. And I think this is the positive thing about the blueprint plans from my point of view, because it's not just about monitoring blueprint. It's about crisis preparedness as well. So it's an access between monitoring and acting. And this is really key from my point of view. And then maybe maybe just in one minute, because we need to move. Thank you. Yeah, you know, I'm so euphoric about the thing. No, the last question was about what do we want to monitor. I think we will start with this project with a monitoring. Irregular migration. It's not about monitoring legal migration in like of working people like trying who want to come to Germany to work. But at the end of the day, I think it's work in progress and we have to see how it develops. Thank you very much. That was a really comprehensive insight. So we will move now from a member states to the EU. Thank you very much, Julia. Thank you. And thank you for the invitation to come and present today. Okay. I'm just looking to share my screen. You tell me when you can see something relevant. Yes, we can see your screen. Okay. Okay. That's splendid. Okay. So, yes, there is has invested heavily in the last couple of years in forecasting migration. We're not going to talk a great deal about the system that we've designed today. I'm going to go rather quickly through that. But I'm really going to try and concentrate a little bit on this nexus between evidence and policy. So that will come towards the end of the presentation. We've covered a lot of ground already, which which I'm very pleased to see. I don't think I need to introduce too much the complexity of migration, you know, except to say that, you know, it's jolly complex and really hard to to predict in in any meaningful way. There's a lot of unknown unknowns. Usually understanding comes after the event. And when we learned about one type of mixed migration flow, let's say about the particular drivers, those lessons are not necessarily transferable to other situations. So really, all of the work is in is in front of us. And I think as Ravenna said at the beginning, there's never going to be an absolute crystal ball. So we have to work together and do the best we can to talk about the future in in at least a sensible way. We have to talk about different types of uncertainty, which with which we're fully in line with, of course, we understand very well that the data are often scattered in low quality and on off. And the now casting is really important than this is full of the epistemic uncertainty. And this is even more so when we when we try and talk about the future. I will I will move on, I think and I'll talk a little bit about our analytical framework, because as we heard earlier, it's best not to have a single tool with lots of very precise forecasts upon which policy is built. Or operational responses, it's best to have a kind of toolbox. And so any so, we've we've built exactly this. So, so we look backwards with traditional data and we do the now casting. And this is where we describe the trends in a descriptive way, and we do our best to uncover the drivers. An example of this is the ASO asylum report, which is available online. What we've been hearing quite a lot about today is the forecasting systems. So this is where ASO has has it has also invested heavily. This is where we use, you know, innovative big data at scale and we try and combine that traditional data to do forecasting. I'll talk more about that. And we've also heard quite a bit about scenarios. This is where we pulled together experts and talk about a range of different plausible futures. And the ASO invested heavily in this and published a report last year, which which we'd like to repeat. And we pulled together large numbers of experts. We didn't ask them, do you think this flow is going to increase or that flow is going to decrease? Instead, we subjected them to a rather rigorous methodology of futurologists and anyway, all the scenarios are included in that report. So I'm going to talk mostly about the middle column of this toolbox, the forecasting. And if we look at the different data sources that we use, one of the main ones is is the GDELT project. So this is a global database of events as reported in print and electronic media. So the best way to think about this data source is when you when you search on Google, you can search the web, you can also search images, you can also search for the news. And so there's data behind that. And so this is the data that we download. We download it every day. And the events are classified into 300 different types of events. And we concentrate on the negative and disruptive events. And we use that to try and estimate a little bit of what we call push factors. So this chart shows us a little bit of push factors that comes from those data for each of the provinces in Turkey. We've also been doing this recently when it comes to Lebanon after the explosion on the 4th of August and also in Belarus at the moment. So this is a really interesting system for looking at the extent to which countries are under this kind of intent to migrate, which we've heard a lot about today. But we're looking at actual events. The second source of data we use that we've also heard quite a bit about today and this is the Google Trends. So these data are available. Every search we do on Google is recorded as we know. And these data are made available to researchers and analysts in one form or another. So we download loads of these data as well and a sudden change in the extent to which a bunch of search topics are used in a certain country. We can use that as an early warning signal as well. So all of these data we scale to a weekly frequency. And we use them to try and predict or at least to correlate with applications for asylum in the European Union. The methodology is quite complicated, but first of all we create some time lags between the events and the applications for asylum in Europe because we don't expect them to happen in the same week. And then we use a machine learning algorithm and adaptive, so that means it's unsupervised elastic net model to sift through these huge amounts of data and uncover some of the correlations. And we notice which events are correlated with applications in the EU and we monitor those and we issue alerts when these events suddenly increase. And these go into the model and help us to some short term forecast. So how to translate and communicate these results where we have the data science, but we don't let the data science lead to building very much. Instead, we've inserted additional layers in between of, let's say migration analysts that check to make sure that the data science is really describing real world events. And we translate the results into easy to understand with beautiful narratives that go into different reports. We've done a joint report with IOM recently, joint reports with member states. And we really think that that's the way to move forward at this time when lots of different organizations have slightly different methodologies. Two of the outputs that we use, well, okay, on the left, this is where policymakers sit, if you like. They would prefer to look at longer term trends. Along the bottom of this heat map on the left, we have weeks. So this is going back 12 months. And on the vertical axis, we have different drivers and different data streams. And where there's a red box, it's highly correlated. So we see here in the bottom left hand corner, this particular migration flow was at the beginning of the year was correlated with those drivers in the bottom left hand corner. But slowly throughout the year, the drivers are slowly changed. And so at the end of the year, this migration flow was correlated with different drivers. From this, we think we can start to inform policymakers about what are really the root causes and underpinning specific migration events. And on the right hand side for operational colleagues, we can do short-term forecasts with a reasonable degree of certainty. But policymakers aren't interested in what's going to happen in a few weeks in advance. We have to try and build a system that has outputs for all our colleagues. So when it comes to the relationship with policy, well, I've mentioned already that we don't allow the data science to leave the building. This is because policymakers tend to talk about the spirit of the regulation, nuances, arguments, loopholes. They don't talk in numbers. So we really have to invest on finding the middle ground so that we can communicate our evidence to them. We do have to manage expectations because as we've heard already today, there's no crystal ball. We tend to deal in probabilities, whereas not just policymakers but also journalists and politicians prefer things to be very much in black and white. But that's not what we talk about at all. We don't talk about certain things. I've mentioned already time horizons. So for policymakers, they require a longer time horizon. We've always already spoken about today, black swans have a huge impact which unforeseen. So we could say that our spring, we could say the war in Syria, we could see COVID-19 as well, which none of us predicted. An important point I'd like to make is we're constantly under pressure to produce or help produce evidence-informed policy. But this is a two-way street. We'd also like to see policy coming out that helps us produce better evidence. So it's fine for the policymakers to come out and say, okay, we've released this new legislative instrument. But if that instrument doesn't contain better data collection, then don't come to us any years' time and say, okay, now we want to understand how well the policy is working. So please bring people like us to the table and let us help you design a policy which will help create better evidence. I've gone over time, so I'll leave it there. Thanks a lot. It's really interesting, Teddy. Thank you. And I think we have a lot of policy makers listening at the moment, so I hope your call has arrived somewhere. Teddy, thank you for this insight into IAZO's work. I think that's really quite special that you're able to share this with us. We do have two questions for you. So one is from Arvid Tseng from the Swedish Migration Agency. We wanted to get some additional information about whether IAZO actually also makes prognosis and assessments on the number of asylum seekers in Europe or coming to Europe, because many EU countries try to predict the number of asylum seekers in their respective countries. So how IAZO works with that. And maybe I can also read you the second question, which is from André Gröga, who presented earlier this morning. And he's referring to the graph on Turkey and would like to know which type of push factors would be included. Or do you expect to capture in the data that you show in the Turkey graph and whether you could give some context why Ankara is showing such high index levels during this time. Okay, fine. Thanks. Okay, so they're both very good questions. So the first question, do we release actual numbers mostly without. Although that can be done and we have done that recently in the context of COVID. So, you know, in April, May, June, numbers of asylum applications launched in Europe fell to historically low levels. And so now we're in a situation where we're doing doing forecasts until the end of the year. And with some reasonably simple arithmetic, we can easily come up with three scenarios for the rest of the year, that they stay low, that they return to normal. Or they return to normal plus all the applicants who were prevented from applying for asylum in April, May and June. So, you know, pushing forward our understanding of the present into the near future is not very hard. You can do that with extrapolations, you can do that with different techniques. It's not something that we do a lot of. Instead, we are looking to try and understand drivers. And we look to forecast those drivers and see how the asylum trends would work. But, I mean, this is a preference that we've chosen. Also, because we don't want to, you know, give the impression that we have a degree of certainty when we're talking about actual numbers. When we talk about the push factor map for Turkey. So there's a lot going on there. So we use 240 different types of events. And I don't know off the top of my head exactly what events were the main drivers. They would be different in each of the provinces. This particular map that I showed is not controlling for population size. So when there's more population, there's more events. But we have done a lot of work also in a joint publication with IOM to control the population size. And what we also do in Turkey is we know very well the number of Syrians registered in each of these provinces. They may not be present, but we know the number of registered Syrians and temp protection. So we can also look at the number of events per capita in that way, which adds a lot of value. Excellent. Thank you very much, Teddy. So then I would like to move to the next speaker, Michael Clemens. Thank you so much for joining us from Washington. We really appreciate that you've made the time so early in the morning. So Michael Clemens will now speak about the effects of economic development on migration, which is sometimes known as the external dimension. Michael, over to you. So I think you're still muted. Try to unmute, Michael. You can't hear me. Oh, I can hear you. Yes. Great. Thank you. I apologize. I'm not one of these professors that is really good at this right now. Thank you so much to you, Julia, to look us to all of you for the chance to learn from you. There's so much expertise about this area from data scientists, demographers, sociologists at this meeting and what can an economic historian like me contribute. I want to talk about what some people would think of as the very long term. I think of it as the medium term, which is the next 10 to 30 years and the relationship between immigration and economic development in migrant countries of origin over that time period. So I'm thinking of children who are about 10 right now when they are making migration decisions, how will their decisions be related to economic development in the countries they come from? Now, for a lot of people, and it's entirely reasonable, this question has a very obvious answer. When you're in a migrant destination country and you see many, many people around who have chosen to not to live and work in the countries they come from where there's limited economic opportunity, but instead to take advantage of opportunities in destination countries, it just stands to reason that when there is more economic opportunity in those places, fewer people will be making that decision. Absolutely entirely reasonable. And I want to argue that there's a sense in which that's absolutely true, but I want to confront that with two facts. So here's one. Julia mentioned that these are coming from two research papers. I'll give a link to my website at the end. This one is from a paper joint with Maria Piamendola at the University of Milan, Bicocca. And this is looking across people here. This is a survey data across individual people. It's from the Gallup world poll. So you're looking at data from 125,000 individuals in 24 low-income countries. So this is Ethiopia is in here. Mali is in here. Afghanistan is in here. That orange hump is the distribution of income at the respondent level where zero is the mean for each country. And that blue line is the probability that a person at each level of income tells the interviewer that they are making active, costly steps currently to imminently prepare to emigrate permanently from the country that they're living in now. Not aspiring to emigrate in an ideal world, not planning to emigrate to the next 12 months, but having said yes to both of those things, making active, costly preparations to be just about to leave, you can see across the income distribution, not only is that not falling or kind of flat, it is enormously rising across the whole income distribution, the probability of roughly triples that people are just about to emigrate. Complimenting that, backing way, way up to the national level now, this is macro data at what a unit of an office is not people, but countries. Here the horizontal axis is country level income, average income per capita, adjusted for prices. And the vertical axis here is the fraction of people born in that country living in any other country. They were born in the country of origin, but they've emigrated. This is all developing countries and it's pooled data for seven different census rounds over the last 60 years, including up to present. And here again you can see countries that are middle income countries at that $10,000 level, which is roughly the level of Tunisia or Philippines right now, really quite developing countries economically. The tendency to be living outside your country of birth is about triple what it is for low income countries at the bottom range. Now, I've been pointing this out for many years. I get many reactions and all of them are intelligent and reasonable and I just want to talk about how to think through this, what to many people is really strikingly counterintuitive relationship. Is this just reflecting high income microstates, little islands? Is it mostly reflecting south-south migration? We can carve those things out. Here's the exact same graph you saw for all immigration pooled together in all of these seven census rounds. If we drop the lowest quartile of population, all countries below $2.5 million of population and we count only emigration to what the World Bank defines as high income countries, the graph looks like this. It's quite similar. The absolute rise across that income range went from 6 percentage points to about 5 percentage points. The relative rise greatly increased instead of the tendency to emigrate roughly tripling across that range and how it rises by roughly a factor of 10. This is definitely not confined to unrepresentative slices of immigration. Another big question that comes up is what is the path of countries over time? As economic development and emigration evolved in a country, do countries tend to follow that path you just saw? The path in that graph is not a snapshot in a moment of time. It's what we call pooled data. It combines the seven census rounds over 60 years, 59 years, and variation across countries. It's combining variation across and within countries over time. What if we isolate the variation just within countries over time and trace their paths? Here is a snapshot of where countries ended up in 2019. This is all developing countries that experienced positive growth over this half century. These are all the developing countries that grew. Again, we have dropped the tiny countries below 2.5 million population. The vertical axis here is counting just emigration to high-income countries. What was the path that these countries took from where they are in this snapshot? Hypothetically, if they were following declining paths, this is just hypothetical arrows that I'm putting here. For example, if a doubling of income per capita were associated with a halving of the tendency to live outside your country, this is what those paths would have looked like between 1970 and 2019. Now I'm going to show you the real data on where these exact same countries traveled in order to arrive at the snapshot you just saw. This is how they really moved between 1970 and 2019. You can see of countries that are still developing countries that there are only three arrows on that entire graph that are downward sloping. The answer is yes, very clearly developing countries tend to follow that pattern you saw in the pool data as they individually, in their own circumstances, economically develop over time. Another conversation is, well, sure this is the very long run. What about the short run? A friend and a brilliant person in Devco once told me, you know, for us the short run is tomorrow, the medium run is three months from now and the long run is next year. You know, as an economic historian, I think of something else. But is this long run relationship, which is over decades that you've been looking at, representative of the short run, well it must be in a particular sense, which is that the long run relationship you just saw is a composition of many short run relationships. So if you drive for four hours and in the short run, each hour you're getting closer to Budapest, leaving Vienna, there is no way that in the long run you're going to end up in Munich. At some point you need to be moving west in each of those short run, single hour relationships in order for the long run relationship to be that you end up in Munich. So if it isn't in fact the case that there's a downward sloping relationship in the short run at any very specific moment in time, in the long run, if countries are to behave as they have in the past, then the average short run relationship must also be positive. That's just a point of math. Another very common reaction, and this is a very reasonable reaction, is this a causal relationship? Are we just looking at a correlation data and not causation? I do want to point out that in the 99 countries where individual level exists for the first graph you saw in this presentation, that upward slope of the tendency to be prepared to migrate across income, that occurs in 93 of the 99 countries. And the arrows that you just saw, there were only three of those countries that are still developing countries, not including Greece, for example. There are only three of those countries that are downward sloping. So the nearly universal experience of development is that these two things are associated, and that certainly places a strong burden of proof on those who would assert that the mechanisms of economic development over the next 10, 20, 30 years are really fundamentally different than they have been in the past, that they will generate a relationship between these two things that is of the opposite sign. That's hypothetically possible. I don't think any of us have a good reason to believe that. Another very useful and interesting reaction is what is the relationship between irregular migration, specifically in economic development, because that's a strong positive interest, policy interest, I should say. The most direct answer is that all the data that you're seeing in this presentation, in the way the data are collected, attempt to comprise both forms of migration, irregular and irregular, lumped together. So they are not necessarily informative about patterns in irregular migration. There's a notable new paper at IZA by Axoy and Pudvara that attempts to measure tendencies of this kind focusing migrant selection, specifically in irregular migrants. And that's an area where a lot more research needs to be done. Of course, irregular migration is inherently difficult to measure. I do want to point out that the irregularity of migration is in part consequence of policy. So if we were to shape policy based on forecasts about irregular migration that are in fact better themselves shaped by policy, there's a little bit of circular reasoning. And it's a little bit like setting a minimum age in Austria at which you can buy vodka at 18 years old, and then interviewing people who are 17 who are illegally buying vodka in extra-legal circumstances and asking, well, what is it fundamentally about them that is making them buy vodka in these extra-legal, irregular circumstances? Is it their lack of education? Is it a fundamental lack of maturity? No, no, the fact that they are 17 and buying irregularly is shaped by the policy which makes the regular channels to buy vodka at 17 unavailable to them. So that's something we certainly need to take strongly into account if we are to study the determinants of regular migration and use it to shape policy which itself determines regularity. Finally, and I hope I have just a couple of minutes to wrap up, a real reaction I got from a brilliant and highly experienced expert in Brussels was, look, I have talked to migrants and they tell me themselves that if they had more economic opportunity they would think they are lying to me. And that's absolutely not. And this is where I want to get to the point that there is a sense in which that's absolutely correct. First of all, we just need to disabuse ourselves of the model in our heads that this is about individual choice. The best way I know how to do this is to show you the relationship between child mortality and emigration. So the vertical axis here you saw before, this is the faction of people in developing countries who live outside their country of birth and the horizontal axis here is no longer economic development, it is child survival. It is the inverse of child mortality. This is the probability that a child born alive in each of the countries of origin makes it to age five. And you can see that people are much more like to leave countries where their children are much more likely to survive. If you think of this as an individual choice that doesn't make any sense and that should make us question whether this is representing incredibly complex structural shifts in this case and very importantly demographic shifts that tend to happen over the course of development or individual choice. It's obviously not the case that people prefer to stay in countries where their children are more likely to die. This is about very complex structural shifts and just to illustrate one of those if you take the individual level data for low income countries only and they were aligned at the mean, they were demeaned. Here we're not aligning them at the mean and we're putting in data at the individual level from 99 different developing countries. This is a graph of 650,000 people's answers to the Gallup world poll data about migration plans. Individual income on the horizontal axis, on the vertical axis, the probability that people are actively preparing to emigrate, that's what the economic historians call the immigration life cycle going up and up and up until you get to the median income of upper middle income countries separate now that exact same curve by education, exactly the same data. That green line is the same information for people who have attained secondary attainment, secondary education that the red line is the same line for people who have not attained secondary education. You can see within each of those groups the association between preparing to emigrate and income is either flat or downward sloping. If you were to ask either of those people if you had more economic opportunity would you say they could say yes and they'd be right, they're not lying to you. However, economic development is also a structural shift. It is not just a moving of people from those curves, it is a shift of people from the red curve who are essentially universally in development and the fact that people who are more educated tend to be more likely to emigrate means that at the same time it can be the case that in aggregate economic development produces more emigration and any given group of people, even every single individual could correctly tell you that more economic opportunity would make them stay at home. Those things are not in conflict, they can both be true nobody's lying to you and I just want to close with this thought that the more people from developing countries that you see around over the next generation in rich countries of migrant destination the more you can be sure that economic development is happening in those places the papers I've referred to are at this link and thank you very much. Absolutely fascinating, thank you so much and thank you so much for also speaking to the non-economists among us in very clear and straightforward terms. I think it touched upon very important policy issues that we are all or policymakers in Europe particularly are grappling with and I also have a question from a policymaker for you namely from Manfred Kohler from the Austrian Ministry of Interior and he's interested whether legal migration from poor countries can really improve these countries structural problems he calls them and as example he puts he refers to brain drain so I think the issue he's referring to here is regarding the attraction of talent to the EU from third countries and what effect that has on developing countries is the way I understand the question. Thank you very much, that's our question. If you could spend the rest of the day talking about that I just I want to point out that there is a fundamental disconnect in any discussions of lawful channels for migration and it's that for the reasons that Mr. Kohler accurately points out developing countries of migrant origin and the constituency there is mostly for lower skilled migration it's exactly the opposite in countries of migrant destination where people are highly suspicious of lower skilled migration and much more welcoming of higher skilled migration I think those concerns are legitimate I do want to point out that there's nothing inherent about the movement of people even skilled people that must cause a brain drain skilled migration if that is what Europe needs and that is what Europe is willing to admit can be planned for and can be invested in and there are experiments going on right now by very forward looking aid agencies GIZ and enable in Belgium in partnering with migrant countries of origin to invest in skills for those people so that number one there isn't a skill drain the skills that are migrating to Europe are skills that have been invested in for that purpose and number two the people who are investing in those skills are the people who will be directly benefiting from them which are the people at the destination so absolutely this is a concern in unregulated migration but there are so many opportunities that have been touched to regulate migration in more constructive ways that legitimate concerns like Mr. Kohler should not be a reason to be suspicious of migration itself inherently thank you very much I think that's really interesting so I think you touched upon an interesting point that I would like to discuss again in the panel discussion namely global skills partnerships and that idea of labour partnerships but before we turn to the panel I would like to give the words to Matias Pryk who will talk about uncertainties and preparing for uncertain migration futures with a specific focus on policy making and policy makers thank you very much Matias can you hear me? we can hear you and we can see the presentation thank you very much Julia for the very kind introduction thank you very much for this wonderful conference I learned so much already today and also this panel I think is really fascinating what I've heard so far I would like to take the opportunity and to talk for the next couple of minutes about the question can we prepare for uncertain migration futures Jakub Bicek my collaborator on the quantum project he has introduced and presented briefly this morning we have worked on a sort of a typology regarding uncertainties and I would like to take it now a little bit further into the policy field let me just start with the key issues basically of this talk the first point is there's a propensity to make predictions to know about the future and this is inherently human we all want to know what's tomorrow when will the COVID situation come to an end finally what will be my next job where will we live in reasonable future when it comes to the collective obviously this turns out to be much more complex and through some explorations and proliferation diversification all what's happening at the most structural levels of course creates impacts and is relevant for migration processes in the future we've also learned this morning and heard several times that the requirements for prediction and forecasting exceed the ability to do so and obviously having this sort of toolbox with different instruments available for making meaningful and reasonable predictions about the futures of course it's a very good practice but it does not prevent predictive failure so the key issue I would like to address here is how do we cope with predictive failure in order to avoid governance failure or how do we de-link basically governance failure from predictive failure I think the key point here is that the challenge is to design some sort of migration policy and governance structures and infrastructures that reflect the likelihood of predictive failure so in some way not only policy makers but all sorts of stakeholders need to prepare for both the trends the likely and the unlikely trends and obviously also the shocks and the impacts of those shocks and unfortunately I think we also have to prepare to be unprepared let me start with this predictive failure governance failure link with the COVID example this is what we see here our scores of an index that has been developed by the Centre for Health Security at John Hopkins University it shows the level of preparedness of 195 countries in the world with regard to global health issues and in particular with regard to pandemics what we see here that there are the highly developed countries at the top of this which suggests that these countries obviously are the best prepared for any sort of global health risk this also index it was first published in October 2019 has received quite high exposure Trump has mentioned it in February saying that the US is the best prepared for anything like the COVID pandemic what we see unfortunately today is that they with past the one million death threshold and we realise that there is some sort of illusion of preparedness and what if thinking in terms of scenarios what may happen if there is a whole panic has come now to some sort of crisis management this crisis management or crisis perception of course is not only in the area of public health but also in we know quite well from others including migration and the perception of the constant crisis in the area of migration I think we all familiar with there are different definitions of a crisis I would like to refer to a definition coming from complexity science which says crisis is a situation of a complex system in particular when the system is temporarily dysfunctional I think that's something we would confirm with regard to the migration system if we may call it like that sort of a constant perception that migration policy and governance do not and does not address and cope with the full complexity of migration and what is the complexity of migration we have heard a lot today already about complex aspects of migration that's the issue of superdiversity and not only more migrants but really more diverse micro populations are migrating the boundaries between diverse micro groups are getting easily blurred increasingly challenging to identify and to target particular groups of migrants and to be migrants we see an increasing landscape of actors that are engaging in the area of migration including NGOs and other actors and we see obviously the whole now notion of biased forces and so what can we do about this complexity when it comes to migration or the adequate responses first of all I would like to start with the understanding that professor a Nigerian geographer has already stated a half century ago he basically says we need to consider migration no longer as linear, unidirectional, push and pull cause and effect movement that is a circular interdependent complex and self-modifying system in which the effect of changes in one part can be traced to the whole system so what we basically need to start with in terms of understanding migration which is that we have to understand migration as part of an adaptive system a narrative complex system Franz Willekens who is also part of the Quantmic Project he has outlined some characteristics of a complex migration system I don't go through this in detail but what we I think have to bear in mind that migration processes are basically building up from the bottom so from the individual, from individual agents and not only migrants and would be migrants but obviously all agents that are involved in migration processes and these processes obviously are sort of triggered by internal and external forces there is constant interaction, exchange of information exchange of resources etc between agents and there is some sort of patterns evolving so the emergence of certain networks but also the emergence of migration patterns at a higher structural level I think is a major feature of complex migration systems what's relevant I think in this kind of system thinking is that systems can be out of equilibrium sometimes they are sometimes they are in a very stable steady state situation but usually there is some sort of disequilibrium or some sort of instability what is also part of the feature of complex system is that they are affected by the epistemic and the LAR3 uncertainties and for that reason there is some limited predictability so in the spirit of Mabugunje who is asking for more system-based thinking when it comes to migration complexes I think one key is that we have to understand migration tribal complexes what tribes migration we are quick in identifying a few factors in fact we have done a systematic tutorial recently as part of the migration project we also mentioned already today where we refued almost 400 studies who have looked into migration tribes and on average those studies study two to three trivers so a very limited number of migration tribes basically our field is very much characterized by a reductionist perspective when it comes to migration tribes so the claim here is that we should more look into more complex tribal configurations and interactions rather than searching for some sort of root causes so when it comes to migration systems and change in equilibria I think you take more of the features in the account that are characteristic of systems more generally so whether they are personal or credible cause some sort of feedback, uncontrolled feedback, cascading effects, extreme events etc unwanted side effects we know all that and it's illustrated here by this bucket type of figure some smaller changes in migration flows we see all the time fluctuations are on the trend but sometimes a sudden shock can shift an entire migration system to a new level this can be a sudden shock or a gradual change of some underlying configuration that is driving migration more generally but there's some sort of tipping point that may occur where a completely new level of migration is reached some sort of regime change when it comes to a particular migration process what is also I think important to understand in particular when it comes to identifying the trivers of migration we have to first understand that migration trivers are usually embedded in the broader configurations or triver networks or triver environments and that we can observe for one and the same outcome, migration outcome very different courses or course of configurations but this is what we also study in another Horizon 2020 project in the MiKNEX project we study so-called INS causations so where individual thriving factors are actually embedded in broader migration complexes so when it comes to sort of you know regime changes and shifts in migration systems we obviously have a whole range of examples where this has happened and we sometimes call these shocks or events like swans like the disaster almost a decade ago in Japan with the nuclear accident in Fukushima so this is often mentioned as a black swan that obviously has led to large-scale population navigations etc but also some partial return as we can see here on the right hand slide but it has led to a new equilibrium, a new balance a new population composition and combination in that region and basically the black swan are not necessarily a black swan but a white swan as Halep would often say also with regard to other major events like COVID it was foreseen, it was mentioned by experts not exactly when it may happen but that it can happen and will happen at some point that makes a black swan a white swan and this is what happened here in Fukushima and in other parts of the world like here in the Aral Sea region where we see similar systemic risk and that's another concept I would like to refer to the systemic risk as a risk that has the potential to destabilize a system including a migration system so migration control guys are coming back to that concept I think is in some ways system imminent it's part of a system adaptation and the systemic change is constant, is ongoing it is an intrinsic part of interconnected system and we also realize that adaptation and the limited control capacity are some sort of some key features of complex migration governance and here I come to the governance aspect I think it is key first to stakeholders including policymakers, scientists, researchers et cetera and others they call this think collectively of how to adapt and mitigate and prevent the manifestations of systemic risk so what does this mean first it requires that we understand the underlying causes of systemic risk that are relevant for migration obviously but in systems in broader systems in interconnected systems obviously an event may easily and quickly trickle to areas that are relevant to migration and migration processes so here a sort of a list of factors that may cause systemic risk I don't go through this in detail I rather focus on the reasons that may cause policy failure because policy and the way governments respond to migration forecast but also migration outcomes so real-time outcomes obviously is part of a migration system and it's an element of an element of uncertainty so in here I think it is key that we understand that migration policy has some limited effectiveness only I mean this is not rocket science it is well-established in literature that migration the element of managing and directing and regulating and controlling migration is only possible within some limits it is I think important to take into account that most policy interventions create some unintended effects and obviously this also has to do with the targeting aspects and policies usually have some target group whether it is the microns origin or the microns type or the asylum seekers some other microns groups are highly skilled that are targeted by particular type policy but it has some still over effects on other groups and also of course regions and areas where microns coming from another aspect of course is the policy receptivity so policies are not so they equally received and internalized and responded to by diverse populations there's the issue of policy timing there is the well documented Matias? I think we need to time to my last slide thank you very much so let me look to the last slide basically how can we get to the starting question how do we prepare for uncertain migration I think it is important that we increase the capacity for good enough for casting so a lot of us are getting today with our theater requirements addressing epistemic and certainly when it comes to knowledge understanding agency human agency I think is very key there is interesting research going on in that area and of course understanding systemic interconnections enhancing smart and mainstream migration governance this is I think a research area that is evolving rapidly mainstream migration in the area of governance so creating multi-actor networks enhancing policy coherence bringing migration policies in line with other migration policies of course also nurturing everyday space academic and public discourses and last but not least we need nevertheless be prepared to be unprepared thank you very much fantastic thank you so much Matias so we have about 12 minutes now for our panel discussion and there are a few questions that still came in for Michael which I would like to translate or transmit so one was from Marie Helenius from the Finnish Museum of Interior and she asked does economic development of a certain country have any effect on the return migration to that country and there is another question from Bernd Weber from Leiden Institute for Economic Research and he is asking what you make of the Schneiderheim paper with where the OECD paper where they are you in regard to the migration hub I think we spoke about this reason Michael but he is asking he would like to hear something to address sort of the fixed effect approach and then there is another question by André Grøga regarding the problem that we don't know the counterfactual so how would EU origins have developed in the absence of intra-EU migration so the time is short I'm not sure whether we have time to answer them all Michael but maybe in a few minutes that would be great thank you can you hear me thank you very much I'm not looking from the finish it's the Finnish Ministry of Interior Finland is that what you said yes about return migration so what you the individual level data that you saw are from people surveyed in their country of origin talking about their their imminent preparation to imminently emigrate the country level data you saw are about net changes in migrant stocks so that is the sum total of emigration and return migration and it combines the movement of people out and the movement of people in so the short answer is the analysis you saw does not directly address the tendency to encourage return migration I can think of extraordinarily clear examples where that has happened Korea is maybe the poster child of that there was a surge of emigration from Korea that accompanied its takeoff into lower middle income upper middle income and now high income status and certainly return migration to Korea at a certain point was very positively associated with migration the same has happened in Mexico in the United States certainly there was a net flow out in recent years the net flow from the United States back to Mexico has been negative that's been true since somewhere around 2006-2007 until just before the crisis and of course in this current crisis all bets are off in those late stages of developments there is very clearly a positive relationship between development at home and the return of people I don't know of good evidence for that in early stages of development if the it seems very clear that if the Gambia behaves like other countries of its income level and again it's such a nearly universal experience that that is quite a reasonable expectation that there will be a net flow out but whether the during that range while Gambia remains a developing country which is going to be true certainly in any reasonable scenario for the rest of our careers and perhaps lifetimes whether the economic growth in the Gambia will encourage return I don't know of good evidence for that specifically the second question was it Bernd Weber was about this paper claiming that that all these relationships are just an illusion and that's not actually the path that countries follow over time there is a in my website which I linked to at the end of the presentation which is just mclim.org it's mclem.org you'll find a link to a paper that goes on for pages about this issue the bottom line is that these results are very highly robust to country fixed effects which is just another way of saying that the paths that individual countries in their own circumstances follow over time are extremely similar to the path to the pattern that you see across countries that is the immigration life cycle is not just a pattern that you observe in a snapshot of countries it is indeed the typical path that average countries have taken over time and in that paper you'll see for example that the cross country relationship between the prevalence of migration and the development level is the percentage change in one versus the percentage change in the other is 0.3 that is a 100% increase in GDP per capita in across a snapshot of countries a doubling of income per capita across a snapshot of countries is associated with a one-third in the tendency of people to live outside it if you'd ignore all information comparing one country to another and just look with it at the path that countries have followed across time you get the same answer of 0.3 one's 0.38 and the other's 0.35 but they are really extremely similar that is we do in fact learn a lot about the path that low income countries are likely to follow as they develop by looking at the experience of countries that are somewhat richer than them in fact when you think about it there should be a very strong burden of proof on anybody who claims the opposite that in fact countries that are somewhat richer than the Gambia might have very high migration but that tells us absolutely nothing about what's going to happen to the Gambia as it develops that's quite a strong claim and I think anybody who studies economic development would agree that our presumption should be the opposite without strong evidence in fact a major technical error was made in that paper unfortunately I talk about it at length in this analysis but this is not really a controversial issue you saw the data that I put on the screen before it shows the arrows the raw data there's no statistical analysis there all those countries have been moving up and to the right it's nearly a universal experience of developments that more and more people tend to live outside as countries develop and if anybody is going to argue that that's just completely illusory or uninformative about the future experience of development I have a very difficult time understanding those assertions I've talked so much that perhaps I should stop there Julia thank you very much so we have five minutes left and we have questions for Teddy Brits and Matthias I'm going to pose the questions to you and then I would ask you to really respond in two minutes so Matthias one question to you is regarding the limited effect of migration policies could the effect increase if migration is understood as an adaptive complex system the question to Brita is in regard to whether Germany's new system would allow stakeholders to react earlier than in 2014-2015 and whether the German presidency has been successful in joining efforts of EU countries and agencies and building up a network which coordinates and validates existing projections forecast and scenario building so maybe choose something that very briefly and Teddy a question to you would be do you have a concrete example of when the information the other provided led to a specific policy change or intervention maybe we can start with Matthias thank you for the question I think it's a very important one the answer is I think there's not a deterministic mechanism that basically leads to a certain migration outcome when it comes to migration policy migration policies and their effectiveness depend a lot on the way they are designed, the way they are communicated and the way they are targeted and I think what we know is that policies that are you know going completely against the current are largely ineffective they are largely ineffective with regard to the initial objectives or the stated objectives of a policy but it does not mean that they don't create the effects they create effects but not necessarily the intended but the unintended and I think that makes it so complex to really monitor and maybe also the multiple ways migration policies can affect migration processes and in the short term but also in the longer term so we know that migration policies have some effects I think that's kind of a stylized fact but with regard to the level of effectiveness I think we have to look really into particular context and the particular migration policy instruments so this is I think we cannot really say anything meaningful in more general terms. Thank you so much Matthias. Prita. I have to unmute myself. So first question the question was Germany will be better prepared for the situation like 2015 I hope so because that's exactly the aim of our new tune and I'm quite confident that we still are already better prepared now by creating this network of experts because from my point of view having this kind of structured dialogue on the migration situation creates a kind of pressure for the decision takers to act and this is from my point of view the most important thing I very much agree on this aspect on the white swarms because I think Covid-19 could have been foreseen as well because there were so many studies before even in Germany in 2012 forecasting exactly a scenario like Covid-19 but it's just about our way of handling this kind of structural risks or systemic risks so this to be brief on the first question and the second one if the German presidency will be successful in creating this network I hope so and I'm quite looking forward to the presentation of Susana and I hope she will tell us something more about the European plans and I'm really proud that all these plans will be presented starting to implement them to the German presidency thank you so much Peter Teddy you have the closing words no thank you very much some responsibility so my question about specific policies that have been based on our work well you know for ASO early warning means as early as possible and so much of our work focuses on events taking place in origin and countries of transit and of course we don't have any say over the policy cycle in those locations what tends to happen is when there's a sudden influx of arrivals or applications for asylum we tend to run our analysis and make adjustments whether we think the influx is likely to continue or whether it's just about to abate and we've done that in for Latin America and several other locations and so this kind of feeds in a little bit into the operational response let's say but as far as policy is concerned so I would echo what British said I'm also proud to be a part of the European project and the way that it's moving forward and I would also congratulate the European Commission for the pact last week because this is a really clear example of the two ways street that I spoke about earlier where people like me and our speakers today we are charged with trying to produce better evidence but we also ask the policy makers to give us the chance to have better data and the European pact that came out last week has several legislative upgrades which includes much better data so we're really licking our lips and we're waiting for that data to come through and we hope that's all approved and endorsed and believe me we'll take forecasting to the next level so final notice watch this space thank you so much unfortunately we need to close this session I would have loved to continue the discussion I'm getting a lot of messages in the chat thanking all the speakers for the interesting presentation and I can only say con that thank you so much for joining us for sharing your interesting insights with us we will take a break now and we will back at 3.45 please do join us again for the last presentations and panels of the day thank you to all of you and see you soon you very much thank you very much