 And more specifically, I would like to focus on migration foreign policy, which is in my view key for any sustainable migration policy fit for the challenges of the 21st century. And to quote John F. Kennedy, the purpose of foreign policy is not to provide an outlet for own sentiment of hope or indignation, it is to shape real events in the real world. And in order to be able to shape three elements are in my understanding essential. One, we need a clear vision, a clear goal of what we want to achieve, which international system we want to create. Two, we need a clear understanding of what the world looks today, but also the challenges and the future ahead. And three, of course, the strategic use of available resources to achieve the goals that we have set. In other words, a project future forward looking migration policy should be evidence based, building on analysis, strategic foresight, and scenario building. Very sorry to interrupt you, Tobias. I didn't know if you wanted to share slides. We can't see them currently. If you did want to share slides, otherwise we could work with Lucas who can who can display your slides if you have some. Just I wanted to make sure that that is working. No, I have not prepared any slides. Okay, that is totally fine. Sorry for interrupting. No worries. As we have already heard today, forecasting is an inherently difficult task and very much the art of the possible. My understanding for two reasons. Firstly, we are operating in a complex world of extreme uncertainty where multitudes of factors may cause or contribute to migration. This is a world we're living in fueled by global power conflict, international conflict induced displacement, the possible disastrous impact of a pandemic as we have heard, an economic depression that might have a destabilizing impact for years on prosperous and poverty-stricken societies alike, climate change, smuggling and human trafficking and countless other challenges that may have a significant impact on global migration. Secondly, we're operating in a field where research is still developing and predictive methods and tools are evolving quickly despite existing present methodological, statistical, technological or even legal limitations. We are therefore very grateful as co-organizers of this event today that so many distinguished experts joined us today to discuss the way forward and exchange best practices. Richard Fontaine, the CEO of the Center for New American Security and former advisor to US Senator John McCain, recently said in a foreign affairs article, ultimately, the unpredictability of world events put a priority on human judgment and undermines Richard's formulas. That means, as was mentioned already a couple of times today, that predictive tools and methods may not be able to detect the black swan amidst all the white swans. It may not be able to predict single or migration events. However, in a world of uncertainty, it might contribute to a better understanding of present and future developments in global migration, thus reducing uncertainty and bringing more light to the complex phenomenon of global migration. This is the reason why also, of course, we in Austria are working on our own predictive models. And to name just one example, I want to highlight our current so-called Mitra's research project, where new methods for monitoring earlier recognition and trend analysis or migration flows are being developed. We are here considering a fusion of information from satellite images, open resource, but also social media data that we want to use to estimate to better predict developments in the geographic setting of North Africa. Unfortunately, I'm not yet in a position to present more concrete results today, but they should be available by the end of the year and will then be evaluated. Starting from these initial observations, I want to make two initial remarks. As Liz Collette said in the morning, exact predictions are not impossible, which is in particular true when it comes to major disruptive events. But I still think the highlight shown today shows that forecasting foresight into never a billion efforts for the work to be undertaken. And what I want to underline even more is that forecasting is not an end to itself, but it should be a basis for our actions or future policy, a guideline on how our limited resources are best used to shape the world we live in. And this brings me to the second part of my intervention today. How can we better link forecasting and foresight into narrow building with concrete migration policy? Ladies and gentlemen, just as forecasting, migration and asylum policymaking is itself also the art of the possible. I've heard some remarks on the pact today. And let's see what it really brings. But the truth also is on a new level, we have for many years been struggling to create a better and fairer asylum and migration system. And given the multitude of drivers of migration, clearly a whole of government approaches needed to achieve improvements on the external dimension of migration. For today's purpose, however, I will restrict myself to the field of migration in a stricter sense. And I think what we need predictive analysis here for is answers or answers might be asking to much at least hints and how and where we can react much more quickly to possible future crisis and help people in need of protection as early as possible already in the region of origin. But also, of course, on the other hand, forecasting can contribute to combating era to era of migration and breaking the business model of smugglers who spread lies about the information and destination countries and cause much suffering along the routes. What we need is building scenarios based on sound migration forecasting tools. And this can be key to a more effective migration management, a more effective migration government. And this should help us guide our interventions and investments. If I may highlight this in the three fields. One example for me is the area of information campaign. Predictive analytics may inform us where social media and online communication campaign with potential migrants are most needed to protect people from false promises and misleading information that might have. Already today, countries like Austria and many other countries in Europe around the world are investing strongly in information campaigns. We do so in countries from North Africa over the Western Balkans to the Silk Road countries. We are focusing on issues like the prevention of illegal migration, but also alternatives that might exist in the country they are present or possibilities for voluntary return. And not only we, but many member states are highly committed to putting an even bigger focus on using information campaigns and modern communication tools. But of course, we would want these campaigns to be even much more targeted geographically and on content. And I think this is where forecasting could be very useful and helpful. If you allow me one side note, it will also be very interesting for us in the future to have more knowledge on what social media platforms can do in terms of predictions and how research and governments could cooperate with them to make better migration prediction. A second field that I want to shed some light on is the connection between migration, forecasting, and scenario building, which could give us clues to pursue more targeted projects and capacity building endeavors in and with countries of origin and transit. Ideally, we could arrive at concrete recommendations on where to better direct our resources to create an environment where one possible scenario on future migration development that is more favorable becomes more likely than another scenario that is less favorable. It might show us where we need to help improve the economic and social situation and resilience of potential migrants in countries of origin and transit, giving life to an effective whole of route approach. Already today, many European countries, including Austria, are helping third countries, for example, like Tunisia, in improving their border management systems, but also in creating economic perspectives as an alternative to irregular migration. We promote volunteer return activities on the Western Balkans, and we will become more active in providing protection in Eastern Africa. We are of the understanding that we are probably moving in the right direction. But as I said, we're living in a world of limited resources and predictive analytics could contribute to making our actions even more effective in terms of reaching the goal of a really better and fair humane international asylum migration system. In short, what we want to achieve and where we hope that forecasting can help us, we hope to move from a reactive migration management system to a proactive migration management system, making better use of all available resources. And last but not least, not only in the third countries, but of course, also on an EU level, we want to be able to detect and predict mixed migration flows as early as possible. So forecasting can help national border forces and front-takes to be prepared to protect our common external border, improve the reception capacities of member states, and increase the general preparedness for new challenges. I think as you can see, the hopes we harbor in new predictive tools are very high, and we hope they can contribute to the creation of a better international asylum and migration system that takes into account the interest of all the stakeholders involved, the migrants and refugees themselves, the countries of first reception and transit, but of course, also the interest and concerns of receiving societies, of us enjoying democratic legitimacy. To this end, we would like to support science in establishing functioning and effective predictive tools. We are very keen on knowing which models have the best or real predictive value, where can we invest more? I think a lot needs still to be done, and we trust that this cooperation between science and government can contribute to making the complex nature of migration a bit less complex, opening the pathway for new and more targeted policies. In an ideal world, we could possibly even formulate proactive migration policy that provides the basis for actions even before a crisis, threats, conflicts, and displacement at large scale on an international level even occur. Ladies and gentlemen, let us move forward in this positive spirit to reach these hopefully common goals. I wanna thank you very much for your attention and I'm happy to answer any questions you might have. Thank you very much. Thank you very much, and Tobias for this intervention and for Austria's position on this. It's a very interesting to hear. You mentioned several things that stuck in my head. First of all, you also mentioned that you acknowledge in the Austrian government acknowledges that this research is evolving, it's sort of exploratory. Many countries are testing out the models and national organizations, they're testing out models. And you also mentioned that there are high hopes for their ability to, in some time down the road, predict migration more accurately. In this exploratory process, is there an exchange at the technical level? Of course, this conference is part of this exchange, but also between the different teams in the different countries developing these models on which approaches work best. That would be one question I would have as somebody that's working in this area and interested in the exchange of best practices. Is this something that is being done at the technical level? Well, just as you said, the very purpose of this conference, which was co-organized by IOM and the Austrian Ministry of Interior, was to gather as much expertise as possible, have an exchange on best practices and find the best way forward. And I think this conference proved very valuable on this one. We are testing in different fields, different models, and we are cooperating. As I said, the Mitras project that we are currently following is a joint venture between different ministries in Austria, but also private stakeholders. I know I cannot speak for others, but there are different branches within our ministry that all look in their respective fields on how forecasting and foresight can be improved. I listened very carefully to the intervention of IASO, and I know there is very good cooperation between our asylum department and IASO, and we also part there of a pilot project if I'm not misinformed. So yes, of course, in the respective fields, in their respective branches, there is cooperation going on. And I think my colleagues in the field of frontiers and illegal migration do the same with their respective colleagues. We are also looking very much forward to developments on the EU level and to a certain degree, we are also following up on the conference where we I think met one or two years ago under the Finnish presidency. So the risk is exchange going on and it's growing. And yeah, that's about it for the moment. Thank you very much. Thank you. Thank you Tobias for those clarifications. I cannot help myself but to make one comment, which is you said, you know, that the Austrian government is supporting science and I just want to second that and speak on my appreciation for that position because as we see in the news every day and on every government around the world, it's doing that these days. So thanks for that. With that, we're moving to our second speaker, Susanna from DG Home at the European Commission. As I mentioned, beginning DG is working on a feasibility study there, trying to pull in various data sources to test the viability of forecasting the EU and we are now excited to hand the floor to Susanna for her presentation. Thank you. Thank you very much, Jasper. I hope that you can hear me well and you can also see my presentation. Good afternoon, ladies and gentlemen. Thank you that you've been with us for so long and I'm extremely happy to present the EU approach, especially that the previous speakers faced already a lot of expectations concerning the content of the pact that was adopted last week. I'm absolutely fascinated by the level of the presentation and the richness of the knowledge that you transmitted today and I'm also a bit intimidated because I see a very big responsibility for the Commission and for DG Home actually to channelize all of this knowledge into policymaking and also into a system which could work for everybody, for all of the stakeholders. And that's why I'm very proud actually that the Commission also got so far that we can also bring some elements into this conversation and we can also present maybe an added value in this since the Commission is always in the crossfire in the migration development and it's very difficult for the Commission to make all of this happy. So I hope that my presentation will give some hope. So just to start, yes, last week the new pact on a migration on this island was adopted with an extremely complex set of documents. There are a lot of legal initiatives and also there are initiatives which are non-legal nature. There is a very rich evidence paper which I would like just to reflect to the previous speakers where they plead for evidence-based policymaking. I think that is the paper, one of the first one which summarizes in a complex manner also all of the statistics in what happened in the last year in the migration field and what characterizes the migratory flows in the stock of migrants and refugees in the EU really with statistics and facts and figures. And there are also recommendations in this pact also concerning migration forecasting and also, for example, the search and rescue operations. We are also very happy actually that all of the developments that we have learned of the last year could really lead to concrete commitments into this pact concerning early warning and forecasting. And there are two pillars basically in all of these documents which include the description of our endeavors and one is the communication itself about the new pact which speaks about the improvement of effectiveness via preparation and foresight and the need of an evidence-based approach. And the other one is a recommendation for a so-called migration preparedness and crisis proof print. It is a self-standing document and this is a commission recommendation. So basically this is the commitment of the commission how to set up and how to set up such tool and how to strengthen the contingency and the preparedness of the EU institutions, the agencies and the member states with anticipation and also how to move forward if there is a crisis and how to coordinate all of the actors. So basically these are the two documents which are relevant but the more detailed rules are laid down in this migration preparedness and crisis blueprint. It is very important to set out that the pact is the fruit basically of a long learning process. 2015, the so-called refugee crisis was also a wake-up call for the member states and also for the commission and the EU agencies and we learned a lot since then and a lot of things have happened also in the field of preparedness, in the field of information gathering and contingency building and also in the provision of a rapid response capacity that we could also see also in the last year. So basically the pact already summarizes what we have built up and what we learned in the last five years. To set up an EU-wide early warning and forecasting system, it's very challenging. I have to be very sincere about it and it's not a silver bullet and it's not the ultimate solution because we are acting in a very complex institutional and political setup but there are many actors also who are pursuing their own political objectives, their own financial objectives as well. So we have to base all of our systems on the understanding how all of these different stakeholders are operating. Now I'm talking about the member states, I'm talking about the external action service, also about the different EU agencies and of course the commission and we cannot forget the third countries, the countries of origin and the countries of transit which have their own interests. And that's why not just a European migration policy but also a such forecasting system, it is not a standalone mechanism but it is just one puzzle in the big picture which has its own place basically and it has their own function in this entire complex machinery in principle. So that is one challenge to work in this environment and in principle to create a broad consensus how the system could work and what would be the result of this system that it would deliver. So we have to very well define the targets what we want to forecast in principle and what should be the output of this product. For the moment we are thinking about of course a multi-agency approach because as already Britta told about the German experience we completely share this sort of multi-agency approach that all of the different stakeholders that bring in their knowledge according to their respective competencies and basically we cannot create and operate any system in a void just with the stakeholders. And that's why we have to build on the member states experiences and also on all of the experiences of the agencies. And I have to say that this sort of cooperation already brought a lot of fruits in the last years also in terms of forecasting because the European Border and Coast Guard Agency and also ASO they developed their own forecasting or early warning methodology and early warning system. And basically there was also one occasion when we could test this system and that was one year ago the so-called caravan-like movements which started to move from Turkey into the EU and from Greece towards the Western Balkans. And with this early warning system provided by the two agencies and also Europe and all of their intelligence and how it was channelized into the political level actually we could achieve with a great efficiency that these caravans didn't start. So that was I think it was one of the best example how the system was completely put already into operation. For the moment what we would target with early warning and forecasting system that would be a short term forecasting. Let's say the next one, three months or six months and we mostly concentrated on irregular migration and within irregular migration to illegal border crossing at the EU external borders. It is also very important really to differentiate what kind of migration we are talking about. So for the EU from operational perspective it has very the most impact for the moment. And of course we have to pay attention what is the action, what is the outcome of that early warning system and how it would feed in into the work of the decision makers in principle. Britta asked already in her presentation how the system would operate and how we could share all of this. So basically the blueprint that I mentioned it sets up a network which is not just a network so it's not 20 people sitting around the table and discussing but it's in principle this is an entire governance of the preparedness and also crisis management. And one part of this preparedness is the monitoring and situational awareness and early warning and forecasting. So actually that you can see on the left side of this presentation. And it would work via a network from the nominated contact points from all of the stakeholders and bringing in their own experiences in principle also for situational awareness and then on that basis for the early warning and forecasting system that would be established in the future. And in the last on the bottom part of the slide with that in the red square you could see what would be actually operational response to that. But what is the action? And basically this is financial support, strengthening of the return, emergency support and constant exchange of information for example or also contingency planning and the contingency increase in within the EU member states and also in third countries and the countries of transit. So basically what we would expect also from this information sharing and forecasting system, these are concrete operational decisions which would be actually then the response at the end of the process which would be the response in the short term. The basis of that system would be actually to have a common understanding about the situational picture in the EU. We are extremely lucky because we didn't start from zero. We actually started from the crisis mode and that was generated also by 2015 and the crisis that all of this information sharing and gathering into a complex situational awareness report. It is started on a platform which is called IPCR platform and it's operated by the general secretariat of the council where the member states and the agencies and the commission, they are uploading all of that information about the situational picture and the commission is obliged on the basis of that information actually to write a report, a migration report with weekly frequency. It is always due on Wednesday. So my team actually is working on this ISA report integrated situational awareness and analysis report today. This report developed into a complete migration and situational awareness report which is really actively fit into the work of the member state authorities every week because there are at least 700 users who are downloading from this web platform this report. This report was also one of the catalystator actually to start with a very complex information gathering process and this is data which is shared on the basis of existing legal instrument and shared with the commission, with Frontex and also with the ASO and they have their respective products that they establish their databases on the basis of a common agreed terminology in principle. And then we also have a qualitative report which are coming from the field which are coming from the EU delegations for example and all of the liaison officer of Europe or the liaison officer of Frontex or the commission also from third countries and which we also base or analysis on open sources. So basically that is the situational picture what we have now and on that basis we would move forward into the next stage which would be this forecasting system. The absurdity of this, the current situation now that we are not in a crisis mode for quite some years but this crisis system of the council is still activated and all of this information exchange also forms part of the crisis system. And the question is and I think that is the political question also to the presidency whether we can have the blueprint and the normal preparedness phase and forecasting phase of the blueprint and the information sharing formally still take place in the crisis system. So I think that is something to look at for the political decision makers actually to move from the crisis mode to the prevention mode and the commission also actually committed that we will provide all of this information and the reports also after the crisis mechanism formally deactivated. We will continue this way. Susanna I'm terribly sorry I think in just a time we have to wrap up. Okay so just an extremely quickly as I said we have a lot of traditional data sources and we also want to look up other data sources which are existing social media data geospatial data and so on and we commissioned a study to a consortium which is called a chorus actually to investigate how forecasting and early warning system would work before data existing data sources and possibly new data sources with machine learning and artificial intelligence technology. So this study is still going on it will be delivered on the 18th of November and that is a good progress and actually there are still a lot of questions to answer basically but this study will be the starting point to start actually the actual feasibility discussions with all of the agencies because we have to still carry on as I said with a multi-agency approach and also with the involvement of the member states because they have to agree to the systems and they have to agree to the outcome of the system. So all of the technical part actually must be exactly understood and agreed with all of us stakeholders. Thank you very much. Thank you very much for Sana for this presentation. Yeah, it seems like you spearheading some really interesting work there that will push the envelope but it sounds like you have quite a task in managing all the political interests of all the various institutions and agencies involved or a big task there and the best of luck to that. There's some good news and some bad news. The good news is we still have almost 150 listeners that are tuned into the session. The bad news is very few of them are asking questions. So I want to remind all the participants to use the possibility to put questions to any of the speakers just write it into the chat function addressed to everyone. And then we can put those questions to the speakers. Don't be shy. I know we have established authorities, state authorities, EU representatives, government representatives, but don't be shy. They're ready for your questions. So please go ahead and ask them and don't leave them all to me which of course I benefit from. Okay. Since there aren't any questions at the moment, I will park my own questions and then in the interest of time, move on to Thomas Liedig and then we'll ask some more questions to Susanna and the other speakers to withstand the discussion. So with this, I would give the floor to Thomas Liedig from the OECD over to you, Thomas. Thank you very much, Jesper. It's a pleasure to share with you some of the results of some work that we actually pursued as a background to our first ever OECD migration and integration ministerial earlier this year and in January, where we had actually the issue of foresight and I'm trying to make the link between actually what we have heard throughout the day and with some of the policy migration management implications that arise from that. So clearly we had a lot of discussions today on this issue of big data forecasting and scenarios but clearly what we put ourselves in the head of the policymakers, how do they feed that uncertainty and the knowledge that is created into their decisions? That is actually a key question that we have to address a little bit less today. But taking a step back, perhaps, before we dig into that, I'd like to recall that actually the development, the future developments that will largely impact on the way how we manage and receive migration can actually be classified in three broad categories. The first is megatrends, the second possible disruptions and the third unforeseen major disruptions. Hopefully this will now get to the next slide. When we talk about megatrends, I will just do this in any case. I will certainly in a minute move on to the next slide. When we talk about megatrends, these are clearly relatively plausible, relatively foreseeable, but we don't really know yet what is, I think my sharing has been interrupted. Can I please get again? Yes, thank you. Could you please try again, Thomas? Thank you. But we don't really know how, when and where they will impact on migration completely. I'm a demographic change in Africa. Which countries in Europe will be most affected? We don't really know. Environmental change, we know it's gonna happen, but how exactly it will materialize and increase migration from which country to which country is really unknown. Plausible disruption. Here we have much weaker signs currently, particularly when we look at the impact that they will have on migration. They are largely unpredictable, but they're completely unpredictable. And we know that they will increase in scale and scope and they could potentially have a large impact on migration, but it's actually unclear when and how this will materialize. And a good example is the whole impact of technological change, which is actually also mega-trend, but how that is creating digital, transparent migrants. How that will impact, for example, on border control on migration management at Portland. And then in the third category, clearly we have those unforeseen major disruptions where we have no or really weak signs currently, which are almost completely unpredictable, but will have a massive and sudden impact on migration. Clearly major conflicts arriving in key origin countries as part of those, as you all know, and probably would take the pandemic a year ago, would have also been part of those unforeseen major disruptions when we look back. So how can policymakers prepare for these developments? And it depends obviously in part of the type of development while we're talking about. Preparing for mega-trends means, first of all, to recognize the importance of these events for migration, to continuously monitor the situation and the policies. Here's something that we have less talked about and which I think is an under-researched area. It's possible thresholds detect an interaction. If you want to create example, there's a lot of debate right now, how will COVID-19 and the asymmetric impact that it's having across countries, for example, here in Europe, how will it impact on inter-European migration? Well, we know that a lot of countries like Spain and Italy, there wasn't much of an increase, much less of an increase in migration, as many feared, following the global financial crisis. But that's not all certain that we won't see a different pattern now, because perhaps a lot of the people who stayed on through the first crisis, they say, well, this is the second crisis in 10 years I'm leaving now. So these threshold effects are not really well researched. And then obviously we need to survey people in origin countries regarding the migration. That's all mega-trend in the bigger picture. Preparing for applause will disrupt. He has clearly prepared, in a way, means first of all to seize the opportunities that are there. So to seize the opportunities notably of big data and then many other artificial intelligence and you name it, but avoid being ruled by that. I think that's very important that policy makers as countries maintain ownership and control over what's happening with those data. Clearly it also means to continuously rethink existing processes and programs in light of these new developments, what opportunities and challenges do arise for my work in this area from these disruptions and to also mainstream it in the administration. I'll get to that point at the end again. And thirdly, to prepare for the unpreceded major disruption. And here we have heard a lot of examples and just now actually to use intelligence and other territory tools to guard emerging signals, early warning systems. And here clearly from a policy perspective, I think it's very important to avoid that we overly base our policy choices on short-term development to avoid the limit of overreaction. To give you one example, that could be a temptation right now to say, well, we could actually scale back the integration services for new rivals, because there's not just new rivals right now. But that would clearly be a mistake because that situation may change relatively quickly. And once you reduce your budget and your planning for that, it's very difficult to adapt it in the future. But it's better an opportunity to reach out to some groups that you have previously not reached and to test new ideas and new tools because this is a good moment to test them. And last but not least, that was also mentioned, to build some contingency plans. There are some plans under the child for migration management and integration capacity in case of major disruption. In addition to all of that, to the specific category-specific preparations, clearly there's a need to improve the overall framework conditions for reaction. And we've actually heard from the example from the European Commission and also from Germany that both at the national government level, there's a need for more policy coordination by also more partnerships beyond the national government. So here, to consider migration and integration as a cross-cutting issue, that connects with other policy domains like education, employment, social protection, trade development, and foreign policies to give you the example of the refugee crisis. It was interesting to observe that as many of you know, as many of you will know, a lot of the push came actually from the decision of Iran and the policy decisions in Iran with respect to the Afghans living on the territory, which pushed a lot of those out. And actually just prior to the crisis, European countries were negotiating the agreement with Iran on the foreign policy side. So, but the link wasn't done between the migration side and the foreign policy side. And we see to all too often, the links also not there still in spite of some improvement between trade and development policies and migration policies. So it clearly implies that these actors sit together and exchange information, not only in times of crisis. And I think it's very good that we have seen just that these mechanisms that have been implemented at a crisis reaction that they continue to be in place at the U level and at some national government. And clearly this is not only a thing for the national government. This requires new partnership beyond the national government to involve cities, social partners, civil society, but also tech entrepreneurs who can alert policymakers about possible new tools that they could use and clearly also micro themselves because these can be important information feeders. Not only in the development of migration policies, but also in the foresight exercise that I think should be done much more systematic in with respect to migration management. Clearly, we need to ensure that these partnerships don't favor some cities, some partners over others that they don't create new inequalities in terms of unfair competition or in country inequalities with respect to dealing with these. And clearly the issue of partnerships with countries of origin also that was mentioned several times. You not only think about central government partnerships but also with local communities and using actually periods of relatively calm what would certainly calm on the migration side when that's going less in terms of regular migration going on to strengthen such partnerships. It's much more difficult to negotiate a return agreement when you have already lots of people that just arrived from that country. When you do it in a period where there's not been a lot of arrivals, it's much easier to do so. But to conclude, how can we ensure that policymakers take uncertainty into account? And actually that link between all the nice tools that we've heard today and discussed and what actually policy action does with that, I think that is the most difficult link to achieve but it's clearly also the most important because without that critical link, all the best instruments will be somewhat in vain. And from all perspectives, I think there's three things that should be done. First of all, to mainstream foresight thinking throughout the world of government to involve both the leaders with the top leadership but also the working level that it's basically ingrained in your DNR of your organization to think about these kind of weak signals and how I can react to those. Policy makers also need to get gradual policy responses in accordance to the certainty. And I think COVID has been a good example of that. In February, certainly the situation was very different from March, April and May and June. And the policy had to adapt to that uncertainty what happened with the pandemic. And last but not least, even in periods where it's relatively calm, it's good to have sort of a crisis room in standby. So you have all those kinds of plans what I would do if that major event would occur to have it in standby because we know there's gonna be new crisis coming. We don't know how and where it will materialize but given that uncertainty that we face and that we will continue to face and that it's likely to increase, we need to maintain those standby crisis rooms. So with that, relatively short thoughts, I would like to conclude my presentation. I invite you to have a look at our four-side publication. It's really downloadable and look forward to the debate with you. Thank you. Thomas, thank you so much for this presentation. Also much appreciated for staying within the time limit. You raised some really important and interesting points there. One very important point that resonated with me is the policy uptake of all these insights and all of these new models, right? If we had the perfect forecasting system with all the relevant actors around the table, how is this knowledge, how is this insight actually then used for policy decision-making? I think a fascinating question and something that could be discussed much further, of course. I think what I would like to do now is actually move on to the last speaker before we delve into the discussion because some questions trickled in for Susanna. So Susanna, you're not off the hook yet. And I'm sure during the next presentation, some questions will come in for Thomas as well. So I think what I would like to do is now move on to Alexander to our last speaker and then open the floor where we'll raise several questions to all of the speakers that were implicated and some of the questions that came in. I hope that works for all of you. Thanks again, Thomas. And now over to Alexander for the last presentation of today and the Data Repertory Council's project on forecasting post-migration. Alexander, over to you. Thank you very much. I hope you can see my screen and thank you very much for inviting me to present today. So I'll be presenting on the work we've been doing in the Danish Refugee Council to explore the use of predictive analytics. Just give you a little bit of background for this work and then go into some details on the models we've been working on and then spend most of my time hopefully on some of the lessons we've learned. So our work in this field was basically initiated about two and a half years ago where we entered into a partnership with IBM with funding from the Danish Ministry of Foreign Affairs. And sort of the scope of the work was to explore the use of predictive analytics and its potential sort of value add in the humanitarian sector. We did this first by trying to build a model to predict mixed migration flows from Ethiopia and later shifted to focusing on predicting forced displacement. And when I talk about forced displacement I'm talking about the refugees, asylum seekers and IDPs. As I mentioned we did this work together and in close partnership with IBM and I think without their support we probably wouldn't have been here today. It was sort of a pro bono partnership where IBM provided the IT infrastructure and the modeling expertise and also sort of the expertise in building user experiences and online use of platforms. What we brought to the table was of course the contextual understanding of what are the drivers of displacement, what are the indicators to look for and the credible data sources. So as I said we ended up building or setting out to build this model on forecasting forced displacement. Seeing can we predict forced displacement one to three years into the future? And we did this around developing a machine learning model around a sort of framework of the known drivers of displacement including more than 120 different indicators in the model and building it on open source data drawing a lot from the World Bank, different UN agencies as well as NGO data. And all the data we gathered was sort of on a national level. And as I mentioned to see can we predict how many will be displaced next year in a given country. We tested it out then in Afghanistan and Myanmar and we got some very encouraging results. So the way we evaluated our model was to see well how well could the model predict sort of known outcomes. So how many forces we displaced would it predict that was in Afghanistan in 2010 and we could of course compare that to how many were actually displaced that year. And when we did that we could see our model had an average margin of error of around 8% in Afghanistan and around 10% in Myanmar which is considered to be sort of quite accurate. We can also see from these results that there were some limitations to the model and I think sort of two noticeable one is that it has a hard time predicting sort of significant shifts or surges in displacement such as the Rohingya crisis in 2017 in Myanmar. And also that it works better in context where you would say the displacement crisis is affecting a large part of the country and it looks less well where displacement is more regionally confined because we're building the model around sort of national level data. And we also applied the model to force the health countries and what we could see there in that context we could compare our model to the current planning figures being used in humanitarian response plans which if you don't know them are sort of the main strategic planning document in the sector. And we could we can see that our model is actually slightly more accurate than the current planning figures that sort of encourage us to see if there is a potential added value here for the wider humanitarian sector in using this model or say similar models. And to try to open up the black box a little bit of machine learning and letting our data sciences out unlike the ASU who wants to lock them in. We developed this online user interface where the users can go in and see the forecast they can access the underlying data but maybe more importantly they can go in and build their own scenarios and tweak the forecast based on their expertise and understanding of the situation. And this platform will be made available to the wider human term community because the vision isn't that this should be sort of a Danish refugee council tool but that it should be used to inform the wider human term sector. Before I go to the lessons learned I just want to give a brief example of how we have used this tool regionally. So we tried to see okay, what is the impact of COVID-19 and analyze what impact does it have on the sort of the drivers of a forced displacement in a different context where we work. We then built scenarios around this analysis to see how would it then impact on forecast displacement. And if we zoom in on the health countries we can see that the baseline forecast in these four countries predicted that around one million additional people would be displaced by the end of 2021. With COVID-19 impact taking into account that number actually increases to two million people. So just an example of how we're currently using that for our strategic planning for the coming year. Just quickly then some of our lessons learned venturing into this field and I'll have sort of five points I want to make and I'll try to be quick. One is, as I mentioned, there are these clear limitations in the model especially the challenge of forecasting these not a certain large scale displacement. So we don't see that these models will be used for sort of standalone tool for decision making but rather that it should be used in scenario building where the experts can sort of weigh in with the assessments of the context and the model can sort of help translate that knowledge into more accurate figures. The other conclusion we have drawn is that when we've touched upon this also other speakers today about how does it inform action and policies for sort of to inform humanitarian action what we see is that we need sort of clear policies for the applications of these models that needs to be also clear use cases in terms of who, when and where these models going through. And also then we think it's important that you have this ability to enable at least not take the staff to engage with the models or for example scenario building to build the ownership and trust in the models which we consider important to turn these models and forecast into actions on the ground. In terms of then also communicating we've also touched upon this today about forecast results. I think for an NGO it's slightly difficult. We need to sort of strike that balance between raising alarm and using the forecast for advocacy and for increasing funding and engagement of donors. While at the same time of course not sort of trying to raise too much alarm in terms of it leading to closure of borders or preventing or maternity for seeking asylum. So just an example when we sent out a press release two weeks ago we had a small second last paragraph on the Sahel analysis of the impact of COVID-19 and obviously that was what made the headlines rather than all the rest of the information in the press release. So really think what we need to focus on when we talk about these results is that yes the placement trends are to some extent predictable. So there really is an excuse for not being prepared. And the second point is that since they are predictable then they also to some extent preventable if you react early on. And I think that is the point we should try to get across in some of this. The fourth point just briefly is that I think we see in the communitarian sector a little bit of sort of a focus at least on strong focus on the risk and ethical concerns around using these profits. And that is very warranted I would say. But we also have to make sure that it doesn't translate into disengagement because I think there are already a number of ways of these new tools and technologies are being used to harm the protection of our people of concern. I think we also owe it to them to explore how can we use these technologies to enhance the protection outcomes. And the last point I want to make and Thomas just mentioned sort of the digitally transparent migrants. I think what we are also seeing is that there are a lot of digitally invisible people around the world. And I think we're seeing a little bit in the communitarian sector it's a little bit sort of a wild west a gold rush and where of course it goes in these modern times and data. And there's a lot of focus on using cold detail records, social media, Google searches and so on. And for good reasons and we've seen a lot of good examples of it today. But from an humanitarian organization focusing on the most vulnerable groups in society I think we're a little bit cautious about these new data sources. Especially because we see a significant portion of these populations being a sort of more data invisible. So just I put some data up here from 2019 around 87% in this year they don't have access to internet on the phones and never use the internet. Around 81% in Bali and you can see that very high numbers there. There is a risk that it will place too much emphasis on these new data sources that we design models that will be biased to more more affluent segments in these populations. And as you can also see in the data and the slide the uses of internet depends a lot on your gender education levels and so on. At the same time these models are being designed in headquarters in the West far away from the affected communities. So I think really a role that we as an NGO needs to play in this space is to ensure the participation of communities in terms of both models development and input and playing the role in terms of making sure that they are visible in the modeling despite this missing digital footprint. I will stop there and I'm very pleased that I got this opportunity to speak with you today and look forward to your questions and comments. Thank you. Thank you very much. Alexander very important points you raised from a different perspective from a different actor and I'm glad we have the Data Reference Council represented in this conference. Some important points that resonated with me is that there are ethical considerations of course in any type of innovative approach when dealing with personal data of individuals but also when transferring to decision-making power through artificial intelligence and so forth. There are ethical considerations. I state here that we consider the second point that I find very important is that there are approaches now that create new invisibilities especially around some vulnerable populations that may not be reflected in some of the approaches and that is an important point to raise. So thank you very much for that. We do have a thicker question also for Alexander but I want to open it up now to the whole panel because now thankfully there are some questions from the floor that trickled in. So we have about 10 minutes but we might go a little over that because the questions are fascinating. So I count on your patience. First two questions to Susana. One from Axel Kreinbrink who works for the Federal Office for Migration and Refugees in Germany, the Banff Office and is a research head there, team lead for research and he would like to know would it be the idea of the EU forecasting system to deliver data for the single member states or only for different parts of our external borders as a whole? And I want to combine this with another sort of yeah, a narrow question also here from Teddy Wilken from ASSO who presents earlier. Will the EU forecasting system take regular migration? For example, visa overstayers and visa free travel into account as well as irregular migration because you mentioned earlier Susana that the system will focus on irregular migrants but Teddy's asking whether that aspect of regular migration that will form part of overall migration will be considered as well. So those two questions for you. Thank you very much. I think the first, of course, is very important to look at the data sources but before we are looking at the data sources we have to determine the objective what we want to focus and then we have to see whether we have sufficient data sources for that or priority lay in the moment on irregular migration and illegal border crossings and not on visa overstayers and that is also my reply to the question from the bump that first, our intention is to start to focus on maybe on a small area and which is well targeted and can be alienated from the rest and then maybe gradually also to expand the system for other aspects like overstayers, secondary movements and also for asylum seekers who are coming from visa-free countries so who don't need to illegally enter the EU they can enter completely legally and then they launch an asylum request. I think whether a single member states can have that focusing on this system I think we have to discuss it when we will have our stakeholder consultation and what would be the requirement for that. Thank you. Great. Thank you very much Susana. There is another question for Alexander. So was there mobile phone data used for the Afghanistan and Myanmar example? That was one question. Then another question is are there will you test the model with additional countries? So will you have the continued IBM support maybe those analysts got tired of the corporate job and moved over to you and I know happily analyzing humanitarian data all day. So will we see additional case studies come out and what is your view on scaling this model to additional cases? Yeah, just the first question on mobile phone data that's the very one. No, we didn't have any mobile phone data in the model and I don't think we foresee using that. One way we are foreseeing in terms of expanding the data source for the model is actually looking more in terms of subnational data points where available in these countries and where relevant, but that's mainly where our sort of expansion lies. In terms of expanding the model, yes, we are indeed the aim is to cover you could say the major displacement producing context and add them to the online user interface within the next sort of three, four, five months. So hopefully we will within the next half year have cover sort of both those displacement context but also countries where we see the potential risk where the model may be able to detect movement. So definitely that's ambition. And we can see that the model performs fairly well when we add new countries. So for the Sahel countries you saw on the slides, there was an average margin of error of around 16%. And that was, there was nothing done there in terms of trying to adapt the model to those specific context. So it is fairly easy to add new countries, as well as just changing a name and a code but then enhancing the models and improving them for the specific countries of both states at the same time. All right. Thank you, Alexander. A question for Thomas as well. You mentioned the importance of involving a multitude of actors around the table. You mentioned, for example, cities, tech companies. How do you best do this? How do you best involve all these multiple actors in decision making process? How do you feed that insight into a policy process? And because linked with another argument that was mentioned earlier, of course, there are intelligence agencies involved in this. There's satellite data being used at times, right? There's personal call data sometimes used. So sometimes sensitive information that is not necessarily made public, but for to incorporate additional actors, right? You would need some more transparency also about the different approaches. So it's just curious on your views on how do you get all these actors around the table and let them feed into the process and whether, yeah, you have good experiences with doing that maybe from other fields as well. Yeah, thanks, Jasper. I think we need to move away from the big data perspective to two issues here. The first is when we do, for example, a foresight exercise and I think it would be very useful to do that on a regular basis. Clearly, when you do this kind of workshop, it can be very fruitful to have those external stakeholders on board because they bring additional views. It's precisely this kind of diversity type of argument that you can apply here very well and which works very well with respect to the out-of-the-box thinking that you want to achieve and the type of forecasting. This is not for sharing the most confidential information. This is for thinking out of the box and identifying weak signals. And I think there in those kind of workshops that can be organized there, it can be very valuable. So that's once on the, let's say, forecasting side or foresight side more specifically. When you look in terms of migration policymaking, clearly in quite a few countries have some kind of advisory board but these often have a very limited role. Sometimes there's experts, for example, in the UK, the Migration Advisory Council is just actually in charge of determining the occupation short assist for labor migration. They really cover only a tiny bit of the market. But in principle, obviously, and it's mainly experts with labor market economists and that's not the type of diversity what I was talking about. But I think you could, but you have that very well established when you talk about the integration field. That's very common to have some kind of integration council that would advise the government on the different aspects. And I think nothing would prevent government from trying it out with respect to migration management as well. That once again, these advisory councils, you don't have to take their decision into account and you have your own intelligence that is a part from those views. But taking that into account can actually be, I think, very fruitful once again to get that holistic perspective. Thank you, Thomas, for this perspective. One last small comment for Alexander and then I wanted to close with more of a bigger picture question to all of you. We do a little round to close this off. But first, a small question to Alexander. Can you distinguish between internal and external displacement in your model? No, we cannot. It was part of the development exercise and to make it a little bit less sensitive, we decided to not distinguish between IDPs and refugees and asylum seekers. So we're not, in that sense, predicting where people go or the direction of the flow, whether they will stay or not. Right, thank you, Alexander. Okay, so to wrap this up, the last session of today, one bigger picture question for me to all of you and I would ask you to respond briefly to it, but give me your quick reaction to the following questions. Let's imagine the EU now is coming up with a comprehensive system. There's a massive resource is also invested in research at this stage. Let's imagine 10 years down the road, we have an EU forecasting system. What, and you are in charge in 10 years and you are the commissioner or yeah, you're the relative commissioner or the council president or somebody in charge and you have to decide whether this funding for the system will be continued or not. How accurate does the system have to be for you to do the thumbs up rather than the thumbs down? One feeling that I have in all this discussion is, how do we actually evaluate the success of such a system? What do we want out of it for this momentum in this field to be continued? So a quick reaction from all of the speakers on what would you expect 10 years down the road to get out of this and what do you think, when do you think it will be worthwhile to have invested this effort in this field? Maybe let's start with Susana and then we move to Tobias and then to Alexander and Thomas. That order was completely random. There was no thought behind that. So Susana, obviously. Thank you very much. I will give you an extremely bureaucratic reply like a good EU official. Pleased to. Probably, if such system is established because it requires a lot of financial resources and of course maybe the reshuffling of the competencies of the agencies or to establish a formal cooperation. So we will have a legal basis for this. And in the legal basis will contain a review clause. So all of the, in every two or three years, basically the efficiency of the system and the delivery of the system must be reviewed. And I think afterwards the necessary adjustment must be made. These kind of IT systems, they are always subject of continuous testing basically. So it's not that they are developed launch and then they are just working because especially if they are working with data models and machine learning, these models always have to be retrained. The results always have to be checked. So there is a continuous validation process and the continuous testing and adjustment process. And in addition to this, which is always subject of a stakeholder consultation, I'm sure that there will be a technical committee with all of the members, it's expressed in this and these changes will be subject to their approval. Basically, in addition to that, there will be this formal evaluation of the systems in every two or three years or so periodically. I would be very happy if in 10 years time it would have really a functional system which can be already by that time really evaluated. But I think it's very realistic in fact that that can be done. And I hope that it will really work according to the accuracy of the member state expert and we have set it and the expert. So how it is agreed. And we have to see whether it really contributes to the decision-making process in some way. I think if something like that is established, it is not abolished because it never happened. So we will never say that noise five weeks ago this system will demolish it and it doesn't have to work. But then the necessary adjustments will be made. Well, I just hope that it will work according to expectation and then it will be able to also to be expense or from a relatively small scope to work on a larger scope. And just from a relatively short-term prediction or early warning it would be able to deliver a longer term also. A longer term forecasting up to two years. And well, and then afterwards we have to see how it is the adjustment but I think once it is established it will be continuously improved but it won't be turned off. Thank you, that is my experience. Thank you very much. We're gonna do some individual forecasting for this forecasting system. And the next forecaster is Tobias Wollander. Tobias, what do you think? Thank you very much. Well, I can give a very brief answer. The system will definitely be around or some kind of system and it will not be turned off because there's basically no alternative as I tried to highlight in my presentation any policy that we undertake in the future should be based on data, on intelligence, on analysis. This can be done either manually or with technologically means and will most likely be a mix. So the question I think in 10 years is not if we turn that system off or not but how much more we can invest to improve it even further. But as I said, like there's basically no alternative to an information-driven approach that helps us better understand the challenges and the problems at hand. Thank you very much. Thank you. Thank you very much. We're seeing the next one in line with Alexander, wasn't it? Yes, thank you. Oh, sorry, I kept on muting myself and muting. I'll try to give you the NGO answer then now that we've got the bureaucratic answer. I'll say if we have a model and models that are being used to protect people rather than borders, then I think we should continue and heavily invest in that. And then I think in terms of the accuracy, I would say for models for forced displacement, I hope we are in a position 10 years from now that we have forecast models that constantly fail because we are able to see the predictions and act accordingly and thereby prevent forced displacement from occurring in the first place. So I hope we see a lot of the failure of these models 10 years from now. No, that was too clever. I didn't expect that answer. That's a good one. Thomas, you were the lucky last. You had all this time to prepare a good answer for this question, so let's hear it. Yes, thank you. So I'll try to add on to what has been said. First of all, I think this is a public good. So there's a clear use of you action in this. So that's already the principal check boxes is marked. I don't think the accuracy of the system should be the most important one. I think that's actually perhaps the least important point in the whole system. I think it would be much more important that it has a good early warning sense for really major things happening, which is not the same of being an accurate tool. I would look also, the decaying case of emergency, how does it adapt to an emergency situation? I mean, like, what's it using in terms of emergency? So when from year to year business, if it doesn't really, if it has some failures, it's less important, but if it must be important in terms of, must show its value added in terms of emergency. Clearly, when you make that decision, you should also look at what's the marginal cost of continuing because building up something new is very expensive, but running it is less expensive. And then last but not least, I think, as I said, it's not only in terms of steady accuracy, that's actually not the key point, but will it bring together different viewpoints, data sources, perspectives that can be used in terms of major disruptions because that's where the user's gonna be and how is it going to be used then by policymakers? I think that's the ultimate question. If you have a perfect system, but nobody uses it, then that is perhaps not useful to continue that. But if it trickles down in the decision process and it's actually used for the day-to-day work of those decision makers, then certainly it should be continued. Thank you very much, Thomas. Very clear point, very concise. We will send that recording to the EU Commission and decision makers in Brussels and in 10 years down the road, you can dig it up and see if we made it. I wanna say a big thanks to all of you speakers for presenting such interesting presentations today. I wanna say a big thanks to the European Migration Network and to the IOM Austria colleagues who have been organizing this fantastic conference. A big thanks to Austrian government for supporting this effort. For me personally, it was a very exciting and a lot of fun to be able to chair some of the sessions today and to listen to all of the interventions. And with this, I will hand over to Julia in Vienna, over to Vienna for some closing remarks. Thank you very much, Jasper. I don't have much to say, except for thank you, Jasper, for co-moderating this day with me. I really, really enjoyed it. I think you really triggered some very interesting conversations. And I would like to also thank all our speakers once more. It was really fantastic. I want to thank them not only for their contributions today, but also for agreeing to contribute to an outcome document of this conference, which is the special issue of the bi-monthly journal, Migration Policy Practice. So please keep a lookout for this forthcoming issue, which will summarize all the conversations we've had today. And also very importantly, not most importantly, thank you so much to my team here in Austria. You were absolutely fantastic. I'm very sorry that you didn't get to meet them. And first and foremost, Lukas Hummer, who's been really the backbone of organizing this event. If you, I'm sure you've all individually been in touch with him. He did a fantastic, outstanding job. We had no technological issues today, which I think is a miracle, but actually it is thanks and due to his hard work and his rigor. Thank you also to, I will now mention my team. It is very small, but I really think it is worth it to Priska Ebner to Alexander Spielefeld, to Martin Stiehler, to Catherine Zink, and to Stefan for their fantastic support. Thank you so much. And finally, thank you to all the participants who are still online, which I think is amazing. It's been an outstanding debate and discussion with fantastic questions and inputs from you. So have a lovely rest of your evening and please do stay in touch. Thank you.