The challenges associated with relatively large and rapidly shifting cross-border migration have created a growing demand for improved analysis of migration such that decision-makers are more proactive and forward-oriented in their program design to anticipate short, medium-, and long-term trends. 

Yet specifically with regard to migration in and out of Sub-Saharan Africa (SSA), the availability and use of innovative data and anticipatory techniques to analyze this data remain scant. This is due to issues of collecting, handling, and processing necessary data across the data lifecycle, as well as coordination to identify data needs and apply anticipatory techniques with the help of multi-sectoral data holders.

To assess these urgent regional data and policy gaps and work towards a roadmap to close them, the Big Data for Migration Alliance, a partnership between the IOM Global Migration Data Analysis Center (IOM-GMDAC), the Joint Research Centre of the European Commission (JRC), and The GovLab, hosted a workshop on April 20th in Brussels. The event featured members of the international migration policy community, academic experts, data scientists, and private data holders who discussed the challenges and opportunities around accessing, re-using, and assessing traditional and non-traditional data to anticipate migration.

Key Takeaways from the Future of Human Mobility Event

The event featured presentations on how data and predictive modeling have been used in labor and mixed migration contexts and brought participants together to devise a roadmap for how anticipatory practices can be leveraged for migration decision-making. From these discussions, five key takeaways emerged for the use of anticipatory methods in human mobility and migration policy:

• Policymakers need to crystallize the purpose of data and data tools to gauge the effectiveness of a migration policy initiative. Before obtaining data or leveraging a data tool for forecasts, stakeholders seeking to examine a forward-looking event should outline the purpose of a data initiative to ensure that stakeholders are all on the same page regarding the collection, use, and analysis of data. This helps ensure the initiative is fit for purpose and mitigates irresponsible or inefficient uses of data and resources. A pre-project proposal helps decision-makers across the academic, private, and public sectors realize the need for data for forecasting and matching the relevant data required for the initiative. 
• Forecasting and data-driven migration work would better serve migrants if it focused on identifying and responding to their needs rather than categorizing their flows. As discussed throughout the workshop, an individual can fit multiple types of migrant categorizations throughout their lifetime. The dynamic nature of migration makes it difficult to pin a label on an individual migrating. Rather than take a categorization-based approach, participants advocated for a needs-based methodology that considers what individuals need in specific situations to improve service delivery and policies.
• Forecasting and early warning models could be an efficient way to use data in response to migration needs. Event participants agreed that anticipatory practices can leverage present-day data to better spot and react to high-priority signals. When used together, high-quality, historical data about migrants, from traditional and non-traditional sources can inform short and longer-term migration events. Thus, unlocking innovative and big data can help launch dignified data-driven policies and programs for target populations.
• Decision-makers need to give en-route and settlement actions as much energy as they do to tracking movements. Political will drives data collection, and while there is much investment in pulling together mobility data, fewer resources are put towards collecting and assessing en-route protection needs and (re)settlement data. In order to help migrants across their journeys, decision-makers should devote similar efforts to using data to improve the social and legal direct responses and structures within which migrants transit and live to ensure safe migration and to empower migrants.
• People—from data holders to government officials to migrants—need to be involved across the development of anticipatory methods. Anticipating migration flows requires connecting with people as much as it requires data. Event participants stressed the need to build connections and actively collaborate with people from migrant communities, private companies, and public sector institutions in order to provide the brain trust, data, and human capital resources that decision-makers need to make accurate and purposeful predictions and policies.

Launch of the “Harnessing Data Innovation for Migration Policy” Handbook

At the event, Marzia Rango, former Data Innovation and Capacity-Building Coordinator at IOM-GMDAC launched the “Harnessing Data Innovation for Migration Policy: A Handbook for Practitioners.” This joint publication by the IOM Global Data Institute (GDI) and the BD4M looks at how big data and other innovative data sources can improve methods for migration forecasting and analysis to obtain high-quality insights in an ethical and policy-relevant manner.

Anticipatory Methods for Migration Presentations

Experts from the labor migration and mixed migration fields gave presentations on the challenges, methods, and policies surrounding anticipatory methods. Below, we provide short summaries of these talks.

• Jose Fabio Jimenez, Head of the Labor Mobility and Markets Unit at IOM, discussed how labor migration in Africa occurs between countries, and free movement policies help encourage the free movement of workers. Yet the skills and certifications of labor migrants vary in title, level, and accreditation across countries, and without a baseline understanding of what skills people actually have and what labor countries need, immigration systems cannot effectively match labor migrant supply to appropriate areas of demand. To this end, he suggested a “whole of government” approach to unlock data sharing across government agencies and design a standardized framework for matching and translating between country-specific skill levels. In addition, private data holders might have more recent data regarding the existence of new jobs and requirements. With this data in hand, forecasting for migration is better slated for accurate and useful predictions.
• Naomi Shiferew, Senior Specialist in Labor Migration and Social Inclusion in West and Central Africa at the IOM Regional Office in Senegal recounted existing free movement policies in Africa, such as the Continental Free Trade Area and the African Union protocol on free movement of persons that facilitate more accessible labor migration in the region. However, she noted that the implementation of these frameworks is impeded by a lack of skilled people, access to legal aid for migrants, and a lack of social protection for them. To correct gaps in the current system, there is a need for traditional sources to track statistics on migrant workers, synchronize labor migration approaches across Africa by paying attention to local and national social and economic demographics, and acknowledge the informal and seasonal work sectors to afford adequate social protections for vulnerable labor migrants. She emphasized that forecasting labor migrant needs can allow decision-makers to work backward and create fit-for-purpose educational policies and skills training programs. 
• Jakub Bijak, Professor of Statistical Demography at the University of Southampton and Lead Researcher of QuantMig considered forward-looking approaches in two dimensions, one that implies the users and policymakers know the level of uncertainty and one where the uncertainty is unknown. Recognizing that we can learn more about the present than the distant future, the challenge is to identify leading indicators and high-priority, high-impact signals occurring right now that will still matter in the future. He emphasized the need for traditional data to complement non-traditional data in order to accurately forecast future scenarios and migration projections.
• Linda Adhiambo Oucho, Executive Director of the African Migration and Development Policy Center, presented a dynamic taxonomy of mixed migrants. She demonstrated how people can fall into multiple migrant categories across their lifetime; for instance, asylum seekers can become labor migrants, who later migrate for family reunification. Yet this fluid status exposes mixed migrants to the risks of trafficking, discrimination, xenophobia, and a lack of legal and social protection that makes it easier to harm and exploit them. 
• Laura Nistri, Regional Data Hub Coordinator at the IOM Office in Kenya, talked about the need to map out baseline journeys for highly fluid mixed migration scenarios that are present in Africa. For instance, when it comes to conflict or climate change, decision-makers could use forward-looking tools to assess the time period of movement and where the hot spots of migration will occur. Further, she stressed the importance of defining why data is being collected and used in order to handle it responsibly. 
• Alexander Kjærum, Senior Analyst at the Danish Refugee Council, presented three forecasting models his team deployed in Africa. Using migration determinants from UN agencies, the World Bank, and other data sources, the project was able to use predictive analytics to estimate the number of displaced people from a given country in a region 1-3 years into the future and within a given administrative region 3-4 months in the future. This insight supports strategic planning and operational responses to migration events. However, he also noted that the models are conservative in their predictions and are not good at catching outlying ‘black swan’ events.
• Tuba Bircan, Professor at the University of Cambridge and Lead Researcher of the HumMingBird Project gave an overview of how non-traditional data, such as cell phone records and satellite data, have been used to track human mobility in Africa. She highlighted the nuances within umbrella categories of migration. For example, both flood and drought-related migration are climate-induced movements; however, when a flood occurs, people tend to leave immediately and later return while during a drought, people wait for as long as possible to leave, and once they leave tend not to come back. She noted that while non-traditional data can shed light on specific insights like these, it needs to be verified with the help of traditional data. As well, all data analysis cannot uncover fully root causes for movement; decision-makers need to engage with affected parties to understand their rationales for migrating.

Overarching Challenges

Ultimately, the presentations highlighted three overarching challenges facing data and data models for anticipatory practices in migration work. Attendees signaled the following challenges as priority needs to address in the field going forward. 

• Standardizing the collection of data: A lack of standardized frameworks for collecting and organizing data across governments leads to datasets that are incomplete and not interoperable. Without coordination and granularity of government data, the caliber of traditional data suffers—which impedes the use of innovative data sources because they still need to be complemented with real numbers to ensure robustness and accuracy. 
• Re-using the data: Silos between the public and private sectors, and their data holders, prevent the fusion of traditional and non-traditional data to improve data-driven analysis. While efforts to engage non-traditional data holders have been on the rise, there remains a lack of awareness of data (re)use initiatives and skepticism of their value to both parties. For example, the private sector has developed its own skills ontology to classify worker credentials which have become popularized and widely adopted. However, if the public sector joins on, issues of who owns and how to refine the classification system can create power asymmetries. At the same time, not adopting a classification framework perpetuates issues in data translation across countries. 
• Assessing the data: Migrants evolve across a spectrum of labels and descriptions across time and circumstances. Efforts to classify migrants fail to account for changing descriptors—for instance, a refugee transitioning into a labor migrant—in a dynamic way. There is a need to build a more in-depth understanding of migration journeys and consider social, economic, and demographic factors that further nuance the data-driven picture of migrants.

***

This event is one of a series of data innovation events the BD4M is hosting throughout 2023. Previously, we discussed some ways that non-traditional and innovative data can be used for labor migration with government officials in Egypt, which can be found here.

Back to the Blog