Future migration trends

Definitions

Early warning systems 

Early warnings systems rely on quantitative and qualitative data to monitor potential drivers and movements of populations in real time to provide short-term estimations in fast-changing contexts (Carammia and Dumont, 2018). The systems allow decision-makers to allocate budgets and resources preventively (Schmidt and Hooper, forthcoming). Early warning systems may establish pre-defined warning thresholds, that when exceeded, kick off a mechanism of coordinated chain action by various stakeholders (Bijak et al., 2017).

Forecasting

Migration forecasting attempts to predict future migration flows and trends using, traditionally, quantitative modelling methods with a medium and long-term horizon (Bijak, 2011; Disney, 2015). This approach statistically models future migration trends based on quantitative data from the past. This type of modelling is possible when rich numerical data are available, for example, on past inflows and outflows, policy changes, as well as various other migration factors and drivers. (Sohst et al., 2020; Bijak, 2016). 

Foresight

This approach tends to use qualitative scenario methods to describe future migration flows and trends. Migration scenarios are qualitative narratives about the future of migration that examine possible structural changes and their consequences for migration (Vezzoli et al., 2017; De Haas et al. 2010). They can be understood as thought experiments of the type “What if…?”. Scenarios are typically developed by a group of experts engaging in systematic group discussions (Sohst et al., 2020).

Foresight approaches largely rely on subjective opinion and judgment of experts. As such methods do not have to take into account statistical information about trends, it can be used when past data are limited, not comparable or infrequent. Some approaches have attempted to combine qualitative and quantitative information in forecasting, for example by using expert-based probabilistic methods (Lutz et al., 1998), or – more recently – Bayesian statistical approaches (Bijak, 2011; Bijak and Wiśniowski, 2010; Bijak and Bryant, 2016), applied recently at the global level (Azose et al., 2016; Sander et al., 2013). 

Data sources

Early warning systems

• The International Organization for Migration’s Displacement Tracking Matrix, implemented in more than 70 countries, collects data that can be used for early warning purposes; 
  
   
• The Early Warning and Preparedness System from the European Asylum Support Office (EASO) provides a regional outlook with an analysis of asylum trends and push–pull factors, as well as risk analysis on a monthly basis; 
  
   
• Multiple EU Members States use predictions of inflows of asylum seekers (European Migration Network, 2014). The sources are based on the analysis of quantitative data on asylum trends, expert knowledge, qualitative border intelligence, media, and contextual data on routes, as noted by Bijak et al. (2017). According to these researchers, while the methods are not very complex, their strength relies in the myriad of sources in which they rely on and that allows to quantify inflows several times a year. The Swedish and Swiss models are among the most consolidated in Europe and some researchers have documented the main steps in the process, as described below (Carammia and Dumont, 2018; Bijak et al., 2017). 
  
   
• The approach from the Swiss State Secretariat for Migration can be regarded as an expert-based model and is focused in year-ahead estimations of asylum flows (Bijak et al., 2017). The process starts by assessing the push factors in sending countries. Country experts are asked to justify and provide an estimate of future flows. Using a wide range of sources, the analysis complements with data on changes along the migratory routes to Europe. Then, pull factors including economic, asylum policy and the welfare arrangements issues is incorporated into the analysis. Finally, analysts put the information together and produce scenarios that vary in the estimated flows.
  
   
• The model from the Swedish Migration Agency is also focused on asylum applications, with a horizon of one to two years, and uses a mix of qualitative and quantitative sources (Bijak et al., 2017). The data cover push and pull factors of migration flows, migration routes, asylum policies. Experts are involved in the process by quantifying the qualitative insights and then a forecast of asylum flows is derived.

Forecasting

Quantitative migration forecasts are often produced by National Statistical Offices and university research institutes and most often focus on a particular region or country. There are a few large providers of population projections at the global level that include projections of international migration:

• The United Nations Population Division of the Department of Economic and Social Affairs has the longest record of production of global population estimates and projections until 2100, including international migration assumptions. Since 1951, it has produced 26 rounds of its global population estimates and projections. To date, the latest version of its series of World Population Prospects (WPP) is the 2019 Revision (UN DESA, 2019). WPP currently covers 233. countries and territories, making it the geographically most complete dataset.
  
   
• The Wittgenstein Centre for Demography and Global Human Capital provides projections of net migration rates per country until the year 2100 based on different scenarios and assumptions (Lutz, et al. 2018). 
  
   
• The US Census Bureau produces net international migration estimates covering  220 countries and territories until the year 2060.

Other datasets have been made available for particular regions, for example, the European Union's EUROPOP, produced by Eurostat, is the most recent edition covering  the period between 2018 and 2100 (EUROPOP, 2018). These datasets from Eurostats provide information at national level across 29 European countries: for each EU-28 Member State and Norway.

Foresight

Data sources for qualitative migration scenarios include:

• The Global Migration Futures project developed systematic exercises with international migration experts, stakeholders and scholars to examine potential future political, economic, social, technological and environmental changes at the global level and their consequences for migration.
  
   
• The Future of International Migration to OECD Countries report analysed key migration drivers (cooperation at the international level, economic convergence versus economic divergence in per capita incomes between OECD and non-OECD countries, and open versus restrictive migration policies). The study prepared four different scenarios of migration by looking at key determinants of global migration flows by 2030. 
  
   
• The Strategic Foresight Unit from the OECD has explored plausible future changes of potentially great significance for migration and integration policy in OECD countries.
  
   
• Future immigration scenarios to the EU. IOM’s GMDAC, in partnership with the Netherlands Interdisciplinary Demographic Institute (NIDI), examined the potential and limitations of using expert opinion on future migration. The pilot study combined two approaches – migration scenarios and Delphi expert surveys – to assess the implications and uncertainty of common migration scenarios for the EU in 2030. The data collected can be explored with a visualisation tool.
  
   
• The UK Government’s Foresight Project developed global migration scenarios with a 50-year frame. The key migration drivers identified were global migration opportunities linked to high global economic growth versus low global economic growth, and the level of inclusion versus exclusion of political, social and economic governance regimes at a local level.

Data strengths & limitations

Early warning systems

• Operational level policy decisions: Early warning system have the potential to change the decision-making from a reactive to a pro-active manner (Bijak et al., 2017). 
• Early warning system do not provide information about the drivers of migration and do not contribute to the explanation of migration trends. 

Forecasting

• Explicit assumptions and guiding theory: The data-driven approaches tend to be explicit in how assumptions within a model can affect future migration flows, that is, they are transparent on how the guiding theory is used to estimate future flows (Sardoschau, 2020). Forecasts also provide a tangible numeric estimate of future flows which is easy to understand. 
• Long time span: Unlike other methods, they are less limited in the time span they can cover since, in principle, past data trends can be extrapolated for many years in the future (Sardoschau, 2020).

However, forecasting is notoriously difficult and unreliable (Bijak, 2016). There are many reasons why migration forecasting is such a difficult task:

• Lack of uniform concepts and definitions for migration: For example, many countries define migration flows differently. In principle, migration involves relocating across an international boundary for a period of time, but the exact operationalization of this concept in practice varies.
• There are many—and unpredictable—drivers of migration: The diversity of motives behind migration flows and the emergence of new types of migration make it difficult to predict. (Bijak, 2016).
• Forecasts are based on different, imperfect and interacting assumptions: Migration forecasting relies on assumptions regarding demographic dynamics, the political, environmental and socioeconomic changes, as well as migration policies. The sheer number of push and pull factors (determinants) and drivers of mobility and immobility, all interacting with one another, makes a comprehensive explanation of migration processes anything but possible. In other words, models are vulnerable to shocks that move away from their assumptions (Sardoschau, 2020).
• Underlying data may be incomplete and not reliable: In many developing countries, empirical evidence about past and current migration flows is almost entirely missing, and for a number of developed countries, data are also incomplete or unreliable (Buettner and Muenz, 2016).

Foresight:

• Educational tool: Migration scenarios allow those that participate in the scenario creation process to question assumptions on migration and imagine, anticipate and prepare for uncertain future migration trends. (Vezzoli et al., 2017).
• Long-term thinking: Involving experts in systematic, iterative and participatory processes to anticipate the future of migration has the potential to reduce short-sighted policies – especially if policymakers are part of the process. This approach is useful to facilitate strategic long-term thinking among executive decision makers rather than providing operational input (Acostamadiedo et al., 2020).

Despite these benefits, scenarios have shortcomings.

• Disagreement among experts: There is often large disagreement and high uncertainty among experts on how migration drivers will affect future migration flows (Acostamadiedo, et al. 2020). 
• Ambiguous impact of migration drivers: The migration scenario reports sometimes produce ambiguous results on the impact of future migration drivers on flows (Sohst et al. 2020). The theoretical model in qualitative scenarios tends to be less explicitly expressed in contrast to quantitative forecasts (Sardoschau, 2020). 
• Difficulty translating input to policy: Migration scenarios are most useful for those that participated in the process, but it is more challenging to translate the scenario narratives into actionable policy. Important attempts have been made to overcome this limitation (Szczepanikova and Van Criekinge, 2018).