Periodic Reporting for period 2 - BAPS (Bayesian Agent-based Population Studies: Transforming Simulation Models of Human Migration)
Reporting period: 2018-12-01 to 2020-05-31
The aim of Bayesian agent-based population studies (BAPS) is to develop a ground-breaking simulation model of international migration, based on a population of intelligent, cognitive agents, their social networks and institutions, all interacting with one another. The ambition of the project is to transform the study of migration – one of the most uncertain population processes and a top-priority EU policy area – by offering a step change in the way it can be understood, predicted and managed. In this way, the project seeks to effectively integrate behavioural and social theory with modelling. To develop micro-foundations for migration studies, model design follows cutting-edge developments in demography, statistics, cognitive psychology and computer science. The project also offers a pioneering environment for applying the findings in practice through a bespoke modelling language, and documenting their provenance in a parsimonious and logical way. At the further stages of the work, Bayesian statistical principles will be used to design innovative computer experiments, and learn about modelling the simulated individuals and the way they make decisions. In particular, we aim to collate available information for migration models; build and test the simulations by applying experimental design principles to enhance our knowledge of migration processes; collect information on the underpinning decision-making mechanisms through psychological experiments; and design software for implementing Bayesian agent-based models in practice. The project seeks to use various information sources to build models bottom-up, filling an important epistemological gap in demography.
Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far
Since the beginning of the project, we have identified the case study to focus on, which is the recent Syrian asylum migration to Europe. We have started assembling the empirical evidence base for the model and completed a meta-data inventory, the first version of which is now available on the project website, and which provides important insights into the trade-offs between different quality aspects of data sources. The first version of the migration model has also been developed and tested, looking at the way in which information spread over social networks shapes the emergence and changes of migration routes. The preliminary results indicate that various aspects of information availability, exchange and reliability are crucial for the formation and persistence of migratory routes. In parallel, we have designed and executed the first round of psychological experiments, focusing on the role of risk and loss aversion in migration and financial contexts, which have now been completed, and indicate similarities between the two contexts. The computational side of the project was devoted to a parallel implementation of the first version of the model in a domain specific language, and on designing a formal description of their provenance, highlighting the differences between different modelling formalisms, and their implications for model-building. Finally, the dissemination highlight during the first 24 months was a successful organisation of a workshop on Uncertainty and Complexity of Migration, held at the British Academy in London on 20-21 November 2018, including project-related talks, short contributed presentations, a public lecture, as well as a policy and practitioner panel and discussion.
Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)
The preliminary work identified some key limitations of the existing state-of-the-art approaches to modelling, which do not make full use of the advances in individual contributing disciplines nor of all the available information. To address these challenges, we are currently working on developing an experimental model building process, progressing gradually from simple to more complex models. These models will have an increasing empirical base covering both statistical data and results of psychological experiments on decision making, which will be integrated through Bayesian statistical meta-modelling. By the end of the project, we expect to deliver a set of recipes and guidelines for the process of building empirically and cognitively grounded social simulations. At a fundamental philosophical level, we will be in a position to answer the question to what extent the use of computational models can reduce the uncertainty and increase our understanding of such a complex social process as migration. At the same time, we expect to be able to answer specific substantive questions on migration, at least in qualitative terms of mechanisms and defining features of migration systems. More generally, we envisage that the project will open up exciting research possibilities beyond demography, and will generate both academic and practical impact, offering methodological advice for policy-relevant simulations.