Search-and-Rescue Operations in the Mediterranean: Are they a ‘pull factor’ for sea crossings of migrants?
Since the large inflows of asylum seekers and other migrants in 2015, migration across the Mediterranean Sea has captured the attention of the public and received extensive media coverage all over Europe. What stimulates or stems sea crossings of migrants? How should the ‘European Rio Grande’ – a pervasive parallel with the equally sensitive US-Mexico border – be controlled to both deter irregular migration and avoid migrants’ casualties during perilous journeys in the Mediterranean? A major controversy revolves around the role of Search-and-Rescue (SAR) operations at sea. Some commentators and political leaders claim that these operations increase departures from Libya and thus serve as a ‘pull factor’, while others condemn the EU and member states for their limited action and alleged breach of humanitarian standards.
Recently, academic research has started to collect and analyse systematic data about the impact of SAR operations on sea crossings – with diverging results. In particular, the papers Migration at Sea: Unintended Consequences of Search and Rescue Operations by Deiana, Dehesri and Mastrobuoni (2019) and Sea Rescue NGOs? A Pull Factor of Irregular Migration? by Cusumano and Villa (2019) reach quite opposite conclusions: the former that SAR operations unintendedly boost migrant smuggling, the latter that they do not affect it. In this post, we briefly assess the strengths and weaknesses of the two studies. We argue that these papers make heroic efforts to adapt the available data to statistical modelling, but they are both limited in different ways and, therefore, cannot provide definitive answers. The relationship between SAR activities and the number of sea crossings of migrants remains an important and unsettled area for future research.
Migration at Sea
Migration at Sea is based on an orthodox econometric analysis of irregular border-crossings in the Central Mediterranean route tapping on detailed records of daily irregular arrivals on Italian shores between 2009 and 2017. The analytical strategy consists of testing whether the number of attempts of crossings (defined as arrivals plus established deaths in transit) and the death risk on the route are influenced by the presence of SAR operations, taking into account sea-wave conditions and the types of vessels used.
The data used are impressive and the analyses conducted with state-of-the-art statistical modelling. The paper reaches two conclusions:
- ‘by reducing the risk of crossing, [SAR] operations likely induced more migrants to attempt to cross, which exposes more people to the risk of death along the passage [and]
- by reducing the costs to traffickers of using unsafe boats, these operations induced a large substitution away from seaworthy wooden vessels and towards flimsy, inflatable boats’ (p. 26).
Our reading of the analyses in this paper is more cautious. The first conclusion relies on an imperfect dependent variable. This is because the estimate of ‘crossings’ does not include would-be migrants that were turned back to Libya, whose number may vary substantially due to the presence of Libyan patrolling – a relevant confounder that is missing from their model. Equally, the political conditions (including internal conflicts) on Libyan soil can intervene to curb or accelerate departures. Another major measurement issue is that the article speaks of ‘SAR intensity’, but the model does not include this variable – only a dichotomous measurement of whether SAR operations were in place or not (whatever their scope). Ideally the models should control for the number of SAR ships in operation at any time to measure ‘intensity’.
The second conclusion assumes that ‘inflatable boats’ are ‘bad boats’ – but this is debatable. In fact, the paper provides evidence of increased purchases of new inflatable boats in the area. A new inflatable boat may be much safer than an old and rusty wooden boat used by smugglers. The variable that would really capture the SAR-induced shift of smugglers to less safe transportation of migrants should be something like the passengers/maximum capacity ratio of boats (ie, to capture overcrowding) rather than a crude description of the nature of the boat.
Sea Rescue NGOs
The policy brief Sea Rescue NGOs asks whether SAR operations by NGOs constitute a pull factor for irregular migration along the Central Mediterranean route and concludes that they do not. Rather, the analysis suggests that weather conditions and policies of ‘onshore containment’ are far more important determinants of departures from Libya than NGOs activities in the Mediterranean. The focus on NGO ships leads the authors to cover a much shorter observation span (January-October 2019) than the Migrants at Sea paper.
This policy brief provides some tentative evidence on the non-relationship between the presence of NGOs at sea and the number of migrants leaving Libyan shores, but this finding is based on analysis of a statistical correlation rather than causality. A ‘pull factor’ argument requires a closer analysis of a specific direction of causation: namely, the effect of NGOs activities on crossing attempts from Libya. The evidence presented does not suffice for a causal interpretation. In particular, there is no consideration of potential issues deriving from what statisticians call ‘simultaneity’. In other words, there might be unobserved factors (for instance, government policies and other political initiatives) that co-determine the number of migrants leaving Libya by sea and NGO-SAR operations. If this is the case, the estimate of the effect of NGOs presence on departures would be biased and typically an instrumental variable strategy should be implemented to correct for such distortion. Moreover, the model specification does not fully control for seasonality – apart from their consideration of wind strength and temperature, perhaps sea wave conditions matter as well. Lastly, even though the paper recognizes the ‘exploratory’ nature of the analysis, the conclusion is limited because it draws on a relatively small number of observations (282 days). As more data become available, further tests should be conducted over a longer period of time.
Given the ongoing policy debates about the role of SAR, the specific methodological issues we discuss in this blog post are not obscure technical points of interest only to experts in research methods. Systematic research on the role and effects of SAR is clearly important and highly relevant to the reform of migration and refugee policies in Europe. It is also important to be clear about the insights and limitations of existing research on this issue. The two papers we discuss in this blog post make valuable preliminary efforts to study the link between SAR activities and sea crossing of migrants. They can be understood as “work in progress” with important findings likely to come to light through further research and scrutiny. This means that their results should be interpreted very cautiously at this stage. However, some commentators have presented results from these papers as concluding rather than opening the debate. For the moment, it is prudent to avoid hasty extrapolations of these early studies.
Note: The authors of the two papers that are discussed in this blog will present their work at a Migration Policy Centre’s Seminar on 10 March 2020.