This article analyzes the key factors guiding the destination choice of Ukrainian refugees in the EU between March and December 2022 in the framework of the activation of the Temporary Protection Directive. To this end, it specifies a migration model, computed for the whole period and two subperiods (March–May and June–December), that captures the influence of social networks. Furthermore, our migration model makes a distinction for the first time between prewar migrant communities and the new social networks that are emerging in wartime circumstances. The estimation of the model, using the Poisson Pseudo-Maximum Likelihood estimator by Santos Silva and Tenreyro (2006), confirmed the importance of both types of social networks in explaining refugees’ choice of destination. The results also revealed that some economic variables (expected earnings and the size of the informal sector) influenced the location choice of refugees since the outbreak of the war, while other socioeconomic and political factors (risk of social exclusion, anti-immigration sentiment, and rule of law) only did so after a few months, when it became increasingly self-evident that the war was going on for a long time.

中文翻译：

---

社交网络如何影响战时难民流动：俄罗斯袭击乌克兰的证据

本文分析了2022年3月至12月在临时保护指令启动的框架内指导乌克兰难民在欧盟目的地选择的关键因素。为此，它指定了一个迁移模型，针对整个时期和两个子时期（3 月至 5 月和 6 月至 12 月）进行计算，以捕获社交网络的影响。此外，我们的移民模型首次区分了战前移民社区和战时环境中出现的新社交网络。使用 Santos Silva 和 Tenreyro (2006) 的泊松伪最大似然估计器对该模型进行的估计证实了两种类型的社交网络在解释难民目的地选择方面的重要性。结果还显示，一些经济变量（预期收入和非正规部门的规模）影响了战争爆发以来难民的地点选择，而其他社会经济和政治因素（社会排斥风险、反移民情绪和法治）几个月后才这样做，当时战争持续很长时间的事实变得越来越不言而喻。