Academic Migration and Academic Networks : Evidence from Scholarly Big Data and the Iron Curtain

Iron Curtain and Big Data are two words usually used to denote completely different eras. Yet, the context the former offers and the rich data source the latter provides, complement each other in a perfect manner enabling the causal identification of the effect of networks on migration. Using Microsoft Academic Knowledge Graph, a scholarly big data source, mapping of academics’ networks is done. Investigation of the effect of the size (synonymous to degree centrality and community capital) and quality (synonymous to eigenvector centrality and reputation capital) of networks by location, i.e. home, destination and foreign is done. Thus the focus is on Eastern European academics behind the Iron Curtain and their networks prior the fall on their probability to migrate from 1989, marking the fall of Berlin Wall, to 2003, marking the year many EE countries signed treaties and held referendums to join the EU. The timing offers unique context in which there was no anticipation of the fall of the Eastern Bloc and together with the data that offers unique rich information, identification is achieved. Approximately 30k EE academics were identified, from which 3% were migrants. From OLS, Logit and Multinomial Logit estimation models there is strong significant evidence on the presence of the cost and signalling channels. An increase in pre-1989 home network size reduces the probability to migrate by 0.1−0.05pp highlighting the increased cost that the academic would face if he is to migrate and his network at home increases in size. However, an increase in home network quality has the opposite, yet more economically insignificant effect, in which an increase in home quality increases the probability to migrate hinting at the stronger signal of the academic’s quality it offers thus increasing the chances of migration. An increase in destination network size for migrants increases the probability to migrate by approximately 7.7 pp highlighting the lower costs of migration due to the increased connections at destination. An increase in foreign network size increases probability to migrate by 0.1pp yet is not significant. I find evidence that implies that for destination and foreign networks, quantity matters more than the quality of the network. Logit results confirm OLS results and MNL results highlight more the cost and signalling channels through different types of migration. Top academics and previous migrants are more likely to migrate, this is especially magnified for those who migrated pre 1989 to an Eastern European country that later became part of the EU, and especially engage in outward migration. Lastly, heterogeneous network effects and joint size and quality effects by fields of study offer insightful results for the interplay of channels especially for Mathematicians, aligning with Borjas and Doran’s (2012) argument and results, and also for academics from Arts and Humanities aligning with Becker et al (2021) argument and results