Terror and Internal Migration in Israel

Abstract

This paper empirically analyses the effect of terror on internal migration between Israeli subdistricts. Using a unique data-set comprising migration flows for the period 1999–2012 and the number of rocket and non-rocket attacks, we test the hypotheses that terror reduces migration into an affected subdistrict and increases migration out of it. According to our results, the effect of terror on migration is asymmetric as we only find evidence for the first hypothesis. This result remains when we use an instrumental variables strategy that corrects for underreporting of the number of rocket attacks. The largest effects of rocket attacks are found for migration into the Southern subdistrict of Ashqelon as well as into other border regions in Northern Israel, while non-rocket attacks also have substantial effects in the more populous centres of Jerusalem and Tel Aviv.

Keywords:

• Israel
• terror
• internal migration
• gravity model

JEL Classifications:

• J61
• R15
• R23

Acknowledgements

The authors would like to thank Claude Berrebi, Mohammad Farzanegan, Aliza Fleischer, Bernd Hayo, Michael Kirk, Esteban Klor, Evelyn Korn, Laura Márquez-Ramos, Guy Stecklov, Christian Traxler, Asaf Zussman, an anonymous member of the Editorial Board and two anonymous referees, as well as the participants of the MACIE brownbag seminar, the CNMS colloquium, the 54th congress of the European Regional Science Congress and the 15th Jan Tinbergen European Peace Science Conference for their comments and suggestions and Annie Roth for proof-reading an earlier version of the manuscript.

Notes

1. There is no general acceptance of a precise definition of terror. Schmid and Jongman (2005) analyse the prevalence of 22 definitional criteria in 101 definitions of terror. The chosen criteria of the Global Terrorism Database meet 5 out of 6 of the most frequently used (appearance in more than a third of definitions).

2. An overview of terror event databases which are publicly accessible is given in Sandler (2014).

3. The data is available from www.btselem.org/israeli_civilians/qassam_missiles.

4. In addition to not covering some attacks, it is also possible that the discrepancy is driven by GTD grouping several rocket attacks in the same event if they occurred at the same place and time.

5. The data is taken from the website www.sikurmemukad.com/lebanon2006.

6. See the online Supplementary Material for details.