The demographic impact of terrorism: evidence from municipalities in the Basque Country and Navarre

ABSTRACT

This paper studies the persistence of the effects of terrorist attacks on the urban structure at the regional level. With this aim, a dynamic differences-in-differences approach is applied to all the municipalities of the Basque Country and Navarre autonomous communities in Spain during the period 1986–2014. The results show that terrorism had a negative and transitory effect on population growth. We also find that incidents with deaths implied more adverse shocks. Terrorist attacks had more significant effects in bigger municipalities and in the provinces with a stronger ideological polarization. Finally, we provide evidence of demographic spatial effects derived from violence.

KEYWORDS:

• urban structure
• population growth
• terrorism
• shock persistence
• spatial effects

JEL:

• J10
• N41
• N91
• R12

ACKNOWLEDGEMENTS

The authors have benefited from the valuable comments of an associate editor (Ernest Miguélez), three anonymous referees, Jesús Crespo, Javier Gardeazábal and Harald Oberhofer, and the participants at the XIXth Encuentro de Economía Aplicada (Seville, Spain) and the 56th European Regional Science Association (ERSA) Congress (Vienna, Austria).

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the authors.

Notes

1. ETA announced that it had ‘completely dismantled all of its structures’ on 3 May 2018.

2. A simple linear interpolation was required to deal with the lack of information for 1997.

3. See https://www.start.umd.edu/gtd/.

4. By proceeding this way, we are not taking into consideration the attacks of extreme right organizations – the Anti-Communist Apostolic Alliance (AAA) and the Basque-Spanish Battalion (BVE), among others – or the state-funded ‘dirty war’ carried out by the paramilitary Antiterrorist Liberation Group (GAL). According to Leonisio et al. (2017), these groups account for up to 7% of the victims related to the Basque conflict, most of them in French provinces.

5. Fertility and migration data were extracted from the RegDat-Dem database (De la Fuente, 2016).

6. This approach can be considered as an alternative to the empirical framework proposed by Davis and Weinstein (2002), later adopted by Brakman et al. (2004) and Sanso-Navarro et al. (2015), based on an instrumental variables estimation. As pointed out by an anonymous referee, the transitory or permanent character of the shocks derived from ETA’s attacks can also be studied using other methods such as stationarity/unit root tests and fractional integration or long-memory models.

7. While considering a small number of factors will lead to biased results, including an excessive number will generate efficiency losses.

8. Appendix A in the supplemental data online presents a study of how its results react to changes in the specification of the empirical model.

9. The magnitude of the estimated coefficients reported in the first two columns of Table 4 suggests that the violence suffered by a neighbouring municipality has a stronger impact on population growth than the violence in a given municipality. This finding may be related to the interpretation of estimated parameters in spatial regression models. As pointed out by LeSage and Pace (2009), it is more complicated than in standard regressions (marginal effects) due to the dependence relationships in the spatial lag terms that generate feedback effects.

Additional information

Funding

This work was supported by the Universidad de Zaragoza/Centro Universitario de la Defensa [grant number UZCUD2015-SOC-04]; the Gobierno de Aragón [grant number S16-ADETRE Research Group]; the Ministerio de Economía y Competitividad [grant number ECO2013-45969-P]; and the Ministerio de Economía, Industria y Competitividad [grant number ECO2017-82246-P].