Socio-economic effects of an earthquake

European Regional Science Association
Vienna | 23-26 August 2016 | ERSA 2016: Cities & Regions: Smart, Sustainable, Inclusive?

Special Session Wed_2_Room TC.3.09 S_S. Counterfactual methods for regional policy evaluation
Francesco Pagliacci, Margherita Russo
Socio-economic effects of an earthquake: Does sub-regional counterfactual sampling matter in estimates? An empirical test on the Emilia-Romagna’s 2012 earthquake
Estimates of macroeconomic effects of natural disaster have a long tradition in economic literature (Albala-Bertrand, 1993a; 1993b; Tol and Leek, 1999; Chang and Okuyama, 2004; Benson and Clay, 2004; Strömberg, 2007; UNISDR, 2009; Cuaresma, 2009; Cavallo and Noy, 2009; Cavallo et al., 2010; The United Nations and The World Bank, 2010). After the seminal contribution of Abadie et al. (2010) in identifying synthetic control groups, with DuPont and Noy (2012) a new strand has been opened in estimating long term effects of natural disaster at a sub-regional scale, at which the Japan case provides plenty of significant economic variables. Although, the same methodology has been applied in estimating the impact of earthquakes in Italy (Barone et al. 2013; Barone and Mocetti, 2014), the analysis has been limited to the regional scale. In our paper we provide a test bed for assessing the relevance of a sub-regional counterfactual evaluation of a natural disaster’s impact. By taking the Emilia-Romagna’s 2012 earthquake as a case study, we propose a comprehensive framework to answer some critical questions arising in such analysis. Firstly, we address the problem of identifying the proper boundaries of the area affected by an earthquake. Secondly, through a cluster analysis we show the importance of intra area differences in terms of their socio-economic features. Thirdly, the identified clusters are adopted to perform a counterfactual analysis based on a pre- and post-earthquake difference-in-difference comparison of average data in clusters within and outside the affected area. Eventually, three frames to apply propensity score matching at municipality level are adopted, by taking the control group of municipalities (outside the affected area): within the same cluster (a), within the same region (b), in all the country (c). The four variables considered in the counterfactual analysis are: total population; foreigner population; total employment in manufacturing local units; employment in small and medium-sized manufacturing local units (0 to 49 employees). All the counterfactual tests largely show a similar result: socio-economic effects are heterogeneous across the affected area, where some clusters of municipalities perform better, in terms of increase of population and employment after the earthquake, against some others. This result sharply contrasts with the average results we observe by comparing the whole affected area with the non-affected one or with the entire region.
Keywords: cluster analysis, counterfactual analysis, Emilia-Romagna
JEL codes: C38, R11, R58