Social Capital Encourages Disaster Evacuation: Evidence from a Cyclone in Bangladesh

Abstract

An essential responsibility of a government is to reduce disaster mortality of citizens by encouraging prompt evacuation. Previous studies document that social capital plays a multifaceted role in facilitating or aggravating evacuation decisions; however, rigorous empirical evidence from developing countries is scarce. Using unique survey data collected from cyclone-affected households in Bangladesh, we examine the association between social capital and the decision to evacuate. We find that those with higher bonding social capital are more likely to evacuate, and the effect is even larger among religious minorities. By conducting a sensitivity test, we confirm that it is implausible to explain these results by unobserved socio-economic status or cyclone damage. We also disentangle five potential mechanisms for the positive effect and demonstrate that it is mainly driven by a reduction in the perceived risk of theft during evacuation. This suggests that bonding social capital compensates for the lack of a well-functioning law enforcement system, contributing to our understanding of the interactive roles of communities and institutions during natural disaster.

Acknowledgements

This paper has been prepared as part of a JICA Research Institute project entitled “Empirical Study on Risk and Poverty in Bangladesh.” All views and interpretations expressed in this paper are those of the authors and not necessarily those of the supporting institutions. The authors would like to express gratitude to Naohiro Kitano, Mahmud Minhaj, Yasuyuki Sawada, and Keijiro Otsuka. The authors are also grateful to Naonori Kusakabe and Data Analysis and Technical Assistance Ltd. for their valuable cooperation in the household survey. The usual disclaimer applies.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplementary Materials are available for this article which can be accessed via the online version of this journal available at https://doi.org/10.1080/00220388.2020.1806245

Notes

1. It should be mentioned that we focus on the evacuation behaviour prior to the landfall of the cyclone. Some disaster-affected people could evacuate or migrate to a non-affected area even after the disaster (Kartiki, 2011; Mallick & Vogt, 2012), and previous studies point out the importance of social capital in this phase as well (Kartiki, 2011). However, this is beyond the scope of this study.

2. See Portes (1998) for the definition by the other researchers.

3. International Strategy for Disaster Reduction (UNISDR, 2004) defines vulnerability as the conditions determined by physical, social, economic and environmental factors or processes, which increase the susceptibility of a community to the impact of hazards. Disaster preparedness refers to pre-disaster activities that are undertaken within the context of disaster risk management and are based on sound risk analysis (UNISDR, 2008). UNISDR (2009) documents resilience as the ability of a system, community, or society exposed to hazards to resist, absorb, accommodate to and recover from the effects of a hazard in a timely and efficient manner, including through the restoration of its essential basic structures and functions. See Birkmann (2006) and Lei, Yue, Zhou, and Yin (2014) for the definitions by the other researchers and institutes.

4. The protective action decision (Lindell & Hwang, 2008; Lindell & Perry, 2004) and social-cognitive preparation (Paton, 2003; Paton, Smith, & Johnston, 2005) models also consider similar frameworks.

5. Maladaptive responses include ignoring an evacuation order and staying home, causing intrinsic and extrinsic rewards such as physical pleasure and approval from community members.

6. Agricultural products, such as paddy, need to be dried after harvest. Therefore, farmers have to protect their products from exposure to heavy rain and inundation during cyclones.

7. This is based on the EM-DAT: The International Disaster Database by Centre for Research on the Epidemiology of Disasters (CRED), accessed on 13 May 2017; for more details see http://www.emdat.be/disaster_list/index.html.

8. For more details about the cyclone damage and post-cyclone rehabilitation, see Mallick and Vogt (2012, 2014)), Saha (2015), Mahmud and Prowse (2012), Sultana and Mallick (2015), and Mallick et al. (2011).

9. Mallick and Vogt (2012) and Saha (2015) find increases in migration after Cyclone Aila from severely affected regions to urban regions. However, Mallick and Vogt (2012, p. 226) claim that only male members of the households migrated. Saha (2015) studies only three of the most-affected villages, which were non-randomly selected, whereas this study uses stratified random sampling to select 24 villages (excluding the villages in Saha (2015)). Therefore, there should be no severe bias in the selection of survey households.

10. The union is an administrative unit in Bangladesh; each union contains multiple villages.

11. It might be insightful to examine the choice of evacuation sites (Lee & Chen, 2018), but given the small sample size of our data, the analysis for locational choice may not provide reliable evidence. Thus, we do not pursue this issue in this study.

12. Supplementary Material 2 explains more details about these measures.

13. The questionnaire was designed to ask the social capital at the time of survey first, and then retrospectively asked about three prior periods including the pre-cyclone.

14. This study employs the OLS even for the binary dependent variable because Oster’s (2019) test is applicable only for the linear regression model. We also estimate the probit model for robustness in Supplementary Material 3.

15. A degree of one suggests that the observables are as important as the unobservables.

16. Regarding the remaining mechanisms, the channel through the availability of evacuation spots predicts an even larger effect of bridging and linking social capital. Social norms to mutual help predict a negative effect of bonding social capital.

17. Trust in the law enforcement system was measured by the question similar with the trust in neighbours. However, it should be mentioned that this question captures the expected ability of law enforcement authorities rather than interpersonal trust. In our sample, 34 per cent of respondents answered affirmatively.

Additional information

Funding

This study was financially supported by KAKENHI [grant number 16K03657] from the Japan Society for the Promotion of Science (PI: Masahiro Shoji).