Heterogeneous effects of natural disasters on migration and household well-being in rural Vietnam: a panel data analysis

Abstract

Most existing studies have often ignored the fact that the impact of natural disasters on households may vary according to their dependence on agriculture. Moreover, no study has examined how natural disasters affect food consumption and cause economic damage and, more importantly, whether this effect varies across households. This study aims to explore the influence of natural disasters on the likelihood of households migrating, on their food consumption, remittances, and economic damage. In particular, the study analyzes whether the effect differs among rural households according to their dependence on agriculture. Assembled from three surveys in 2012, 2014, and 2016, we use a unique balanced panel dataset of 1929 households to investigate the influence of natural disasters on migration and household well-being in rural Vietnam. To account for unobservable time-invariant factors that may affect both migration and household well-being, we adopt a fixed-effect estimator. Analyzing a year of natural disasters, we discovered that these do, in fact, cause rural household migration. Further investigation reveals a heterogeneous effect. While natural disasters increase the probability of migration for those with average dependence on agriculture, this is not the case for those with low or high dependence. Natural shocks, however, have an increasing effect on remittances for the former group but not for the latter. Natural shocks cause economic damage for those who are moderately or heavily dependent on agriculture, but not for those who are less reliant on it. In addition, it was found that such shocks reduce the food consumption of those who are heavily reliant on agriculture. The findings imply that migrating may be an effective coping strategy for households with a moderate reliance on agriculture. In light of the finding that households with high reliance have a higher risk of falling into poverty after natural disasters, our study suggests that supportive policies should be prioritized for this vulnerable group. Also, a number of households in this group may face obstacles to migrating. Local government policies supporting their migration should be efficiently implemented, especially for those living in regions prone to natural disasters.

Appendices

Appendix 1: Sample of Vietnamese Communes in the VARHS survey

Source: Berlemann and Tran (2020).

Appendix 2

To analyze the influence of natural disasters on migration and household well-being, our model specifications are as follows:

$$Y_{{ijt}} = \alpha _{1} + \alpha _{1} L_{{jt}} + \alpha _{2} X_{{ijt}} + \alpha _{3} T_{t} + \mu _{i} + \varepsilon _{{it}}$$

(1)

Y_ijt represents the outcomes, including a dummy variable for migration, remittances, economic damage, and food consumption per capita. Ljt is the number of natural disasters, such as droughts, typhoons, and floods, that took place in a specific commune j 12 months ago. We expect that natural disasters in the previous year, rather than in the current year, may affect the decision of households to migrate. It is difficult for households without any resources (such as money, assets, etc.) to migrate; they must have means at a certain minimum level. For this reason, households are more likely to migrate in the year following natural disasters.

We also control for X_ijt, they are various characteristics of household i in commune j at time t, including household size, number of working-age members (working age is defined as 15–55 years for women and 15–60 years for men), number of girls aged under 5, number of girls between the ages of 5 and 15, number of women between the ages of 15 and 60, number of boys aged under 5, number of boys between the ages of 5 and 15, and number of men between the ages of 15 and 60. T_t stands for year-fixed effects.