Natural disasters' impact, factors of resilience and development: A meta-analysis of the macroeconomic literature

• We meta-analyze the macroeconometric studies on the effects of natural disasters.
• We analyze separately studies on disaster direct and indirect costs.
• Direct cost studies are affected by data (decades included) and publication bias.
• Indirect cost studies are affected by publication bias.
• Need to better report results in primary studies (observations and dataset).

We systematize 64 primary studies published in 2000–2013 on the macroeconomic impact of natural disasters by providing OLS and generalized ordered probit meta-analyses for 1858 and 1991 regressions, respectively. We investigate how the reported results in the primary studies are influenced by the empirical design, the estimation technique, and/or publication bias. We analyze primary studies on disaster direct costs and indirect costs separately. According to our meta-analysis, disasters on average have a negative impact in terms of direct costs and an insignificant impact in terms of indirect costs.MST and FAT–PET–PEESE estimates motivate the meta-analyses showing the need for a multivariate approach to consider strong systematic research heterogeneity. Time-based characteristics of the data and publication bias strongly impact on the results of the primary studies, thus implying the need for authors to carefully consider the selection of time period and for research institutions to understand the sources of selection bias in disseminated results. We argue that further research is necessary on the heterogeneity of results of indirect cost studies and suggest that future studies on the macroeconomic impact of disasters should explore more often the mitigation role of education, investment and openness by including these as explanatory variables.