Abstract
This study sheds light on possible spatial aggregation bias and mismeasurement in natural disaster impacts on economic growth when adopting the synthetic control method. Using the impacted U.S. areas of Hurricane Katrina in 2005 as a case study, we compared trends of real GRDP per capita, population, and GRDP at New Orleans-Metairie metropolitan statistical area and Louisiana State, 1992 - 2017. This study empirically demonstrates that improper spatial units of analysis could distort the causal inference in the synthetic control method. Additionally, per capita growth as indicators of recovery from natural disasters is likely to misinterpret economic growth when affected regions experience a sharp decrease in population. These biased results arouse special attention to possible underestimation of natural disaster impacts, particularly when analyzing economically lagging regions. Also, seemingly increasing per capita indicators do not always support evidence of creative destruction after natural disasters.
