Abstract
In this study, we analyze impacts of large natural disasters on county per capita income between 1990 and 2012 in the Contiguous United States. Using a difference-in-differences model with fixed effects and controlling for serial correlation, we find that the incidence of large disasters in a county significantly reduces its income as compared to its neighboring counties.
Similar content being viewed by others
Notes
Instead of using income, Belasen and Polachek (2008) and Deryugina (2013) use earnings and find different results. Balasen and Polachek (2008) find that hurricanes have a positive impact on the growth rate of earning per worker in the short run. Deryugina (2013) finds no significant impact on the level of average earnings in the short run but a positive impact in the long run.
Xiao (2011) investigates the long-run impacts of flooding for both low and high damages using both one and two controls to match each treatment. The long run impact is only significant for the flooding with low damage and using one-treatment-to-two-control match.
The definition of a large disaster will be discussed in the next section.
If a county shows up in the treatment group twice, two county fixed effect coefficients are specified. For instance, if a county is hit by a disaster in 1994 and another one in 2005, two fixed effect coefficients are specified for that county to capture the potential changes occurred in the county between 1994 and 2005.
The main data source of this dataset is the “Storm Data and Unusual Weather Phenomena” from the National Climatic Data Center (NCDC). Due to NCDC’s change in reporting procedures, every event listed in NCDC’s storm dataset that had exact damage values assigned was entered into the database from 1995 onward. However, only selected events with property or crop damage higher than $50,000 are recorded from 1990 to 1995 (HVRI 2013).
We do not use crop damage because damage amounts are mainly based on self-report and are potentially influenced by government subsidy and crop/livestock insurance programs, whereas the property damages are based on insurance reports.
We try to normalize the property damage by county GDP of the previous year as in Cavallo et al. (2013). We find systematic bias in the sample selection, because it tends to select small rural counties into the treatment.
While physical measurements of disaster intensity like wind speed for hurricanes are preferred, these measures are not generalizable in the sense that they may not be applicable to other disaster types or regions. As a result, we use the concept of relative damage to measure the direct damages from multiple disasters.
BEA also provides annual economic data like income transfers, employment and population. However, all these variables are endogenous to income. The population data are more problematic because it is essentially imputed from the decennial census data, which tend to smooth out short-run impacts. Consequently, none of these variables are included in the baseline analysis.
The inverse Chi-squared statistic is 6453.6 (p value =0.001), which rejects the Fisher-type unit-root test, indicating that at least one panel is stationary.
We do not use the sum of relative damages from multiple disasters occurred in the same year, because it may blur the distinction between the treatment and control groups. If the sum is used, it is possible that the treatment and the corresponding control counties may be hit by some common disasters due to the spatial average problem discussed later. Moreover, SHELDUS does not record natural disasters with property damage less $5000 between 1990 and 1995, using the sum of damages may generate inconsistency in data over time.
As a robustness check, we include all these candidate control counties as the control counties instead of creating the counterfactual using their mean. The results are similar.
The event of hurricane Katrina is not completely excluded from the treatment because some counties at the periphery of the affected region are included.
As a robustness check, we also select sample at other percentiles, but results are consistent.
Disaster aids are unavailable at the county level.
We do not run regression to investigate the third possibility due to the absence of migration data.
We do not use one year before the treatment event because we expect that large disasters may have immediate impacts on per capita income.
Further decrease in the cutoff value results in insufficient number of observations.
References
Anttila-Hughes JK, Hsiang SM (2013) Destruction, disinvestment, and death: economic and human losses following environmental disaster. Available at SSRN 2220501
Banzhaf HS, Walsh RP (2008) Do people vote with their feet? An empirical test of Tiebout’s mechanism. Am Econ Rev 98:843–863
Belasen AR, Dai C (2014) When oceans attack: assessing the impact of hurricanes on localized taxable sales. Ann Reg Sci 52:325–342
Belasen AR, Polachek SW (2008) How hurricanes affect wages and employment in local labor markets. Am Econ Rev 98:49–53
Bertrand M, Duflo E, Mullainathan S (2004) How much should we trust differences-in-differences estimates? Q J Econ 119:249–275
Boustan LP, Kahn ME, Rhode PW (2012) Moving to higher ground: migration response to natural disasters in the early twentieth century. Am Econ Rev 102:238–244
Brown C, Meeks R, Ghile Y, Hunu K (2013) Is water security necessary? An empirical analysis of the effects of climate hazards on national-level economic growth. Philos Trans Ser A Math Phys Eng Sci 371:20120416
Cavallo E, Noy I (2011) Natural disasters and the economy–a survey. Int Rev Environ Res Econ 5:63–102
Cavallo E, Galiani S, Noy I, Pantano J (2013) Catastrophic natural disasters and economic growth. Rev Econ Stat 95:1549–1561
Coffman M, Noy I (2012) Hurricane Iniki: measuring the long-term economic impact of a natural disaster using synthetic control. Environ Dev Econ 17:187–205
Dell M, Jones BF, Olken BA (2012) Temperature shocks and economic growth: Evidence from the last half century. Am Econ J Macroeconomics:66–95
Deryugina T (2013) The role of transfer payments in mitigating shocks: evidence from the impact of hurricanes. Available at SSRN
Horowitz JK (2009) The income–temperature relationship in a cross-section of countries and its implications for predicting the effects of global warming. Environ Res Econ 44:475–493
Husby TG, Groot HL, Hofkes MW, Dröes MI (2014) Do floods have permanent effects? Evidence from the Netherlands. J Reg Sci 54:355–377
HVRI (2013) The spatial hazard events and losses database for the United States, version 12.0. Hazards and Vulnerability Research Institute. Columbia, SC: University of South Carolina
Kousky C (2014) Informing climate adaptation: a review of the economic costs of natural disasters. Energy Econ 46:576–592
Misiewicz J (2011) Fat-tailed distributions: data, diagnostics, and dependence. RFF
Nielsen-Pincus M, Moseley C, Gebert K (2014) Job growth and loss across sectors and time in the western US: the impact of large wildfires. For Policy Econ 38:199–206
Strobl E (2011) The economic growth impact of hurricanes: evidence from US coastal counties. Rev Econ Stat 93:575–589
Tobler WR (1970) A computer movie simulating urban growth in the Detroit region. Econ Geogr 46:234–240
Vu TB, Hammes D (2010) Dustbowls and high water, the economic impact of natural disasters in China. Asia Pac J Soc Sci 1:122–132
Xiao Y (2011) Local economic impacts of natural disasters. J Reg Sci 51:804–820
Author information
Authors and Affiliations
Corresponding author
Electronic supplementary material
Below is the link to the electronic supplementary material.
Rights and permissions
About this article
Cite this article
Mu, J.E., Chen, Y. Impacts of large natural disasters on regional income. Nat Hazards 83, 1485–1503 (2016). https://doi.org/10.1007/s11069-016-2372-3
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11069-016-2372-3