Long-term effects of natural disasters

From SKAIPedia
Jump to navigation Jump to search

Contributors: Tatyana Deryugina

For the purposes of this overview, "long-run" is defined as five years or more post-event. One study (Strobl 2011) does not explicitly show long-run estimates but mentions them in the text.

Virtually all multi-event studies discussed below focus on area-level outcomes (e.g. country-level). The long-run effects of a disaster on affected individuals may be qualitatively and quantitatively different from the effects on the affected area if (as is often the case) a disaster affects population flows.

Findings of the studies discussed below may also differ because of differences in what constitutes an extreme event. Some studies use physical disaster attributes, such as wind speed, whereas others use the consequences of an extreme event, such as fatalities. Studies also set different thresholds for what constitutes an extreme event and vary in the sample of events and affected areas they analyze. Finally, estimation methods may also differ across studies.

Published papers

Multiple events

  • Strobl (2011)[1] estimates the effects of major hurricane strikes on the economic growth rates of US coastal counties. His sample consists of hurricanes making landfall in 1970-2005. He finds that such events cause county-level growth rates to fall by 0.45 percentage points in the year of the strike, but past years' strikes have no such effect. He examines in- and out-migration responses by income and finds that a hurricane strike increases both in- and out-migration by about the same amount, leaving the population unchanged on net. Almost 30 percent of a hurricane's effect on contemporaneous growth is driven by richer individuals moving away from the hurricane-affected county in the aftermath. At the state level, where quarterly GDP data are available, the negative growth impacts are present for one quarter and are subsequently offset by positive growth effects, resulting in no growth effects at the annual level. The growth effects of a hurricane cannot be detected in national GDP data.
  • Cavallo et al. (2013)[2] look at the country-level economic growth effects of large disasters, focusing on earthquakes, tsunamis, floods, and windstorms that occurred in 1970-2000. They define a large disaster as one where the share of population killed exceeds some percentile of the worldwide distribution. The authors then use synthetic control to construct a counterfactural for each country affected by a disaster and consider up to ten years post disaster. On average, catastrophic disasters (>99th percentile of population share killed) appear to have substantial short- and long-run effects on GDP per capita. The authors show, however, that these results are driven by two cases where there was radical political change in the aftermath: the Islamic Iranian Revolution in 1979 and the Sandinista Nicaraguan Revolution in 1979. Milder disasters (>90th or >75th percentile of population share killed) have no detectable effects on GDP in either the short or long run.
  • Deryugina (2017)[3] studies the short- and long-run county-level impact of US hurricanes that made landfall in 1979-2002, considering up to 10 years post-landfall. She finds that population and average wages/salaries are unchanged, while the employment rate declines by 0.6-0.8 percentage points 5-10 years after a hurricane. Per-capita transfers from government increase significantly and persistently, driven by increases in unemployment insurance, public medical payments, and family assistance. Stronger hurricanes (Category 3+) have more negative wage impacts and increase government transfers by more than weaker hurricanes.
  • Karbownik and Wray (2019)[4] use differences-in-differences to study how in-utero and early-childhood exposure to hurricanes affects labor market outcomes in the long run. They link early-life exposure to longer-term outcomes by combining World War I draft records and 1940 Census data. White men exposed to hurricanes in utero or early childhood had earnings that were about 5% lower. There was no effect on labor market participation while migration and education can only account for a small portion of the decrease. The authors therefore conclude that physiological factors are at play.
  • Boustan et al. (2020)[5] use decennial Census data to estimate the effects of a variety of natural disasters on county-level outcomes in the US. The authors define a "severe" disaster as one that causes 25 or more deaths and consider disasters occurring in 1930-2010. They estimate that a severe disaster decreases the net in-migration rate by 1.5 percentage points. Using data from 1970-2010, they further show that severe disasters decrease house values, rents, and family incomes, as well as increase the poverty rate.

2005 US Hurricane Season (Hurricanes Katrina and Rita)

  • Sacerdote (2012)[6] tracks the short- and long-term outcomes of students displaced by Hurricanes Katrina and Rita. Among students displaced by the hurricanes who remain in Louisiana, he finds that math and English test scores decline substantially in the year after the hurricanes (by 0.07-0.20 standard deviations, depending on the sample). Among displaced students from Orleans Parish, test scores fall by 0.10 standard deviations in the year after the hurricane but are higher than baseline three to four years post-hurricane (0.05-0.18 standard deviations higher). College-going is not statistically different from that of earlier cohorts from the same high schools.
  • Basker and Miranda (2014)[7] examine the effect of Hurricane Katrina on business survival. They focus on Mississippi, where impacts were much more localized, damaged infrastructure was quickly repaired, and the population was not reduced, and consider businesses in the retail, restaurant, and hotel sectors. They find that businesses that suffered "severe" or "catastrophic" damage were 30 percentage points less likely to survive than businesses with "limited" or "moderate" damage. This effect is particularly pronounced for firms that are small and those that are less productive.
  • Deryugina, Kawano, and Levitt (2018)[8] estimate the effect of Hurricane Katrina on short- and long-term outcomes of victims initially living in New Orleans. They track individuals over time (1999-2013) and space and measure outcomes using IRS tax data, including W-2 forms provided by employers for non-filers. A control group of individuals from 10 similar but unaffected cities serves as the counterfactual. Hurricane Katrina lowered incomes by about $2,300 in 2006, but in the long run (2008-2013), victims' earnings were actually higher than absent the storm. One quarter of the households was displaced by Hurricane Katrina for at least a year; of these households, about two-thirds returned to New Orleans by 2013. Withdrawals from retirement accounts and self-employment also increased throughout the post-hurricane period. Update of Social Security Disability Insurance (SSDI) increased temporarily.
  • Raker et al. (2019)[9] estimate the long-term mental health consequences of Hurricane Katrina. Their sample contains of low-income mothers who were initially living in New Orleans and were interviewed both before and after Hurricane Katrina. Twelve years after the hurricane, one in six respondents had post-traumatic stress symptoms (PTSS), and PTSS rates in earlier surveys were even higher. Rates of non-specific psychological distress (PD) were elevated throughout the post-hurricane period and did not show a reduction over time. The authors also examine predictors of persistent PD and PTSS, in combination and alone.
  • Deryugina and Molitor (2020)[10] estimate the effect of Hurricane Katrina on short- and long-term (2005-2013) mortality of elderly and long-term disabled victims initially living in New Orleans. In the year of the hurricane, the mortality of this population increased by about 0.55 percentage points. However, mortality then declined to below what it would have been absent Hurricane Katrina. By the year 2013, cumulative survival of Hurricane Katrina victims was 2.07 percentage points higher than the counterfactual. Most of the mortality decline can be explained by victims relocating to lower-mortality areas.
  • Groen, Kutzback, and Polivka (2020)[11] estimate the effects of Hurricanes Katrina and Rita on short- and long-term earnings of storm-affected workers. The authors use the Longitudinal Employer-Household Dynamics (LEHD) data merged with American Community Survey (ACS) data and use propensity score matching to select control counties that resemble hurricane-treated counties. They estimate short-term (1-year) earnings losses averaging $298.4 per worker (s.e.=191.4) and long-term (6 years later) earnings gains of $792.3 per worker (s.e.=141.8). Losses are larger among workers whose residence or workplace suffered damage, but long-term gains are largely similar. The authors attribute short-term earnings losses to job separations and long-term earnings gains to wage increases in affected areas.

Other single-event studies

  • Coffman and Noy (2012)[12] use the synthetic control method to estimate the effect of Hurricane Iniki, which hit the Hawaiian island of Kuaui in 1992. Using other Hawaiian islands, which were unaffected by the storm, as potential controls, they estimate that Kauai did not recover from the hurricane 18 years later, and that its population was 12 percent smaller than it would have been absent Iniki.
  • Hornbeck (2012)[13] estimates the short- and long-term effects (through 1992) of the 1930's American Dust Bowl. He finds that counties whose land was more eroded by the Dust Bowl suffered persistent declines in agricultural productivity relative to less-eroded counties in the same state. Land values in more-eroded counties fell by 30 percent more in 1930-1940 relative to less eroded counties and remained about 30 percent lower in 1978-1992. Agricultural revenues also declined by a similar amount, and the declines persisted. Adjustments in the form of crop-switching and switching from crop to animal farming were estimated to be slow, and the total amount of farmland did not decline substantially. Instead, there was a substantial decline in population of the more eroded counties relative to less-eroded counties.
  • Hornbeck and Naidu (2014)[14] study the effects of Great Mississippi Flood of 1927 through 1970. Counties that were flooded experienced large and persistent outflows of the black population relative to non-flooded counties in the same state. Landowners in flooded counties increased the capital intensity of agricultural production, likely due to the decrease in the supply of black labor that the landowners were dependent on prior to the flood. Farm sizes also increased by more in flooded counties, likely reflecting faster agricultural modernization.
  • Caruso and Miller (2015)[15] study the long-run effects of being affected by the 1970 Ancash Earthquake in-utero or when young and on the intergenerational consequences of such effects. Individuals living in Peruvian states not affected by the earthquake and individuals who were in the affected state but born later or earlier serve as controls. Using the 1993 Peruvian Census, the authors find that men who were in-utero during the earthquake have 0.5 less years of schooling than controls, while women have 0.8 less years of schooling. The education of women who were in their first or second year of life is reduced significantly but by slightly less than women who were in-utero; the effects of experiencing an earthquake in the first or second year of life for men is not significant. Both genders also become parents about four and a half years earlier if affected in-utero or early life. Women who were affected in-utero or when very young also have less educated spouses and are 3 percent more likely to become single mothers or divorce. Using the 2007 Peruvian census, they find that children of women who were in-utero during the earthquake had 0.4 fewer years of education. There is no effect for children of affected men.
  • Lynham, Noy, and Page (2017)[16] use the synthetic control method to estimate the long-run effects of a powerful tsunami that hit the Hawaiian city of Hilo (on the island of Hawaii) in 1960. Using other Hawaiian islands, which were unaffected by the tsunami, as potential controls, they estimate that unemployment was 32 percent higher 15 years later than it would have been absent the tsunami and that population was 9 percent lower.
  • Tahir, Daniels, and Das (2021) [17] study the consequences of a strong 2005 earthquake in Pakistan that killed over 70,000 people. Victims of the earthquake also received cash aid equal to 150% of their annual consumption expenditure. Using data collected four years after the earthquake, the authors find no difference in household income among less versus more exposed households. However, children who were less than 3 years old at the time of the earthquake had significantly lower height-for-age scores, while children who were aged 3-11 and had mothers without primary education had significantly lower test scores.
  • Alves, de Andrade Lima, and Emanuel (2021) [18] show that businesses affected by the 2011 Rio de Janeiro landslides reduced the number of employees by 2.2%-6.4% in the five years following the disaster, with effects growing over time. However, wages remained unchanged and there was no increase in the probability of businesses closing. For a few years following the disaster, affected businesses were more likely to receive subsidized credit, suggesting that disaster aid was helpful in limiting closures. Negative effects were larger for smaller businesses and for ones in the manufacturing, retail, and wholesale sectors.
  • Eskander and Barbier (2022) [19] estimate the long-term consequences of the 1970 cyclone in Bangladesh, which killed over 300,000 people. They use the Bangladesh Household Income and Expenditure Survey in 2000, 2005, and 2010, and assume that respondents were born in the same region they are currently residing. The authors conclude that the cyclone lowered surviving child victims' education levels, self-reported adult health, as well as adult solvency and consumption.

Working papers

  • Hsiang and Jina (2014)[20] estimate the effect of tropical cyclones in 1975-2008 on country-level economic growth. They measure cyclone impacts using wind speed exposure scaled by total country area, which accounts for the fact that larger countries typically have a smaller share of their area affected by such events. They find that cyclones have a negative effect on economic growth that persists for about 15 years. A one-standard-deviation increase in wind speed exposure decreases GDP by 3.6 percentage points two decades later, while a "one-in-a-hundred" country-year event decreases it by almost 15 percent. Effects are very similar for countries with below-median versus above-median GDP per capita. Other comparisons sometimes yield quantitatively non-trivial but not statistically significant differences in cumulative impacts.



  1. Strobl, Eric. 2011. "The economic growth impact of hurricanes: Evidence from US coastal counties." Review of Economics and Statistics. Link
  2. Cavallo, Eduardo, Sebastian Galiani, Ilan Noy, and Juan Pantano. 2013. "Catastrophic natural disasters and economic growth." Review of Economics and Statistics. Link
  3. Deryugina, Tatyana. "The fiscal cost of hurricanes: Disaster aid versus social insurance." American Economic Journal: Economic Policy. Link
  4. Karbownik, Krzysztof, and Anthony Wray. 2019. "Long-run consequences of exposure to natural disasters." Journal of Labor Economics. Link
  5. Boustan, Leah Platt, Matthew E. Kahn, Paul W. Rhode, and Maria Lucia Yanguas. 2020. "The effect of natural disasters on economic activity in US counties: A century of data." Journal of Urban Economics. Link
  6. Sacerdote, Bruce. 2012. "When the saints go marching out: Long-term outcomes for student evacuees from Hurricanes Katrina and Rita." American Economic Journal: Applied Economics. Link
  7. Basker, Emek and Javier Miranda. 2014. "Taken by storm: business financing and survival in the aftermath of Hurricane Katrina" Journal of Economic Geography. Link
  8. Deryugina, Tatyana, Laura Kawano, and Steven Levitt. "The economic impact of Hurricane Katrina on its victims: Evidence from individual tax returns." American Economic Journal: Applied Economics. Link
  9. Raker, E. J., S. R. Lowe, M. C. Arcaya, S. T. Johnson, J. Rhodes, and M. C. Waters. 2019. "Twelve years later: The long-term mental health consequences of Hurricane Katrina." Social Science & Medicine. Link
  10. Deryugina, Tatyana, and David Molitor. 2020. "Does when you die depend on where you live? Evidence from Hurricane Katrina." American Economic Review. Link
  11. Groen, Jeffrey A., Mark J. Kutzbach, and Anne E. Polivka. 2020. "Storms and jobs: The effect of hurricanes on individuals’ employment and earnings over the long term." Journal of Labor Economics. Link
  12. Coffman, Makena, and Ilan Noy. 2012. "Hurricane Iniki: measuring the long-term economic impact of a natural disaster using synthetic control." Environment and Development Economics. Link
  13. Hornbeck, Richard. 2012. "The enduring impact of the American Dust Bowl: Short-and long-run adjustments to environmental catastrophe." American Economic Review. Link
  14. Hornbeck, Richard, and Suresh Naidu. 2014. "When the levee breaks: black migration and economic development in the American South." American Economic Review. Link
  15. Caruso, Germán, and Sebastian Miller. 2015. "Long run effects and intergenerational transmission of natural disasters: A case study on the 1970 Ancash Earthquake." Journal of Development Economics. Link
  16. Lynham, John, Ilan Noy, and Jonathan Page. 2017. "The 1960 tsunami in Hawaii: long-term consequences of a coastal disaster." World Development. Link
  17. Andrabi, Tahir, Benjamin Daniels, and Jishnu Das. 2021. "Human capital accumulation and disasters: Evidence from the Pakistan earthquake of 2005." Journal of Human Resources. Link
  18. Alves, Pedro Jorge, Ricardo Carvalho de Andrade Lima, and Lucas Emanuel. 2021. "Natural disasters and establishment performance: Evidence from the 2011 Rio de Janeiro Landslides." 'Regional Science and Urban Economics'. Link
  19. Eskander, Shaikh MSU, and Edward B. Barbier. 2022. "Long-term impacts of the 1970 cyclone in Bangladesh." World Development. Link
  20. Hsiang, Solomon M., and Amir S. Jina. 2014. "The causal effect of environmental catastrophe on long-run economic growth: Evidence from 6,700 cyclones." NBER Working Paper 20352. Link