The Impact of Perceived Background Risk on Behavioral Health: Evidence From Hurricane Katrina

Abstract

I explore the hypothesis that Hurricane Katrina in 2005 raised perceived background risks, which had spillover effects on behavioral health outcomes of mental health, substance use, and health insurance. I explore this hypothesis using Behavioral Risk Factor Surveillance System data for [2002][2003][2004][2005][2006][2007]. I use a difference-in-difference model to estimate the effect that Hurricane Katrina had in the non-damaged storm surge region, in 90-day intervals leading up to and after the hurricane, compared to areas impervious to hurricanes. Within non-damaged counties at risk of any storm surge, I find causal evidence that Katrina increased poor mental health days by 8.8% for three months after Katrina, and increased smoking by 10% over nine months, which translates to 39.0 million extra days of poor mental health and 1.1 million extra smokers. Results suggest that perceived background risk increases may have important spillover effects on health for people far from the actual disaster, and highlight a hidden cost of the government's failures in managing Hurricane Katrina. ways. While I can control for observable population differences brought about by Katrina evacuees moving disproportionately into the control and treatment groups, unobservable population differences could introduce bias into the analysis.I mitigate possible confounding from unobserved differences in migration and economic climate after Katrina in several ways. First, I exclude counties that were directly damaged by Katrina, or that were damaged by two other hurricanes making landfall shortly after Katrina. These counties are likely to have been most significantly affected by migration and economic factors. My treatment group is therefore areas that are vulnerable to hurricanes but were not directly damaged by these hurricanes. In the analysis, I then control for individual-level observable characteristics associated with migration and economic factors, such as race/ethnicity, employment status, and household income. To get a sense of the magnitude in which unobserved factors may enter into the analysis, I check the balance of the data before and after Katrina based on key demographic characteristics. I do not observe substantial variation in observables before and after Katrina, suggesting unobserved differences are also likely small. To proxy control for any remaining unobservable differences in migration or economic climate, I control at the county level for number of evacuees per 2 capita and distance to New Orleans in the post period, and at the state level for unemployment rates. In these ways, I can isolate the effect of perceived background risk changes on behavioral health outcomes. This paper is related to two strands of literature. The first strand explores the effects of background risk on risk aversion and decision-making in the presence of risk. New risk information, such as from an information campaign on radon exposure, has been found to raise perceived background risk (Smith and Johnson 1988). One study direc...

Environmental disasters and mental health: Evidence from oil spills in the Peruvian Amazon

“…The studies on Hurricane Katrina give useful insights of the magnitude of the damages caused by these events. Pesko (2018) explores the effects of this hurricane on outcomes related to behavioral health and specifically on substance use, including smoking. Based on a difference-indifference approach, he finds causal evidence that this natural hazard increases poor mental days by 18.8 percent for the first month after Katrina.…”

“…While causal studies on environmental disasters and mental health are rather limited, there are some recent causal studies on natural disasters or disasters that are not directly related to human activity. For instance, Pesko (2018) explores the effects of Hurricane Katrina on outcomes related to behavioral health and specifically on substance use, including smoking. Based on a difference‐in‐difference approach, he finds causal evidence that this natural hazard increases poor mental days by 18.8% for the first month after Katrina.…”

“…To our knowledge, the closest study to ours in Economics isPesko (2018), which studies hurricane Katrina and focuses on broad mental health issues and on specific substance abuse.2 In recent years there has been some controversy that a number of oil spills in the North-Peruvian oil pipeline may have been caused intentionally. The specific motivation for doing this is unclear, but some claim that this may occur either to steal oil or as a form of protest by communities surrounding the oil pipeline.…”