What Drives Households to Buy Flood Insurance? New Evidence from Georgia

Summary

Benefiting from access to detailed data on the federally run National Flood Insurance Program for the entire state of Georgia, USA, we analyze residential flood insurance purchasing behavior in the state over more than three decades (1978-2010). The demand for flood insurance on an extensive margin, based on take-up rates, is found to be relatively price inelastic. Aligned with the behavioral economics literature, recent flood events temporarily increase purchases, but this effect fades after 3 years. We also find that the proportion of developed area in floodplains has a significant positive impact on insurance take up rates. Contrary to what is often assumed, we do not find evidence that insurance purchase and mitigation efforts are substitutes. Educated individuals, individuals over the age of 45, and African-Americans are, all else equal, more likely to purchase flood insurance.

Situation

Insurance is one of the most widely recognized risk transfer tools for ex ante management of weather disasters such as floods. In the US, since 1968 the National Flood Insurance Program (NFIP) offers flood insurance coverage to residents and small businesses in communities that adopt minimum floodplain management policies. While the program has been in place for over forty-five years, the low take-up rate for flood insurance remains a vexing issue. Academic research on its operation and the demand for flood insurance through the NFIP is fairly recent and non-existent for the state of Georgia.

Response

We analyze the demand for flood insurance across 153 counties in Georgia for the period 1978-2010. We estimate a demand equation in which the number of flood insurance policies-in-force in a county is a function of socio-demographic variables, price of insurance, and flood risk characteristics of the county. We collected county-level data on NFIP policies-in-force (number of policies, flood insurance premium collected, and flood insurance coverage) from FEMA. We obtained a GIS file with floodplain maps and land cover data for the state of Georgia from Georgia's GIS clearinghouse and overlaid FEMA floodplain maps onto the land cover layer to calculate the proportion of developed area of a county that falls in a floodplain. Data on total flood damage per capita in previous years was collected from SHELDUS, a county-level hazard dataset derived from the National Climatic Data Centre. The socio-demographic variables: Age, Education, Income, Occupancy and Race come from the Bureau of Economic Analysis and the U.S. Census Bureau.

Impact

As we would expect, the counties with higher proportion of developed land within floodplains purchase more flood insurance policies. We also find the price elasticity of flood insurance to be fairly low (at -0.30, -0.15) suggesting that those exposed the most want and do purchase coverage. At a time when issues of affordability are at the forefront of the NFIP's reform debate, the finding that an increase in the price of premiums does not highly impact the take-up of flood insurance can help policymakers make informed policy decisions. Other determinants of risk perception, such as having experienced recent flood events, have a significant positive effect on the number of policies-in-force, supporting the hypothesis of the availability heuristic. A recent flood event can be easily brought to mind and therefore heightens the perceived probability of a future flood, which leads to purchasing flood insurance. We found that this effect vanishes after 3 years, however; memory is short. Demographic variables such as education and age are found to have a significant effect on the number of policies-in-force in a county. It is interesting to note that age groups above 45 are more likely to buy flood insurance which may be suggestive that these age groups are more risk averse than their younger counterparts and that therefore awareness campaigns regarding the role of insurance in reducing losses should target the younger population. We also tested the impact of race and found that areas with higher concentration of African-Americans had, all things being equal, a higher demand for flood insurance. FEMA flood risk awareness campaigns have never segmented target audience by race or age. Our analysis suggests that it might be important to do so.

State Issue

Economic Well-being for Individuals and Families

Details

• Year: 2015
• Geographic Scope: State
• County: Clarke
• Program Areas:
  • Agriculture & Natural Resources