News stories after Hurricane Harvey are bringing national attention to the unfortunate reality that many flooded homes were not covered by flood insurance. A Washington Post story reported that only 17 percent of homeowners in affected areas had a current policy. Some of these flooded properties lie outside the FEMA designated 100-year floodplain, or Special Flood Hazard Area (SFHA). Estimates from past events show that 20 to 30 percent of all flooded properties lie outside the floodplain. Still, much of the flood costs are borne by homeowners within the floodplain who chose not to insure despite prior knowledge of risks. Nationwide, about 50 percent of these high-risk SFHA homes are currently uninsured. Are these homeowners overly optimistic about future flood risks? Or is there some more rational reason for low insurance take-up?
In search of answers to these questions, we recently undertook a survey of homeowners in the floodplains of Anne Arundel County, Maryland—a county on the western shore of the Chesapeake Bay and home to the state capital, Annapolis. Homeowners who took the survey reported their individual perceptions of flood risks, whether or not they had flood insurance, and whether they have experienced a flood in the past. They also answered several questions about their risk preferences and socioeconomic backgrounds. While we are still in the process of analyzing the data, one early finding from the 236 responses is telling: homeowners who have experienced a flood in the past have significantly greater beliefs about the probability of having their homes flood in the future and are more likely to own an insurance policy. The median household that has experienced a flood perceives a 2 percent (or 1-in-50-year) annual chance of flood, while the median household that has not experienced a flood perceives a 0.5 percent (or 1-in-200-year) annual chance.
The Distribution of Risk Perceptions among Homeowners Depends on Flood Experience
While it may seem obvious that experiencing a flood increases one’s perceived flood risk, the implication of this pattern is important. It shows that each homeowner is highly uncertain about his or her individual risk and relies largely on personal experiences to infer the probability of future flooding. When floods occur very rarely, relying on personal experience inevitably leads to substantial underestimation of risks. Indeed, 60 percent of surveyed homeowners believed they had an annual flood risk of less than 1 percent—and this share is even higher (70 percent) among homeowners who never experienced a flood.
Why does it take personal experience with a disaster to sharpen people’s understanding of risks? It may be entirely rational for people to update their risk perceptions in this way if they do not have complete faith in FEMA’s 100-year floodplain designations—though this causes underestimation of flood risks. If rational belief updating is the sole reason for low insurance take-up, then the best strategy to increase take-up is to improve the accuracy and credibility of current flood maps so that homeowners rely less on their experiences and more on expert risk estimates.
On the other hand, it may not be true that all optimism over flood risk is a rational response to uncertainty. An alternative to rational optimism is biased optimism, which occurs when someone’s self-assessment is at odds with both expert estimates and with personal experience. Biased optimism can stem from one of several sources. Psychologists point to cognitive dissonance as one source of overconfidence, defined as the tendency to maintain false beliefs to preserve consistency with one’s previous actions—such as the decision to buy a home in a high-risk area. Another possibility is that biased attention to recent events leads people to underestimate disaster risks when none have occurred recently.
Economists have shown that one way of detecting bias in a population is to measure the share that believes it is better-off than at least half of all other members of the same population. In short, if perceptions are rational, we should expect no more than 50 percent of all homeowners in the SFHA to think that they have lower flood risk than at least half of all other SFHA homes. (The actual test is slightly more nuanced, but this explanation captures the basic idea.) For example, 75 percent of homes cannot all be lower risk than at least 50 percent of all other homes—so if 75 percent of homeowners think they are lower risk than 50 percent of all other homeowners, at least some of those homeowners must be biased in their self-assessments. Each participant in our survey was asked to estimate his or her likelihood of having lower flood risk than at least half of all other SFHA homes in Anne Arundel County—their responses were revealing.
How Probable Do Homeowners Think It Is That They Have Lower Flood Risk than Other SFHA Homes?
If risk perceptions are unbiased, then we should expect the average home to report a probability of 50 percent. The actual average, however, is 59 percent, with one out of every five homeowners reporting that they are lower risk than most other SFHA homes with 100 percent certainty. This pattern of responses suggests that there exists at least some biased optimism among homeowners in our sample. This is important to note because biased homeowners may be less swayed by improved risk information and communication. Other tools, such as mandatory insurance requirements, may be more effective if bias is sufficiently pervasive.
There are many policy concerns raised by the impact of Hurricane Harvey and the devastating flooding in Texas. Some of these are related to where communities are allowing homes to be built. Others are related to further local land use and floodplain management decisions. But once again, after a major disaster, we are discussing the low rates of flood insurance among damaged properties. Gaining a better understanding of why homeowners fail to purchase and how to change that decisionmaking—through financial incentives, mandates and better enforcement of mandates, better communication of risks, and a host of other options—are critical concerns moving forward.
Preliminary results from our Maryland survey suggest that high-risk homeowners exhibit both rational and biased optimism over flood risks. Our ongoing analysis aims to determine which type of optimism plays a greater role in predicting insurance take-up. The predominant source of optimism dictates the best policy for aligning the risk perceptions of high-risk homeowners with the assessments of flood experts. Stay tuned for a more in-depth presentation of our findings.