Association for Public Policy Analysis & Management

Climate change is predicted to increase the frequency, intensity, location and/or duration of many extreme events.  The pattern of predicted changes in extremes varies around the world.  While some dry areas are getting dryer and some wet areas wetter, there are some locations—notably in part of the US and Europe—where both extremes are predicted by climate models to become more common.  This could lead to weather patterns such as occurred in 2014-2015, when Texas jumped between drought conditions to flooding and back to drought again.  Or when 2002 brought floods to Europe, followed the next year by drought.  Managing the risk of both flooding and drought simultaneously can be challenging, particularly when the time between extreme events is short. 

 Insurance is and will become an increasingly important tool for managing disaster risks.  Insurance can provide greater and faster payouts to households, protecting them financially, speeding rebuilding, and limiting economic multiplier effects in communities.  In this way, insurance coverage for disasters is one component of increasing resilience.  Still, around the world, take-up rates for disaster insurance, when voluntary, have been found to be quite low.  Previous studies in other domains have suggested that individuals may be excessively focused on the near-term, as well as unduly influenced by recent events.  In these ways, their risk assessments may be systematically biased, perhaps leading to suboptimal risk management choices. The decision to purchase disaster insurance could be subject to these same biases.

If homeowners myopically focus on recent events, they could mistakenly presume that a drought indicates that flooding is less likely.  In the American west, the may not actually be the case as models predict increases in both extremes.  It is also the case that droughts can actually increase the likelihood that the next extreme rainfall event causes flooding since water cannot infiltrate the soil as well.  Following hyper-dry conditions, particularly after wildfires, the first rain can produce substantial runoff leading to flooding.  Drought and flood are thus not always independent either.

In the United States, flood insurance is available through the federal National Flood Insurance Program.  This paper tests whether homeowners in the American west alter their flood insurance purchases in response to droughts and floods over the 2000-2014 time period.  Does the occurrence of drought conditions lead homeowners to drop flood insurance?  Do floods cause them to purchase more?  This paper tests these questions using a panel data set of flood insurance take-up rates in counties in northwest, west, and southwestern states (as defined by the National Centers for Environmental Information).  The paper uses county fixed effects models to control for time-invariant aspects of a county and is able to control for several other confounders, such as the requirement that recipients of disaster aid insure, previous flood claims, and previous damages.