Association for Public Policy Analysis & Management

Flooding has long been a threat to communities, but risk management is a challenge for cities, state and the federal government. In 1990, the US Federal Emergency Management Agency created the Community Ratings System (CRS) to provide incentives to local governments to improve flood resilience. The CRS measures the community flood management efforts, and flood insurance premiums are discounted accordingly. A number of factors shape community participation, including flood risk, fiscal capacity, and socio-economic characteristics, and about 1,200 counties and cities voluntarily participate in the CRS, but far more communities do not participate.

In this paper, we explore the factors shaping community CRS participation, and in particular we assess the competing phenomenon of policy diffusion versus free riding. Policy diffusion via spillover effects or a contagion model would suggest that neighboring communities would be more likely to adopt resilience efforts once one local community led the way, all else equal. City officials or concerned citizens could see the resilience efforts in a nearby community, want similar protection, and perhaps more easily follow in their footsteps by borrowing from their policy framework. Therefore, we would see contagion effects following early adopters.

Conversely, if tougher codes raise costs or at least are perceived to raise costs for businesses and residents, flood management efforts could put a community at a competitive disadvantage in terms of economic development or population expansion. Consequently, nearby communities could benefit from this competitive advantage versus a nearby early adopter. Further, the neighbor’s risk mitigation efforts could reduce potential damage from floods for the nearby communities who do not participate so free riding incentives would suggest that they not act.

Using FEMA data on CRS participation, US Department of Transportation data on flood risk at a one-kilometer cell level, local finance data from the Census of Governments, and Census data for socio-economic factors, we develop a basic model of policy adoption for CRS participation. We then enhance this model by adding in a spatial error and lag term to capture whether neighbors' adoption helps explain adoption in a community. It is possible that some negative correlation happens -- especially if "downstream" communities want to free ride on their upstream neighbors' proactive mitigation. In addition, we examine state border effects as flood risks typically correlate very strongly on either side of a border, even when local policies may differ greatly.

Overall, the results can contribute to the policy diffusion literature at the local community level, test for free rider activity, and inform policymakers about the factors shaping the success of the CRS program.