Posted on June 28, 2023
Renowned pioneer in modern geography Waldo Tobler posited his first law of geography in 1970: “Everything is related to everything else, but near things are more related than distant things.” Seemingly simple and intuitive, this observation is the basis for the complex spatial mathematical models that were developed decades later, such as spatial autocorrelation and interpolation. However, the lesser-known second law of geography Tobler proposed in 1999 has profound implications when we examine the effects of policies and boundaries on people and environments: “The phenomenon external to a geographic area of interest affects what goes on inside.”
Researchers at Resources for the Future (RFF) have leveraged these principles to analyze the effects of a long-standing law that’s intended to address some of the challenges associated with coastal development: the Coastal Barrier Resources Act of 1982. Coastal barrier landforms, such as dunes and wetlands, provide critical services. In addition to protecting a rich and productive diversity of coastal resources (fisheries, wildlife, and recreational opportunities), these barriers break waves and slow floodwaters. These latter processes protect coastal communities from storm-surge flooding and sea level rise—hazards that will increase in frequency and intensity with climate change. The Coastal Barrier Resources Act aims to protect undeveloped coastal barriers, preserve the ecosystem services that coastal barriers provide, and reduce federal expenditures on damages from natural disasters. To achieve these goals, the policy removes all federal incentives and financial assistance for development within a set of designated coastal barriers. This decreased support has increased the cost of developing these coastal lands by transferring some of the risk from federal taxpayers to individual property owners.
The law has established the Coastal Barrier Resources System (CBRS), which is a set of designated areas along the Atlantic and Gulf Coasts where federal infrastructure spending, the availability of federal flood insurance, and federal aid after a disaster are prohibited. The Coastal Barrier Resources Reauthorization Act of 2000 includes provisions for digitally mapping the system, and the public now can explore CBRS areas in the interactive Coastal Barrier Resources System Mapper.
With the rapidly increasing need for coastal community resilience against climate change impacts, these coastal barriers are now more valuable than ever. In a new working paper, Hannah Druckenmiller, Yanjun (Penny) Liao, Sophie Pesek, Margaret Walls, and Shan Zhang analyze the long-term outcomes of applying CBRS designations to coastal land. Using spatial data science tools, the authors identified whether and how much these long-term outcomes can be attributed to the passage of the Coastal Barrier Resources Act—both the CBRS areas themselves and neighboring areas. The study examines the impact of the policy on development density, property values, damages from flooding, local tax revenues, and demographic change.
One approach to identifying the effects of the Coastal Barrier Resources Act could have been to simply compare outcomes inside and outside of CBRS designations. However, this approach could not have identified whether the policy itself caused those outcomes, because differences between CBRS and non-CBRS areas could have existed before the policy was implemented. So, to isolate the causal impact of the policy, the authors developed a novel method to select specific comparison areas (known as “counterfactuals”) that would have closely resembled CBRS areas at the time those CBRS areas were designated. Figure 1 shows CBRS areas (left column) beside their corresponding counterfactual areas (right column).
Figure 1. Selection of Areas within and outside the Coastal Barrier Resources System, for Comparison When Evaluating the Policy Impact of the Coastal Barrier Resources Act
“The CBRS areas were selected based on a specific set of criteria consisting of various geomorphic and development features,” Liao says. “A proper comparison between areas where the policy was applied and the counterfactuals must account for the nonrandom characteristics that led to the designation of CBRS areas in the first place, which makes those areas distinct from any other average coastal location.”
The study combined a machine learning technique called “regionalization” with propensity score matching to identify areas that are suitable for comparison. Differences in present-day outcomes between the CBRS designations and counterfactual areas can be attributed to the policy. To identify areas that have comparable features to CBRS sites, the authors wrangled and leveraged a wealth of high-resolution data on land cover, elevation, infrastructure, socio-demographics, and proximity to protected areas.
Almost 40 percent of the US population lives in coastal areas, according to the National Oceanic and Atmospheric Administration. How might past and potential CBRS designations affect these communities? What are the spillover effects? After all, land use changes don’t occur in a vacuum, and Waldo Tobler’s assertion can apply: “The phenomenon external to a geographic area of interest affects what goes on inside.”
In their analysis, the authors identified neighboring areas that could have been affected by the policy by drawing a two-kilometer buffer around both CBRS and counterfactual areas. Because we again can reference Tobler’s insight, “Everything is related to everything else, but near things are more related than distant things,” the authors used what’s called a spatial lag model to examine how policy impacts could vary according to a site’s distance from the coastal barrier.
“The core challenge of this project was that CBRS designations were hand-selected by natural resources planners, meaning that we could not consider the CBRS selection as random in our analysis,” Druckenmiller says. “We needed a way of finding counterfactual areas that closely resembled the CBRS areas at the time of their designation, so that we could make inferences about the effects of the policy based on a divergence in outcomes between the two types of land areas. We developed a method using spatial machine learning that relied on identifying counterfactual areas that were statistically indistinguishable from the CBRS areas when those areas originally were selected.”
These creative and novel methods allowed the researchers to answer important questions about the impacts of CBRS designations, including on neighboring communities. First, they’ve found that the Coastal Barrier Resources Act has been effective in achieving its main goal: the policy has decreased development density within the designated areas by about 85 percent. Second, some of this development has been displaced to neighboring areas. Third, by conserving natural lands within the CBRS, the policy has provided natural amenities and flood protection to neighboring communities, which show up as higher property values and reduced flood damages.
“The importance of protecting land through methods like designating CBRS areas can be very important for communities,” Walls says. “People in coastal communities experience myriad spillover effects from protected lands, such as changes in local revenues derived from property taxes and reduced susceptibility to the disastrous effects of hurricanes.”
The authors estimate that the current system of CBRS lands has led to a 7 percent reduction in claims to the National Flood Insurance Program, which adds up to approximately $112 million per year. Finally, the authors estimate roughly zero effect of the Coastal Barrier Resources Act on revenues from county property taxes in counties that contain CBRS lands. While some revenue loss occurs because of the reduced development on CBRS lands, those losses largely get offset by the increased development and property values on neighboring lands.
These methods tackle the central problem that comes with a nonrandom assignment of comparison sites, which complicates a broad class of place-based policies. Thus, the techniques in this study could be adapted to assess the impacts of, for example, designating conservation areas; programs for economic development that target specific areas like Opportunity Zones, Promise Zones, or state-designated enterprise zones; or policies that distribute aid based on community risk of natural hazards.
