Conference Agenda

With California facing ever-increasing losses due to wildfires, there is a need to examine all policy options in order to reduce risks and keep homes insurable. One recent action taken by the insurance commission has been to reward communities that engage in risk reduction activities through the Firewise USA program. While the state mandates insurers to give discounts to residents of participating communities, the effectiveness of this program has not been studied, and insurers cite this lack of study to provide very low and widely varying discounts. Here, we compare Firewise USA communities to similar communities that do not participate in a differences-in-differences design. We find that the probability that a community experiences a fire after becoming a Firewise USA site decreases by 12.5 percentage points, implying that insurance premiums should be reduced by about 6\% in participating areas, compared to an average reduction of just 2.4\%. Wealthier areas and those more exposed to fire risk are found to be more likely to join the Firewise USA program. The findings suggest that community risk reduction activities could play a key role in reducing losses for wildfires, but more could be done to increase the access to the program for poorer communities.

This paper identifies the real estate market responses to rainfall and fire risk in Hawai‘i by leveraging wide variations in precipitation patterns and fire risk exposure. Using repeat-sales housing transaction data between 2000 and 2019, we document three key findings. First, rainfall shocks depress property values, highlighting the disruptive impact of extreme precipitation. Second, wildfire risk also reduces property values, underscoring the market’s sensitivity to fire-related hazards. Third, the negative impact of rainfall shocks is moderated in areas designated as fire zones, suggesting that buyers value the fire-mitigating benefits of increased rainfall. Our results are robust to various specifications of rainfall measures and definitions of both fire risk and fire incidence. Our findings contribute to the understanding of how compound climate risks are capitalized in real estate markets.

Forest pathogens are an increasingly important yet understudied driver of economic welfare losses. This paper estimates the causal impacts of White Pine Blister Rust (WPBR), a non-native forest disease, on wildfire activity, recreation behavior, and visitor welfare in the United States. We combine high-resolution RustMapper projections of WPBR invasion and establishment risk with nationally representative U.S. Forest Service National Visitor Use Monitoring data and administrative wildfire records. Identification exploits sharp ecological thresholds in disease risk using a geographic regression discontinuity design. We find that WPBR exposure significantly increases local wildfire activity, particularly once establishment risk is reached. Near ecological host boundaries, establishment risk increases the number of wildfires within 5–10 km by approximately 1–3 additional fires over a five-year period, implying meaningful increases in suppression costs. These ecological effects translate into large economic responses in recreation markets. Visitors exposed to WPBR risk shorten trips, are 40–45 percentage points less likely to stay overnight, substitute toward shorter day-use activities and closer destinations, and substantially reduce spending. At narrow bandwidths, invasion risk lowers per-trip expenditures by roughly \$900–\$970, driven primarily by declines in lodging, restaurant, and recreation-fee spending. Visitor satisfaction declines most sharply for facilities and environmental quality. Taken together, the results imply economically meaningful revenue losses for gateway communities and highlight forest disease as an underappreciated driver of both wildfire risk and recreational welfare losses. The findings underscore the potential economic returns to proactive forest health and invasive species management as part of climate adaptation policy.

Managing drought risk is critical to mitigating the negative effects of water

scarcity on well-being. This paper presents causal evidence that politicians with

reelection motives respond to drought risk by drilling more water wells than those

without. I exploit close elections in a regression discontinuity design to estimate the

effects of reelection incentives on the provision of public water wells, conditional on

measures of precipitation. I find that the provision of water wells in municipalities

with first-term mayors is higher on average than in cities where mayors face term

limits, as long as the frequency of dry years is sufficiently high.

As climate change intensifies, frequent river flooding from intense short-duration rainfall will cause significant damage. Ecosystem-based disaster risk reduction employs green infrastructure, utilizing forests and retarding basins to prevent water concentration in rivers and reduce peak volumes. However, this approach requires a much larger spatial scale than gray infrastructure, which may disadvantage some residents. Therefore, understanding residents’ preferences in the watershed is crucial when planning flood control using green infrastructure. This study analyzed resident preferences for establishing retarding basins in the upper reaches of a river basin to mitigate flood damage downstream. A discrete choice experiment (DCE) assessed how the distance to mitigation areas affects perceived benefits. In March 2024, a web questionnaire targeting residents of municipalities in the upper reaches, which had previously experienced flooding, was administered, yielding 2,263 responses. Respondents provided opinions on constructing a retarding basin in the upstream area to mitigate downstream flood damage. This scenario assumed that construction was under consideration. Analysis of DCE data using a random parameter logit model revealed that benefits decrease as the distance to the damage reduction increases, even with the same degree of damage reduction. Furthermore, a latent class model identified two classes with different preferences. The results of this study will inform cost-effective plans for installing retarding basins and building consensus among residents regarding their implementation.

Data quality assessment is crucial to ensure reliable results in applied research. In this study, we examine the effects of data quality by analyzing paradata, specifically focusing on response times in both the survey and the DCE tasks. Unlike previous studies that rely on less precise, section-level timings or fixed thresholds, we use detailed time measurements that allow for a more sensitive detection of both, unusually fast and unusually slow responses. In addition, we propose a multidimensional classification of respondents that combines these behavioural measures with self-assessment variables such as “not understanding the DCE” and “finding the DCE difficult to understand”, providing a cognitive dimension to the evaluation of engagement and response quality. Our findings indicate a high prevalence of inattentive behavior, with almost 30% of respondents completing the DCE too quickly, spending one second or less on at least two choice tasks and failing to read the attribute descriptions. The results also indicate that speeding is not evenly distributed across respondents. Individuals under the age of 40 and those with medium–low level of education are more likely to rush through the DCE, suggesting that inattentiveness is associated with socio-demographic characteristics. Hence, we encourage the use of paradata in future studies to complement that information obtained via survey responses.

In August 2006, 528 cubic meters of hazardous waste were illegally dumped across seventeen sites in Abidjan, Côte d'Ivoire, affecting over 100,000 people. This paper examines the long-term causal impacts of toxic exposure on women's health behaviors, labor supply, fertility, and education.

Using 2021--2022 Demographic and Health Survey data, we compare women who lived within 3 kilometers of dumping sites in 2006 to those 7--20 kilometers away, employing inverse probability weighting. Fifteen to sixteen years after the crisis, exposed women are 13.7 percentage points more likely to use modern contraceptives (p = 0.075) and to have been tested for HIV (p = 0.059). We find suggestive evidence of 12--14 percentage point increases in employment (p = 0.094--0.107). However, we find no significant effects on fertility, healthcare utilization, or education.

The pattern---significant health behavior changes without detectable fertility or education effects---suggests the crisis generated lasting shifts in health consciousness rather than altering ultimate outcomes. Findings are consistent with post-crisis interventions expanding contraceptive and HIV testing access in underserved areas, combined with economic coping strategies that increased female labor supply.