Abstract: Elephant populations have declined by half since 1979. In response, activists have promoted destroying confiscated and stockpiled ivory to “send a message” to reduce elephant poaching and ivory demand. As a result, more than 280 tons of ivory has been destroyed between 1989 and 2017. This is the first paper to estimate the causal effect of the amount and location of these destructions on the elephant poaching rate across African and Asian countries with elephants. I use data from CITES’ Monitoring of Illegal Killing of Elephants program from 2003 to 2019, paired with information on ivory destruction events. The main result is that the destruction of ivory does not reduce poaching rates. On the contrary, in African countries with elephants, ivory destructions increase poaching rates, with negative spillover effects from in-country events on the rest of the continent. This suggests the negative supply shock from the destructions dominate and incentivize poaching by increasing the (illicit) ivory price. For sites in Asia there is no evidence that elephant poaching rates respond to ivory destructions.

Abstract: Negative environmental impacts of violent conflict have been observed worldwide. Whether or not active global conflicts are declining in number remains hotly debated, the number of countries entering post-conflict periods is on the rise, and the impact of this transition on land cover changes remains poorly understood. In Colombia, though large-scale armed conflict has concluded, the post-conflict period represents an ongoing threat to forest conservation, putting at risk commitments to meet global conservation goals and even those stipulated in the peace accord. This paper aims to assess land cover change associated with the Colombian conflict in the Andes-Amazon region between 1988 and 2019. First, we use the Landsat archive to map land cover and characterize the spatial patterns of change at the regional level. Second, to empirically identify the effect of conflict on land cover change, we employ a difference-in-difference approach using local conflict events data. During conflict (1988–2011), land cover in the Andes-Amazon remained relatively stable, however during the post-conflict period (2012–2019), the conversion from forest to agriculture increased by 40%. We find that forest cover surrounding conflict events (1 km radius) decreased significantly, on average by ~ 19% during conflict, which accelerated to ~ 30% in the post-conflict period. Similarly, agriculture expansion is most substantial during the post-conflict period, but exclusively in municipalities with population below the 50th percentile. Landscape metrics show that in peripheral municipalities (<50th), agriculture occurs in clumped distributions during the conflict period. Meanwhile, during the post-conflict period, this expansion happens more quickly, with significantly greater agricultural patch sizes than during the conflict period. We conclude that a slow implementation of conservation governance, the emergence of illegal land markets, and illicit land uses (i.e., coca and illegal cattle ranching) may accelerate land cover change in the coming years.

Abstract: Climate change is expected to increase the intensity and spatial extent of storms. Research has looked to natural defensive structures, such as mangrove forests, as cost effective ways to protect communities from cyclones and storm surges. In this study, we want to understand to what extent mangroves can protect local communities from tropical cyclones of different intensities. To do so, we use remotely sensed data on nighttime lights (NTL) and mangrove forest locations to study whether the presence of a nearby mangrove can protect local communities and provide guidance on size and proximity thresholds for protection. At a global scale, we find evidence that nearby mangroves can offset the shock from a cyclone if the mangrove is large enough and the community is close enough. For an average cyclone strike, mangrove patches as small as 101ha (1 sq. km) were able to offset the shock for communities within 17.5km. This result holds for high-intensity cyclone strikes (SSH Cat. 4-5), however the protective radius of a mangrove shrinks. At a minimum communities need to be within 1km of a mangrove patch at least 178ha to be protected from a high-intensity cyclone strike. This research highlights the role of mangroves as natural defensive structures in protecting against (even high-intensity) cyclones, and provides global guidelines on distance to and size of mangrove forests needed for protection from cyclones.

Abstract: [manuscript is being actively updated]

Abstract: Previous research on the value of beaches and dunes has focused its attention to the Eastern Coast of the United States, and has been limited in sample size and availability of measurements. We offer the first hedonic analysis of beach width, dune width, and dune height on a dynamic coastline with seasonal variability. We use panel data on beach width from a probabilistic climate emulator and cross-sectional measures of dunes on 13,140 property transactions between January 2005 and February 2020 in the Pacific Northwest of the United States. Across the period, the estimates show that the average capitalization of dune height ranges from 176 to 21,322$/m, with higher marginal values for shorter dunes. Dune width capitalizes negatively at -1,197$/m. Beach width varies with proximity to the shoreline, and buyers of properties closest to the ocean place the highest marginal value on beach width. For properties within 50m of the shoreline beach width capitalizes at 4,531$/m. At the average, an additional meter of beach width capitalizes at 293$/m, and it declines to 0$/m approximately 600m from the shoreline. Previous studies have noted difficulty in obtaining accurate measures of beach width due to their dynamic nature, and have suggested an aggregate may better capture buyers’ expectations. We explore this in our second analysis by comparing panel and cross-sectional measurements. Our analysis suggests using an average across multiple observations in time, whether panel or cross-sectional, is a better fit and produces less sensitive estimates than the use of a measurement from a single point in time.