Performing Department
(N/A)
Non Technical Summary
Damages from invasive species are a growing problem for the agricultural sector and the rural communities that become exposed. The invasion of the spotted lanternfly in 2014 serves as one example of the speed at which invasive species--estimated to cause economic damages of $137 billion each year--can spread and damage rural and agricultural economies. Mitigation efforts have succeeded in slowing, but not preventing, the spread of the new spotted lanternfly--described by some as the worst invasive pest since the spongy (formerly, gypsy) moth 150 years ago. As a result, it is crucial to understand the economic damages to farm operations and rural communities, as well as study to what degree farmers are adapting to the new pest, and by which means. This project will use recently available data on the expansion of the spotted lanternfly across U.S. counties, along with the planned release of the 2022 Agricultural Census in 2024, and employ econometric analysis methods to study how crop yields, crop revenue, land values, and agricultural employment are affected following the emergence of spotted lanternfly. In addition, this project will examine the degree to which farmers are switching to different crops, or are increasing their pesticide use in response to the new pest. Finally, this project will use the above estimates to quantify the projected effects of future spotted lanternfly expansion under different scenarios. These projected estimates will highlight the regions where damages might be high, potentially necessitating prioritizing control efforts in those locations.
Animal Health Component
100%
Research Effort Categories
Basic
(N/A)
Applied
100%
Developmental
(N/A)
Goals / Objectives
This project's primary goal is to estimate the economic damages and impacts on agricultural operations due to a changing environment, caused by the spotted lanternfly (SLF) (Lycorma delicatula--an invasive species which was first detected in Pennsylvania in 2014. The key dimensions this study will focus on are: agricultural production, employment in agriculture and forestry, and agricultural land values. Additionally, this project will examine whether farmers are adapting to the presence of the SLF by changing their crop choices, or by increasing their pesticide use. To make the estimation results helpful in designing and prioritizing efforts to slow the spread of the SLF, this project will also use the estimation results to predict damages in locations where the SLF has not yet managed to establish.Damages from invasive species are rapidly growing domestically and globally. Scientists warn about the projected growth in the number of invasive species due to increased international trade, land use changes, climate change, and inadequate prevention and control measures. Quantifying the full scope of damages from all invasive species in the U.S. is challenging, yet previous estimates agree on an order of magnitude of over a hundred billion dollars a year--with USDA NIFA citing annual damages of $137 billion. A 2023 global assessment estimated that damages from invasive species have "quadrupled every decade since 1970," and are now at roughly $423 billion a year.The concerns regarding the magnitude of the threat from invasive species, and the speed at which they can spread are all exemplified in the case of the SLF. The SLF is especially damaging to specific tree fruits, hops, nut trees, and nursery stock--with over 70 plant hosts flagged as susceptible to damages from the SLF. Despite efforts to control its spread, the quarantine area has grown by two orders of magnitude from 130 squared km in 2014 to over 24,000 squared km in 2019. By 2023, there were confirmed reports of sightings in 25 U.S. states, and spatiotemporal modeling predicts that there is a high probability that the SLF will expand throughout most northeast and southeast parts of the U.S., as well as establishing itself on the west coast, likely in California, from where it will expand north.Current estimates of the potential damages from SLF infestations approximate by using survey responses, and applying those to cross-sectional data from the 2017 US Agricultural Census. As a result, management decisions are not informed by estimates that consider either the counterfactual trajectory that infested locations might have experienced in the absence of an SLF invasion, or the potential adaptation strategies farmers might employ to reduce their exposure or damages from the SLF. Quantifying these damages more accurately and precisely is important because there are various approaches to slowing the spread of the SLF, with varying success rates and costs. It is therefore important to understand whether the avoided damages from slowing the spread justify the costs of control operations.To address the knowledge gap on the economic damages of the SLF, this project will assemble panel data on the key outcomes of agricultural production, employment, land values, and pesticide use. Using newly available data on the spread of the SLF, the analysis will estimate how these outcomes evolve after the successful invasion of the SLF in comparison to locations not yet infested by the SLF, but that are likely to experience invasion.There three specific aims of the projects are:Aim 1: Estimate the realized damages from the establishment of the SLF. I will use the existing data on SLF establishment to define exposure to the treatment of interest, and use regression analysis to estimate damages to agricultural production, changes in agricultural and forestry employment, and responses in agricultural land markets. I will use data at both county and state levels to complement each analysis. For land markets, I will also use geo-coded transaction data on agricultural land sales. These results will provide the first large-scale estimates of damages from the SLF relative to a control group of yet-to-be invaded locations.Aim 2: Estimate the degree to which farmers are adapting to the presence of the SLF by changing crop choices and/or increasing their insecticide use.Using a similar approach to Aim 1, I will estimate whether cultivated acres are switching away from the most SLF-susceptible crops following exposure. In addition, I will examine if insecticide use is increasing in response to SLF establishment. These results will provide the first large-scale analysis on whether farming practices are changing in a way that might also help explain if realized damages differ substantially than previous damage predictions.Aim 3: Project the extent to which damages might propagate across space as the SLF continues its expansion up to 2050. A key concern with the SLF is that we have not yet seen the full extent of the damages it can cause because its population levels and their spatial extent have not reached equilibrium levels. I will use the results obtained from Aims 1 and 2 to estimate the potential damages and responses in the locations projected to see the SLF become established by 2050. These results will help inform current slow-the-spread efforts and highlight potential damages hot spots.
Project Methods
This project proposes to use the staggered expansion of the SLF to estimate the damages it has caused, and the responses by the farmers in the SLF-established locations. An ideal experiment would randomly disperse SLF populations and compare areas with SLF populations to those without. For obvious reasons, this is neither feasible nor ethical. A simple comparison of how the outcomes of interest have changed in the locations with established SLF populations may fail to account for secular trends in those outcomes due to global trade, technological changes, changes in consumer preferences, or short-term weather variation. Even if we account for many time-varying characteristics, the analysis may recover biased estimates due to unobserved factors that are systematically correlated with both the expansion of the SLF and the outcomes of interest.The sudden and unexpected emergence of the SLF provides a natural experiment, however, which approximates the above mentioned ideal experiment. Over time, the SLF has managed to establish itself in new locations via dispersion mechanisms that are not entirely predicable ex-ante, and are not a simple deterministic function of baseline values of the outcomes of interest. In other words, SLF expansion follows a quasi-random process through which more locations phase into the treatment group. This gradual expansion also results in different treatment intensities--as locations that experienced an early establishment of an SLF population will likely experience greater damages over time.In order to make correct inferences that allow for a causal interpretation, it is crucial to compare the evolution of the outcomes of interest in the locations that became exposed to the SLF, to those in locations that have not become exposed yet. A complicating feature of such a comparison, however, is that some locations that are not classified as having an established SLF population might indeed have one, and they are already experiencing the treatment effect of interest. The inclusion of erroneously classified locations in such a comparison would incorrectly interpret the estimated damages as secular trends in those outcomes and lead to attenuated estimates. It is therefore important to define the control group in such a way that takes into consideration both spatial spillovers across county borders, as well as delayed or incorrect classification of the true SLF presence in a location.The primary empirical analysis proposed here uses regression analysis to link how the outcomes of interest (the dependent variables) change in the years after the establishment of the SLF, relative to how those outcomes are changing over time in places in which the SLF has not yet become established. For this difference-in-differences comparison to recover a causal treatment effect, the key assumption is that outcomes would have evolved along parallel trends in the absence of SLF exposure. It is also possible to conduct a triple-differences analysis comparing exposed to non-exposed counties, before and after exposure, and susceptible to non-susceptible crops. However, such an analysis might suffer from a violation of the stable unit treatment value assumption (SUTVA) because farmers might adapt by switching from susceptible to non-susceptible crops. Indeed, this potential violation is the primary objective of interest in Aim 2.A complementary analysis will use regression analysis to estimate how the growing share of SLF-exposed area affects the outcomes of interest. This approach can also be interpreted as a difference-in-differences research design with a continuous, instead of discrete, treatment. In this case, the parallel trends assumption is even stronger as it needs to hold for different levels of the continuous share of SLF-exposed area. This additional analysis is still of value, despite the stronger parallel trends assumption, due to data limitations regarding the number of agricultural outcomes and the temporal resolution of currently available county-level data.Using the estimation of the damage functions, I will use the modeled expansion of the SLF to predict future damages.I will use the data from 2017 to 2022 to calculate a mean baseline for the outcomes of interest. I will then use the estimates of the damages I generate in Aims 1 and 2 to calculate the changes to those outcomes in those projected counties under the assumption that the baseline values represent a meaningful approximation to those values averaged up to 2050.Using more detailed data on the projected expansion, I will be able to calculate the projected damages by probability bins (high, medium, and low establishment probabilities), as well as by decade of projected expansion (2030, 2040, and 2050).