Research

Agricultural runoff from excessive fertilizer use significantly contributes to nonpoint source water pollution, causing harmful environmental and economic impacts such as algal blooms and health risks. Cover cropping—planting non-cash crops during off-seasons—has emerged as a promising conservation practice to mitigate these effects, supported by payment for ecosystem services (PES) programs like the USDA’s Environmental Quality Incentives Program (EQIP). This study assesses the social value of EQIP in improving water quality by evaluating the effectiveness of cover cropping and EQIP's role in promoting adoption in the Midwest. Leveraging a unique 17-year, satellite-derived dataset on plot-level cover crop adoption and harmonized water quality metrics, we find that a one percentage point increase in upstream cover crop adoption reduces total nitrogen in surface water by 0.9%, though no significant effect is found for phosphorus. Analysis of EQIP's Mississippi River Basin Initiative shows an initial 8.7 percentage point increase in cover crop adoption, with the impact diminishing over time. Overall, EQIP’s cover cropping practices recover 48% of their implementation costs through reductions in nitrogen pollution, suggesting partial but significant economic returns in terms of environmental benefits.

This study investigates the short-run effects of the U.S.-China trade war on U.S. agricultural futures prices, focusing on five primary commodities: soybeans, corn, wheat, rice, and oats. Initiated in early 2018 by President Trump, the trade war resulted in substantial tariffs imposed by both countries, severely impacting the U.S. agricultural sector. To mitigate farmers’ losses, the U.S. government introduced $28 billion in trade aid packages for farmers. This paper utilizes daily futures price data for these grains from 2004 to 2020 and comprehensive supply and demand factors. Due to the non-stationarity of the data, first-difference regressions are employed to quantify the price effects of tariffs and government payments. The findings indicate that a 25% Chinese tariff on U.S. soybeans led to a significant decrease in soybean and wheat futures prices, highlighting the severe short-term impacts of trade barriers on agricultural markets. Additionally, the analysis reveals that the massive trade aid payments had mixed effects on futures prices, challenging the assumption that such payments would not further distort the market.

In this paper, we developed an open-source package to analyze the overall trend and responses of both carbon use efficiency (CUE) and corn yield to climate factors for the contiguous United States. Our algorithm enables the automatic retrieval of remote sensing data through Google Earth Engine (GEE) and U.S. Department of Agriculture (USDA) agricultural production data at the county level through an application programming interface (API). Firstly, we integrated satellite products of net primary productivity and gross primary productivity based on the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor, and climatic variables from the European Centre for Medium-Range Weather Forecasts. Secondly, we calculated CUE and commonly used climate metrics. Thirdly, we investigated the spatial heterogeneity of these variables. We applied a random forest algorithm to identify the key climate drivers of CUE and crop yield, and estimated the responses of CUE and yield to climate variability using the spatial moving window regression across the U.S. Our results show that growing degree days (GDD) has the highest predictive power for both CUE and yield, while extreme degree days (EDD) is the least important explanatory variable. Moreover, we observed that in most areas of the U.S., yield increases or stays the same with higher GDD and precipitation. However, CUE decreases with higher GDD in the north and shows more mixed and fragmented interactions in the south. Notably, there are some exceptions where yield is negatively correlated with precipitation in the Missouri and Mississippi River Valleys. As global warming continues, we anticipate a decrease in CUE throughout the vast northern part of the country, despite the possibility of yield remaining stable or increasing.

Government intervention in insurance markets, such as flood, wildfire, and crop insurance, is a common response to the growing risks associated with climate change. As the frequency of extreme weather events increases, these aggregate shocks necessitate policies that help stabilize incomes and mitigate the economic fallout. However, subsidized insurance programs may also reduce incentives for ex-ante adaptation. In this paper, we examine this trade-off in the context of the U.S. Federal Crop Insurance Program (FCIP) and its impact on farmers' adaptation to climate change. Using data on insurance contracts, yields, farm incomes, input expenditures, and weather realizations over three decades, we find that higher levels of insurance coverage are associated with both larger yield losses and reduced input use during extreme weather events. This suggests that subsidized crop insurance may diminish incentives to mitigate weather shocks. Yet, we also find evidence that higher insurance coverage effectively stabilizes farm incomes during extreme weather events. An optimal policy design would minimize this empirically observed trade-off between income stabilization and incentives to adapt. To explore this, we develop a structural model of insurance choice and production decisions to quantify the net welfare effects of subsidized insurance and simulate alternative policy designs.

This paper analyzes the management strategies employed by Michigan highbush blueberry growers to combat Spotted Wing Drosophila (SWD), an invasive vinegar fly originating from East Asia that poses a significant threat to fruit crops. A dynamic structural econometric model is developed to study growers' decisions related to fly and larva monitoring as well as insecticide application. The model is applied to a comprehensive dataset comprising daily decision records of blueberry growers in Michigan. The findings provide insights into the effectiveness of various management strategies and their implications for economic outcomes in the agricultural sector.

This project incentivizes smallholder farmers in rural India to adopt agricultural practices that improve soil carbon sequestration. We carry out a full RCT that pays farmers as a function of measured improvements in soil organic content in a context with liquidity constraints. The RCT lays the groundwork for developing a larger-scale program that links small farmers to commercial firms providing carbon credits. The project will also explore the potential of satellite data to validate the adoption and impact of regenerative agricultural practices, which will be essential for any scale-up.

The significant role of agriculture in the rise of atmospheric carbon dioxide is well acknowledged, with carbon credit markets like Verra emerging as a market-driven solution to decrease carbon emissions. Despite this, small-scale farmers in developing countries have scarcely tapped into these markets for better carbon sequestration. One of the challenges in connecting India's smallholder farmers with carbon credit firms is posed by the cost-effectiveness of soil testing and monitoring in areas dominated by small fields, where testing costs could exceed the carbon benefits. This study proposes using satellite data and weather data to estimate yields and soil organic carbon (SOC) in small fields in India. We develop a Python pipeline to process open access Sentinel-2 imagery at 10-meter resolution, extracting multiple vegetation indices from its 13 spectral bands per pixel, aggregating this data at the field level, and refining the dataset to eliminate anomalies and smooth time series fluctuations. We also utilized the Copernicus’ ERA5 Climate Reanalysis database to calculate Growing Degree Days (GDD) and Extreme Degree Days (EDD), as well as Total Precipitation as additional variables. By integrating this with precise ground truth data from Odisha, India, and employing the machine learning techniques of Random Forest, Long Short Term Memory Models, and Convolutional Neural Networks, we aim to predict rice yield and SOC accurately. For the purpose of model training, this project utilizes a pre-intervention baseline soil analysis from mid 2022 for the ground-truth SOC data and survey from 2019 to 2022 for the ground-truth yield data. This data is presented on two levels: surface and subsurface levels. For future works, we intend to enhance accuracy and broaden our research scope by integrating different data sources. Emphasizing satellite data for SOC and yield predictions could significantly streamline and economize monitoring efforts in global carbon markets and inform agricultural policy decisions, offering a convergent, scalable, and cost-effective solution for sustainable development.

A practical and reliable way to estimate field-scale evapotranspiration (ET) and CO2 fluxes can significantly help with the optimization of water use and other sustainable practices in precision agriculture and ecosystem restoration. AmeriFlux is a "Big Data" framework updated through a tower-network that provides ecosystem measurements including water, greenhouse gas (GHG) and energy fluxes. Its sites are located in North, Central and South America, but they are limited to one or a few points in the region. The main focus of our research is to develop an effective and wide-ranging methodology for field-scale hydrological and carbon flux estimations based on the integration of AmeriFlux data and satellite images. The AmeriFlux data we use cover a pair of commercially farmed, adjacent rice fields located in Lonoke County, Arkansas for the period 2016-2018. We first illustrate that there is significant and numerically large correlation between the ET measurements and CO2 fluxes made at the AmeriFlux sites and land surface products derived from satellite remotely sensed data (normalized difference vegetation index (NDVI), air temperature, precipitation, and surface pressure) derived by Landsat-8 and Sentinel-2 sensors. Linear regression and random forest models were then developed for predictions. We will explore the spatial and temporal pattern of the data in the future analysis, as well as integrate with local high resolution geophysical data to better understand the effect of the soil spatial heterogeneity, which is known to impact plant development. We envision that the integration of such methodology with eco-hydrological models will enable capabilities to better estimate water use efficiency and carbon storage potential at the field-scale.