Grants and Contracts Details
Mounting political interest and technological innovations have led to a rise in renewable energy investment. Legislatures in important agricultural states have enacted aggressive renewable energy standards including Minnesota’s 26.5% 2030 goal, Indiana’s 10% 2025 goal, Illinois’ 25% 2026 goal, and Ohio’s 8.6% 2026 goal (NCSL, 2021). The EIA projects the capacity of solar photovoltaic (PV) to rise by 31% from 2021 to 2022. Unlike wind turbines, solar is a unique renewable energy source that limits alternative land uses. Land constitutes the largest holding of wealth in the agricultural sector and accounts for the majority of its productivity. As sale values represent expected long-run income streams, solar’s impact on farmland values is of considerable concern. However, there has been little work in this area and the causal impacts of solar installations are not widely understood. This information is important for a variety of stakeholders including agricultural producers, policymakers, and landowners particularly as the agricultural workforce continues to age. As the overall solar footprint is small, estimating localized effects of solar requires detailed, clean, and geolocated agricultural land sales information. The ERS considers the agricultural economy, the rural economy, resources, and the environment as important areas of interest. Through the subsequent impact on land and agricultural wealth, solar may affect these important topic areas. Solar and crop production share sunlight as a common input. Unlike conventional crop production, solar is likely less affected by soil quality, insect infestation, and disease pressure. This means that opportunities from solar are potentially higher for owners of marginal cropland. Further, solar may compete with alternative ecological land uses such as set-aside programs like CRP. Many of these programs involve returning land to its “natural” state to improve soil quality or lessen the environmental impact of agriculture. While “green” energy production indirectly conveys environmental benefits, it may be at odds with the environmental goals of the set-aside programs. It may also affect the payments farmers can receive from these programs (FSA, 2018). This cooperative agreement aims to estimate the causal impact that utility-grade solar energy has on local per-acre farmland values and quantify the potential differences these effects may have across land of differing agricultural quality. Developing a literature on these impacts will help policymakers assess renewable energy standards’ feasibility, palatability, and consequences. It is also relevant for rural communities as they consider solar energy projects proposals in their areas. Impacts of Solar Installations on Local Farmland Values [CA#] Nicholas J. Pates, University of Kentucky, Lexington, KY ERS Investigator(s): Steven Ramsey 2 APPROACH: Common factors can motivate the placement of solar and conventional farms. Field soil characteristics, for example, will be important for both alternative uses in agricultural and ecological spaces. Since solar is less dependent on soil quality, the potential of solar energy is especially high for owners of marginal agricultural land. However, marginal farmland may have long-term CRP contracts that may limit the benefit of installing solar panels. We will separately consider the local effects of solar installations based on local quantile National Commodity Crop Productivity Index benchmarks. Identification is important for establishing a causal link between solar farms and land values. If CoreLogic provides the sufficient coverage, we propose using an unsupervised machine learning algorithm to pre-classify geospatial clusters of agricultural land. We will cluster by field-level characteristics, including slope, elevation, crop productivity indices, and soil makeup. We will also include environmental controls such as average annual solar irradiance, historic growing-degree and extreme-degree days, field location, and distances to important infrastructure like roads, railroads, large elevators, and high voltage transmission lines. To account for solar lease contract expirations, we will also consider the age of the nearest plant at the time of sale. Our first step will be to stratify our sample across these variable metrics. The idea of pre-stratifying the sample is first to ensure that parcel sales occur within the same local “market” and have roughly similar crop and energy environmental profiles. This step is akin to blocking techniques used in matching studies. If CoreLogic does not provide sufficient coverage, we will utilize propensity score matching and include it as a balancing and matching variable in the secondary matching procedure discussed in the next paragraph. This will address the identification issue by ensuring that matched observations have similar likelihoods of being within a neighborhood of a solar installation based on the factors potentially confounding the analysis. We will divide these subsamples into treatment and control groups to estimate the local causal impact of solar installations on per-acre sale values. These are parcels sold inside and outside of a neighborhood of a solar farm, respectively. We will then construct synthetic control groups using a technique known as genetic matching (GM). GM falls within a branch of machine-learning techniques and maximizes total parameter balance between treated and control groups. From here, we can estimate the effect of the proximity to solar using regression adjustment. This will control for the time of sale, local basis patterns and the other observables mentioned earlier. With these results, we will estimate the impact that solar installations have on average land values within each distinct land market. Our modeling strategy is computationally intensive and requires expansive, up to date, and geolocated land sales information. For this reason, we require CoreLogic data which is the most comprehensive and up-to-date dataset on observed real estate sales data in the US. Our research team has tested this approach on the ZTRAX dataset, a noisier, less complete subset of this dataset which does not offer consistent indicators to identify agricultural property, confounding property characteristics, and arms-length transactions. In addition, the ZTRAX data has spatial gaps in property transactions related to non-reporting assessment departments and ends in 2016.
|Effective start/end date||9/20/22 → 12/31/23|
- Economic Research Service: $35,000.00
Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.