Expanding the coverage and accuracy of parcel-level land value estimates

Planning for cost-effective conservation requires reliable estimates of land costs, spatially-differentiated at high resolution. Nolte (2020) provides a county-by-county, parcel-level estimation approach that dramatically improves estimates of fair market value for undeveloped land across the contiguous Unites States. Much undeveloped land of conservation interest is under threat of conversion to agricultural use or is already agricultural. This paper demonstrates the value of accounting for additional variables that affect agricultural productivity and demand for undeveloped land, as well as the benefit of modeling at scales corresponding to regional agricultural markets. We find that countywide median home value, climatic variables, and several parcel-level soil type variables contribute substantially to predictive power. Enlarging the set of predictors and the geographical scale of modeling improves accuracy by approximately 15 percent and, relative to a more restricted modeling benchmark adapted from Nolte (2020), extends coverage into 376 counties occupying 1.35 million km(2). To assess the practical benefits of our modeling approach, we simulate the protection of 30 percent of US lands via government purchasing, modeled after the Biden administration's 30x30 initiative. Using our proposed modeling strategy, the purchasing agency saves approximately $15 million per year, or 4 percent of the USDA's annual land easement budget.

Automated summary

This paper presents an estimation approach at the county and parcel level to improve the accuracy of estimating the market value of undeveloped land in the United States. By considering factors that affect agricultural productivity and demand for undeveloped land, such as median home value, climatic variables, and soil type, the predictive power is substantially increased. Expanding the set of predictors and the geographical scale of modeling further improves accuracy and extends coverage to more counties.