Spatial panel models for the analysis of land prices

The paper investigates the determinants of cropland value in 12 selected Midwestern U.S.A. States in the years 1971- 2009. We adopt the Ricardian Present Value Model (PVM) as the theoretical framework, and therefore focus on the relationship between land value and cash rents, expecting to find a positive one. In order to model the spatial effects that characterize the data, we estimate a spatial dynamic panel data model with fixed individual effects. The employed dataset represents an improvement with respect to earlier studies because it refers to a rather homogeneous sample of States and only to cropland rather than farmland in general, and also excludes the value of buildings from the value of farmland. Also, net, rather than gross, cash rents per acre of cropland are used, as this reflects the net return to the landowner. Our results allow the computation of short and long run cropland value elasticity to cash rents, whose close-to-1 value appears to support the PVM. We also highlight the importance of taking spatial effects into consideration when addressing this field of analysis.