Economic problems that require micro-level analysis within a discrete-choice framework are often spatial processes, where standard models result in inconsistent parameter estimates. Existing methods for spatial discrete-choice models become infeasible in micro-level data applications due to large sample sizes. The authors propose an extension of Klier and McMillen’s (2008) generalized moments estimator to multinomial choice models with spatial lag dependence that is computationally simple.

Simulations indicate that the proposed estimator captures accurately the degree of spatial dependence in the data, provided spatial dependence is not too high. The methodology is employed to analyze the conversion process to various land uses using parcel-level data from a rural-urban fringe county within a large metropolitan area. The authors find evidence of positive spatial dependence of about 0.36—a result consistent with the widely-accepted idea that land-use conversion is a spatial process. This suggests that uncoordinated local land-use policies designed at a small scale, while attempting to manage growth at a local level, may fragment urban development and result in suboptimal land-use patterns at a regional level.