Regional scientists frequently work with regression relationships involving sample
data that is spatial in nature. For example, hedonic house-price regressions relate
selling prices of houses located at points in space to characteristics of the homes as
well as neighborhood characteristics. Migration, commodity, and transportation flow
models relate the size flows between origin and destination regions to the distance
between origin and destination as well as characteristics of both origin and destination
regions. Regional growth regressions relate growth rates of a region to past period ownand
nearby-region resource inputs used in production.
Spatial data typically violates the assumption that each observation is independent
of other observations made by ordinary regression methods. This has econometric implications
for the quality of estimates and inferences drawn from non-spatial regression
models. Alternative methods for producing point estimates and drawing inferences for
relationships involving spatial data samples is the broad topic covered by spatial econometrics.
Like any sub-discipline, spatial econometrics has its quirks, many of which
reflect influential past literature that has gained attention in both theoretical and applied
work.
This article asks the question — what should regional scientists who wish to use
regression relationships involving spatial data in an effort to shed light on questions of
interest in regional science know about spatial econometric methods?