Spatial effects within the agricultural land market in Northern Ireland

Abstract

The importance of dealing properly with spatial effects, such as spatial autocorrelation, in cross-sectional econometric estimation has become more widely recognised in recent years. Spatial autocorrelation is similar in many ways to serial correlation, but while the latter is ordered on a one-dimensional time axis, the former is ordered in two dimensions. The multidirectional nature of spatial dependence means that specialised techniques are needed for diagnostic testing and estimation purposes. This paper uses these specialised diagnostics to test for spatial effects within a hedonic pricing study of the agricultural land market. The tests indicate that spatial autocorrelation (in the form of spatial lag dependence) and spatially distinct sub-markets (or spatial heterogeneity) are present. Ignoring these effects in the estimation process is likely to lead to biased parameter estimates. Consequently, we re-specify the hedonic model to allow for these spatial effects. The presence of spatial lag dependence suggests that there is circularity of price setting within the agricultural land market. This means that agricultural land prices are not solely determined by the inherent characteristics of the land, but tend to reflect also the average local price per acre.