One-directional adjacency matrices in spatial autoregressive model: A land price example and Monte Carlo results

In the context of spatial econometrics, we discuss the specification of one-directional effects, not mutual dependencies. Using an empirical study (a spatial autoregressive model of land price data in Fukui Prefecture, Japan) and Monte Carlo simulation results (contiguity matrices built based on regular lattices using the rook criteria), we show that spatial dependencies might not be recognized if such dependencies are assumed to be reciprocal.

► We discuss the specification of one-directional effects in spatial econometrics. ► An empirical study shows the setting reveals the hidden spatial autocorrelation. ► Monte Carlo results show that the dependency may not be found in misspecified models.