Estimating spatial spillover in housing construction with nonstationary panel data

A model is proposed in which building contractors have regional preferences so that housing construction in different regions are imperfect substitutes. The model hypothesizes spatial and national spillovers in construction. Although the government does not engage directly in housing construction, it influences regional housing markets by auctioning land to contractors. Contractors are hypothesized to use housing-under-construction as a buffer between starts and completions. Spatial panel data for Israel are used to test the model and investigate the determinants of regional housing construction. Because the spatial panel data are nonstationary, we use spatial panel cointegration methods to estimate the model. The estimated model is used to calculate impulse responses which propagate over time and across space.