Analyzing housing price determinants in Izmir using spatial models

Abstract

The vast majority of the studies on house price dynamics focus on either structural/locational/demographic variables in a cross-sectional setting (i.e., hedonic price modeling) or on the impact of macroeconomic fundamentals in a time series framework. In this work, we argue that both approaches fall short of providing adequate information as cross-sectional analyses largely ignore the macro-dynamics, whereas time series approaches fail to incorporate the cross-sectional dimension. Current work combines both dimensions in a panel framework and provides, in this way, a methodological contribution as well as more informative analyses as it captures the impact of a wide array of variables. Thus, this study examines the housing prices in Izmir/Turkey by adopting the above-mentioned dimensions with both panel and spatial panel regressions. The study area consists of 212 neighborhoods located in different districts of Izmir. The period of analysis covers 30 months between 2017 and 2019. As an outcome of the empirical analyses, both structural/demographic, and macroeconomic variables were found evidently important. Hence, it is understood that all dimensions (structural/locational/demographic, macroeconomic) should be incorporated into comprehensive modeling. A high spatial dependence and positive spatial spillover effects were also detected.