M. B. Palacios Navarro, F. Tusell Palmer, M. J. Bárcena Ruiz, P. Menéndez
During the first years of the present century housing prices in Spain have experienced substantial increases. As the real state bubble began to deteriorate after the onset of the financial crisis of 2008, widely different figures have been given on how much housing prices have already dropped. In this paper we obtain an index as the estimate of the price level in a semi-parametric model of readily available offered prices in the city of Bilbao (Spain). We fit a model to the log price per squared meter to capture the influence on house prices of several attributes, the effect of time and the influence of the location. The suitably normalized profile of the effect of time provides an estimate of the price index. To include the effect of the location, two approaches have been used: including the postal code as a categorical regressor and using geographically weighted regression, with each house geocoded to UTM coordinates, and a backfitting algorithm.
Palabras clave: hedonic models, geographically weighted regression,price index, spatio-temporal data
Programado
XC1a Pósters (Estadística)
18 de abril de 2012 12:00
Salón Madrid
