Estimation of Real Estate Prices Using State Space Modeling and Kalman Filtering

• Main Author: Agnar Darri Lárusson 1986-
• Other Authors: Háskóli Íslands
• Format: Thesis
• Language: English
• Published: 2017
• Subjects: Tölfræði; Fasteignaviðskipti; Markaðsmál; Fasteignamat; Iceland
• Online Access: http://hdl.handle.net/1946/29099

This study uses the Kalman filter in state space form to estimate the market price of Icelandic real estate. The data are obtained from Registers Iceland (RI), and includes roughly 32500 instances of sales informations located within the greater capital area, ranging from January 2010 up to and including June 2017. The hedonic approach is taken, wherein sales price is modeled using explanatory variables, which includes housing characteristics such as size, location and type. Several other methods exist for estimating real estate prices using the hedonic approach. For the Icelandic market, one such method is used by RI for taxing purposes. The aim of the study is to achieve perfomance results comparable to RI’s models using the aformentioned methods with as few explanatory variables as needed. The result is a model with relatively few variables which performs on par with RI.