Helsingin yliopisto - Helsingfors universitet - University of Helsinki ID 2009-227
Faculty of Social Sciences
Department of Economics
Rouvinen, Annina
Työn nimi-Arbetets titel-Title
Constructing hedonic regression models for quality adjustment in a house price index
General Economics
Työn laji-Arbetets art-Level
Master's thesis
Aika-Datum-Month and year
Sivumäärä-Sidantal- Number of pages
93 s., 2 liites.
The thesis examines the construction of hedonic functions as a method of quality adjustment in price measurement from a theoretical and an empirical perspective. Hedonic price regression models are constructed for houses in Helsinki, using Statistics Finland’s data set on house prices and characteristics. The models are applied for quality adjustment in house price indices. In the theoretical part, we overview the mainstream microeconomic model of hedonic functions and present an alternative theoretical framework for hedonic functions, put forward by recent advances in the hedonic field. We start with introducing the main concepts of the subject: the hedonic hypothesis, on which the theory of functions rests; the characteristics space, which is the domain of hedonic functions; and the implicit characteristics markets, where hedonic functions are formed. Producer and consumer optimization behaviours in the implicit characteristics markets are analysed. In particular, we compare the formation and the properties of hedonic price functions in two different implicit characteristics market environments – in perfect and imperfect competition – and address the plausibility of the two market equilibrium models. We find justifications for concluding that the market structure for the majority of heterogeneous products is of imperfect competition. In such markets hedonic functions are formed in a complex equilibrium process, which implies that hedonic functions can change over time, they can take different forms in different market areas, and that negative coefficients in hedonic functions cannot be ruled out. Some econometric issues concerning the construction of hedonic regression models are addressed, with the main results that profound knowledge of the heterogeneous product is vital for selecting the explanatory variables for a hedonic regression equation, and that choosing the functional form for a hedonic regression equation is not a theoretical, but an empirical matter. In the empirical research, we illustrate the findings of the theoretical overview by constructing hedonic regression models for house prices. In particular, our purpose is to find empirical evidence for the implications that imperfectly competitive markets were found to have for hedonic functions and address the sensitivity of the hedonic quality adjustment to omitted variables. Hedonic regression models are estimated separately for four areas in Helsinki, using the method of ordinary least squares. We report on the parameter estimation results, and choose two hedonic regression models for each area, one with a narrower set of explanatory variables and another with a broader set of variables. Applying the models for quality adjustment, we calculate quarterly house price indices for the time period 2002Q1 to 2008Q4 for each area, and compare the quality adjustment performances of the two hedonic regressions. We conclude that the findings of the empirical study on hedonic house price regressions support the view that hedonic functions are a result of a complex market equilibrium process, and therefore hedonic functions can differ in different market areas for the same heterogeneous product and the parameters or the functional forms do not need to be stable over time.
hedonic regression analysis hedonic function quality adjustment house prices price index hedoninen regressioanalyysi hedoninen funktio laadunvakiointi asunnot - hinnat - Helsinki hintaindeksit
Säilytyspaikka-Förvaringsställe-Where deposited
Muita tietoja-Övriga uppgifter-Additional information