|
Abstract:
|
This paper proposes a model of house prices where a house price level in one period continuously affects the house price growth in the subsequent period. Past methods for estimating tiered house price indexes have defined tiers discretely (i.e., high, medium, low price tiers). In these frameworks small changes in value can lead to potentially large changes in expected price trajectories if the change alters the observation's classification. The discrete tier methods are also subject to substantial estimation biases because the classification of a property's tier is correlated with the endogenous variable. The contributions of this paper are: 1) defining a continuous house price tiering specification; 2) defining a data generating process consistent with this specification; 3) developing a methodology for consistent estimation of tiered house price growth in the context of both for repeat sales and hedonic house price models; 4) illustrating the use of a tiered model for forecasting; and 5) examining key 'bubble' markets for evidence of market segmentation during the boom and bust phases of the recent housing cycle.
|
