Pricing the residential housing of Shanghai and Shenzhen

Abstract

[ACCESS RESTRICTED TO THE UNIVERSITY OF MISSOURI AT AUTHOR'S REQUEST.] The price of a real estate depends on its location. In China, an urban high-rise typically includes multiple dwellings, which share the same characteristics such as the age of the estate, the neighborhood amenities, and the surrounding environments, etc. First, with price series of Shanghai's residential housing market during the period 2003 - 2005, we develop three hedonic pricing models by incorporating district dummies, proxy variables like the distance to CBD and the teacher student ratios, and estate fixed effect respectively to control for location (and neighborhood) fixed effect. We find that estate dummy variables provide us with small enough location classification and allow us to better examine the differentiation across properties. Second, in China an urban complex of high-rise buildings (estates) typically includes rental apartment units and owner-occupied units of similar quality. We use the rent data for inference of estate-specific fixed effect, which in turn helps to estimate the unobserved rent of a unit sold in the same real estate property. With data of two major cities in China, Shanghai and Shenzhen, we develop a Bayesian approach to conduct joint finite-sample inference of price and rent, and analyze the factors that determine the price-to-rent ratios of the units sold. A crucial feature of our rent-based pricing model lies in its heterogeneity across properties. We detect that the estate-fixed effect is the dominating factor of the cross-estate mean of the price-to-rent ratio and an important source of the cross-unit variations in the ratio.