Identifying the Micro-Dynamic Effects of Urban Street Configuration on House Price Volatility Using a Panel Model

Abstract

For the majority of households, housing represents the main component of wealth, and the “wealth effect” of housing on consumption is significant (Case et al. 2005). Furthermore, the housing market not only impacts the economy through wealth effects, but also has influences on other markets (e.g., the mortgage market, mortgage insurance, mortgage backed bonds, and consumer durables). Therefore, understanding the sources of housing market price volatility has great significance on economy and the welfare of society. Despite the importance of the housing market, most empirical analyses mainly focus on exploring the macrodeterminants of house price movements over time by using aggregated data (e.g., real income, real GDP, tax rates, interest rates, population, construction costs, and consumer price index) (Holly and Jones 1997; Hort 1998; Takala and Barot 1998; Meese and Wallace 2003; Harter-Dreiman 2004; Riddel 2004). At the micro-level, there are also researchers who identify the price impact of one-time changes of building characteristic, measuring the added value for housing price over time by hybrid repeat sales models and hedonic models (McMillen 2003; Case et al. 2006; McMillen and Thorsnes 2006; Tian 2006; Noonan 2007). However, hybrid repeat sales and hedonic models have been criticized for the basis of sample selection and underestimation (Polinsky and Rubinfeld 1977; Polinsky and Shavell 1982). Recently, Iacono and Levinson (2011) examined the micro-dynamic relationship between accessibility and land value over time by first difference estimation. They used property sales data of Minneapolis–St. Paul region for the years 2000 and 2005 and measured of accessibility combined with regional accessibility data for their respective years to predict changes in prices over time in a particular location. However, they failed to find any significant micro-dynamic effects of accessibility on land value over time. They also noted that measuring changes in accessibility should be described more accurately than they were able to. Although these approaches could address the problem of omitted variable, they did not consider the unobserved heterogeneity variation over time (e.g., housing policy change and construction cost change), and assume the macrofactors to be constant, which also may lead to an estimation bias.