2007
Abstract:
In principle, spatial and temporal variations in the price of real estate within an
urban area offer an excellent source of data with which any number of economic or
policy questions might be addressed. Be it the value of proximity to public goods {
schools, parks, pollution, etc. { or proximity to private goods { retail and consumption
activities, for example { these locational amenities should be capitalized into property
prices. It is on this premise that much of the empirical literature on access or proximity
rests. Unfortunately, in a complex urban setting, common empirical approaches to
measuring the value of proximity can be highly sensitive to choice of subsample and to
the parameterizations of proximity itself. This is a direct result of an urban context
in which distinct housing submarkets can exist even within small areas, variations
across which swamp the relatively simple controls employed in traditional hedonic
analysis. This paper demonstrates that external heterogeneity significant complicates
traditional hedonic analysis and may preclude its use in complex urban land markets.
In this setting, appeals to the Law of Large Numbers may be inappropriate. This
paper reports the extent to which commonly-used empirical approaches produce widely
inconsistent estimates using the example of a valuation exercise of proximity to mass
transit. The paper proposes a more robust approach that o®ers a greater degree of
confidence in parameter estimates. In contrast to the instability of parameter estimates
under traditional models, the more robust, nonparametric approach yields a consistent
finding of no significant capitalization of light rail into single-family home values.
