shipping and even rent at the
mall. This concept is the basis
for the adjustment grids used
by real estate valuers. In a sense,
the traditional appraisal grid is
a hedonic model without the
regression and with significantly
less data.
Properly estimating the
contribution of each attribute in
a hedonic model requires some
technique that can measure the
dependence of a variable (price)
on one or multiple other variables
(time, beds, baths, cityscape
views). The tool used for this
job is regression analysis. At its
simplest, regression analysis is the
familiar practice of finding a line
of best fit through a scatter plot. In
the case of real estate, this could
translate to finding a line of best fit
through a scatter plot of property
sales with price on the y-axis and
time on the x-axis. The equation
of the line would provide a rough
estimation of the dependence in
general of price on time—what
is commonly referred to as the
market trend. Multiple regression
analysis generalizes this method to an arbitrary
number of independent variables.
Bedrooms, Bathrooms and Beyond
Prospective buyers of homes and retail centers
do not simply consider the physical attributes of
properties. Neither does a residential appraisal
only consider room count and cabinetry. The
power of the hedonic pricing model comes
from its ability to reach beyond bedrooms and
bathrooms to include any other factor that may
contribute to value. The model can be used to
price attributes that are external to the property
but potentially influential to value, such as school
district, zoning, highway access and proximity to a
commercial center.
In the past three decades, hedonic price models
have been applied to a wide variety of conditions
to quantify their influence on property values.
Subjects include hazardous sites, incinerators,
prisons, groundwater contamination, high-voltage
transmission lines, pig farms, freeway noise,
pipelines, light rail, substations, utility-scale solar
farms and wind farms. Also popular are less
location-specific factors such as air quality, water
quality and ambient noise.
Natural resources are not market goods, so their
value cannot be observed directly from prices.
The hedonic regression has become an especially
popular tool for researchers studying the extent
to which stakeholders are willing to pay (or
not willing to pay) for natural resources such
as wetlands, air quality and open space. Along
with other methods such as qualitative choice
modelling and contingent valuation surveys,
hedonic regressions are widely used in Natural
Resource Damage Assessments (NRDA). The
NRDA process, through CERCLA (Superfund)
and the Oil Pollution Act, assesses and restores
“bugs, bunnies and flowers,” along with other
resources impacted by events such as accidental
petroleum releases and shipwrecks.
These models are similarly useful for pricing
public goods and services such as energy and
transmission infrastructure, hazardous site
cleanup and transportation corridors. Like natural
resources, these are public goods and services
not traded in an open market. Researchers must
indirectly value them to conduct cost-benefit
analyses for government accountability and to
suburban Southern California. But
how much does each additional
square foot add? How about each
degree of an ocean viewshed?
Using hedonic regression analysis,
answers to these questions may be
estimated to a dollar amount or
percent indication.
Hedonic regression analysis
brings together a 20th century
theory of consumer demand (the
hedonic model) and a 19th century
statistical technique (regression
analysis). Hedonic models need
not employ regressions, and
regressions need not be hedonic
price models. A hedonic model is
one that represents a single good
or service as consisting of a set
of attributes—with each attribute
subject to its own demand in the
market (and thus having its own
implicit price). Members of the
IRWA may have heard appraisers
refer to real property ownership
as a bundle of rights. Likewise, a
good such as a pair of brand name
blue jeans may be considered a
bundle of attributes—branding,
quality of fabric, color, labor,