MIT LIBRARIES
:
3 9080 02618 2060
'.!''.'
''
..'
..
:')
1*1
I »1
.M415
no.05-27
2006
l
!•;« Mi'.;
il!
Digitized by the Internet Archive
in
2011 with funding from
Boston Library Consortium Member Libraries
http://www.archive.org/details/doeshazardouswas00gree2
neWEY
HB31
v
-"
.M415
2&
1^?o k
Massachusetts Institute of Technology
Department of Economics
Working Paper Series
DOES HAZARDOUS WASTE MATTER?
EVIDENCE FROM THE HOUSING MARKET AND THE
SUPERFUND PROGRAM
Michael Greenstone
Justin Gallagher
Working Paper 05-27
October 31 ,2005
Revised: September, 2006
Room
E52-251
50 Memorial Drive
Cambridge,
MA 02142
paper can be downloaded without charge from the
Social Science Research Network Paper Collection at
This
http://ssrn.com/abstract=840207
OF'
200S
uffififiS^
Does Hazardous Waste Matter?
Evidence from the Housing Market and the Superfund Program*
Michael Greenstone and Justin Gallagher
September 2006
We thank
Daron Acemoglu, David Autor, Maureen Cropper, Esther Duflo, Dick Eckhaus, Alex Farrell,
Gyourko, Paul Joskow, Matthew Kahn, David Lee, Jim
Poterba, {Catherine Probst, Bernard Salanie, Randall Walsh, Rob Williams, Catherine Wolfram, and
especially Ted Gayer for insightful comments. The paper also benefited from the comments of seminar
participants at ASSA, BYU, UC-Berkeley, UC-Davis, UC-Santa Barbara, CEMFI, Colorado, Columbia,
HEC Montreal, Kentucky, LSE, MIT, NBER, Rand, Resources for the Future, SITE, Stanford, Syracuse,
Texas, and UCLA. Elizabeth Greenwood provided truly outstanding research assistance. We also thank
Leila Agha, Brian Goodness, Rose Kontak, William Li, and Jonathan Ursprung for exemplary research
assistance. We especially thank Katherine Probst for generously sharing data on the costs of Superfund
clean-ups.
Funding from the Center for Integrating Statistical and Environmental Science at the
University of Chicago and the Center for Energy and Environmental Policy Research at MIT is gratefully
acknowledged.
*
Don
Fullerton, Jon Gruber, Jon Guryan, Joe
*
Does Hazardous Waste Matter?
Evidence from the Housing Market and the Superfund Program
Abstract
This paper uses the housing market
develop estimates of the local welfare impacts of Superfund
to
sponsored clean-ups of hazardous waste
sites.
hedonic model predicts that they will lead
We
show
that if
to increases in local
consumers value the clean-ups, then the
housing prices and
new home
construction,
as well as the migration of individuals that place a high value on environmental quality to the areas near
the
improved
sites.
hazardous waste
sites
We
compare housing market outcomes
chosen for Superfund clean-ups
missed qualifying for these clean-ups.
We
in
to the areas
find
that
the areas surrounding the
surrounding the 290
sites that
Superfund clean-ups are
first
400
narrowly
associated
with
from zero local changes in residential property
values, property rental rates, housing supply, total population, and the types of individuals living near the
sites. These findings are robust to a series of specification checks, including the application of a quasiexperimental regression discontinuity design based on knowledge of the selection rule. Overall, the
preferred estimates suggest that the local benefits of Superfund clean-ups are small and appear to be
substantially lower than the $43 million mean cost of Superfund clean-ups.
economically small and
statistically indistinguishable
Michael Greenstone
Justin Gallagher
MIT, Department of Economics
50 Memorial Drive, E52-359
Department of Economics
549 Evans Hall, MC 3880
Cambridge,
and
MA 02142-1347
NBER
mgreenst@mit.edu
UC Berkeley
CA 94720-3880
Berkeley,
justing@econ.berkeley.edu
Introduction
The estimation of
individuals' valuations of environmental amenities with revealed preference
methods has been an active area of research
more than
for
models outlining revealed preference methods
to
three decades.
There are
now
theoretical
recover economically well defined measures of
willingness in a variety of settings, including housing markets, recreational choices, health outcomes, and
the consumption of goods designed to protect individuals against adverse environmentally-induced
outcomes (Freeman 2003 and Champ, Boyle, and Brown 2003 contain reviews). The application of these
approaches, however,
undermine the
often accompanied
is
by seemingly valid concerns about misspecification
of any findings. Consequently,
credibility
many
determine individuals' valuations of environmental amenities.
Hazardous waste
concern.
to
are skeptical that markets can be used to
1
an example of an environmental disamenity that provokes great public
The 1980 Comprehensive Environmental Response, Compensation, and
became known
danger
sites are
as Superfund,
gave the
EPA
the right to place sites that pose an
at
those
sites.
Through 2005, approximately $35
Liability Act,
which
imminent and substantial
public welfare and the environment on the National Priorities List
remedial clean-ups
that
(NPL) and
to initiate
billion (2005$) in federal
monies
and an unknown amount of private funding has been spent on Superfund clean-ups, and yet remediations
are incomplete at roughly half of the nearly 1,600 sites.
2
The combination of
these high costs and the
absence of convincing evidence of its benefits has made Superfund a controversial program (EPA 2006).
This paper uses the housing market to estimate the welfare consequences of Superfund sponsored
clean-ups of hazardous waste
proximate
to the
counterfactual
is
Superfund
likely to
polluted ones in the US.
in the
1
sites.
sites in the
The empirical challenge
absence of the clean-ups
is
is
be especially challenging, because the
that the evolution of
unknown. The development of a
sites
assigned to the
For example, what would have happened to housing prices
NPL
in
are the
valid
most
Love Canal, NY,
absence of the famous Superfund clean-up there?
Further, the increasing reliance on stated preference techniques to value environmental amenities
to dissatisfaction
with the performance of revealed preference techniques. See
Hausman (1994)
for discussions
2
housing prices
Throughout
the paper,
Hanemann
of stated preference techniques.
monetary figures are reported
in
2000
S's, unless
otherwise noted.
is
(1994)) and
surely related
Diamond and
To
solve this problem,
EPA
rule that the
we implement
used to develop the
NPL
first
conduct 400 clean-ups. After cutting the
list
in 1983.
The
of candidate
sites
implemented the Hazardous Ranking System (HRS)
on the risk
it
compare
initial
HRS
To
exogenous change
show
that if
improvement
in a local
We
979) to focus the comparisons
structure the analysis,
from 15,000
EPA
each
to
site a
690, the
EPA
score from
placed the 400
outcomes between 1980 and 2000
scores above and below the 28.5 threshold.
1
only allocated enough
sites
we model
amenity
in areas
to
invented and
to
1
00 based
HRS
with
scores
We
near sites that had
also implement a regression discontinuity
among
sites
with scores near the threshold.
the consequences of a quasi-experiment that leads to an
in the context
of the hedonic method (Freeman 1974; Rosen 1974).
consumers value the clean-ups, then there are two empirical predictions.
at the site
money
1983, making them eligible for Superfund remedial clean-ups.
in
the evolution of housing market
design (Cook and Campbell
We
NPL
initial
EPA was
that assigned
posed, with 100 being the most dangerous. The
exceeding 28.5 on the
knowledge of the selection
a quasi-experiment based on
First, the
should lead to increases in the demand and supply of local housing and,
in turn,
increases in the prices and quantities of houses. Second, the improvement should lead to sorting such that
the share of the population living near the
quality increases.
The implication
is
that
improved
sites that places a
high value on environmental
an exclusive focus on housing prices as
experimental hedonic studies (Chay and Greenstone 2005; Linden and Rockoff 2006)
in
previous quasi-
may obscure
part of
the welfare gain.
The
the
NPL
results suggest that individuals place a small value
and subsequent clean-up. Specifically,
we
on a hazardous waste
find that a site's placement
on the
site's inclusion
NPL
is
on
associated
with economically small and statistically indistinguishable from zero local changes in residential property
values, property rental rates, housing supply, total population, and the types of individuals living near the
site.
These findings are robust
measured 7
that the
(in
mean
1990) or 17
(in
to a
wide variety of specification checks, and they hold whether they are
2000) years after placement on the NPL. Overall, these findings suggest
local benefits of a
Superfund clean-up as measured through the housing market are
substantially lower than our estimated average cost of $43 million per Superfund clean-up.
The conventional hedonic approach compares
2
areas surrounding
NPL
sites
with the remainder of
the US.
HRS
In contrast to the
suggest that gains
in
research design, the conventional approach produces estimates that
property values exceed the
produce a number of puzzling results
is
mean
undermine confidence
that
evidence that the conventional approach
is
However, these regressions
costs of clean-up.
likely to
with other determinants of housing market outcomes.
approach's validity. Further, there
in the
confound the
also
effect
Notably, the
of the presence of a
HRS
NPL
site
research design appears to
greatly reduce the confounding.
The study
is
conducted with the most comprehensive data
researchers on the Superfund program and
Superfund hazardous waste
and census-tract
2000.
site
level
Consequently,
hedonic
sites as
its
effects.
of 2000, the
The
sites that
file
ever compiled by the
resulting database has information on
narrowly missed placement on the
housing market outcomes for 1980 (before the release of the
study
this
literature,
which
is
is
a substantial departure
entirely
EPA
first
all
1,400
initial
NPL,
NPL), 1990, and
from the previous Superfund/hazardous waste
comprised of examinations of one or a handful of
collectively covers just 30 different sites (Schmalensee et
or other
al.
sites
and
1975; Michaels and Smith 1990; Kohlhase
1991; Kiel 1995; Gayer, Hamilton, and Viscusi 2000 and 2002; Kiel and Zabel 2001; McCluskey and
Rausser 2003; Ihlanfeldt and Taylor 2004; Messer
The paper proceeds
how
the
HRS
research design
housing market outcomes.
provides some
summaiy
may
findings, respectively. Section
I.
II
from the
statistics.
VI
I
2004; and Farrell 2004).
3
provides background on the Superfund program and
allow for credible estimation of the effects of Superfund clean-ups on
Section
interpretation for the results
Section
as follows.
et al.
discusses
HRS
how
to use
research design.
Sections IV and
V
hedonic theory to provide an economic
Section
III
details the data sources
report on the econometric
and
methods and empirical
interprets the results, while VII concludes.
The Superfund Program and
a
New
Research Design
Using EPA estimates of the probability of cancer cases and the costs of Superfund clean-ups, Viscusi and
Hamilton (1999) find that at the median site expenditure the average cost per cancer case averted by the clean-up
exceeds $6 billion. This health effects approach requires knowledge of the toxics present and the pathways they
travel, the health risk associated with a toxic by pathway pair, the size of the affected population, the pathwayspecific exposure,
and the willingness
associated with each step,
we
to
pay
to
avoid mortality/morbidity.
think this approach
is
Due
to the state
of
scientific uncertainty
unlikely to produce credible benefit estimates.
A. History
and Broad Program Goals
Before the regulation of the disposal of hazardous wastes by the Toxic Substances Control and
Resource Conservation and Recovery Acts of 1976,
burying them
practices.
in the
ground.
Love Canal,
NY
Throughout the 1940s and 1950s,
industrial firms frequently disposed of wastes
perhaps the most infamous example of these disposal
is
was
this area
a landfill for industrial waste, and
21,000 tons of chemical wastes were ultimately deposited there.
found high concentrations of dangerous chemicals
safety of this area
prompted President Carter
relocation of the 900 residents.
in the air
and
to declare a state
The Love Canal
by
After
soil at
New York
more than
state investigators
Love Canal, concerns about
of emergency in 1978 that led
the
to the
incident helped to galvanize support for addressing the
legacy of industrial waste, and this led to the creation of the Superfund program in 1980.
The centerpiece of the Superfund program, and
4
hazardous waste
sites.
These multi-year remediation
this paper's focus, is the
efforts
aim
to reduce
NPL
by the end of 2005 and thereby chosen
permanently the serious, but
1,552 sites have been placed
not imminently life-threatening, dangers caused by hazardous substances.
on the
long-run remediation of
for these long-run clean-ups.
The next subsection
describes the selection process, which forms the basis of our research design.
B.
Site
Assessment
As of
& Superfund Clean-Ups Processes
1996,
inclusion on the
more than 40,000 hazardous waste
NPL. Since
final step
(HRS), which
is
had been referred
*
original
of the assessment process
is
reserved for the most dangerous
HRS
the application of the
sites.
evaluated the risk for exposure
The Superfund program
also
EPA
for possible
EPA
follows a
sites.
The
EPA
to
Hazardous Ranking System
developed the
standardized approach to identify the sites that pose the greatest threat to
The
to the
there are limited resources available for these clean-ups, the
multi-step process to identify the most dangerous
The
sites
HRS
humans and
in
1982 as a
the environment.
chemical pollutants along three migration
funds immediate removals, which are short-term responses to environmental
emergencies aimed at diminishing an immediate threat. These actions are not intended
environmental problem and are not exclusive to hazardous waste sites on the NPL.
to
remediate the underlying
'pathways': groundwater, surface water, and
air.
The major determinants of risk along each pathway
for
a site are the toxicity and concentration of chemicals present, the likelihood of exposure and proximity to
humans, and the
of the potentially affected population. The non-human impact
size
HRS
plays a minor role in determining the
HRS
The
From 1982-1995,
NPL. These
EPA
sites are the
assigned
all
first
step
a site
is
is
placed on the NPL,
summarized
construction of
all
in the
The
site
as the Basis of a
This paper's goal
challenge
that
is
that
sites
NPL
risk.
score of 28.5 or greater to the
it
test
scores and their role in assignment to the
generally takes
many
is
is
The
final step is the site's deletion
most polluted
in the
covary with both proximity to hazardous waste
is
to
complete.
remedy
it.
The
This
that
from the NPL.
New Research Design
to obtain reliable estimates
sites are the
it
problem and how best
is
complete, the immediate threats to health have been removed, and
on housing market outcomes
cannot be tested directly,
years until clean-up
NPL.
receives the "construction complete" designation once the physical
the long-run threats are "under control."
hazardous waste
HRS
Record of Decision (ROD), which also outlines the clean-up actions
clean-up remedies
HRS Scores
with a
100 being the highest level of
only ones that are eligible for Superfund remedial clean-up. The Data Appendix
are planned for the site.
1982
sites
a further study of the extent of the environmental
is
assessment
C.
to 100, with
hazardous waste
provides further details on the determination of HRS
Once
also considered but
score.
produces a score that ranges from
the
is
of the effect of Superfund sponsored clean-ups of
in
areas surrounding the sites.
US, so
sites
it is
likely that there are
and housing
unobserved factors
Although
prices.
notable that proximity to a hazardous waste site
is
The empirical
this possibility
associated with lower
population densities, lower household incomes, higher percentages of high school dropouts, and a higher
fraction of
mobile homes among the housing stock.
Consequently, cross-sectional estimates of the association between housing prices and proximity
to a
"
hazardous waste
site
may be
severely biased due to omitted variables."
In fact, the possibility of
Cross-sectional models for housing prices have exhibited signs of misspecification in a
5
number of
other settings,
confounding due to unobserved variables has been recognized as a threat to the use of the hedonic method
to
develop reliable estimates of individuals' willingness
This paper's challenge
invention (Small 1975).
market outcomes near Superfund
A
may
sites in the
NPL
feature of the initial
Through an
remedial action.
dangerous
sites (Section
exactly
400
105(8)(B) of
sites
sites that
posed the greatest
A
were referred
initial
It
the
risk.
HRS
EPA
central role of the
First,
NPL, because
In the first year after the
and investigated as potential candidates for
EPA winnowed
assessment process, the
EPA
to
this list to the
develop a
budgetary considerations caused the
EPA
NPL
690 most
of "at least" 400
to set a goal
of placing
to
provide a scientific basis for determining the 400 out of the 690
Pressured to
to the
initiate the
690 worst
sites,
clean-ups quickly, the
and
their scores
it
is
NPL
initial
EPA
developed the
HRS
were ordered from highest
published
in
to
September 1983
scores exceeding 28.5. See the Data Appendix for further details.
housing market outcomes near
HRS
HRS
was applied
limited to sites with
reasons.
to the
score of 28.5 divided numbers 400 and 401, so the
The
and clean-up.
on the NPL.
about a year.
lowest.
NPL
assignment process that has not been noted previously by researchers
CERCLA),
The EPA developed
was
absence of their placement on the
Although the Superfund legislation directed the
sites.
its
develop a valid counterfactual for the housing
to
provide a credible solution to the likely omitted variables problem.
legislation's passage, 14,697 sites
in
is
pay for environmental amenities since
to
HRS
score provides a compelling basis for a research design that compares
sites
with
unlikely that sites'
the 28.5 threshold
was
scores therefore reflected the
initial
HRS
scores above and below the 28.5 cut-off for at least three
scores were manipulated to affect their placement on the
established after the testing of the 690 sites
EPA's assessment of the
risks
posed by each
was completed. The
site
and were not based
on the expected costs or benefits of clean-up.
Second, the
HRS
scores are noisy measures of risk, so
including the relationships between land prices and school quality,
Chay and Greenstone 2005; Deschenes and Greenstone 2006).
it
is
possible that true risks are similar
and climate variables (Black 1999;
is an alternative
1988). Other potential sources of biases
air pollution,
Incorrect choice of functional form
source of misspecification (Halvorsen and Pollakowski 1981; Cropper
et al.
of published hedonic estimates include measurement error and publication bias (Black and Kneisner 2003;
Ashenfelter and Greenstone 2004).
above and below the threshold. This noisiness was a consequence of the
scientific uncertainty
health consequences of exposure to the tens of thousands of chemicals present at these sites.
any evidence
there wasn't
that sites
with
Register specifically reported that the
28.50 do not present a significant risk
HRS
"EPA
to
scores
has not
human
below 28.5 posed
made
little
risk to health.
6
will
score,
which allows
compare outcomes
health, welfare, or the
environment" and that a more
"near" the 28.5 cut-off.
smoothly around the regulatory threshold, then
An
eventual
additional feature of the analysis
NPL
status but
is
site
was on
Finally,
it
the
NPL in
(i.e.,
1982)
is
possible to
This
is
We
HRS
score
9
Specifically,
we
unobservables are similar or changing
make
that an initial score
it.
7
a highly nonlinear function of
causal inferences.
above 28.5
because some
is
sites
8
highly con-elated with
were rescored, with the
The subsequent analysis uses an
was above 28.5
indicator
as an instrumental variable for
order to purge the potentially endogenous variation in
important to emphasize that
Superfund remediations.
6
is
is
not a perfect predictor of
variable for whether a site's initial
whether a
it
If the
determining whether they ended up on the NPL.
later scores
NPL
for a quasi-experimental regression discontinuity design.
at sites
Federal
a determination that sites scoring less than
Third, the selection rule that determined placement on the
HRS
Further,
The
informative test would require "greater time and funds" (Federal Register, September 21, 1984).
the
about the
sites that failed to qualify for the
NPL
NPL status.
were
ineligible for
investigated whether these sites were cleaned-up under state or local
A
recent history of Superfund's makes this point. "At the inception of EPA's Superfund program, there was much
be learned about industrial wastes and their potential for causing public health problems. Before this problem
could be addressed on the program level, the types of wastes most often found at sites needed to be determined, and
their health effects studied. Identifying and quantifying risks to health and the environment for the extremely broad
range of conditions, chemicals, and threats at uncontrolled hazardous wastes sites posed formidable problems.
Many of these problems stemmed from the lack of information concerning the toxicities of the over 65,000 different
industrial chemicals listed as having been in commercial production since 1945" (EPA 2000, p. 3-2).
to
7
the crude nature of the initial HRS test is by the detail of the guidelines used for determining
The guidelines used to develop the initial HRS sites were collected in a 30 page manual. Today, the
analogous manual is more than 500 pages.
The research design of comparing sites with HRS scores "near" the 28.5 is unlikely to be valid for sites that
received an initial HRS score after 1982. This is because once the 28.5 cut-off was set, the HRS testers were
One way
the
HRS
to
measure
score.
encouraged
minimize
and simply determine whether a site exceeded the threshold. Consequently,
pathways once enough pathways are scored to produce a score above the threshold.
As an example, 144 sites with initial scores above 28.5 were rescored and this led to 7 sites receiving revised
scores below the cut-off. Further, complaints by citizens and others led to rescoring at a number of sites below the
cut-off. Although there has been substantial research on the question of which sites on the NPL are cleaned-up first
(see, e.g., Sigman 2001 ), we are unaware of any research on the determinants of a site being rescored.
to
testing costs
testers generally stop scoring
programs and found
that they
were frequently
untouched.
left
programs, a typical solution was to put a fence around the
The point
health hazards.
activities at
II.
non-NPL
Using Hedonics
An
explicit
commonly used
individuals.
is
To
sites in
to
Value Changes
to infer the
its
in
and place signs indicating the presence of
NPL
a clean local
environment does not
The hedonic
exist.
economic value of non-market amenities
like
price
method
It
new
The purpose of
equilibrium.
their optimizing decisions.
then discusses
environmental quality due to a Superfund clean-up leads agents to
decisions and the resulting
made
how
an improvement
discussion
is
strategy to infer the welfare consequences of Superfund clean-ups using decennial
Brief Review of Equilibrium
in the
it
This
in local
and profit-maximizing
alter their utility
this
is
environmental quality to
empirical implementation has generally been in cross-sectional settings where
reasonable to assume that consumers and producers have already
A
exceeded the clean-up
sites drastically
Local Environmental Quality Due to Superfund Clean-ups
section briefly reviews the cross-sectional equilibrium.
A.
were targeted by these
the sites that
scope and cost.
market for
date,
site
that the remediation activities at
is
Among
to devise an empirical
Census
data.
Hedonic Model
Economists have estimated the association between housing prices and environmental amenities
at least since
Ridker (1967) and Ridker and Henning (1967).
(1974) were the
differentiated
first to
good
is
give this correlation an economic interpretation.
described by a vector of
house, these characteristics
public services
(e.g.,
hazardous waste
(1)
The
site).
However, Rosen (1974) and Freeman
may
characteristics,
its
include structural attributes
local school quality),
(e.g.,
C =
In the
(c h c 2 ,..., c n ).
lh
i
a
In the case of a
number of bedrooms), neighborhood
and local environmental amenities
Thus, the market price of the
Rosen formulation,
house can be written
(e.g.,
distance from a
as:
Pi=-P(c ,,cia,...,c ta ).
i
partial derivative
implicit price.
It is
holding constant
all
of P(») with respect to the
the marginal price of the
lh
j
th
characteristic, <9P/<5cj, is referred to as the marginal
j
characteristic implicit in the overall price of the house,
other characteristics.
8
In the hedonic
schedule (HPS),
model, the locus between housing prices and a characteristic, or the hedonic price
generated by the equilibrium interactions of consumers and producers.
is
markets are competitive,
that
consumption of the numeraire,
consumers rent one house
all
X
(with price equal to
1),
at the
market price, and
It is
assumed
depends on
utility
and the vector of house characteristics:
(2) u = u(X,C).
The budget
constraint
is
expressed as
1
-P-
Maximization of (2) with respect
each of the characteristics
Cj (e.g.,
It is
By
=
to the
to satisfy {d\J/dc )
}
local environmental quality)
X
I
0,
where
I is
income.
budget constraint reveals that individuals choose levels of
(8U/9x) = 8P/9Cj. Thus, the marginal willingness to pay for
must equal the marginal cost of an extra unit of Cj
convenient to substitute the budget constraint into
inverting this equation and holding
willingness to pay for
c, is
all
(2),
characteristics of the
which gives u =
house but
in the
market.
u(I- P, c h c 2 ,..., c n ).
constant, an expression for
j
obtained:
(3)B = B (I-P,c ,C.;,u*).
J
J
Here, u*
is
J
the highest level of utility attainable given the budget constraint and
quantities of other characteristics.
the
maximum amount
Heterogeneity
to differences in the
HPS
This
that an individual
is
would pay
for different values
due
declining marginal rate of substitution between
in locations
where
utilities
The other
and
Cj
X
their marginal willingness to
marginal implicit price, which occur
consumers'
the optimal
is
it
reveals
utility constant.
preferences and/or incomes leads
depicted
in
Figure la, which plots the
of three consumer types. The consumers are denoted as types #1, #2, and #3,
and potentially there are an unlimited number of each type.
houses
of Cj, holding
to differences in
chosen quantities of a characteristic. This
Cj
is
referred to as a bid (or indifference) curve, because
in individuals' bid functions
and bid curves for
Cj
would be lower
2
at
at sites
side of the market
is
Cj',
Cj
,
and
(because
pay for
3
Cj
Each bid function reveals
,
X
Cj
=
is
I
-
P).
The
three types choose
equal to the market determined
respectively.
Given market
prices,
these
with higher or lower levels of local environmental quality.
comprised of suppliers of housing services.
suppliers are heterogeneous due to differences in their cost functions. This heterogeneity
differences in the land they own.
the standard
For example,
it
may
We
assume
may
result
that
from
be very expensive to provide a high level of local
By
environmental quality on a plot of land located near a steel factory.
function,
(4) O,
we can derive
(cj5 C,\ n),
its
offer curve for the characteristic
inverting a supplier's profit
Cj!
= O,
where n
is
the
maximum
available profit given
its
cost function and the
HPS. Figure la depicts
curves for three types of suppliers. With this set-up, individuals that live in a house that they
be both consumers and suppliers and their supplier self would rent to their consumer
The HPS
point on the
is
HPS,
offer
own would
self.
formed by tangencies between consumers' bid and suppliers' offer functions. At each
the marginal price of a housing characteristic
equal to an individual's marginal
is
willingness to pay for that characteristic and an individual supplier's marginal cost of producing
the consumer's perspective, the gradient of the
HPS
it.
From
with respect to local environmental quality gives the
equilibrium differential that compensates consumers for accepting the increased health risk and aesthetic
disamenities associated with lower local environmental quality.
environmental quality must have lower housing prices to
attract potential
reveals the price that allocates consumers across locations. Thus, the
effects
of a marginal change
in a characteristic.
From
Put another way, areas with poor
HPS
homeowners, and the
HPS
can be used to infer the welfare
the suppliers' perspective, the gradient of the
HPS
reveals the costs of supplying a cleaner local environment.
B. Wliat are the
Consequences of a Large Change
in
Environmental Quality
in the
Hedonic Model?
This study assesses the impacts of Superfund remediations of hazardous waste
to
cause non-marginal improvements
fleshes out the hedonic
suppliers,
model
and social welfare.
affect welfare since
in
environmental quality near the
to describe the theoretical impacts
Any
site.
of these clean-ups on consumers,
impacts on the labor market are ignored, because wage changes don't
any gains (losses) for workers are
of a constant
HPS
does not
HPS may
which intend
This subsection extends and
offset
by losses (gains)
for firms
We
The Impacts of an Amenity Improvement on Consumers and Suppliers.
where the overall
sites,
shift in
response to the increased supply of "clean"
be valid because to date only 670 Superfund
remediated. They are located in just 624 of the 65,443
10
US
census
tracts,
sites
(Roback 1982).
focus on the case
sites.
The assumption
have been completely
which constitutes
a small part of
the
US
housing market.
Now,
from
3
Cj'
to
Cj
consider the clean-up of a hazardous waste
as in Figure la in the
rental price of housing near the
neighborhood surrounding the
improved
rental rate exceeds their willingness to
become too expensive, given
The
result
is
that
pay for the clean-up.
their preferences
consumers
move
3
cj
house with
at
a price of p 3
change locations, but
their utility
One consequence of
will
have greater unobserved
based on
is
move
clean-up.
If residential
are a pure benefit for
this case, the
It
is
its
10
,
in the
utility.
The
previous rental price and environmental
the key result
is
this taste-based sorting is that the residents
Cj (i.e.,
that
pi
and
where they
site,
their original
Cj
Cj').
will
some consumers
will
and
p.
of the improved neighborhood
environmental quality and/or higher incomes." Thus, the marginal
owners near the
site are the
We test
for this taste -based sorting below.
only agents whose welfare
is
affected by the
and commercial land markets are perfectly integrated, then the higher
landowners because the change
all
that the
to restore the equilibrium.
unchanged because they choose locations with
taste for
HPS
Consequently, their neighborhood has
near the newly cleaned-up
resident will be less tolerant of exposure to hazardous waste.
In this set-up, land
evident from the
chosen and optimal values of p and
So assuming zero moving costs
.
It is
between communities
site
their originally
will
site.
and income, and the clean-up reduces their
will migrate
some type #3 consumers
Additionally,
consume
to a
increases local environmental quality
For type #1 consumers, the increase
site will rise to P3.
type #1 consumers that had chosen the improved
quality will
site that
supply of land for residential purposes
is
in
environmental quality
is
rental rates
costless for them.
In
fixed.
possible that the residential and non-residential land markets are not perfectly integrated,
perhaps due to zoning laws, which are costly to change (Glaeser and Gyourko 2003).
increase in rental prices
is still
a pure benefit for
owners of residential land near the
site.
In this case, the
The higher
rents
we assume zero moving costs although this surely isn't correct. In the presence of moving costs,
made worse off by the amount of the moving costs. See Bayer, Keohane, and Timmins (2006) on the
impacts of moving costs on the valuation of air pollution.
11
See Banzhaf and Walsh (2005) and Cameron and McConnaha (2005) for evidence of migration induced by
environmental changes.
In principle, the new residents' incomes could have a direct effect on individuals'
For simplicity,
renters are
valuations of living
in the
community.
We
ignore this possibility here because this will not create any social
benefits as long as the benefits from living near high
income individuals
11
are sufficiently linear.
for residential land will cause
land to residential usage.
when
some owners of
non-residential land to find
it
profitable to convert their
Presumably, the pre-clean-up rental rate of the converted land had been higher
in the non-residential sector
and/or there
may be
costs associated with conversion (e.g., legal fees
associated with rezoning), so the benefits for owners of converted land are smaller than for owners of land
that
was already used
Ultimately, the benefits of conversion determine the shape
for residential housing.
of the supply curve of residential land near the
and the welfare gain for these land owners.
site
The
empirical analysis tests for supply responses.
To summarize,
there are four predicted impacts of an amenity improvement.
land (and housing) near the improved
sorting.
site will increase.
We
Census data
after
we
how
discuss
the Full Welfare Effects.
contrast to the preceding discussion, here
we
prices so that local environmental quality
is
suppliers.
Consequently,
the improved sites and elsewhere.
The
A
Total
i
benefits for
Consumer
with decennial
The
full
it
is
after
it
£ [Bitci/"
L
and
A
less expensive.
is
now
in local
change
sum of
environmental quality.
in relative prices
necessary to account for the
consumers can be expressed
WTP =
welfare benefits are the
all
In
allow for the possibility that the remediation alters relative
05
',
WTP
could affect
all
of agents living near
pos,
[Pi
(Cij*
as:
C.{, u/)
post
,
- Bi(cij*
C. a *)-
pr
Pi
indexes a household and there are n households
clean-up and "post"
Fourth, the
12
-
where
likely to increase.
to test these predictions
consumers' and suppliers' willingness to pay (WTP) for the change
(5a)
is
develop a formal expression for the welfare effects of a Superfund clean-up.
A General Expression for
consumers and
of
Second, consumers will respond with taste-based
Third, the supply of residential land (and housing) near the site
entire welfare gain accrues to land owners.
First, the price
all
pre
V
post
C,/,
and Cy
prc
u,*)]
C.;/)],
,
(i=l,...,
adjustments are complete.
the difference in her valuation of exposure to Cy
,
prc
13
n) in the country.
"Pre"
Thus, each consumer's
minus the difference
is
WTP
before the
is
equal to
in the rental rates at
12
See Bartik (1988) and Freeman (2003) for more extensive discussions of the general welfare impacts of nonmarginal amenity improvements.
13
For simplicity,
Pre
C.y
).
we assume
We make
that
consumers do not adjust
their
the analogous assumption about suppliers.
12
consumption of the other characteristics so C_y* Pre =
these values of cy.
The
(5b)
A
benefits for suppliers can be expressed as:
Total Supplier
pos,
WTP =
posl
£k [P (c k/
I [Tk(c
k
*
-
where k indexes a supplier and
k
kj
there are
m
C. kj *)- P k
,
post
,
C. k/)-
(k=
prc
(c k
i k (c k
m)
1,...,
>
rc
,
>
re
,
C. k /)]
C. kj *)],
So each supplier's
cost of producing a house with characteristics C.
x k (C) is supplier k's
suppliers in the country.
WTP
is
equal to the pre and post
difference in price minus the difference in costs.
Thus, the societal change
transfer
from buyers
in
to sellers so
it
welfare
is
the
at
the
new and
the change in suppliers' costs.
The implementation of
consumers' bid functions and
suppliers' cost functions.
all
(5a) and (5b).
Consequently, the
cancels out.
difference in consumers' total willingness to pay
sum of equations
this
total
change
The
in
article,
this
showed
The
it
is
hedonic approach to
difficulty of this task
that taste-based sorting
cost functions for
all
undermines
this
suppliers in the
Can We Learn about
first
efforts to infer
approach requires estimates of bid functions for
economy. This
Rosen (1974) proposed
is
a
a 2-step
how
In a recent paper, Ekeland,
demand (and supply)
who
consumers' bid functions from the HPS.'
all
3
consumers and
tremendous amount of information, and there
is
to the task.
Census Data?
decennial census data on housing and demographic variables can
approach for estimating bid functions (and offer curves).
clear that nothing can be learned about the structure of preferences in a single cross-section"
15
two reasons.
underscored by Epple (1987) and Bartik (1987)
the Welfare Effects of Superfund Clean-ups from Decennial
This subsection considers
14
was
consensus that existing data sources are not up
C.
at least
impossible to observe the same individual facing two sets of prices in a cross-
Second, the implementation of
a
all
estimation of even a single individual's/taste type's bid function has proven to be extremely
challenging, because
14
welfare equals the
approach requires reliable estimates of
estimating the value of amenity changes has not met with great empirical success for
section.
a
is
old levels of environmental quality minus
Three decades after the publication of the original Rosen
First, the
price change
Heckman and Nesheim (2004)
He
later wrote, "It is
(Rosen 1986,
functions in an additive version of the hedonic model with data from a single market.
13
p. 658).
outline the assumptions necessary to identify the
be used
to learn
There are
about the welfare effects of Superfund clean-ups.
at least
two features of these
data that merit noting because they affect the form and interpretation of the subsequent empirical analysis.
The
first
feature
is
that census tracts are the smallest unit
This means that
across the 1980, 1990, and 2000 censuses.
is
it
of observation that can be matched
infeasible to observe individuals over
time and therefore to obtain estimates of their bid and cost functions. Consequently,
we now
consider the
impacts of a clean-up in the context of census tract-level demand and supply functions for residential
land,
which
are
determined by the bid and cost functions of local consumers and suppliers.
We begin
perfectly inelastic,
with the case where the supply curve for residential land near a hazardous waste
which
is
likely to
1
is
depicted in Figure lb with S and
be the case
D
1
in the short-run,
and equilibrium outcome
site
downward
sloping.
This
consider an exogenous
raises current residents' valuation
and, as sketched out in the previous subsection, with free migration
individuals with even higher valuations of environmental quality will
demand curve
is
Now,
(Pi, Qi).
The improvement
increase in environmental quality due to a clean-up.
of living near the formerly dirty
and demand
site is
for residential housing near the
improved
move
in.
This
site shifts out.
is
The
net result
is that
the
depicted as D" and causes
prices to increase to P 2 but leaves quantities unchanged.
With a
parallel shift in the
of the shaded areas A] and
residential plots of land
challenge
is
and
to accurately
A
2
in
measure the change
and the remediation will lead
(Pi, Q2),
which
is
to
is
likely to
changes
in
be more
elastic
to the
this
supply curve, the
is
the
sum of
is
times the number of
a practical perspective, the
new
in prices.
prices only, so they
The gain
welfare
in
B 2 and A 2
,
may have
.
is
is
Previous
understated
improvements.
evident that with census-tract data the development of a
14
site.
equilibrium combination
the shaded areas B],
method have generally examined
sum
Figure lb depicts the unrealistic polar
no change
(potentially dramatically) the welfare gain associated with amenity
It
the
conversion of non-residential land,
With
reflects a substantial gain in quantities but
applications of the hedonic
due
quantities.
consumer surplus and
in price
is
house or residential land prices near the improved
and
.
the welfare gain
From
in prices
perfectly elastic as S 2
entirely an increase in
HPS,
mean change
This equals the
Figure lb.
in the
entirely accrues to suppliers or landowners.
In the longer nan, supply
case where supply
demand curve and no change
full
welfare measure requires
knowledge of the shapes of the supply and demand curves.
separately identifying supply and
demand over
We
are
unaware of a credible strategy
between censuses.
the 10 year periods
for
In this situation,
precise welfare calculations require ad hoc assumptions about the elasticities of supply and demand,
except for the case where neither prices nor quantities change. In
changes
and quantities, so our primary conclusion
in prices
subsequent analysis finds small
fact, the
is
Superfund remediations did not
that
substantially increase social welfare.
The census
tract-level
demographic data can also be used
taste-based sorting in response to remediations.
An
increase in the
to test the theoretical prediction
number of high income
people that are likely to place a high value on environmental quality
would provide complementary evidence
population shifts near the
The second
1990, and 2000.
sites
that the clean-ups are valued.
would suggest
we would
like to
ensure that
we have
However, the
2000.
By
first
then,
NPL was
that they are only available in 1980,
in the future
site's
placement on the
and homeowners
An immediate measurement
NPL
will naturally
of the impact on prices would
all years.
released in 1983, and housing prices cannot be observed again until 1990 or
some of the clean-ups
will
have been completed, and the time
have been greatly reduced. For
this reason, the
to
completion for the others
measurement of the impacts of the
designation with 1990 or 2000 Census data will overstate the properly measured benefits.
III.
A.
is
captured the impact of the clean-up on the value of housing services in
(relative to 1983) will
NPL
near the remediated sites
In contrast, a failure to find these
measure the impact of a
immediately after the announcement so any benefits are
discount them by the rate of time preference.
individuals or
that the clean-ups did not lead to substantial welfare gains.
feature of the data that merits highlighting
Ideally,
in areas
of
Data Sources and Summary
Statistics
Data Sources
We
constructed the most comprehensive data
contains detailed information on
hazardous waste
characteristic,
sites
with 1982
all
hazardous waste
HRS
file
ever compiled on the Superfund program.
sites
scores below 28.5.
placed on the
We
also
NPL
by 2000, as well
compiled housing
price,
and neighborhood demographic information for areas surrounding these
15
It
as the
housing
sites.
This
subsection describes the data sources.
More
provided
details are
in the
Data Appendix and Greenstone
and Gallagher (2005).
The housing, demographic and economic data come from Geolytics's Neighborhood Change
Database, which includes information from the 1970, 1980, 1990, and 2000 Censuses.
1980 data predate the publication of the
each of the hazardous waste
We use the
which
are
sites
first
and used
NPL
this
in 1983.
We
collected the longitude and latitude for
information to place
all sites in
a unique census tract.
Geolytics data to form a panel of census tracts based on 2000 census tract boundaries,
drawn so
that they include
approximately 4,000 people in 2000. Census tracts are the smallest
geographic unit that can be matched across the 1970-2000 Censuses.
entire country in tracts in 2000.
Geolytics
fit
The Census Bureau placed
US Census Bureau
The
result
is
that the
analysis
remaining areas of the country cannot be matched
housing price data
NPL
The
is
restricted to the
in 1980, 1990,
before January
1,
to a
and 2000.
tracts that
sites
which were tested for inclusion on the
The
each observation
tracts,
is
and 353 of the 1982
where the weights
are centered at the sites.
sites (but
The
initial
all
NPL.
further reduced
HRS
first
sites.
conducts the
tracts that share a
excludes the tracts that contain the
comprised of the weighted average of
sites).
variables across these
are the 1980 populations of the tracts.
unit of observation in the third grouping
means across
sites,
The second implements an analysis among census
border with the tracts that contain the hazardous waste
neighboring
NPL
would have
analysis uses three different groupings of census tracts.
analysis at the census tract level.
have non-missing
This sample includes 985 of the 1,398 sites listed on the
2000 and 487 of the 690
The subsequent
so the 1970
tract,
tracts that include these areas.
48,147 out of the 65,443 2000 census
the sample to include just 37,519 census tracts, 708 of the
case,
1970 and 1980,
that in
2000 census
addition of the sample restriction that 1970 housing prices be nonmissing
In this
is
only traded areas that were considered 'urban' or belonged to a metropolitan area.
and 1980 values of the Census variables are missing for 2000
The
the
1970, 1980, and 1990 census tract data to the year 2000
census tract boundaries to form a panel. The primary limitation of this approach
the
Importantly, the
is
the land area within circles of varying radii that
For these observations, the census variables are calculated as the weighted
the portion of tracts that fall within the relevant circle.
16
The weights
are the fraction of each
by
tract's land area within the relevant circle multiplied
radius,
we
that
is
it
its
1980 population.
would value a clean-up and making the area
plausible that residents
the subsequent tables,
we
in
housing prices aren't required for clean-ups
also collected a
number of
and 18.2
The same source was used
various issues of the Federal Register.
completion of remediation
these milestones.
file that
(i.e.,
to
NPL
Information on each
site's size in acres
is
that
the
were placed on the
limitation of the
which
is
was
HRS
initiation
NPL
composite
NPL
listing.
of clean-up,
for sites that achieved
RODs.
the
Finally,
we
and estimated actual
initiated
Greenstone and Gallagher's (2005) Data
the costs of clean-ups (also see Probst and
US
NPL
the
GIS determined
unavailable.
We
sites.
1,398 hazardous waste sites in the 50
comprised of the 690 hazardous waste
A
ROD,
comes from
conducted with two samples of hazardous waste
Sample" and includes
covariates
All
Konisky 2001).
Statistics
The analysis
Columbia
The mean 1980 values of
determine the dates of
construction complete), and deletion from the
Appendix provides more information on
NPL
In
(12), respectively.
reported the dates of the release of the
costs for sites that reached the construction complete stage.
Summary
that
pathway scores, were obtained from
collected data on the expected costs of clean-up before remediation
B.
17
variables about the hazardous waste sites.
scores, as well as separate groundwater, surface water, and air
provided a data
enough so
large
S349 and S796 million and the mean (median) number of census
tracts that are at least partially inside these circles are 9.9 (8)
We
sites so
to pass a cost-benefit test.
focus on circles with radii of 2-miles and 3-miles.
the housing stocks in these circles are
EPA
In choosing the optimal
attempted to balance the conflicting goals of requiring houses to be near enough to the
implausibly large increases
The
16
by January
1,
2000. The second
sites tested for inclusion
circle
approach
is
on the
is
initial
The
first is
states
"1982
called the "All
and the
HRS
District
Sample" and
of
is
NPL.
that street address level data
on housing prices and the
assign a census tract's average to the portion of the tract that falls within the circle,
equivalent to assuming that there
no heterogeneity in housing prices or other variables within a tract.
EPA's and scientific community's positions on the distance from
a Superfund site that the contaminants could be expected to impact human health.
The 1982 Federal Register
reports, "The three-mile radius used in the HRS is based on EPA's experience that, in most cases currently under
investigation, contaminants can migrant to at least this distance. It should be noted that no commentators disagreed
17
is
Our use of a 3-mile
radius
is
is
consistent with the
with the selection of three miles for technical or scientific reasons" (Federal Register July 16, 1982).
17
Table
column
presents
1
(1) are
summary
missing housing price data
which
the
1
is
982
NPL
from the All
HRS
on the hazardous waste
Sample and are limited
sites in these
census
to sites in a
samples. The entries in
tract for
which there
is
1980, 1990, and 2000. After these sample restrictions, there are 985
in
more than 70% of
statistics
the sites placed
Sample. The column
NPL
on the
(2) entries are
complete housing price data. Column
(3) reports
by 2000. Columns
based on the 487
(2)
and
(3) report data
sites located in a
on the remaining 189
incomplete housing price data (generally due to missing 1980 data). 14
nonsites,
from
census tract with
sites located in
census tracts with
sites are outside
of the continental
United States and were dropped from the sample.
A
Panel
75%
of
all
NPL
sites
443 of the 676
that
exceeds the 400
sites that
Congress
NPL. Panel B demonstrates
C
Panel
NPL
set as
Sample eventually were placed on
an explicit goal, because, as
The median
sites only.
effects
mean HRS
that
we have
NPL.
the
discussed,
This number
some
due
to a
site size
few very
D
are achieved
is
The modest
large sites.
reveals that the clean-up process
reported, rather than the
198 (16) of the
NPL
sites in
designation by 2000 (1990).
between 1980 and 2000.
We
For
with
them
for
measured
in acres,
which
is
available for
ranges between 25 and 35 acres across the samples. The means are
on property values are likely to be confined
Panel
sites
scores are similar across the columns.
reports on the size of the hazardous waste sites
substantially larger
yet.
HRS
1982
(1) reveals that about
Together, columns (2) and (3) demonstrate
received this designation in the 1980s.
sites in the
NPL. Column
scores below 28.5 were rescored and then received scores above the threshold qualifying
initial
the
reports on the timing of the sites' placement on the
size
to relatively small
is
slow.
geographic areas around the
until the different
have not reached
(2) received either the construction
this reason,
will also assess
sites
suggests that any expected
sites
The median time
mean, because many
column
of most
we
how
all
sites.
milestones
of the milestones
complete or deleted
focus on changes in housing prices and quantities
rental rates
change as
sites
move through
the different
stages of the clean-up process.
Panel
costs
among
E
reports the expected costs of clean-up for
sites that are construction
NPL
sites,
and F
details
expected and actual
complete or deleted. The expected costs are measured before any
remediation activities have begun, while actual costs are our best estimates of
18
total
remediation related
expenditures assessed after the
variables have been reported for the
(2)), the
mean and median expected
Among
same
In the
sites.
1982
obtain an estimate of the
55%.
mean
We multiply
the overall
HRS
Sample
HRS
it
mean expected
EPA's
costs of administering Superfund.
Superfund clean-ups
A
in the
the
focus on
(i.e.,
HRS
Sample of $43
does not include the legal costs or deadweight loss
Nevertheless,
it
include the site's share of
contrasted with the estimated benefits of
it is
(2)
and
(3) across the panels reveals that the sites with
conditional on scoring above and below 28.5 are remarkably similar.
various cost variables are comparable in the two columns. Consequently,
Moreover, the
cost variables.
to the
changes
similarity of the
results
complete housing price data are similar
may be
representative for the entire 1982
sites in
The mean
in the test
column
This
million.
complete housing price data are similar on a number of dimensions. For example, the mean
subsequent results
column
remainder of the paper.
comparison of columns
that the sites without
time these
first
cost of $27.5 million by 1.55 to
associated with the collection of funds from private parties or taxes, nor does
the
we
that
is
Sample, the mean actual costs exceed
actual costs of clean-up in the 1982
estimate of costs understates the true costs, because
believe this
and $15.0 million.
costs are $27.5 million
the construction complete sites in the 1982
the expected costs by about
We
construction complete.
site is
column
HRS
to the
HRS
Further, the
column
how
column
(2)
the scoring
(1) sites with the other sites suggests that
from the application of the
HRS
research design to the
1
size
and
conclude
(2) sites, suggesting the
and
scores are a few points lower, but this comparison
in the
to
scores
Sample.
(1) are similar to the sites in
over time and changes
HRS
median
seems reasonable
it
and without
it
may be
982
HRS
(3) in size
is
and the two
not meaningful due
was conducted.
Overall, the
reasonable to assume that the
Sample
are informative about
the effects of the other Superfund clean-ups.
We now
graphically summarize
some other
features of the
geographic distribution of the 985 hazardous waste
Sample. There are
a national
NPL
program.
sites in
sites
45 of the 48 continental
The highest proportion of
Belt"), reflecting the historical concentration of
Figure 2 displays the
with complete housing data in the All
states,
demonstrating that Superfund
sites is in the
heavy industry
19
two samples.
Northeast and Midwest
in these regions.
NPL
is
genuinely
(i.e.,
the "Rust
3A
Figures
and 3B present the geographic distribution of the 487
HRS
data in the 1982
The
below 28.5
fewer
sites are in
states.
For example, there are not any below 28.5
we emphasize econometric models
to
be
the
487
HRS
1982
sites in the
statistically indistinguishable
of
distributions
initial
sites
HRS
reflect
EPA
comparable
distribution looks approximately normal, with the
Minnesota,
mitigate the influence of these
state fixed effects.
scores where the bins are 4
Sample. Notably, the
and
To
housing prices that include
in
sites in
across the country pose a problem for
market shocks.
changes
for
Figure 4 presents a histogram of the
among
scores
both categories are spread throughout the United States, but the
identification in the presence of localized housing
shocks,
HRS
sites in
The unequal
and Delaware.
Florida,
with complete house price
Figure 3 A (3B) displays the distribution of sites with 1982
Sample.
exceeding (below) 28.5.
sites
considered
risks to
human
HRS
HRS
points wide,
scores within 4 points
health
(EPA
1991).
modal bin covering the 36.5-40.5 range. Further,
The
there
isn't
obvious bunching just above or below the threshold, which supports the scientific validity of the
HRS
scores and suggests that they weren't manipulated.
16.5 and 40.5.
NPL
This set
is
and constitutes the regression discontinuity sample that
Data from
A. Least Squares Estimation with
Here,
we
housing prices and
(6)
Yc2000
(7)
1
variable
NPL
the
X
c i9so'n
site that
any of the Superfund
vector
X
c
i98o
X
+
c i98
t")
'P
+
exploit in the subsequent analysis.
Methods
is laid
1
out in the following system of equations:
£ C 2000,
c 2000,
log of the median property
l(NPL c2 ooo) equals
hazardous waste
we
the Entire U.S.
This approach
listing.
= © l(NPL c2 ooo) +
is
scores between
discuss a conventional econometric approach to estimating the relationship between
(NPL C 2000) =
where y C2ooo
HRS
centered on the regulatory threshold of 28.5 that determines placement on the
IV. Econometric
for
Importantly, 227 sites have
value in census tract c in 2000.
The
indicator
only for observations from census tracts that contain (or areas near) a
has been placed on the
sites in
column
(1)
NPL
of Table
by 2000. Thus,
1,
this variable takes
not just those that were on the initial
includes determinants of housing prices measured in 1980, which
20
on a value of
may
1
NPL. The
also determine
NPL
status.
e c2 ooo
A
and
r| C 2ooo
are the unobservable determinants of housing prices
few features of the
X
vector are noteworthy.
variables to avoid confounding the effect of
that
X
may be due
C i98o,
NPL
to
to adjust for
status.
Fourth,
in
we
permanent differences
that include a full set
many
we
(e.g.,
housing prices across
in
and the possibility of mean
tracts
However,
wealthy neighbors as an amenity, which seems
which variables belong
in the
the variables available in the
The
X
we emphasize
results
of state fixed effects.
number of bedrooms, school
to identify the bid function.
in these variables
include the 1980 value of the dependent variable, y c8 o in
quality,
exclusion restriction
this
limited to housing and
is
and
other similar variables are generally excluded on the grounds that they are
needed
1980 values of the
changes
status with "post-treatment"
applied hedonic papers, the vector of controls
neighborhood characteristics
status, respectively.
restrict this vector to
Third, to account for local housing market shocks,
reversion in housing prices.
from specifications
Second,
NPL
First,
NPL
and
"demand
is
shifters"
and are
invalid if individuals treat
In the subsequent analysis,
likely.
Income and
air quality)-
we
are agnostic about
vector and report estimates that are adjusted for different combinations of
Census
The Data Appendix
data.
coefficient 9 measures the effect of
NPL
status
1980 mean property values and the other covariates.
appreciation between census tracts with
NPL
sites
determinants of housing prices that covary with
the full set of covariates.
on 2000 property values, after controlling for
Specifically,
and the
NPL
lists
rest
it
tests for differential
of the country.
status, then the estimates
formally, consistent estimation requires E[e C 2ooo r)c20oo]
=
0.
of
housing price
If there are
will
unobserved
be biased. More
Ultimately, this "conventional" approach
places a lot of faith in the assumption that linear adjustment for the limited set of variables available in the
Census removes
B.
A
all
sources of confounding.
Quasi-Experimental Approach based on the 1982
HRS Research Design
This subsection discusses our preferred identification strategy that has two key differences with
the conventional one.
HRS
First,
we
limit the
sample
Sample with complete housing price
EPA judged
to
be
among
data.
to the
census tracts containing the 487
Thus,
all
observations are from tracts with
the nation's most dangerous in 1982.
21
sites in the
If,
for
1982
sites that the
example, the P's differ across
tracts
with and without hazardous waste sites or there are differential trends in housing prices
without these
sites,
then this approach
more
is
likely to
in tracts
produce consistent estimates. Second,
with and
we
use an
instrumental variables (IV) strategy to account for the possibility of endogenous rescoring of sites.
More formally, we replace equation (7) with:
l(NPL c200 o) = X cl980 'n + 5 l(HRS c82 > 28.5) +
(8)
where l(HRS c82 > 28.5) serves
tracts
with a
site that
predicted value of
estimate of
site
IV
.
having a 1982
For
HRS
has a 1982
HRS
score exceeding the 28.5 threshold.
the probability of
NPL
demonstrate that the
We
1
for census
then substitute the
the estimation of equation (8) in the fitting of (6) to obtain an
IV framework,
8rv
is
identified
from the variation
in
NPL
status that
is
due
to a
score exceeding 28.5.
provide a consistent estimate of the
0i V to
This indicator function equals
as an instrumental variable.
l(NPL c2 ooo) from
In this
ti c200 o,
listing
first
HPS
gradient, the instrumental variable
without having a direct effect on housing prices.
condition clearly holds.
The second condition
The next
must
affect
section will
requires that the unobserved
determinants of 2000 housing prices are orthogonal to the portion of the nonlinear function of the 1982
HRS
score that
E[l(HRS c82 >28.5)e c2000 ] =
We
NPL
>
HRS
IV estimator
is
consistent if
also exploit the regression discontinuity design implicit in the 1(«) function that determines
28.5) s c2 ooo]
1982
In the simplest case, the
c i 98 o.
0.
eligibility in three separate
included in
X
not explained by
is
X
^
c i9 8 o
ways
to obtain
over the entire 1982
to partial out
HRS
IV estimates
Sample.
that
In the
allow for the possibility that E[l(HRS c82
first,
a quadratic in the 1982
HRS
score
is
any correlation between residual housing prices and the indicator for a
score exceeding 28.5.
This approach relies on the plausible assumption that residual
determinants of housing price growth do not change discontinuously
at the
regulatory threshold.
The
second regression discontinuity approach involves implementing our IV estimator on the regression
discontinuity sample of 227 sites with 1982
HRS
assumption
"neighborhood" of the regulatory threshold. More formally,
it
is
is
that all else
E[l(HRS c82 >
is
held equal
28.5) e c200 o|16.5
Recall, the
HRS
score
in the
< 1982
is
scores between 16.5 and 40.5.
HRS < 40.5] =
Here, the identifying
0.
a nonlinear function of the ground water, surface water, and air
22
migration pathway scores. The third regression discontinuity method exploits knowledge of this function
by including
pathway scores
the individual
approaches are demanding of the data and
the vector
in
this is reflected in
X
c i9 S o-
All three regression discontinuity
higher sampling errors.
V. Empirical Results
A. Balancing
of Obsei-vable Covariates
This subsection examines the comparisons that underlie the subsequent least squares and quasi-
NPL
experimental IV estimates of the effect of
NPL
whether
status
on housing price growth.
status
We
and the l(HRS cS 2 > 28.5) instrumental variable are orthogonal
Formal
predictors of housing prices.
seems reasonable
impossible, but
it
covariates across
NPL
status or
Elder, and Taber 2000).
presume
1(HRS C S2 >
28.5)
observable
to the
presence of omitted variables bias are of course
tests for the
to
begin by assessing
that research designs
may
suffer
that
balance the observable
from smaller omitted variables bias
Further, if the observables are balanced, consistent inference does not
(Altonji,
depend on
functional form assumptions on the relations between observable covariates and housing prices.
Table 2 shows the association of NPL status and l(HRS c8 ? > 28.5) with potential determinants of
housing price growth measured
headings
Column
in the
(2) displays the
border with a
tracts
985 census
in 1980.
tracts with
means
in the
containing one.
tract
Column
NPL
(5)
and (6) repeat
hazardous waste
41,989 census
Columns
with hazardous waste sites with 1982
Columns
(3)
HRS
means
in pairs
Column
in the least
of the
(7)
compares the means
square approach.
tracts with a
NPL
columns
site.
The
variables listed in the
and complete housing price
tracts that neither contain a
90 and 137
NPL
in the 181
site
row
data.
nor share a
and 306 census
tracts
below and above the regulatory
The remaining columns report p-values from
tests that
are equal. P-values less than 0.01 are denoted in bold.
in
columns
(1)
entries indicate that
Moreover, the
sites
and (4) report on the means
this exercise for the
first six
means of the
scores below and above the 28.5 threshold, respectively.
threshold in the regression discontinuity sample.
the
(1) reports the
tracts
with
and (2)
to explore the possibility
of confounding
1980 housing prices are more than
NPL
sites
20%
lower
in
have lower population densities, lower
household incomes, and mobile homes account for a higher fraction of the housing stock (8.6% versus
23
Overall, the hypothesis of equal
4.7%).
Due
determinants of housing prices.
means can be
to this
confounding of
(8)
and
(9)
compare
all tracts
level for 21
status,
it
may
of the 27 potential
be reasonable to assume
with hazardous wastes that have 1982
and above the 28.5 regulatory threshold and those
is
NPL
1%
produce biased estimates of the effect of NPL
that least squares estimation of equation (6) will
Columns
rejected at the
in the
immediately evident that by narrowing the focus
HRS
status.
scores below
regression discontinuity sample, respectively.
to
It
these tracts, the differences in the potential
determinants of housing prices are greatly mitigated (see especially population density and percentage of
mobile homes). This
means cannot be
means
is
especially so in the regression discontinuity sample
3%
rejected at the
are substantially reduced for
level for
many of the
any of the 27 variables.
where the hypothesis of equal
Notably, the differences in the
variables, so the higher p-values
do not simply
reflect the
smaller samples (and larger sampling errors).
One
variable that remains a potential source of concern
are greatly reduced in the 1982
However, Panel
entirely
B
of Table 4
HRS
in
we
1980 housing prices. The differences
Sample, relative to columns (1) and
(2),
but they are not eliminated.
Greenstone and Gallagher (2005) demonstrates that these differences
disappear after adjustment
Nevertheless,
is
for
the
1980 housing, economic, and demographic variables.
2000 housing prices below.
control for 1980 housing prices in the regressions for
Overall, the entries suggest that the above and
below 28.5 comparison, especially
discontinuity sample, reduces the confounding of
NPL
status.
It is
in the regression
not a panacea, however, so
we
also
adjust for observables.
B. Least
Squares Estimates of the Impact of Clean-ups on Property Values
Table 3 presents the
first
ever large-scale effort to
property value appreciation rates. Specifically,
it
test the effect
of Superfund clean-ups on
reports the regression results
from
fitting
4
least squares
versions of equation (6) for 2000 housing prices. In Panel A, 985 observations are from census tracts that
contain a hazardous waste site that had been on the
18
The 985 NPL
sites are located in
NPL
892 individual census
24
at
any time prior
tracts.
to
2000.
18
The remainder of the
In the regressions, observations
on
tracts that
sample
have
a
is
comprised of the 41,989 observations on the
NPL site
nor are adjacent
to a tract
The remaining Panels use
with a
NPL
circles
around the
Panel
site in the
NPL
A
NPL site.
with a
These
two panels
last
In Panel B, the observations from each tract
sample are replaced with the observations based on the 3-mile radius
C
and
D
are identical to
HRS
limited to the hazardous waste sites in the 1982
2000.
with complete housing price data that neither
slightly different samples.
Panels
sites.
tracts
A
Sample
and B, except
that
are included so that they can be
that the set
were ever on the
compared
of
NPL
NPL
sites is
by January
1,
subsequent quasi-
to the
experimental results.
The
indicator.
entries report the coefficient
and heteroskedastic-consistent standard error on the
All specifications control for the natural log of the
parameter should be interpreted as the growth
rest
of the country. The exact covariates
of the table and are described
The Panel
NPL
listing
results
more
first
row
in tracts
statistically significant
because
it is
large
show
B
all
between 1980 and 2000.
median housing prices grew by 4.0%
NPL.
placed on the
The column
criteria.
We
from
report p-values
that there is a perfectly inelastic supply curve.
(It
NPL
sites
the
bottom
(4) estimate
would
of 6.7%
Specifically, the
is
in In
easily be judged
the most reliable one,
unobserved state-level determinants of housing price growth.
As
this
at the
7.3% (measured
to
All of these estimates
test
assumes
on the
that all
II,
this is
to the
square root of the number of
25
summarize
NPL
sites.
the
indicator are
that the benefits
equivalent to assuming
program benefits occur
sample, the 1980 aggregate value of the housing stock
have a weight equal
sites to
based on the assumption
discussed in Section
also
NPL
tests that the coefficients
Superfund clean-ups pass a cost-benefit
housing market.) In
site, relative to
approach finds a positive association between
explores the growth of housing prices within 3 miles of the
that
NPL
near a
Data Appendix.
in the sites' tracts
are entirely reflected in local housing prices.
include multiple
in areas
price in 1980, so the reported
each specification are noted in the row headings
that this least squares
by conventional
housing prices.
enough so
site
housing prices
in
detail in the
indicate that
with a
adjusted for
Panel
total gain in
more
and housing price increases
estimates in the
points)
A
in
in
mean housing
NPL
is
in the local
$874 million and
the
mean
cost of a clean-up
is
$39 million, so we
whether the change
test
in
housing prices exceeds 4.5%.
All of the estimates are statistically different from zero and imply that the placement of a site on
the
NPL
column
associated with a substantial increase in housing prices within three miles of the
is
(4) specification indicates a precisely estimated gain in prices
ups pass
this cost-benefit test
The own census
D
in
cannot be rejected
tract results in
Panel
C
in
of 10.6%. The null that the clean-
are similar to those in A.
The 3-mile radius
emphasizing
unreliable.
drawing any
definitive
that three features
First,
it
conclusions
of the estimates also
or
policy
implications,
however,
it
NPL
status
to
is
own
confounded by many variables.
worth
is
of the evidence presented so far suggest that the Table 3 estimates
Table 2 demonstrated that
seems reasonable
all
richer
test.
the 8 3-mile radius sample point estimates exceed the
because
circle results
The point estimates from the
specifications are about twice as large as those in B, and, if taken literally,
Before
The
any of the specifications.
indicate large increases in housing prices, like those in Panel B.
indicate that Superfund passes this cost-benefit
site.
may be
Second, 6 of
census tract estimates. This seems suspicious,
expect the impact on housing prices to be greater closer to the
especially in light of their relatively small size (recall, the
median
size is less than 30 acres).
sites,
Third, the
point estimates from the 3 -mile samples are unstable across specifications, so the exact choice of controls
plays a large role in any conclusions.
ranges from 4.6%
to
23.2%.'
For example,
in
Panel D, the implied increase
We now
relationship
turn
to
between 1982
probability that a site
housing prices
9
Quasi-Experimental Estimates of NPL Status on Housing Prices from the 1982
C.
in
the
HRS
preferred
quasi-experimental
scores and
NPL
was placed on
the
NPL
status.
approach
HRS Sample
and begin by assessing the
Figure 5 plots the bivariate relation between the
by 2000 and
its initial
HRS
score
among
the
487
sites in the
also noteworthy that the point estimate on the NPL indicator is quite sensitive to the choice of functional form
two controls: the number of housing units and number of owner occupied units in both Panels B and D. This
likely reflects the fact that the values of these variables differ substantially between the observations on the 3-mile
circles and the census tracts. It also underscores the importance of unverifiable functional form assumptions when
the variables are not balanced across the areas with and without NPL sites.
'
It is
for
26
1982
HRS
from
the estimation of nonparametric regressions that use Cleveland's (1979) tricube weighting function
Sample. The plots are done separately for
and a bandwidth of
O.5.
20
1982
HRS
HRS
score as in Figure 4.
scores.
The
of
NPL
above and below the 28.5 threshold and come
sites
Thus, they represent a moving average of the probability of
The data
points represent the
mean
figure presents dramatic evidence that an initial
status.
HRS
A
score below 28.5.
associated with an
2005).
83%
NPL
is
NPL
HRS
from the census
a border with these tracts. In Panels
tracts containing the
total
housing stock
The
In the fifth
column
sample
in the four
controls in the
first
column, the 1982
HRS
specification details are noted in the
property values in
A
its
at
487 hazardous waste
comprised of the average of
D, the sample
each
site's
score and
as in
sites
its
is
all
score above 28.5
is
in
2000. In Panel A,
sites in the
1982
HRS
variables across tracts that share
comprised of the land area within
circles with
The means of the 1980 values
longitude and latitude.
census
square are added to the column (4) specification.
column
with 1982
row headings
results suggest that a site's
own
with an
samples are $71, $525, $349, and S796 million, respectively.
same
comprised of the 227
The Panel
sites
four columns are identical to those in the four specifications in Table 3.
(6), the controls are the
that is
is
C and
of 2 and 3 miles that are centered
of the
by 2000.
a powerful first-stage relationship
Sample. In Panel B, each observation
radii
NPL
(Greenstone and Gallagher
Table 4 presents IV estimates of the effect of NPL status on housing prices
the observations are
a strong predictor
by 2000 among
of the figure reveals that a
increase in the probability of placement on the
evident that there
It is
statistical version'
is
were placed on the
Virtually all sites with initial scores greater than 28.5
status across
4-unit intervals of the
score above 28.5
Again, rescoring explains the nonzero probability of placement on the
initial
same
probabilities in the
NPL
(4),
HRS
at the
but the sample
the regression discontinuity
scores between 16.5 and 40.5.
bottom of the
placement on the
tract, relative to tracts
is
The sample and
table.
NPL
has
little
impact on the growth of
with sites that narrowly missed placement on the
The smoothed scarterplots are qualitatively similar with a rectangular weighting function
and alternative bandwidths.
27
In
(i.e.,
equal weighting)
NPL. The
point estimates indicate an increase in prices that ranges from
associated t-statistics less than two.
may
be the most credible, so
it is
The regression
0.7%
to
5.6%, but they
all
have
discontinuity specifications in columns (5) and (6)
notable that they produce the smallest point estimates (although they are
also the least precise).
Panel
B
specifications in
presents the
columns
(4)
-
(6)
conventional levels for any of them.
outside the site's
own
census
C
and
D
Panels
the
NPL
Thus, there
summarize the
housing prices.
sample
In the
that the gain in
is little
evidence of meaningful gains
total
gain
3 -mile radius circle
in
housing prices associated with a
samples.
They
The threshold housing
housing prices
in
columns
(4)
all six
-
site's
price gains are
NPL
placement on
12.3% and 5.4%.
designation has
(6) specifications, the point estimates all
little
on
effect
imply price increases
of the 2-mile specifications and the most reliable 3-mile ones, the null
housing prices exceeds the break-even threshold
is
rejected at conventional significance
levels.
Overall, these quasi-experimental estimates suggest that Superfund clean-ups
benefit
test.
that
at
also report whether the clean-ups pass
results provide further evidence that the
smaller than 2%. Further in
reliable
range between -0.6% and 1.5% and zero cannot be rejected
cost-benefit tests analogous to those in Table 3.
circle
most
the
tract.
by using the 2- and
The
The point estimates from
adjacent tract results.
fail to
pass this cost-
This finding further undermines the credibility of the results from the conventional approach
suggested that the benefits substantially exceeded the costs.
Figure 6 provides an opportunity to better understand the source of these regression results.
plots the nonparametric regressions of
covariates) against the 1982
HRS
2000
residual housing prices (after adjustment for the
score in the 2-mile radius sample.
21
column
It
(4)
The nonparametric regression
is
estimated separately below (dark line) and above (light line) the 28.5 threshold. The graph confirms that
there
21
is little
association between
2000 residual housing
prices and the 1982
Figure 6 provides a qualitative graphical exploration of the regression results.
prices and 1982
HRS
is
The
same caveat
applies to Figure
7.
supported by Table 2's finding that the covariates are well balanced
A
HRS
comparison of
between housing
score has not been
However, the meaningfulness of this
among
above and below the regulatory threshold, especially near the regulatory threshold.
28
score.
relationship
scores cannot be exactly inferred from this graph, because the
adjusted for the column (4) covariates. This
graph
HRS
sites
with 1982
HRS
scores
the plots at the regulatory threshold
placement on the
equal
NPL
there.
is
of especial interest
in light
of the large jump
apparent that the moving averages from the
It is
and
left
radius circle sample.
Panel
A
(i.e.,
columns
from the
reports the results
pathway scores
individual
however,
may
it
To
4, 5,
and 6) from Table
to the vector Xciggo
of controls
of which are
isolate the effect of
NPL
D
The
this indicator
tests
whether the
effect
of the
intuition
is
NPL
status is also
NPL
plots of land to build
and the indicator for 2000
NPL
status.
score exceeding 28.5.
be
tract
Of the
70 have
the In
measured as of 1990. Taken
NPL
designation has
is
new
the top quartile
may be
houses.
23
among
statistically different
sites in the
sites
The
greater in higher density areas
Specifically, the specification
is
the interaction of
latter variable is treated as
endogenous and
population density exceeding 4,053 and sites with
from
zero.
It
seems
that the
absence of a price effect
NPL by
is
would be
not due to an
directly estimate supply responses below.
122 tracts with a population density exceeding 4,053, 63 have a
placed on the
tracts in this
All of the estimates of the interaction are negative, and none
abundance of undeveloped plots of land, although we
""
is
designation differs in the 2-mile circles with a
that the price response to clean-ups
instrumented with the interaction of indicators for
to
2000
on land values (rather than housing values),
status
includes an indicator for these tracts and the additional variable of interest, which
judged
the
be appropriate to adjust for these variables. In Panel C, the dependent variable
where there are fewer undeveloped
HRS
and
which adds the
B adds
Panel
in all four specifications.
population density exceeding 4,053 per square mile, which
a
(1) specification
on the growth of property values.
Panel
sample.""
on the 2-mile
4.
together, these specifications provide further support for the Table 4 finding that the
effect
fit
These variables may be affected by the clean-ups so they
of the 1990 (rather than 2000) median house value and
now
right are virtually
third regression discontinuity style approach,
values of the controls as separate covariates.
be endogenous.
all
For conciseness, we only present estimates from the column
most robust specifications
the three
little
probability of
at the threshold.
Table 5 presents the results from a series of specification checks,
may
in the
HRS
score exceeding 28.5 and
2000.
"
Notably, Davis (2004) finds a 13% house price response to the outbreak of a cancer
Nevada, which has a population density of just 5 people per square mile.
29
cluster in Churchill County,
We
conducted a number of other specification checks. These included using the
In
of the mean
median) house price as the dependent variable, using a fixed effects style approach where
(rather than the
2000 and 1980 house prices
the difference between the Ins of
is
the dependent variable (rather than
controlling for 1980 In prices), controlling for the fraction of census tracts within the 2-mile circles with a
boundary change between 1980 and 2000, and adding the 1970 values of the controls (including the
1970 housing prices) as separate covariates
to adjust for
mean
years
NPL
has
little
effect
of
reversion or pre-existing trends in the
subsample where these variable are available. These specification checks
finding that a site's addition to the
In
all
lead to the
same
qualitative
on the growth of nearby housing prices nearly 20
24 25
later.
D. Quasi-Experimental Estimates of Stages of Superfund Clean- Ups on Rental Rates
We now turn
20%
roughly
of
all
to using the In
it
possible to
is
rental rates as the
outcome
variable.
Rental units account for
housing units and generally differ on observable characteristics from owner occupied
homes. This outcome's appeal
so
median
abstract
is
that rental rates are a
measure of the current value of housing
from the problem with the housing price outcome
expectations about time until the completion of the clean-up are unknown.
Further,
we
that
can
services,
individuals'
test
whether
the impact on the value of local housing services varies at different stages of Superfund clean-ups.
Table 6 presents separate estimates of the effect of the different stages of the remediation process
on the
In
median
rental rate.
observations per county.
We
stack equations for 1990 and
The 1980 housing
2000
In rental rates, so there are
two
characteristics variables are calculated across rental units,
24
The own census tract sample regression results for some of these specification checks are presented in Greenstone
and Gallagher (2005). That version of the paper also reports on a test of whether there was greater housing price
appreciation near sites where the groundwater was heavily contaminated and residents use well water for drinking.
We
assumed
that clean-ups
would be highly valued
evidence of differential house price appreciation
in these areas,
however
this test also failed to find significant
Greenstone and Gallagher 2005). Additionally,
we would have liked to test whether the effects of clean-ups differed for large sites or ones where the estimated costs
of clean-up are high (so called "mega" sites) but the size and estimated cost data are only available for NPL sites.
in these areas (see
RODs ultimately receive a "no further action"
remove the environmental risk and/or the risk naturally
Sample where all RODs received the "no further action"
them. The regression results are virtually identical to those
Probst and Konisky (2001) find that approximately 14.5% of
classification
dissipated.
when removal
activities
There are eleven
classification so
sites in the
no remediation
presented in Table 4
when
were
sufficient to
1982
HRS
activities took place at
the observations from near these sites are dropped.
30
rather than across
owner occupied
the controls listed in the
The
are equal to
a
ROD
row headings
for sites that
prices.
The
effects of
by 1990 or 2000 were: placed on
The instruments
the
NPL
but no
ROD had been
are the interactions of the indicator for a 1982
The
these three independent indicators.
allow for clustering
are equal.
at the site level,
The number of sites
There
is
some evidence
speed through which a
site
and
table reports the point estimates
issued; issued
HRS
NPL,
score above 28.5 and
their standard errors,
that higher voter turnout
moves through
HRS
score
is
is
which
also listed in brackets.
and per capita income are associated with the
the clean-up process
and the stringency of clean-ups (Gupta
1995 and 1996; Viscusi and Hamilton 1999; Sigman 2001). Consequently, the two stage
strategy
of
along with the p-value from an F-test that the three point estimates
each category and the mean
in
all
by three independent indicator variables. They
status is replaced
but were not completely remediated; and "construction complete" or deleted from the
respectively.
al.,
above analysis of housing
are allowed to differ in 1990 and 2000.
NPL
indicator variable for
1
units as in the
unlikely to purge these sources of endogeneity, so
it
et
least squares
appropriate to consider these three
is
parameter estimates associational or descriptive.
There are a few important findings.
NPL
for either 7 or 17 years, but the
estimates from the
NPL
revise
reliable specifications suggest that there is
little
effect
on
The
rental rates near these
site's
placement on
leads to an immediate reduction in the value of housing services near the site as nearby residents
upwards
their expectation
of the risk they face from the
negative, and zero cannot be rejected for any of them.
26
category have been on the
has not developed a remediation plan for them yet.
specifications, the point estimates for the "Construction
been
"NPL Only"
This finding undermines a key feature of the popular "stigma" hypothesis that a
sites.
the
more
EPA
First, sites in the
fully
remediated and yet there
is little
effect
26
site.
Complete or
This finding
on rental
Second,
NPL
in the
more
reliable
Deletion" category are
is telling,
because these
sites
all
have
rates.
states that a site's placement on the NPL causes nearby residents to revise their expectation
upwards permanently so that the value of nearby housing services declines relative to before its
listing on the NPL, even after remediation is completed. Harris (1999) reviews the stigma literature and McCluskey
and Rausser (2003) and Messer, Schulze, Hackett, Cameron, and McClelland (2004) are case studies that present
evidence that prices decline immediately after the announcement that a local site has been placed on the NPL.
The stigma hypothesis
of a
site's health risk
31
any of the
Third, the null that the three parameter estimates are equal cannot be rejected in
specifications.
This result demonstrates that the approximately zero effect on housing prices
the averaging of a positive effect at fully remediated sites and a negative effect at sites
is
incomplete or hasn't been
that
initiated.
is
not due to
where remediation
Overall, these results are consistent with the housing price findings
Superfund clean-ups have small effects on the value of local housing services.
E. Quasi-Experimental Estimates
If
2000
demand
Demand Shifters
consumers value Superfund clean-ups, then the clean-ups should lead
the population near
quality.
of the Effect ofNPL Status on
Table 7
shifters
respectively.
tests
NPL
sites is
comprised of individuals
whether there were changes
measured by demographic
The
in the
that place a higher value
income and wealth
characteristic of residents,
entries report the parameter estimate
the
NPL
would be more willing
Panel
C
NPL
We
is little
effect
on
dummy
for
NPL
NPL
sites,
status
had hypothesized that families with
designation leads to changes
outcomes. Finally, the
suggests that there
population near
details.
to live in these areas after the clean-ups.
provide any meaningful evidence that the
shifter (and racial justice)
total
on environmental
designation on the measures of income and wealth are
inconsistent across specifications and imprecisely estimated.
children
and
by
education) of residents,
(i.e.,
and standard error on the
from four specifications. The row headings provide specification
The estimated impacts of
to migration, so that
However, Panel
in the
B
fails to
demographic demand
of the point estimates across specifications
instability
in
total population.
Notably, this table's qualitative findings are unchanged by the inclusion of 1980 housing prices
and housing characteristics as covariates.
Overall, there
associated with changes in variables that proxy for shifts in
F.
is
little
demand
evidence that the
for
NPL
designation
is
environmental quality.
Quasi-Experimental Estimates of the Effect of NPL Status on Supply
An
that
increase in the supply of housing units in the vicinity of a
Superfund clean-ups increase the value of the surrounding land.
NPL
In
site
Table
would provide evidence
8,
we
test this possibility
with the 2- and 3-mile radius samples, using the same four specifications from Tables 5 and
32
6.
These
results are also inconsistent across specifications.
of the
NPL
designation has
effect
little
The most reasonable conclusion
on the supply of housing.
regressions of 2000 residual housing units after adjustment for the
radius sample.
that
There
is little
exceed 28.5, especially
assignment
that the
Figure 7 plots nonparametric
column
(2) covariates
evidence of an increase in housing units near
at the
is
sites
with
from the 2 mile
HRS
initial
scores
regulatory threshold.
VI. Interpretation and Policy Implications
This paper has shown that across a wide range of housing market outcomes, there
evidence that Superfund clean-ups increase social welfare substantially.
is
little
In light of the significant
resources devoted to these clean-ups and the claims of large health benefits, this finding
is
surprising.
This section reviews three possible explanations.
First, the individuals that
low willingness
individuals a
to
good
pay
choose to
live
avoid exposure to hazardous waste
to
that they don't value highly.
of the population with a high
that the population average
WTP
WTP
It is
sites.
substantial.
the population that lives near these sites,
and
a
In this case, society provides these
possible (and perhaps likely) that there are segments
to avoid exposure to hazardous waste sites.
is
may have
near these sites before and after the clean-ups
It
may even be
However, the policy relevant parameter
this is the
is
the
parameter that the paper has estimated.
the case
WTP
of
27
Second, scientific has not found decisive evidence of substantial health benefits from the cleanups of hazardous waste
consumers believe
sites
(Vrijheid 2000; Currie, Greenstone and Moretti 2006).
that the reductions in risk are small
Consequently,
and rationally place a low value on them.
Of
course, the discovery of large health improvements in the future could cause consumers to increase their
valuations of the clean-ups.
2
Third, the sites with initial
27
A
popular theory
is
that sites
rental rate results in Section
18
Another possibility
HRS
scores less than 28.5 also received complete remediations under
become permanently stigmatized when they
are placed
on the NPL.
Recall, the
V D are inconsistent with this theory and would lead to a rejection of this hypothesis.
consumers are imperfectly informed about the location of Superfund sites and their
media often devote extensive coverage to local Superfund sites
and their clean-ups. Further, at least a few states (e.g., Alaska and Arizona) require home sellers to disclose whether
there are hazardous waste sites in close proximity.
clean-ups.
is
We think this
that
is
unlikely, because local
33
s
programs. In
state or local land reclamation
and below 28.5
would have received
sites
this case, a
same
the
zero result
treatment.
is
to
We
conducting an extensive search for information on remediation activities
From
these investigations,
ambitious and costly
at sites
with
we concluded
initial
Further,
among
the
was roughly S3
40%
of the
This
million.
Superfund clean-up.
sites
is
investigated this possibility by
at
these sites."
that the clean-up activities
scores exceeding 28.5.
evidence of any remediation activities by 2000
be expected since both the above
at
roughly
60%
were dramatically more
we were unable
For example,
of the
where there was evidence of clean-up
sites
to find
with scores below 28.5.
average expenditure
efforts, the
about S40 million less than our estimate of the average cost of a
This difference
is
not surprising, because the state and local clean-ups were often
limited to restricting access to the site or containing the toxics, rather than trying to achieve Superfund'
goal of returning the
so
it
may be
Nevertheless,
site to its "natural state."
some remediation took
place at these
sites,
appropriate to interpret the results as the impact of the extra $40 million that a Superfund
clean-up costs.
In
our view, the most likely explanations are that the people that choose to live near these
don't value the clean-ups or that consumers have
reduce health
risks.
In either case the results
residents' gain in welfare
less
little
mean
from Superfund clean-ups
sites
reason to believe that the clean-ups substantially
that given the current state
falls
well short of the costs.
of knowledge, local
The implication
ambitious operations like the erection offences, posting of warning signs around the
sites,
is
that
and simple
containment of toxics might be a more efficient use of resources.
This paper has provided an important piece of what would constitute a
Superfund's benefits.
It is
full
accounting of
possible that there are other benefits of these clean-ups that are not captured in
the local housing market, including health and aesthetic benefits to individuals that do not live in close
proximity to Superfund
19
Specifically,
we
filed
sites,
reductions in injuries to ecological systems, and protection of ground water.
freedom of information act requests with the
followed any leads from these documents.
We
EPA
personnel and state and local environmental
these searches, there
turn
up
is
for information
no centralized database about these
officials.
sites
different information.
34
web
site
on these
and the
sites
sites
and
of state
Additionally, we contacted national and
Although we expended considerable effort in
so we cannot be certain that further efforts wouldn't
departments of environmental quality and used internet search engines.
regional
EPA
also searched the Superfund
Further, the discovery of compelling evidence of health benefits might cause local residents to increase
their valuations,
and
this
would presumably be
reflected in the housing market.
VII. Conclusions
This study has used the housing market to develop estimates of the local welfare impacts of
Superfund sponsored clean-ups of hazardous waste
sites.
housing market outcomes in the areas surrounding the
Superfund clean-ups
clean-ups.
We
to the areas
surrounding the 290
The
first
basis of the analysis
is
400 hazardous waste
sites that
a
comparison of
sites
chosen for
narrowly missed qualifying for these
find that Superfund clean-ups are associated with economically small and statistically
indistinguishable from zero local changes in residential property values, property rental rates, housing
supply, total population, and the types of individuals living near the
series
sites.
These findings are robust
to a
of specification checks, including the application of a quasi -experimental regression discontinuity
design based on knowledge of the selection
rule.
Overall, the preferred estimates suggest that the local
benefits of Superfund clean-ups are small and appear to be substantially lower than the $43 million
mean
cost of Superfund clean-ups.
More
broadly, this paper
makes two
experiment that improves a local amenity
findings are that
new home
if
contributions.
in the
value on the amenity will
it
models the consequences of a quasi-
The key
context of the hedonic model.
consumers value the amenity, then there
construction.
First,
theoretical
will be increases in local housing prices
and
Further, there will be taste-based sorting such that individuals that place a high
move
to areas
where they can consume
it.
Second,
body of research (Black 1999; Chay and Greenstone 2005) demonstrating
it
that
contributes to a growing
it
is
possible to identify
research designs that mitigate the confounding that has historically undermined the credibility of
conventional hedonic approaches to valuing non-market goods.
Perhaps most importantly,
this
paper has demonstrated that the combination of quasi-experiments
and hedonic theory are a powerful method
to use
markets to value environmental and other non-market
goods.
35
DATA APPENDIX
This data appendix provides information on a number of aspects of the data set that
conduct the analysis for
this paper.
Due
space constraints,
to
this is
we compiled
to
an abridged version of the data
appendix that is available in Greenstone and Gallagher (2005). The longer data appendix includes details
on the variables on: the size of the hazardous waste sites; whether a site has achieved the construction
complete designation; and the determination of expected and actual remediation costs. It also includes a
discussion of how we placed hazardous waste sites in 2000 Census tracts and a lengthier discussion on the
1982 HRS sample.
I.
Covariates in Housing Price and Rental Rate Regressions
The following are the control variables used in the housing price and rental rate regressions. They are
listed by the categories indicated in the row headings at the bottom of these tables. All of the variables
measured in 1980 and are measured at the census tract level (or are the mean across
tracts, for example tracts that share a border with a tract containing a hazardous waste site).
are
1980 Ln House Price
mean value of owner occupied housing units
In
in
1980 (note: the median
1980 Housing Characteristics
housing units (rental and owner occupied)
of total housing units (rental and owner occupied)
is
sets
of census
unavailable in 1980)
total
%
total
that are
occupied
housing units owner occupied
% of owner occupied housing units with bedrooms
% of owner occupied housing units with bedroom
% of owner occupied housing units with 2 bedrooms
% of owner occupied housing units with 3 bedrooms
% of owner occupied housing units with 4 bedrooms
% of owner occupied housing units with 5 or more bedrooms
% of owner occupied housing units that are detached
% of owner occupied housing units that are attached
% of owner occupied housing units that are mobile homes
% of owner occupied housing units built within last year
% of owner occupied housing units built 2 to 5 years ago
% of owner occupied housing units built 6 to 10 years ago
% of owner occupied housing units built 10 to 20 years ago
% of owner occupied housing units built 20 to 30 years ago
% of owner occupied housing units built 30 to 40 years ago
% of owner occupied housing units built more than 40 years ago
% of housing units without a full kitchen
% of housing units that have no heating or rely on a
stove, or portable heater
% of housing units without air conditioning
% of housing units without a full bathroom
1
all
all
fire,
all
all
Note: In the rental regressions
versions of the variables.
housing units with
in
Table
6, the
For example, the
bedrooms."
owner occupied
first
variable
1980 Economic Conditions
36
is
variables are replaced with renter occupied
replaced with the
"%
of renter occupied
mean household income
% of households with income below poverty line
unemployment
rate
% of households that receive some form of public assistance
1980 Demographics
population density
% of population Black
% of population Hispanic
% of population under age 18
% of population 65 or older
% of population foreign bora
% of households headed by females
% of households residing in same house as 5 years ago
% of individuals aged 16-19 that are high school drop outs
% of population over 25 that failed to complete high school
at least
% of population over 25 that have a BA or better
(i.e.,
16 years of education)
Assignment of HRS Scores and their Role in the Determination of the NPL
30
The
The HRS test scores each pathway from to 100, where higher scores indicate greater risk.
individual pathway scores are calculated using a method that considers characteristics of the site as being
II.
included in one of three categories: waste characteristics, likelihood of release, and target characteristics.
The
final
pathway score
is
a multiplicative function of the scores in these three categories.
for example, that if twice as
many people
are thought to be affected via a
pathway then
the
The logic is,
pathway score
should be twice as large.
The
(1)
HRS
final
=
Score
HRS
[(S
2
score
^
is
calculated using the following equation:
+ S 2 SW + S 2 a )
v
/
3]
\
where Sgw, S sw and S a denote the ground water migration, surface water migration, and air migration
31
pathway scores, respectively.
As equation (1) indicates, the final score is the square root of the average
of the squared individual pathway scores. It is evident that the effect of an individual pathway on the total
HRS score is proportional to the pathway score.
It is important to note that HRS scores can't be interpreted as strict cardinal measures of risk.
A
number of EPA studies have tested how well the HRS represents the underlying risk levels based on
32
cancer and non-cancer risks.
The EPA has concluded that the HRS test (at least from the late 1980s
version) is an ordinal test but sites with scores within 4 points of each pose roughly comparable risks to
,
human
health
,
(EPA
1991).
33
From 1982-1995, the EPA assigned all hazardous waste sites with a HRS score of 28.5 or greater
NPL. Additionally, the original legislation gave every state the right to place one site on the NPL
without the site having to score at or above 28.5 on the HRS test. As of 2003, 38 states have used their
to the
,0
The capping of individual pathways and of attributes within each pathway is one limiting characteristic of the test.
There is a maximum value for most scores within each pathway category. Also, if the final pathway score is greater
than 100 then this score is reduced to !00. The capping of individual pathways creates a loss of precision of the test
since all pathway scores of 100 have the same effect on the final HRS score but may represent different magnitudes
of risk. See the EPA's Hazard Ranking System Guidance Manual for further details on the determination of the
HRS
31
32
i3
score.
In 1990, the
EPA
The EPA
states that the
but that "the
HRS
HRS test so that also considers soil as an additional pathway.
EPA studies that have examined this issue.
early 1980s version of the HRS test should not be viewed as a measure
revised the
See Brody (1998) for a
list
it
of
does distinguish relative risks
among
sites
and does identify
significant risk to public health, welfare, or the environment" (Federal Register 1984).
37
sites that
of "absolute
appear
risk",
to present a
these sites would have received a HRS score above 28.5. Six of these
were included on the original NPL released in 1983, but due to their missing HRS
scores these six sites are excluded from this paper's analysis.
In 995 the criteria for placement on the NPL were altered so that a site must have a HRS score
greater than 28.5 and the governor of the state in which the site is located must approve the placement.
There are currently a number of potential NPL sites with HRS scores greater than 28.5 that have not been
proposed for NPL placement due to known state political opposition. We do not know the precise
exception.
It is
unknown whether
"state priority sites"
1
number of these sites because our Freedom of Information Act request
was denied by the EPA.
Primary Samples of Hazardous Waste Sites
The paper relies on two primary samples of hazardous waste
Sample" and the "1982 HRS Sample."
for information about these sites
III.
sites,
which we
label the "All
NPL
A. All NPL
Sample
The All NPL sample includes NPL
sites located in the
50
US
states
and the District of Columbia
Although there are NPL sites located in US
sample because the census data from these
areas differs from the data for the remainder of the country. Further, the sample is limited to sites that
were listed on the NPL before January 1, 2000 to ensure that site listing occurred before any data
collection for the 2000 census. There are 1,398 sites in this sample.
that
were placed on the
territories
B.
1982
on the
HRS
NPL
before January
1,
2000.
such as Puerto Rico, they are not included
HRS Sample
The second sample
initial
consists of the
690
in the
sites that
were tested between 1980 and 1982
National Priority List announced on September
scored exceeding 28.5 were placed on the
extensive discussion about
some of the
details
8,
for inclusion
1983. In this sample, sites that received a
NPL. See Greenstone and Gallagher (2005)
first NPL and this sample, more
for a
surrounding the
more
generally.
IV. Matching of 2000 Census Tracts to 1980 and 1990 Censuses
The census
is
tract is
used as the unit of analysis, because
available in the 1980, 1990 and 2000
US
Census.
boundaries are fixed so that the size and location of the census
The
census data.
Information on
tracts using
is
the smallest aggregation of data that
year 2000 census
in the text,
tract is the
same
for the
the 1980 and 1990 census tracts
at:
www.geolytics.com
outline of their approach
is
as follows.
block level data. Their documentation
were adjusted
to
fit
the
2000 census
tract
1980 and 1990
fixed census tract data boundaries were provided by Geolytics, a private
how
can be found on their website
An
it
As noted
tract
company.
boundaries
.
mapped 1990 census tracts into 2000 census
"The basic methodology was to use the smaller
Geolytics
states,
blocks to determine the population-weighted proportion of a 1990
tract that
was
later redefined as part
of
34
A 1990 street coverage file was used to weight populations of 1990 blocks included in
a 2000 tract."
2000 census tracts when the 1990 blocks were split among multiple census tracts. The assumption is that
local streets and roads served as a proxy for where populations were located. Block level data for 1980
were unavailable.
This complicated the mapping of 1980 tracts into 1990 tracts.
However, the
correspondence between 1980 tracts and 1990 blocks is "very good." As such "splitting a 1980 tract into
1990 tracts had to be done spatially, meaning based solely on the 1990 block to 1980 tract
correspondence."
35
34
Appendix J: Description of Tract Remapping Methodology of Geolytics Data Users' Guide
Change Database (1970-2000), page J3.
35
Ibid,
page
J4.
38
for
Neighborhood
V. Neighbor Samples
We
use two approaches to define the set of houses outside each site's tract that
We
by the clean-up.
The
contains the
first
site.
approach defines the neighbors as
GIS software was
adjacent neighbors.
and the median
may be
affected
refer to these sets of houses as "neighbors."
In the
is 7.
1982
all
census tracts that share a border with the tract that
used to find each primary census tract
HRS
maximum number
sample, the
The population of each adjacent census
housing characteristics, and demographic variables for each
tract
tract
and extract the identity of
of neighboring census tracts
was used
when
to
is
its
21
weight the housing price,
calculating the
mean
adjacent
neighbor values.
of the
The second approach defines neighbors based on circles of varying radii around the exact location
GIS software is used to draw a circle around the point representing the site (generally the
site.
center of the
site,
but sometimes the point associated with the street address).
sample, the GIS program draws circles with radii of
from
used
all
census tracts that
to calculate the
variables.
To
product of
its
fall
within
its
mean housing
1
mile around each of the
For example
sites.
values, housing and demographic characteristics, and
population and the portion of
tracts
included in the
For the 2 (3) mile ring the
and 8 (18.2 and 12).
1
1
site,
mile
data
1-mile radius circle (including the tract containing the site) are
calculate these weighted means, each census tract within the circle
number of census
in the
For a given
its
total area that falls
mile ring for a
maximum number
site is
of neighbor
39
is
within the circle.
economic
weighted by the
The maximum
37 and the mean and median are 3.9 and
sites is
3.
80 (163), with a mean and median of 9.9
REFERENCES
Todd
Altonji, Joseph G.,
and Christopher R. Taber, "Selection on Observed and Unobserved
NBER Working Paper No. 7831, 2000.
E. Elder,
Variables: Assessing the Effectiveness of Catholic Schools,"
Ashenfelter, Orley and Michael Greenstone, "Estimating the Value of a Statistical Life:
of Omitted Variables and Publication Bias," American Economic Review,
Banzhaf, H. Spencer and Randall P. Walsh,
"Do People Vote with Their
XCIV (May
Feet:
An
The Importance
2004), 454-60.
Empirical Test of
Environmental Gentrification," Mimeograph, 2005.
Timothy J. "The Estimation of Demand Parameters
Political Economy, 95 (1987), 81-88.
Bartik,
in
Hedonic Price Models." Journal of
Timothy J. "Measuring the Benefits of Amenity Improvements
Economics, 64 (1988), 172-83.
Bartik,
Bayer, Patrick, Nathaniel Keohane, and Christopher Timmins.
in
Hedonic Price Models." Land
"Migration and Hedonic Valuation: The
Case of Air Quality." Mimeograph, February 2006.
Black,
Dan
A. and
Thomas
J.
Black, Sandra,
"Do
"On the Measurement of
XXVII (Dec. 2003), 205-20.
Rneisner,
Journal of Risk and Uncertainty,
Better Schools Matter?
Journal of Economics,
CXIV (May
Brody, Thomas M., "Investing
Management of Risk,"
in
Job Risk in Hedonic
Wage Models,"
Parental Valuation of Elementary Education," Quarterly
1999), 577-99.
Cleanup:
A New
Look
at
Decision Criteria for the Comparison and
Dissertation, Northwestern University, 1998.
Cameron, Trudy Ann and Ian T. McConnaha, "Evidence of Environmental Migration," Mimeograph,
2005.
Champ,
Kevin J. Boyle and Thomas C. Brown, eds., A Primer on Nonmarket
The Netherlands: Kluwer Academic Publishers, 2003).
Patricia A.,
(Dordrecht,
Chay, Kenneth Y., and Michael Greenstone, "Does Air Quality Matter?
Market," Journal of Political Economy, CXIII (Apr. 2005), 376-424.
Cleveland, William
the
American
Cook, Thomas
S.,
Evidence from the Housing
"Robust Locally Weighted Regression and Smoothing Scatterplots," Journal of
Statistical Association,
D.,
Valuation,
LXXIV
(Dec. 1979), 829-36.
and Donald T. Campbell, Quasi-Experimentation: Design and Analysis Issues for
Field Settings, (Boston,
MA: Houghton
Mifflin, 1979).
Cropper, Maureen
for
L., Leland B. Deck, and Kenneth E. McConnell, "On the Choice of Functional Form
Hedonic Price Functions," Review of Economics and Statistics, LXX (Nov. 1988), 668-75.
Currie, Janet, Michael Greenstone, and Enrico Moretti, "Are Hazardous
Waste
Sites
Hazardous
to
Human
Health? Evidence from Superfund Clean-Ups and Infant Health," Mimeograph, 2006.
Davis, Lucas, "The Effect of Health Risk on Housing Values: Evidence from a Cancer Cluster,"
40
American Economics Review,
XCIV
(Dec. 2004), 1693-1704.
Deschenes, Olivier and Michael Greenstone, "The Economic Impacts of Climate Change: Evidence from
Agricultural Profits and Random Fluctuations in Weather," American Economic Review,
(forthcoming) 2006.
Diamond, Peter A. and Jerry A. Hausman, "Contingent Valuation: Is Some Number Better Than No
Number?" Journal oj'Economic Perspectives, VIII (Autumn 1994), 45-64.
Ekeland, Ivar, James
J.
Heckman, and Lars Nesheim,
"Identification and Estimation of
Hedonic Models,"
Journal of Political Economy, CXII (Feb. 2004), S60-S109.
Environmental Protection Agency, "National
Federal Register 5598 - 5605, Vol
Environmental
Protection
"
Environment,
Agency,
Priorities List for
No
28;
Feb
Superfund:
20
56,
Uncontrolled Hazardous Waste Sites,"
11, 1991.
Years
of Protecting
Environmental Protection Agency, 2006
Differentiated Products," Journal of Political
"A
Health
and
the
SUPERFUND BENEFITS ANALYSIS
Epple, Dennis, "Hedonic Prices and Implicit Markets: Estimating
Farrell, Alex,
Human
http://vvwvv.epa.gov/superfund/action/20years/index.htm ," 2000.
Partial Benefit-Cost Test
Demand and Supply
Functions for
Economy, 95 (1987), 59-80.
of NPL Site Remediation: Benefit Transfer with Met-Analytic
Hedonic Data," Mimeograph, University of California, Berkeley, 2004.
Federal Register Rules and Regulations, "National Oil and Hazardous Substance Contingency Plan; The
National Priorities List;
Amendment," Various
Issues.
Federal Register Rules and Regulations, "Environmental Protection Agency. 40
Oil and Hazardous Substances Contingency Plan
[SWH-FRL
2163-4] 47
FR
CFR
Part 300 National
31 180," (July 16, 1982).
40 CFR Part 300 [SWHand Hazardous Substance Contingency Plan; National
Federal Register Rules and Regulations, "Environmental Protection Agency.
FRL-2646-2] Amendment
Priorities List," Vol. 49,
A
to National Oil
No. 185 (September 21, 1984).
"On Estimating Air Pollution Control Benefits from Land Value
of Environmental Economics and Management, I (May 1974), 74-83.
Freeman,
—
,
Myrick,
III,
Studies, Journal
The Measurement of Environmental and Resource Values, (Washington, DC: Resources for the
Future, 2003).
W. Kip Viscusi, "Private Values of Risk Tradeoffs at Superfund
Housing Market Evidence on Learning about Risk," The Review of Economics and Statistics,
Gayer, Ted, James T. Hamilton, and
Sites:
LXXXI1
—
,
(Aug. 2000), 439-51.
"The Market Value of Reducing Cancer Risk: Hedonic Housing Prices with Changing Information,"
Southern Economic Journal,
Glaeser,
Edward
L.
Affordability."
LXIX
(Oct. 2002), 266-89.
and Joseph Gyourko. "The Impacts of Building Restriction on Housing
Federal Reserve Board of New York. Economic Policy Review June 2003. pp. 2141
39.
Greenstone, Michael and Justin Gallagher, "Does Hazardous Waste Matter? Evidence from the
Housing Market and the Superfund Program," NBER Working Paper No. 1 1790, 2005.
Gupta, Shreekant, George
Van Houtven, and Maureen
Environmental Regulation?
An
Analysis of
EPA
Cropper,
"Do
Benefits and Costs Matter in
Decisions under Superfund," Analyzing Superfund:
Economics, Science, Law, Richard L. Revesz and Richard B. Stewart
(Washington, DC:
eds.,
Resources for the Future, 1995).
—
,
"Paying for Permanence:
An Economic
RAND Journal of Economics, XXVII
Analysis of EPA's Cleanup Decisions
(Autumn
Halvorsen, Robert and Henry O. Pollakowski, "Choice of Functional
Journal of Urban Economics,
X
at
Superfund
Sites,"
1996), 563-82.
Form
for
Hedonic Price Equations,"
(Jul. 1981), 37-49.
Hanemann, W. Michael, "Valuing the Environment through Contingent Valuation," Journal of Economic
Perspectives, VIII (Autumn 1994), 19-43.
Harris,
"
John
D.,
"Property
Values,
Stigma,
and
Superfund,"
Environmental
Protection
Agency,
http://www.epa.gov/superfund/programs/recycle/propertv.htm ," 1999.
Ihlanfeldt, Keith R.
and Laura O. Taylor, "Externality Effects of Small-Scale Hazardous Waste
Evidence from Urban Commercial Property Markets," XLVII (January 2004),
1
Sites:
17-39.
"Measuring the Impact of the Discovery and Cleaning of Identified Hazardous Waste
on House Values," Land Economics, LXXI (Nov. 1995), 428-35.
Kiel, Katherine A.,
Sites
Economic Benefits of Cleaning Up Superfund
The Case of Wobum, Massachusetts," The Journal of Real Estate Finance & Economics, XXII
Kiel, Katherine A. and Jeffery Zabel, "Estimating the
Sites:
(Mar. 2001), 163-84.
Kohlhase, Janet
XXX (Jul.
E.,
"The Impact of Toxic Waste
Sites
on Housing Values," Journal of Urban Economics,
1991), 1-26.
Linden, Leigh L. and Jonah E. Rockoff, "There Goes the Neighborhood? Estimates of the Impact of
Crime Risk on Property Values from Megan's Law,"
McCluskey,
Jill J.
NBER Working Paper No.
12253, 2006.
and Gordon C. Rausser, "Hazardous Waste Sites and Housing Appreciation Rates,"
Journal of Environmental Economics
and Management,
XLV
(Jul.
2003), 166-176.
Messer, Kent, William Schulze, Katherine Hackett, Trudy Cameron, and Gary McClelland, "Stigma: The
Psychology and Economics of Superfund," Working Paper, 2004.
Michaels, R. Gregory and V. Kerry Smith, "Market Segmentation and Valuing Amenities with Hedonic
Models: The Case of Hazardous Waste
Sites,"
Journal of Urban Economics,
XX VIII
(Sep. 1990),
223-42.
Probst, Katherine N. and
David M. Konisky, Superfund 's Future: What Will
Resources for the Future, 2001).
42
It
Cost? (Washington, DC:
Ridker, Ronald, The
—
Economic Cost of Air Pollution, (New York, NY: Prager,
1967).
A. Hcnning, "The Determinants of Residential Property Values with Special Reference
Pollution," Review of Economics and Statistics, XLIX (May 1967), 246-57.
and
J.
Roback, Jennifer, "Wages, Rents, and the Quality of Life," Journal of Political Economy,
XC
to
Air
(Dec.
1982), 1257-78.
Rosen, Sherwin, "Hedonic Prices and Implicit Markets: Product Differentiation
Journal of Political Economy,
—
,
LXXXII
in
Pure Competition,"
(Jan. 1974), 34-55.
"The Theory of Equalizing Differences," Handbook of Labor Economics, Orley Ashenfelter and
eds., (Amsterdam: North-Holland, 1986).
Richard Layard,
Schmalensee, Richard, Ramachandra Ramanathan, Wolfhard
Ramm, and Dennis Smallwood, Measuring
External Effects of Solid Waste Management, (Washington DC: Environmental Protection Agency,
1975).
Sigman, Hilary, "The Pace of Progress
Journal of Law and Economics,
at
XLIV
Superfund
Sites: Policy
Small, Kenneth A., "Air Pollution and Property Values:
Statistics,
Goals and Interest Group Influence,"
(Apr. 2001), 315-44.
A
Further
Comment," Review of Economics and
LVII (Feb. 1975), 105-7.
W. Kip and James T. Hamilton, "Are Risk Regulators Rational? Evidence from Hazardous
Waste Cleanup Decisions," American Economic Review, LXXXIX (Sep. 1999), 1010-27.
Viscusi,
Vrijheid, Martine, "Health Effects of Residence
Near Hazardous Waste Landfill
Sites:
A Review
of
Epidemiologic Literature," Environmental Health Perspectives Supplements 108 No. SI (March
2000: 101-12.
Zimmerman, Rae,
"Social Equality and Environmental Risk," Risk Analysis, XIII (Dec. 1993), 649-66.
43
Table
1
:
Summary
Statistics
on the Superfund Program
All
NPL
w/
Sites
1982
non-Missing House
HRS
w/
Sites
House
Price Data
1982
HRS
Sites
Price Data
Data
Price
(1)
(2)
(3)
985
487
189
306
95
332
312
111
#1981-1985
985
406
# 1986-1989
340
14
9
# 1990-1994
166
4
3
# 1995-1999
73
2
2
44.47
43.23
15.54
16.50
310
334(25)
42,560
10,507(35)
405,760
Number of Sites
1
982
HRS
Score Above 28.5
A. Timing of Placement on
Total
B.
Mean
Mean
Scores
|
Scores
|
HRS > 28.5
HRS < 28.5
HRS
w/
Missing House
non-Missing
NPL
97
Information
41.89
C. Size of Site (in acres)
Number of sites
Mean (Median)
920
with size data
1,186(29)
Maximum
195,200
D. Stages of Clean-Up for
Median Years from
NPL
NPL
97
Sites
Listing Until:
ROD Issued
4.3
4.3
Clean-Up
5.8
6.8
12.1
11.5
12.8
12.5
Initiated
Construction Complete
NPL
Among
Deleted from
1990 Status
NPL Only
Sites
NPL
by 1990
ROD Issued or Clean-up Initiated
Construction Complete or Deleted
2000 Status
NPL Only
Among
Sites
NPL
394
335
22
100
31
210
68
16
7
137
15
by 2000
3
ROD Issued or Clean-up Initiated
370
119
Construction Complete or Deleted
478
198
E. Expected Costs of Remediation (Millions of 2000 S's)
753
293
# Sites with Nonmissing Costs
Mean (Median)
$28.3 ($11.0)
$27.5 ($15.0)
95
th
$89.6
Percentile
F.
Sites
75
95
$29.6 ($11.5)
$146.0
$95.3
Actual and Expected Costs Conditional on Construction Complete (Millions of 2000 S's)
w/ Both Costs Nonmissing
Mean (Median) Expected Costs
Mean (Median) Actual Costs
The estimated
cost information
operating unit associated with a
site.
477
203
69
$15.5 ($7.8)
$20.6 ($9.7)
$32.0 ($16.2)
$17.3 ($7.3)
$21.6 ($11.6)
Notes: All dollar figures are in 2000 S's.
12/31/99.
33
Column
is
$23.3 ($8.9)
(1) includes information for sites placed on the
calculated as the
See the Data Appendix
sum
across the
for further details.
first
NPL
Record of Decisions
before
for each
t
u
i
—
,
<
>o
e
e
=
3 o
—
CO
> ST
>
co On
00 ro rf
=f
-3-
*
o
»n
ID
?p
o
o
©
©
J3
> X
"c3
to
>
Oh
m
(U
<n?
-
>
P-,
^
i
O
*3"
on
V
a!
<*j
•n
NO "1
*— oo
A CM
53
sc
V
in
©
©
© ©
© ©
©
© © ©
m
—mm
o oo
^-CNOONC^CNOCNrnONOO
inON^j-inootn^^inooin^
ONr-ONONr-,fNOOCNOOO
-3-
m
w
Co
o o
cn
CN
CN
CN
»n
<*j
r»
NO
NO
CN
3"
oM
"T,
cn
S5
ON
2
o
^tcncn —^ONNOrnr^ONin'—
— -^-r-~00NOr-inr«-)'rr —
,^
;o\-=roNNoomNo—immm
™
oo £j
"o'o'oooooooo'o
CN 4.
in
t~ -r
6 in
r^-
rj-
„
1-1
L
z2
ioONND-^-in'-'CNinOooo
'3-
rm
o
no
o
o
__
«
m
~ no
^
rl
r4
o'
r^"ONoONr~oomor~-r—<C*-lONONONinOONON—
^
_.t^-*r-~'^r—'mr-—HCNoom
""ooo'o'oooo'ooo
oo
-3-
©©©©©©©
©©©©©©©
© © © © © © ©
©©©©©©©
£3
"3-
-t
— oo
t^ i— ©
in — © ©
oo On O © ©
3 m
p
o o o o © © ©
r-~
0O©
SON--
© ©
cn cn no
on on in
m©©©©©©0©—'NO©
qooooooinoqooo
0©©©©©©o'©0©©
On OO
«
vo oo oo
no
oo'oooooo'o'ooo
NO NO in
no'
in
o'oooooo'ooo'cJo
On©©©©©NOno© —
f>
_
— o
-3-
t>
00 oo
On ON
-
oo
cn r— oo o on
incNONm-*moor-~oom^-CN
~-*
"3-
«-l
m
no
oo
o o o o o
o
r~
minON^t-r-~-NO'^-mONDr-~r-~
r- oo oo Nf
©
o ©
©
©
oo ©
© in
-
>q
od in
CN
00
o
©
©
©
—
>
a
.
i
i
ON ON in
f« r^ -r
—
oo 00
t
CL,
A
«
i
^
_=3
_2
—
i
•* ^r in
o
ro NO in
o
rn ON NO NO
o o cn — in
3 O O O O
rj-
^r
0O
On
—
O
o
oo
i-oONNO^rin—'CNincoooo
m
m
in
m
O
ON in ON ^1" ^-i
in oo
oo
oo no oo oo
CN — in ro
o
m
m
O
o o o o o
>n
od
CN
C 3
o<
CN
*
,
A
c?
3"
oo
B4
o>
i
m
1
01
'
Pi <**
C
B
3 3
r
in
—
cn
r- o
r-
>n
CN
—
3
£ H 22
0O „
^
1
o
N
\
s"5
'5
"
—
3 m
— 3
*
00 -r
© On o -r
o 33 3
"^
oo
i
-f
i
f~-
oo -t
3
—
t o
i
<*
cn
oo r^
C>
o 3
m
oo
•n
ON
fN -t DC oo r
in
CN ^t- = OO
i
3
— 3 3 3 3 3 3
.
"
no
rn on
r- cn cn
m no
o\ in io t
t
^—
in m
o o cn
o O O O 3 O
X
in
DC
CN
m
V
CN NO
m'
00
">
3C
3 —
"3"
Ch
a.
n oo n -'ONinoooininm
N
;^ — oONONONOor-in —^ooo
r^-NOO—-rnOrninNO
^j
^r t—
i-OONND-^in—
>cNin o od o
"o'o'o'ooo'o'o o o ©
"* £>
NO 'H
m
On
rn
^r rn
cn in
*
ON 00 3 3
„ © CN
C
© © 3 3 3
NO
cn
N©
"™ ©'
ixj
^^
CO cn
NT)
C^J
r--
i>
00
-J
o
3
—
-r
,_,
moincNNDm^roONroT
r^mcNcN'-'^)-r--cNcNr--in
— r-OinoocNcNr-~r-M-m
oONNOTrin—
'is t oooo
CN
ON
ON
NO
CN
Z Q
o
u Z
*
o ^r
» © on
rj ^r
m oo
m rr^ on — —
—
in
o cn
m
© © © ©' o
t-~
o'oooooooo'oo
03
i
i
x>
u 3
3 —
K O
j CN
Dh
>i
Z X>
-r
t
,
in
'
1
f-
3
3
3
3
—
-f o o
©
in NO
no
"*
°
3; r© °°
O °V
m
NO 2 -
0O
22
-3-
CN oo oo "*
m m o
3 9 « ^c in-"N t ot» m
o u ~
3 d o © ©' 3 O © © © ©
o
o
r
oo
o
-
oo -f oo
DC
"3-
in no
^r
cn ^r oo on cn p-
r-~
in
—
•
in
no cn
— *
m
On NO
rcn
in
ci
©'
on
CN
—
i
© O O
U
O
oo
H
o
o
On O
On O
— CN
-o x
J -J
0*
Z Z
P-.
u
_> _>
CO
<u
-=
u
Oi
B C B
a
DUD
M — —
~
cn
o
-3
c
o
U
<
a
'3 o
m z
03
—
—3
^
!
n-l
7> T5^
o
o ©
on o oo
On O On
— CN rl
N° ^°
hZe
S?
OJ
Sc
g
01
a
y u
R
—
A
a
g
"S -c
g
Bu
33
o
—
—
'c
<
2
^
itsD
^
B
oo
—
<a
x
B
Q.
O
Oh
o
CD
o
U
©
00
—
E
Q
55
its
S?
a >
a
_o
a,
s
ca
o>
a, -a
.2 s
•yO
N^O
O^ O^
ra
E
^O
s
o>
Z
S
>n
CN
oo
A
-s^O
^O
s
s
-
'
— —
<.
;
'
ii
ii
1
i
i
O
lO ID OO VO
— r— co
m
o
r~ in
©'©'©©'
JS
<
m
o © o o
-3-
no
r--
a
w m
.a
P-S *
O ON — —
O
O O
— © ©
o o o o
O
©
O ©
© ©
e ci
o
o
o o
e o o
©
© ©
© ©
©
m ©
© ©
©
© o © ©
-h
c*1
Tfr
o
o
o
o
Tf
(N
in
in
©'
On
oo
on
[U
in in —
m
^ —
©
— m m
©"
~. -.
© © © (N
tt
Eoo-g g
HON « £
ft
vo
m
*r rcn oo © oo
\t (M N o
©'©©'©'
[
t
a>
r^j
ctf
c*i
a>
©
<=>.
r~"3-
-^-
<U
— A
i
o o
o
n
~ © ©
s:
—
©
~ ©' ©
CNl
o-
(N
r^
in
h «i
©
o ©
©
m
©
©
©
on
m
in
oo
in
On rs ON
oo <N 00
00 rN) oo
•—
°°„
On"
bO
Q
tl
c:
—
m m
© - © ©
o\
no
00
in oo no
t^ ^i
r^ (N on
CM
~-<
©"
N t
—
— ©
©'©©©
"d"
CM
-c
1
«
>>
o
in ~* in ©
— oo in in
m
^t — ©
m
en <n —
©'©'©'©'
in
m
— oo >o
~
—
© oC
© P
— m
NO c NO ~ r- r^r t— (N _e t— r) <N
»— r- in
rn r| ro
c; ^_T
(N CNl
(N
rM BJ
3
m
— —
r--
t:
.a
s s
r-
c
©
~
© © ©
© ©
m
o
53
h-<
°
U H _C 3
u c '3 p
S a a t
O
r^ m in in
n— m
on oo in no
o
m no
rs (N —
-:-:<=>©. rn
O O O (N <( ©'©'©'©
CTn
^D 'O
oo in
^f
r*~i
r--
t
^--
B
_o
3
oq .5
-2
a
o
<
<
-a
=
O
pa
in
A
E
o
V
3
Oh
xO v= xO
O^ o^ ox
o
3
O
Q
9J
tn
r.
u
-I
o
o
w o S 1 8
S Z S £ £
S s?
x n? s° n°
O
CT*
CT»
cT»
K
—
" U U
x
g-H
Table
Least Squares Estimates of the Association Between
3:
NPL
Status and
(1)
A. All
1(NPL
NPL
Sample,
Own
R-squared
1(NPL
NPL
(2)
(3)
(4)
Census Tract Observation
Status by 2000)
B. All
House Prices
0.040
0.046
0.073
0.067
(0.012)
(0.011)
(0.010)
(0.009)
0.579
0.654
0.731
0.779
Sample, 3-Mile Radius Circle Sample Obsevation
Status by 2000)
0.030
0.059
0.113
0.106
(0.011)
(0.013)
(0.012)
(0.011)
Ho: > 0.045, P-Value
0.076
0.726
0.999
0.999
R-squared
0.580
0.652
0.727
0.776
C. Restrict
1
(NPL
NPL
Sites to those in
1982
HRS
Sample,
Status by 2000)
R-squared
D. Restrict
1(NPL
NPL
Sites to those in
1982
HRS
Own
Census Tract Observation
0.071
0.076
0.085
0.057
(0.016)
(0.015)
(0.014)
(0.013)
0.581
0.655
0.732
0.780
Sample, 3-Mile Radius Circle Sample Observation
Status by 2000)
0.046
0.145
0.232
0.194
(0.015)
(0.022)
(0.023)
(0.022)
Ho: > 0.054, P-Value
0.302
0.999
0.999
0.999
R-squared
0.580
0.653
0.728
0.777
1980 Prices
Yes
Yes
Yes
Yes
1980 Housing Characteristics
No
No
No
Yes
Yes
Yes
No
No
Yes
Yes
No
Yes
1980 Economic and Demographic Variables
State Fixed Effects
42,974 in Panels A and B and
Panels C and D. In Panel A/B (C/D) 985 (332) observations are from an area containing a hazardous
that had been on the NPL at any time prior to the 2000 observation on housing prices. The difference
The sample
Notes: The table reports results from 16 separate regressions.
42,321
waste
in
site
between A/B and C/D
inclusion in the
initial
the
NPL
unit of observation
within circles centered
is
is that C/D drops observations from areas with the 663 NPL sites that were not tested for
NPL. The remainder of the sample is comprised of the 41,989 observations on census tracts
with complete housing price data that neither have a
A/C
size
is
at the site
NPL
site
nor are adjacent to a tract with a
the tract that contains the site and in
with a radius of 3 miles.
B/D
it
is
NPL
site.
based on the census
In Panels
tracts that fall
For these circle-based observations, the dependent and
tracts inside the circle, where the weight is the
independent variables are calculated as weighted means across the
fraction of the tract's land area inside the circle multiplied
by
the tract's 1980 population.
coefficient and heteroskedastic-consistent standard error (in parentheses) on the
squared
statistic.
Panels
B and D
also report p-values from tests of whether the
The
entries report the
NPL indicator, as well as
NPL parameters multiplied
Rby the
the
1980 aggregate value of the housing stock exceeds $39 million (Panel B) and $42 million (Panel D), which is our
The aggregate values of the housing stocks in B
and D are $874 and $796 million, respectively. The controls are listed in the row headings at the bottom of the
best estimate of the cost of the average clean-up in these samples.
table.
See the text and Data Appendix for further
details.
Table
4:
Two-Stage Least Squares (2SLS) Estimates of the Effect of NPL Status on House Prices
(2)
(1)
Own
A,
1(NPL
Status by 2000)
(3)
(4)
(5)
(6)
Census Tract
0.037
0.043
0.056
0.047
0.007
0.027
(0.035)
(0.031)
(0.029)
(0.027)
(0.063)
(0.038)
B. Adjacent Census Tracts
1(NPL
Status by 2000)
0.066
0.012
0.011
0.015
-0.006
0.001
(0.035)
(0.029)
(0.025)
(0.022)
(0.056)
(0.035)
C. 2 Mile Radius from Hazardous
1(NPL
Status by 2000)
Ho: > 0.123, P-Value
Waste
Sites
0.018
0.013
0.018
0.003
0.018
-0.008
(0.032)
(0.029)
(0.026)
(0.023)
(0.053)
(0.033)
0.001
0.000
0.000
0.000
0.024
0.000
D. 3 Mile Radius from Hazardous Waste Sites
(NPL
Status by 2000)
0.056
0.036
0.027
0.001
-0.027
-0.005
(0.038)
(0.031)
(0.026)
(0.022)
(0.052)
(0.034)
Ho: > 0.054, P-Value
0.482
0.282
0.153
0.008
0.058
0.041
1
980 Ln House Price
Yes
Yes
Yes
Yes
Yes
Yes
1
980 Housing Characteristics
No
Yes
Yes
Yes
Yes
Yes
No
No
No
No
Yes
Yes
Yes
Yes
No
No
Yes
Yes
Yes
No
No
No
No
Yes
No
No
No
No
1
1980 Economic
& Demographic Vars
State Fixed Effects
Quadratic in
1
982
HRS
Score
Regression Discontinuity Sample
Notes:
The
entries report the results
house price)
is
contains the
site
from the
site
the dependent variable throughout the table.
(Panel A), tracts that share a border with the
calculated as weighted
is
comprised of multiple census
means across
NPL
status
and
HRS
this variable is
The
ln
Yes
(2000 median
units of observation are the census tract that
(Panel B), the areas within a circle of 2 mile radius
tracts,
In Panels
B-D where
the dependent and independent variables are
where the weight is the fraction of the tract that fits
1980 population. The variable of interest is an indicator
the relevant census tracts
the Panel's sample selection rule multiplied
for
The
site
(Panel C), and the areas within a circle of 3 mile radius from the site (Panel D).
the unit of observation
with a 1982
No
from 24 separate instrumental variables regressions.
by the
tract's
instrumented with an indicator for whether the
score exceeding 28.5.
The
tract
had a hazardous waste
site
and heteroskedastic consistent
(B-D) the samples sizes are 487
entries are the regression coefficients
standard errors (in parentheses) associated with the
NPL
indicator.
In Panel
A
(483) in columns (1) through (5) and 227 (226) in column (6). Panels C and D also report p-values from tests of
whether the NPL parameters multiplied by the 1980 value of total housing exceeds $43 million, which is our best
estimate of the cost of the average clean-up.
The 1980 aggregate values of the housing stock
roughly $71, $525, $349, and $796 million (2000
further details.
$'s).
See the notes to Table
3,
in the four
the text and the Data
panels are
Appendix
for
Table
5:
Further
2SLS
Estimates of the Effect of NPL Status on 2000 House Prices, 2-Mile Radius
Sample
(1)
A.
1(NPL
Add
Status by 2000)
Add
-0.033
-0.021
0.028
0.021
(0.050)
(0.034)
(0.053)
(0.061)
-0.01
-0.046
-0.020
(0.044)
(0.031)
Controls for 2000 Covariates
Status by 2000)
0.018
(0.032)
C. Dependent Variable
1(NPL
is
Status by 2000)
1
(0.020)
1990 Housing Prices
Status by 1990)
D. Does Effect of NPL Status Differ
1(NPL
w
jzl
Controls for Ground Water, Surface Water, and Air Migration Pathway Scores
B.
1(NPL
(2)
in Tracts in
0.019
-0.019
-0.000
-0.040
(0.048)
(0.033)
(0.070)
(0.056)
Top
Quartile of Population Density?
0.048
0.015
0.027
0.006
(0.036)
(0.027)
(0.056)
(0.041)
-0.072
-0.043
-0.040
-0.040
(0.072)
(0.047)
(0.047)
(0.067)
1980 Ln House Price
Yes
Yes
Yes
Yes
1980 Housing Characteristics
No
No
No
No
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
No
Yes
No
No
Yes
1(2000 NPL)* 1 (Top Quartile Density)
1
980 Economic and Demographic Variables
State Fixed Effects
Quadratic in 1982
HRS
Score
Regression Discontinuity Sample
Each panel reports parameter estimates and standard errors from 4 separate regressions for 2000 housing
differ from the Table 4 specifications in the following ways: A. adds controls for the individual
pathway scores that are used to calculate the HRS score; B. adds controls for the 2000 values of the covariates; C.
tests for the impact of NPL status on housing prices in 1990; and D. allows the effect of NPL status to differ in
census tracts in the top quartile of population density. In all panels, the sample is the 2-mile radius one and the
sample sizes are 483 in columns (1) through (3) and 226 in column (4). See the Notes to Table 4 and the text for
Notes:
prices.
The panels
further details.
Table
6:
2SLS
Estimates of Stages of Supcrfund Clean-ups on Rental Rate Growth, 2-Mile Radius
Sample
(1)
(2)
(3)
(4)
0.124
-0.019
-0.043
-0.053
(0.046)
(0.034)
(0.049)
(0.052)
0.101
-0.023
-0.054
-0.083
(0.030)
(0.022)
(0.042)
(0.032)
0.059
-0.001
-0.028
-0.037
(0.032)
(0.021)
(0.041)
(0.032)
P- Value from F-Test of Equality
0.25
0.48
0.36
0.31
1980 Rental Rate
Yes
Yes
Yes
Yes
1980 Housing Characteristics of Rental Units
No
No
No
No
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
No
Yes
No
No
Yes
l(NPLOnly)
[115 Sites,
Mean HRS =
40.2]
l(ROD& Incomplete Remediation)
[329 Sites, Mean HRS = 44.3]
1
NPL Deletion)
Mean HRS = 41.6]
(Const Complete or
[214 Sites,
1980 Economic and Demographic Variables
State Fixed Effects
Quadratic in 1982
HRS
Score
Regression Discontinuity Sample
Notes:
is
The
from 4 separate instrumental variables regressions. The In (median rental rate)
There are two observations per county, one for 2000 and one for
entries report the results
the dependent variable throughout the table.
NPL status has been replaced by three independent indicator variables. They
by 1990 or 2000 were placed on the NPL but no ROD had been issued, issued a ROD but
remediation was incomplete, and "construction complete" or deleted from the NPL, respectively. The instruments
are the interactions of the indicator for a 1982 HRS score above 28.5 and these three independent indicators. The
1990.
Here, the indicator variable for
are equal to
1
for sites that
table reports the instrumental variables parameter estimates
The
three indicators of clean-up status.
parameters are equal.
The
2000. The sample sizes
details.
in
effect
of
columns
all
and standard errors clustered
at
the site level for the
table also reports the p-value associated with a F-test that the three
of the controls
listed in the
row headings
(1) through (4) are 966, 960, 960,
are allowed to differ in
and 452, respectively. See the
1990 and
text for further
Table
7:
IV Estimates of 2000
NPL
Status on
2000 Demand
Shifters,
2-Mile Radius Sample
(2)
(3)
(4)
2,758
1,514
-1,233
-563
(1,232)
(1,289)
(3,078)
(2,255)
-0.007
-0.005
0.008
0.003
(0.003)
(0.003)
(0.007)
(0.004)
0.002
-0.000
-0.009
-0.010
(0.006)
(0.007)
(0.018)
(0.013)
0.000
-0.000
0.002
0.001
(0.001)
(0.001)
(0.003)
(0.002)
-0.001
-0.003
-0.014
-0.006
(0.004)
(0.004)
(0.009)
(0.006)
-0.015
-0.015
-0.006
-0.008
(0.008)
(0.007)
(0.018)
(0.010)
,946
355
-2,304
-348
577)
(559)
(1,613)
(876)
1980 Dependent Variable
Yes
Yes
Yes
Yes
State Fixed Effects
No
Yes
Yes
Yes
No
No
No
No
No
No
Yes
Yes
(1)
A. Income and Wealth
Household Income
[1980 Mean: 42,544; 2000
%
-1980 Mean:
14,318]
Public Assistance
[1980 Mean: 0.078; 2000 -1980 Mean: 0.000]
% College Graduates
[1980 Mean:0.135; 2000 -1980 Mean: 0.081]
B. Demographics
Demand
Shifters
% Population Under Age 6
[1980 Mean: 0.086; 2000 -1980 Mean: -0.018]
% Population Over Age 65
[1980 Mean: 0.106; 2000 -1980 Mean: 0.019]
% Black
[1980 Mean: 0.088; 2000 -1980 Mean:0.027]
C. Total Population
Total Population
[1980 Mean: 19,517;
Quadratic
1982
in
2000- 1980 Mean:
HRS
1,526]
Score
Regression Discontinuity Sample
Notes: The entries report the results from 28 separate instrumental variables regressions. The 2000 values
of the variables underlined
The unit of observation is the
The dependent and independent
weighted means across the census tracts within the 2 mile radius circle, where
of the tract within the circle multiplied by the tract's 1980 population. The
in the first
column
are the dependent variables.
area within a circle of 2 mile radius from the hazardous waste
variables are calculated as
the weight
is
the fraction
variable of interest
was placed on
the
is
an indicator that equals
NPL by
2000 and
1
for observations
this variable is
site.
from
tracts
with a hazardous waste
site that
instrumented with an indicator for whether the tract had
HRS score exceeding 28.5. The entries are the regression coefficients
and heteroskedastic consistent standard errors (in parentheses) associated with the NPL indicator. The
samples sizes are 483 in columns (1) through (3) and 226 in column (4). The 1980 and 2000-1980 means
of the dependent variables are reported in square brackets. See the text and previous tables for further
a hazardous waste site with a 1982
details.
Table
8:
2SLS
Estimates of the Effect of 2000
NPL
Status on
Housing Supply, 2- and 3- Mile Radius
Samples
(2)
(3)
(4)
344
60
-889
-260
(151)
(151)
(369)
(202)
1,094
309
-907
-56
(330)
(286)
(702)
(373)
1980 Dependent Variable and Ln House Price
Yes
Yes
Yes
Yes
1980 Housing Characteristics
No
No
No
No
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
No
Yes
No
No
Yes
(1)
Total Housing Units
2 Mile Radius
from Hazardous Waste
Sites
[1980 Mean: 7,363; 2000 -1980 Mean: 1,013]
3
Mile Radius from Hazardous Waste Sites
[1980 Mean: 16,431; 2000- 1980 Mean: 2,187]
1980 Economic and Demographic Variables
State Fixed Effects
Quadratic in 1982
HRS
Score
Regression Discontinuity Sample
The entries report
number of housing
The dependent variables
where the units of observation are the areas
within a circle of 2 and 3 mile radius from the site. The dependent and independent variables are calculated as
weighted means across the relevant census tracts where the weight is the fraction of the tract that falls within the
circle multiplied by the tract's 1980 population. The variable of interest is an indicator for NPL status and this
variable is instrumented with an indicator for whether the tract had a hazardous waste site with a 1982 HRS score
exceeding 28.5.
The entries are the regression coefficients and heteroskedastic consistent standard errors (in
parentheses) associated with the NPL indicator. The samples sizes are 483 in columns (1) through (3) and 226 in
column (4). The means of the dependent variable in 1980 and the mean change between 2000 and 1980 are reported
in square brackets in the first column. See the notes to Table 3, the text and the Data Appendix for further details.
Notes:
are the
the results from 8 separate instrumental variables regressions.
units.
The
results are reported for the cases
Figure
la:
Bid Curves, Offer Curves, and the Equilibrium Hedonic Price Schedule
in a
for Local Environmental Quality
Hedonic Price Schedule
Price
of
Supplier #3
// ^^Consumer
#3
Land
P3
3
.
J
Figure lb: Welfare Gains
Due
to
Local Environmental
Quality
Amenity Improvements with Two Supply Curves
Price
Quantity of Land for
Residential Housing
Hedonic Market
1
u
co
g
o
C3
-a
u
J2
"5.
«
XJ
o
o
o
**<x$fcr
7
1
Xi
I
00
0)
d
Q-,
Z
—
<
co
co
—i
<L>
c
r3
£
oo
CO
3
— o
=
s
N
m
—
o
tN
—
a
-
IT)
0O
1
a
B
3>
u
J3
On
C«-<
pl,
~O
y.
to
o
e
"C
CL
u
0O
3N
,
—
a
cd
E
o
3
u
CO
X>
•J
D.
E
o
cc_
60
'J
nJ
CO
CO
xl
Q,
J
—
o
00
O
u
a
z
,
<
xi
-
XI
=jj
c
CO
CO
E
a
-
CO
o
a
00
Jj
.c
O
2
,+5
S
Figure
3:
Geographic Distribution of Hazardous Waste Sites
A. Sites with 1982
HRS
B. Sites with 1982
Notes:
tract
The 1982
HRS
Sample
is
in the
1982
HRS
Sample
Scores Exceeding 28.5
HRS
Scores Below 28.5
comprised of the 487 hazardous waste
sites that
were placed
in a
census
with nonmissing housing price data in 1980, 1990, and 2000. 306 (181) of these sites had 1982
scores above (below) 28.5.
HRS
1
1
J
On
Z
y
.S
re
«
~S
a
"°
g
-Mi?
= U
i-
O '0 D,
a
2 u
<U — u~>
S £fod
'
c/}
-w
"Si '3
> re
* u
o w
'S
<U
S
*j
Q>
U
""
'S >
^
u. 00
•a
ti
"3
•5
w
H
.
S
co
•— 00
<u
—
5
J
1N ^
-s %
£ 2
g
OX)
«
J3
—
—U
o
X
o
u a o
u o
c
00 G
J3
C
E
-
u -3
e
3 c
z re
c
3
o
a
03
3 s
« *s
00
a;
a>
re
2
— £ SI
-_J
rN -S
00
o
u
00
00
oi
*-.
o
3
_re
42
a
(U
o
a
3 u
o o
5
" u
re
u
T3
a .a
00
s
U
T3
,p
5U IS&
(N
00
ON
!!
13
-S
i>
.as
o
3>
S^oo
re
re
o °
3 DO
-C
s
00
re
<u
re
H
o
o
.S
c
H
=* _n
-5
J3
Aousnbojj
00
Z
o
o
O
re
o
00
o
—
o
3
.i
CJ
Figure
5:
Probability of Placement on the
NPL
HRS
by 1982
Score
T" —
05
o
o
o
-
™CO>>
'
-Q
_l
o.
^-
.
1
z
o
c
<° -
l»CD
O
TO ^J-
Q.
_
'
y
1«£
!"-
/T
/m
o
a
<- -
o
-
m*r
mf*
i
m
i
1
1
1
20
40
60
80
1982
HRS
Score
NPL status and the 1982
487 sites in the 1982 HRS sample. These plots are done separately for sites below
(dark colored line) and above (light colored line) the 28.5 threshold. They come from the estimation of
nonparametric regressions that use Cleveland's (1979) tricube weighting function and a bandwidth of 0.5.
Notes: The figure plots the bivariate relation between the probability of 2000
HRS
The
4.
score
among
the
data points present the
See the text for further
mean
details.
probabilities in the
same
4-unit intervals of the
HRS
score as in Figure
Figure
6:
2000 Residual House Prices
after
Adjustment for Column 4 Covariates, 2-Mile Radius Sample
(N _
in _
w
CD
o
T
-
m
i
m
T
House
.05
i
il
m
J
-'1
m
B
-.05
m
\
m
Residue
i
2000
-.1
m
c\j
_
T
i
40
20
1982
HRS
80
60
Score
Notes: The figure plots the results from nonparametric regressions between 2000 residual housing prices
from the 2 mile radius sample
after
4 (except the indicator for a
HRS
adjustment for the covariates
in the
score above 28.5) and the 1982
column
HRS
(4) specification of
scores.
Table
The nonparametric
regressions use Cleveland's (1979) tricube weighting function and a bandwidth of 0.5.
These plots are
done separately for sites below (dark colored line) and above (light colored line) the 28.5 regulatory
threshold. The data points are based on the same 4-unit intervals of the HRS score as in Figures 4 and 5.
See the text for further details.
Figure
7:
2000 Residual Housing Units
after
Adjustment
for
Column 4
Covariatcs, 2-Mile Radius Sample
o
CD
~
T
o
o
CN
T—
w
*i
<=
3
o
o
CO
°>o
.Soto
*3-
X
_o
-
x*^"
f^
CD
3
-o
rk
m
o
O
~~*"^
*-^~^
m
,
o
—"•—-^» 1> .^p-—
J-~-»
^
s*
I
1
§2
o2
I
-
ti
CM °°
o
o
CM
T
-
*"
o
o
CD
I
i
I
m
"
40
20
1982
Notes:
The
figure plots the results
from the 2 mile radius sample
after
8 (except the indicator for a
HRS
HRS
80
fiO
Score
from nonparametric regressions between 2000 residual housing
adjustment for the covariates
in the
score above 28.5) and the 1982
column
HRS
units
(2) specification of Table
scores.
The nonparametric
regressions use Cleveland's (1979) tricube weighting function and a bandwidth of 0.5.
These plots are
done separately for sites below (dark colored line) and above (light colored line) the 28.5 regulatory
threshold. The data points are based on the same 4-unit intervals of the HRS score as in Figures 4-6. See
the text for further details.
3521
sJ
*~*
*—
'7