The Effect of Pollution on

the Value of Houses

Econometric Analysis
Walter Sosa-Escudero
Spring 2009

A really classic paper:

Harrison, D. and Rubinfeld, D., 1978. Hedonic
prices and the demand for clean air, Journal
of Environmental Economics and
Management, 5, 81-102

Motivation

How can we measure the willingness to pay for

clean air?
Standard problem in public finance: free riding.
No incentives to reveal willingness to pay
If pollution affects prices of houses, this can be
used to measure willingness to pay. Families in
fact pay more (less) to live in less (more)
polluted places.

One strategy: compare values of

houses with different levels of
pollution.

There is a problem with this strategy.

Methodology

A hedonic model to explain what determines

the value of the houses.
The model is used to decompose how different
characteristics of a house contribute to the total
price.
In this hedonic model the level of pollution is
included as one of the characteristics who may
be explaining the value of the houses.
The role of the regression model is to isolate
the contribution of pollution from other
competing factors.

Data and Variables

Data: classic paper by Harrison and Rubinfeld (1978).
Explained variable:

VALUE: average value of occupied houses in
Boston (thousands of $).
Explanatory variables

NITOX: concentration of nitrogen oxides (parts per
million, annual average concentration).

CRIME: crime rate in the locality (crimes per capita,
in %).

Variables

ROOMS: Average rooms per dwelling.

AGE: proportion of housing built before 1940.

DIST: average distance to five major

employment centers in the Boston area (km).
ACCESS: index of accessibility to highways of
the radial Boston area.
TAX: tax rate ($ / $ 10,000).
PTRATIO: ratio of students per teacher.

Summary statistics
Variable |
Obs
Mean
Std. Dev.
Min
Max
-------------+----------------------------------------------------value |
506
22.53281
9.197104
5
50
crime |
506
3.613525
8.601545
.0063
88.9762
nitox |
506
.5546951
.1158777
.385
.871
rooms |
506
6.284634
.7026172
3.561
8.78
age |
506
68.5749
28.14886
2.9
100
dist |
506
3.795043
2.10571
1.1296
12.1265
access |
506
9.549407
8.707259
1
24
tax |
506
408.2372
168.5371
187
711
ptratio |
506
18.45553
2.164946
12.6
22

The Hedonic Model

valuei = + 1 crimei + 2 nitox i + 3 roomsi + 4 agei
+ 5 dist i + 6 accessi + 7 tax i + 8 ptratio i + i
i = 1, ..., 506

Ex-ante conjectures
E[valuei ]
Since,
j =

x ji

What signs do we expect for j?

Positive Coefficients : 3 , 6

Negative Coefficients: 1, 2 , 4, 7, 8

Coefficients without conjecture: , 5

OLS estimation
regress value crime nitox rooms age dist access tax ptratio
Source |
SS
df
MS
-------------+-----------------------------Model | 28064.0746
8 3508.00932
Residual |
14652.221
497
29.48133
-------------+-----------------------------Total | 42716.2956
505
84.586724

Number of obs
F( 8,
497)
Prob > F
R-squared
Adj R-squared
Root MSE

=
=
=
=
=
=

506
118.99
0.0000
0.6570
0.6515
5.4297

-----------------------------------------------------------------------------value |
Coef.
Std. Err.
t
P>|t|
[95% Conf. Interval]
-------------+---------------------------------------------------------------crime | -.1834488
.0364887
-5.03
0.000
-.25514
-.1117576
nitox | -22.81088
4.160742
-5.48
0.000
-30.98569
-14.63607
rooms |
6.371512
.3923866
16.24
0.000
5.600571
7.142453
age | -.0477499
.0141018
-3.39
0.001
-.0754564
-.0200434
dist | -1.335269
.2001468
-6.67
0.000
-1.728507
-.942031
access |
.272282
.072276
3.77
0.000
.1302777
.4142863
tax | -.0125921
.0037702
-3.34
0.001
-.0199995
-.0051847
ptratio | -1.176787
.1394154
-8.44
0.000
-1.450703
-.9028705
_cons |
28.40667
5.365948
5.29
0.000
17.86393
38.9494
------------------------------------------------------------------------------

OLS estimation
regress value crime nitox rooms age dist access tax ptratio
Source |
SS
df
MS
-------------+-----------------------------Model | 28064.0746
8 3508.00932
Residual |
14652.221
497
29.48133
-------------+-----------------------------Total | 42716.2956
505
84.586724

Number of obs
F( 8,
497)
Prob > F
R-squared
Adj R-squared
Root MSE

=
=
=
=
=
=

506
118.99
0.0000
0.6570
0.6515
5.4297

-----------------------------------------------------------------------------value |
Coef.
Std. Err.
t
P>|t|
[95% Conf. Interval]
-------------+---------------------------------------------------------------crime | -.1834488
.0364887
-5.03
0.000
-.25514
-.1117576
nitox | -22.81088
4.160742
-5.48
0.000
-30.98569
-14.63607
rooms |
6.371512
.3923866
16.24
0.000
5.600571
7.142453
age | -.0477499
.0141018
-3.39
0.001
-.0754564
-.0200434
dist | -1.335269
.2001468
-6.67
0.000
-1.728507
-.942031
access |
.272282
.072276
3.77
0.000
.1302777
.4142863
tax | -.0125921
.0037702
-3.34
0.001
-.0199995
-.0051847
ptratio | -1.176787
.1394154
-8.44
0.000
-1.450703
-.9028705
_cons |
28.40667
5.365948
5.29
0.000
17.86393
38.9494
------------------------------------------------------------------------------

Distance from downtown (dist):

-----------------------------------------------------------------------------value |
Coef.
Std. Err.
t
P>|t|
[95% Conf. Interval]
-------------+---------------------------------------------------------------crime | -.1834488
.0364887
-5.03
0.000
-.25514
-.1117576
nitox | -22.81088
4.160742
-5.48
0.000
-30.98569
-14.63607
rooms |
6.371512
.3923866
16.24
0.000
5.600571
7.142453
age | -.0477499
.0141018
-3.39
0.001
-.0754564
-.0200434
dist | -1.335269
.2001468
-6.67
0.000
-1.728507
-.942031
access |
.272282
.072276
3.77
0.000
.1302777
.4142863
tax | -.0125921
.0037702
-3.34
0.001
-.0199995
-.0051847
ptratio | -1.176787
.1394154
-8.44
0.000
-1.450703
-.9028705
_cons |
28.40667
5.365948
5.29
0.000
17.86393
38.9494
------------------------------------------------------------------------------

Expected value decreases in $ 1335 per kilometer

from downtown Boston.

Crime rate:
-----------------------------------------------------------------------------value |
Coef.
Std. Err.
t
P>|t|
[95% Conf. Interval]
-------------+---------------------------------------------------------------crime | -.1834488
.0364887
-5.03
0.000
-.25514
-.1117576
nitox | -22.81088
4.160742
-5.48
0.000
-30.98569
-14.63607
rooms |
6.371512
.3923866
16.24
0.000
5.600571
7.142453
age | -.0477499
.0141018
-3.39
0.001
-.0754564
-.0200434
dist | -1.335269
.2001468
-6.67
0.000
-1.728507
-.942031
access |
.272282
.072276
3.77
0.000
.1302777
.4142863
tax | -.0125921
.0037702
-3.34
0.001
-.0199995
-.0051847
ptratio | -1.176787
.1394154
-8.44
0.000
-1.450703
-.9028705
_cons |
28.40667
5.365948
5.29
0.000
17.86393
38.9494
------------------------------------------------------------------------------

An increase in 1 % in the crime rate, decreases the

average value of houses in $183

The effects of pollution

Pollution (nitox):
-----------------------------------------------------------------------------value |
Coef.
Std. Err.
t
P>|t|
[95% Conf. Interval]
-------------+---------------------------------------------------------------crime | -.1834488
.0364887
-5.03
0.000
-.25514
-.1117576
nitox | -22.81088
4.160742
-5.48
0.000
-30.98569
-14.63607
rooms |
6.371512
.3923866
16.24
0.000
5.600571
7.142453
age | -.0477499
.0141018
-3.39
0.001
-.0754564
-.0200434
dist | -1.335269
.2001468
-6.67
0.000
-1.728507
-.942031
access |
.272282
.072276
3.77
0.000
.1302777
.4142863
tax | -.0125921
.0037702
-3.34
0.001
-.0199995
-.0051847
ptratio | -1.176787
.1394154
-8.44
0.000
-1.450703
-.9028705
_cons |
28.40667
5.365948
5.29
0.000
17.86393
38.9494
------------------------------------------------------------------------------

An increase in one unit in the index of concentration of

nitric oxide will decrease the average value of
houses in $ 22,810.

Suppose the government can implement a

policy that reduces pollution in 5% in a
certain neighborhood. What would be the
expected increase in the value of houses in
that neighborhood
Given a level of contamination x, a 5%
reduction implies a decreas in
contamination in 0.05 x.
According to our estimates, this reduction
produces an increas in the expected value
of houses in $1140x (22810 * 0.05* x).

For a neighborhood with average

contamination (0.55ppm) this implies an
increase in the value of houses of $627.
What is the social benefit of implementing
this policy? Suppose each of the houses
increases its value in $627 and that there
are N hosues. Is the cost of reducing
contamination in 5% greater than $627 N
(this is more or less the maximum families
should be willing to pay to have pollution
decreased).