Microsoft Word - regulate33.docMatthew E. Kahn1
UCLA and NBER
April 2008 Preliminary and Incomplete
[email protected]
1 I thank seminar participants at USC, Research Triangle, and the
summer 2007 UBC Real Estate Conference. I thank Jeff Zabel for
kindly sharing data with me. Ed Glaeser provided useful
suggestions. This research was supported by the Richard S. Ziman
Center for Real Estate at UCLA.
2
Introduction
Housing supply regulation raises the cost of building new housing.
The
consequences of housing supply regulation have been documented in a
number of recent
studies including Fischel 2000, Mayer and Somerville 2000, Quigley
and Raphael 2004,
Glaeser, Gyourko and Saks 2005, Schill 2005). By limiting supply in
some of the most
desirable cities in the United States, such regulation may
contribute to the extraordinary
price appreciation that has been observed in major coastal cities
such as San Francisco,
Los Angeles, Boston and New York City (Gyourko, Mayer and Sinai
2006).
Why do housing supply regulations differ across space? One
explanation focuses
on the median voter’s narrow self interest. Home owners have a
financial incentive to
discourage new construction because it reduces the scarcity value
of their asset (Fischel
1999). Richer communities may engage in fiscal zoning to keep the
poor out. Minimum
lot zoning reduces the likelihood that new entrants will be much
poorer than incumbents.
Communities may also enact housing supply regulation to preserve
and enhance
local quality of life. Environmentalist communities are especially
likely to pursue such
goals. Environmentalists may seek to block local growth to preserve
local public goods
such as open space, bike paths and clean air and to preserve the
character and culture of
their community.
Academics have also posited that communities may justify
anti-growth sentiment
using environmental concerns as a politically correct justification
for less noble reasons.
“The actual issues that lead people to oppose homebuilding are hard
to discover. By far the most frequent objections that growth
opponents raise have to do with environmental impacts. These range
from harm to wildlife to destruction of natural resources to
increases in air pollution. Yet to label all protest as
environmentalism would be a mistake. Many growth opponents use
environmental arguments to mask other motives such as fears of
property tax increases or anxieties about keeping their
community
3
exclusive. Environmental rhetoric has become a valued currency for
public debate with much greater voter appeal than arguments that
appear more narrowly self interested. As a result people who are
not environmentalists in any sense borrow it for their own
purposes.” (Frieden 1979 page 8)
While the environmental “activism” hypothesis is intriguing, it is
challenging to
test because of the difficulty of identifying credible measures of
community
environmentalism. Put simply, what evidence can be collected to
document that
Berkeley, California’s population is a “greener” community than a
Republican
community in Orange County California?
This paper uses California and national level data to test
community
environmentalism’s role in determining new housing supply.
California provides an
ideal setting for studying this issue because of its clear spatial
variation in where
environmentalists do (i.e Berkeley) and do not live. It is well
known to be an innovative,
regulatory leader as demonstrated by Gov. Arnold Schwarzenegger
recently signing the
first cap on greenhouse gas emissions (AB32).
Detailed political registration and voting data on binding
California propositions
provide a rich data set for identifying environmentalist
communities. California is the
nation’s leading state in voting on direct environmental
initiatives (Kahn and Matsusaka
1997, Matsusaka 2005). In recent research, I have documented that
data from political
markets such as a community’s share of Green Party members is
useful for proxying for
community environmentalism (Kahn 2007, Kahn and Morris 2007). In
these papers, I
have documented that in communities whose population votes and
registers as “greens”
that such communities are more likely to purchase Toyota Prius
vehicles, have
environmentalist stores locate within their borders and feature a
workforce that
4
disproportionately commutes using public transit, walking and
biking. The
environmentalism indicators I introduce below are notable because
they are based on
actual choices of community members rather than stated attitudes in
surveys. I also
examine a second national level data set. While the national data
offers a larger sample,
my community environmentalism measures are not as high quality as
my California
measures.
This paper proceeds in four steps. First, I sketch a dynamic
process through
which environmentalist communities form in some areas but not in
others. Second, I
present my estimation strategy for testing for environmentalism’s
role in determining
new housing supply. Section Three discusses the data sources. The
results are
presented in Section Four and Section Five concludes. The paper’s
major findings are
Where do Environmentalist Communities Form?
This paper’s core goal is to test whether there is less new housing
construction in
environmentalist communities. To begin to study this issue requires
an understanding of
where such communities form within urban areas. Within metropolitan
areas, greens are
not “randomly assigned” across communities.
People differ with respect to their support for environmental
causes. For reasons
that lie outside this paper’s focus, some people are
environmentalists and others are not.
The heterogeneous population Tiebout sorts into sub-communities.
This sorting provides
a source of variation to test this paper’s main hypothesis.
5
Within a metropolitan area, environmentalists are likely to cluster
in high density
areas, featuring opportunity to walk, located near public transit
stations and in high
environmental amenity areas such as Berkeley, Santa Monica and
Santa Cruz. Small
initial differences in exogenous spatial attributes such as
proximity to the ocean can have
a social multiplier effect. As environmentalists move to a nice
community, green
businesses such as organic restaurants would be more likely to
locate near this
community. Such “endogenous” green amenities would only further
encourage
environmentalists to move to this community. Such a green community
would act as a
club financing “green” infrastructure such as bike lines. A social
multiplier effect takes
place such that a community that develops a “green” reputation
attracts stores that cater
to this group and this attracts more “like-minded” people
(Waldfogel 2007). Community
social interactions would only reinforce this dynamic as a type of
social multiplier effect
feeds on itself.
Why would an environmentalist community engage in “extra” zoning?
The
median voter may believe that growth threatens the area’s natural
capital and overall
quality of life. The residents of the area may believe that due to
their environmentalism
they have been able to overcome tragedy of the commons problems
such as litter that
degrade quality of life in other communities. They may fear that
this social ethic could
be “diluted” if growth takes place. In this sense, environmentalist
areas may be like
religious communities and as such are club goods. Iannaccone (1992)
and Berman
6
(2000) have argued that club member utility is higher if the size
of the group is larger and
if the average devotion to the cause is higher. Housing supply
limitation measures may
help to self-select only those environmentalists who are committed
to the cause and this
raises the average devotion to the group.
To test for environmentalism’s role in determining new housing
production across
different communities, I will use data from California and the
nation as a whole. In one
set of results I will present below, the dependent variable will be
the count of annual new
single detached housing permits in a place in a given calendar
year.
New Housingjlt = αl + γ Greenjlt-n + φ Xjlt + U jlt (1)
Controlling for other baseline community characteristics, (X), and
geographical
fixed effects (such as county or metropolitan area fixed effects),
I test whether γ is less
than zero. This would indicate that there is less new housing
production in
environmentalist communities. This regression approach is an
indirect way for
documenting environmentalism’s consequences for housing supply
limitations. Below, I
will present some direct evidence where the dependent variable will
be a measure of the
housing supply regulations that a place has enacted at a point in
time (based on the
Wharton Survey).
In interpreting estimates of equation (1), there are three distinct
reasons why I
may find that γ is negative. The first explanation is the
“treatment effects” hypothesis.
The presence of environmentalists creates a voting coalition that
seeks policies to limit
growth. The second explanation emphasizes selection effects. Greens
locate in
7
communities that for exogenous reasons are difficult to build new
housing. A third
explanation is that greens locate in “hippie communities” that
block growth for reasons
independent of environmental concerns.
There are plausible reasons to question the exogeneity of “Green”
in equation (1).
The unit of analysis in the California regressions will be a
place/year such as Santa
Monica in 1996. The error term in equation (1) will reflect
unobserved demand factors.
Greens could be agglomerating in the most desirable areas of a
metropolitan area or their
costly actions may make an area a more attractive place to live. In
both the “selection”
and “treatment” cases, developers will seek to build in such
places. In this case, the
presence of Greens signals that a place has high quality of life
for exogenous reasons. If
greens cluster in the best parts of a metropolitan area, then E( U
jt-n | Green) will be
greater than zero. Environmentalists may cluster in places with
high unobserved (to the
econometrician) quality of life and their actions may improve an
area’s quality of life
increasing the demand to live there. The net effect of either the
selection or the
treatment mechanism would be that E( U jt | Green) would be
positive and OLS estimates
of equation (1) would be biased against my core hypothesis that γ
is less than zero.
Conversely, it is possible that E(U|green)<0. Environmentalists
could cluster in
“hippie” places where anti-business sentiment creates red-tape that
hinders development.2
In this case, the environmentalists may not be causing the hurdles
to development. They
2 “First, just navigating the process adds explicit financial and
time costs. The effects of those direct costs are similar to those
of fees. Second, both the outcome and the length of the regulatory
process are uncertain. Developers do not know the extent to which
local authorities will demand costly changes in project density,
design, or type before granting a final approval. For risk adverse
developers, this uncertainty will reduce the level of new
construction above the direct effect of higher mean construction
costs (Mayer and Somerville 2000).”
8
may merely settle in such areas. In this case, my estimates of
equation (1) will over-
estimate the effect of “environmentalism” because the
environmentalism measures I
present below may in part reflect the omitted variable of
“anti-business” ideology. I will
address this below by creating some measures of community
“liberalism” and testing
whether controlling for such community ideology measures reduces
the effect of my
environmentalism measures. Clearly, the validity of this test
hinges on the absence of
collinearity. I need there to be environmentalist communities that
are not liberal and
vice-versa.
instrumental variables strategy for reducing the correlation
between the error term in
equation (1) and the environmentalism indicators. Below, I will
contrast OLS and IV
estimates of this equation. The instrument set will be lagged
political party registration
variables from the year 1972 and measures of the city’s urban in
1970. I will discuss the
validity of such variables as instruments below. In equation (2), I
present the “first
stage” of TSLS.
Greenjt = α + γ Politicsjt-n + φ historyjt-n + ψ jt (2)
In equation (2), a California city j’s time t environmentalism is
modeled as a
function of the community’s past political party of registration
shares “Politics” and of its
past attributes such as its lagged population density. The lag
structure in equation (2) will
be 18 years when the variable I instrument for is 1990
environmentalism and it will be 8
years when I instrument for year 2000 environmentalism in
estimating equation (1).
9
If the error term in equation (1) has a permanent component, then
my political
instruments are unlikely to resolve the environmentalism
endogeneity issue. For
example, if greens always cluster in the most beautiful part of a
metropolitan area and if
this beauty is a fixed effect, then my instrument set (the lagged
political variables) will
also be correlated with the error term in equation (1).
Data Sources
Identifying California Greens
The percentage of each census tract’s voters who are registered
with the Green
Party.3 The Green Party is well known for its environmentalism
activism, and California
is the state with the highest count of Green Party registered
voters both in absolute terms
and as share of all registered voters.4 But even in California, the
Green Party’s
membership is small. Across 7002 California census tracts in the
year 2000, the average
tract’s Green Party share is 0.009 and the median is 0.005. This
makes Green Party
membership a costly political choice, because in California,
members of this party lose
the right to vote in another party’s primary election.
In California, voters have the opportunity to participate in
lawmaking through
ballot initiatives (Matsusaka 2005). Many of these initiatives are
related to
3 The Berkeley IGS (see http://swdb.berkeley.edu/) provides data
for each California census tract's count of registered Green Party
Voters in the year 2000. It is important to note that voting
precincts and census tracts spatially overlap but they do not
coincide. To translate the voting precinct data into census tract
data, The Berkeley IGS takes the precinct data (there are over 1700
Precincts in Los Angeles County alone) and uses a statistical
procedure based on ecological inference to create the census tract
data. 4 See http://cagreens.org/platform/ecology.htm
10
environmental issues. Voting patterns based on these binding votes
is informative about
a community’s environmentalism.
Here is a brief summary of three propositions I study.
Proposition 185 in 1994: This measure imposes a 4 percent sales tax
on gasoline not diesel fuel beginning January 1, 1995. This new
sales tax is in addition to the existing $.18 per gallon state tax
on gasoline and diesel fuel and the average sales tax of
approximately 8 percent imposed by the state and local governments
on all goods, including gasoline. Revenues generated by the
increased tax will be used to improve and operate passenger rail
and mass transit bus services, and to make specific improvements to
streets and highways. The measure also contains various provisions
that generally place restrictions on the use of certain state and
local revenues for transportation purposes.
(www.calvoter.org/archive/94general/props/185.html)
In the results I report below I will use factor analysis to extract
an environmentalism
factor based on a place’s share of registered Green Party voters in
1992 and its share of
voters who voted in favor of Proposition 185 in 1994. This factor
will represent the early
1990s environmentalism indicator for a place.
The year 2000 place environmentalism index is based on factor
analysis of four
variables. The place’s Green Party registered voter share is the
first indicator. The next
two are the share of the place that voted in favor of the following
propositions.
Proposition 12 in March 7, 2000, The 2.1 billion dollar "Safe
Neighborhood Parks, Clean Water, Clean Air and Coastal Protection
Bond Act of 2000" (2000 Bond Act). This proposition authorizes the
state to sell $2.1 billion of general obligation bonds to fund many
designated programs. About $940 million of the bond money would be
granted to local agencies for local recreational, cultural, and
natural areas. The remaining $1.16 billion would be used by the
state for recreational, cultural, and natural areas of statewide
significance. Proposition 12 will help make our parks safer, keep
our water free of pollution, improve air quality, and preserve our
natural resources.http://ca.lwv.org/lwvc.files/mar00/pc/prop12.html
Proposition 13 in March 2000: This proposition authorizes the state
to sell $1.97 billion of general obligation bonds to spend on
programs designated to provide: Safe Drinking Water ($70 million)
Flood Protection ($292 million) Watershed Protection
11
($468 million) Clean Water and Water Recycling (355 million) Water
Conservation ($155 million) Water Supply, Reliability and
Infrastructure ($630 million)
http://ca.lwv.org/lwvc.files/mar00/pc/prop13.html.
My final measure of environmentalism is the count of hybrids
registered in each
California zip code divided by the zip code’s count of households.
Given the reputation
of these cars as “green vehicles,” and the fact that hybrid owners'
savings in fuel costs are
far smaller than price differential between hybrids and comparable
conventional models,
hybrid ownership represents a costly "green" choice and is a strong
indicator of
environmentalist beliefs. Using factor analysis on these four
measures of place
environmentalism yields an intuitive index. Based on 349 places in
California, the top
ten “greenest” places are: Albany, Berkeley, Fairfax, Belvedere,
Piedmont, Mill Valley,
Larkspur, Portola Valley, Sausalito and Palo Alto. Other notable
places include; Santa
Monica (ranked #12), Malibu ranked #17 and Santa Cruz ranked #23.
The “brownest”
five are; Woodlake, Montclair, Ripon, Gonzales, and Escalon.
The instrument set will include a city’s political party
registrations such as the
1972 share of the city’s registered voters who were registered
Democrats. In estimating
equation (2), I expect that a city’s “liberalness” persists over
time such that cities with
larger share of liberal voters (i.e registered voters from the
Democratic Party and the
Peace and Freedom Party) in 1972 have larger environmentalism
scores in the 1990s.
In a series of case studies about California, Frieden (1979) argues
that
some home owner communities voice environmentalist views as a
political correct means
of achieving their NIMBY goals. My measures of environmentalism are
based on
political choices not on stated rhetoric. It is possible that a
community of “fake greens”
will voice words justifying their anti-growth stance based on
environmental arguments
12
but based on political data not reveal any “true” environmental
leanings. My
environmentalism measures are unlikely to suffer from this
problem.
Identifying Greens: National Measure #2
In previous work based on California data, I have documented
that
Representatives whose constituents are environmentalists (based on
their initiative voting
and political affiliations such as being members of the Green
Party) are more likely to
have sharply pro-environment voting records (Kahn 2007). To measure
a
Representatives’ “pro-environment” voting record, I use data from
the League of
Conservation Voters.
The annual League of Conservation Voters’ (LCV) “Scorecard”
determines which
roll call votes are important pieces of environmental legislation
and identifies what is the
“pro-environment” vote on each specific issue (see
www.lcv.org).
“This Scorecard represents the consensus of experts from 19
respected environmental and conservation organizations who selected
the key votes on which Members of Congress should be graded. LCV
scores votes on the most important issues of the year, including
environmental health and safety protections, resource conservation,
and spending for environmental programs. … Dedicated
environmentalists and national leaders volunteered their time to
identify and research crucial votes.”
A Representative’s voting record is likely to be positively
correlated with
unobserved constituent ideology if Representatives vote partially
based on their
constituent’s preferences. If Representatives voted their own
ideology on environmental
bills, then their voting records would not be informative about the
preferences of the
median voter in their district.
13
It is important to note that my measures of environmentalism are
based on costly
“revealed preference”. This matters because researchers such as
Friedan (1979) have
argued that self interested home owners embrace environmentalism,
not out of “true”
ideology but instead to cloak themselves in a politically correct
coating of
environmentalism that allows them to achieve their goals of banning
new construction
without explicitly stating their “true” motivations.
Measuring New Housing Supply
I focus on the count of new construction being permitted and
observed in different
physical jurisdictions at different dates. The first data set I
examine is annual building
permits for each Californian city are based on Zabel and Peterson
(2006) data obtained
from the California Industry Research Board (CIRB)
http://www.cirbdata.com/. The
dataset includes the total number of permits granted each year for
different types of
housing structures for more than 400 FIPS places over the period
1990-2004, as recorded
by the CIRB. This represents the incorporated subset (with minor
exceptions) of all
California FIPS places and encompasses the majority of all land
within FIPS boundaries.
Within the sample, places that featured a large share of new
development in 1990 also
had high levels of new development in 2004. The correlation between
the 1990 and 2004
new development share (defined as place j’s count of new
permits/total sample new
permits) is .818. For the nation as a whole, I use data from the 5%
IPUMS Census of
Population and Housing micro data. The micro data reveals whether a
specific household
lives in a single detached home and the age of the housing
unit.
14
Evidence from California
In Table One, I report California housing permit regressions using
place level data
from 1990 to 2004. The unit of analysis is a place/year. The Table
reports four OLS
regressions of equation (1). Standard errors are clustered by
place. In each of these
regressions, I control for the place’s share of housing units that
are owner occupied, and
the place’s population size, average income and population density.
County fixed
effects and year fixed effects are included in each regression. The
environmentalism
indicator is based on a place’s environmental factor constructed
using information on its
Green Party registration share in 1992 and the share of votes in
favor of Proposition 185
in 1994.
Table One reports four OLS estimates of equation (1). In columns
(1) and (2) the
dependent variable is the log(1+total single family home permits).
The results in these
columns differ because I weight the regressions by a place’s 1990
population in column
(1) while in column (2) the results are unweighted. In all of the
California regressions I
report, the standard errors are clustered by place. All else equal,
a standard deviation
increase in a place’s environmentalism reduces its single family
permits by 34% in the
population weighted regression and by 24% in the unweighted
regression (see column
(1)). In columns (3) and (4), I report results where the dependent
variable equals a
place’s total log(1+total housing units). Environmentalist places
are issuing a larger
share of their new permits as multi-family dwellings. The results
in column (3) indicates
that a standard deviation increase in community environmentalism
reduces total housing
15
permits by 20%. This estimate is smaller than the estimate of 34%
reported in column
(1) for single detached housing units. These findings are robust to
controlling for other
community political ideology measures. I continue to explore the
robustness of these
results to including more place level controls.
In future work, I will incorporate place specific measures of
available infill
development land. This data has been created in a project led by
John Landis (see
http://infill.gisc.berkeley.edu/index.html). In future work, I will
create measures of a
California place’s “liberalism” based on its voting on propositions
that have no obvious
environmental component. I will study how the environmentalism
coefficient is affected
by controlling for this measure of place liberalism.5
To provide some evidence on the causal link between place
environmentalism and
regulatory intensity, I use data from the Wharton Regulation Index
(see
http://real.wharton.upenn.edu/~gyourko/Wharton_residential_land_use_reg.htm).
Gyourko, Saiz and Summers (2008) have created a measure of
regulatory severity. In a
place level regression of their Wharton Residential Land Use
Regulation Index on my
year 2000 environmentalism indicator (weighted by place population)
yields:
reg WRLURI p2 [w=Pop]
5
http://en.wikipedia.org/wiki/California_Proposition_22_(2000)
16
Source | SS df MS Number of obs = 179
-------------+------------------------------ F( 1, 177) = 54.15
Model | 32.8429086 1 32.8429086 Prob > F = 0.0000 Residual |
107.349181 177 .606492546 R-squared = 0.2343
-------------+------------------------------ Adj R-squared = 0.2299
Total | 140.192089 178 .787596007 Root MSE = .77878
------------------------------------------------------------------------------
WRLURI | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
p2 | .6183516 .0840287 7.36 0.000 .4525247 .7841786 _cons |
.8955058 .0629615 14.22 0.000 .7712539 1.019758
------------------------------------------------------------------------------
And, including county fixed effects yields:
Linear regression, absorbing indicators Number of obs = 179 F( 1,
135) = 38.66 Prob > F = 0.0000 R-squared = 0.5524 Adj R-squared
= 0.4098 Root MSE = .6818
------------------------------------------------------------------------------
WRLURI | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
p2 | .8852921 .1423755 6.22 0.000 .6037171 1.166867 _cons |
.8192679 .0651948 12.57 0.000 .6903327 .948203
-------------+----------------------------------------------------------------
fips | F(42, 135) = 2.284 0.000 (43 categories)
Variable | Obs Mean Std. Dev. Min Max
-------------+--------------------------------------------------------
WRLURI | 179 .5866114 .7194855 -1.268611 3.592063 p2 | 179 .0801588
.8992566 -1.640422 3.164381
(april15b.do)
In addition to using place/year level data, I also use household
level Census data
from the 2000 IPUMS data 5% sample. This allows an examination of
whether
observationally identical households live in different housing
types depending on their
community’s degree of environmentalism.
In Table Two, I report three linear probability models using the
year 2000 Census
Micro data. The sample includes all households who live in a PUMA
(the geographical
unit) whose centroid is within 30 miles of the closest central
business district (CBD).
The regression equation is reported below.
17
Housing Consumption = MSA + demographics + B*Distance to CBD +
δ*Green + U
In column (1), the dependent variable is a dummy that equals one if
the household
lives in a house that was built between 1990 and 2000. If
environmentalist communities
are slowing growth, then all else equal, in these communities
people should be less likely
to live in new single detached housing. In Table Two, I control for
the household’s
income and socio economic status and the community (the PUMA)
physical distance
from the CBD and metro area fixed effects. Holding all of these
factors constant, I focus
on the estimates of the year 2000 environmental factor. As shown in
Table Two, a
standard deviation increase in a PUMA’s environmentalism factor is
associated with a
1.3 percentage point reduction in the probability that the
household lives in new housing
a 6.3 percentage point reduction in the probability that the
household lives in a single
detached home. In the third column of Table Two, I present some
evidence that my
measure of community environmentalism correlates with behavior that
one would expect
that environmentalists would engage in. Controlling for a PUMA’s
distance to the CBD,
an extra standard deviation of community environmentalism increase
the probability that
a household head commutes using public transit, bicycle or walking
by 1.6 percentage
points. This is a large effect given that the sample average for
workers commuting by
these modes is 4.5%.
18
In this section, I contrast OLS and IV results using my California
sample from
1990 to 2004 and from 2000 to 2004. Table Three reports two
estimates of my first stage
regression. In column (1), the dependent variable is a place’s
environmental factor based
on the 1990s data. The explanatory variables include metro area
fixed effects and data
from the 1970 Census and the 1972 Party Registration data. As shown
in column (1),
richer places, and places with a larger share of Democrat
registered voters in 1972 and
places with more Peace and Freedom voters and fewer American
Independent voters are
statistically significant positive correlates in determining a
community’s
environmentalism level based on the 1990s indicator. Reading the
platform for these last
two parties indicates that these results are intuitive. The Peace
and Freedom party is a
socialist party while the American Independent party opposes
immigration, abortion and
would end the federal income tax.6 Column (1) does feature some
surprises. Neither
distance to the CBD nor community density is statistically
significant.
In column (2) of Table Three, I report a similar regression but
this time using the
data on the community’s year 2000 environmentalism factor as the
dependent variable.
Controlling for a place’s log of average household income and its
population density and
distance to the CBD, communities with more Democrats, Green Party
members and
Peace and Freedom registered voters (based on 1992 registration
data) are more likely to
be environmentalists. The results reported in Table Three should
boost one’s confidence
that lagged political registration variables do predict community
environmentalism.
6 In the instrumental variables results I report below, I do not
use lagged place per-capita income as an instrument. It is included
in Table Three to provide some descriptive evidence concerning
which places are “green”.
19
The instrument set can be thought of as identifying which places
are historically
liberal communities. Given that liberal communities are often
environmentalist
communities, this logic satisfies the first condition for being a
valid instrument set.
If the error term in equation (1) does not contain a persistent
fixed effect, then transitory
demand shocks embedded in this error term will be uncorrelated with
this instrument.
Tables Four and Five present a set of OLS and IV results for
comparable samples.
Table Four focuses on data from 1990 to 2004 and instruments for
place
environmentalism in 1990. Table Five presents OLS and IV estimates
of equation (1)
using data from 2000 to 2004 and instruments for environmentalism
in the year 2000.
Table Four’s results are based on 279 urban places in California
for which I have data on
their permits over the years 1990 to 2004, the place’s census
demographics in 1990, its
environmentalism index based on the 1990s data and I have data on
its 1972 political
party registration data.7 In Table Four, I include metropolitan
area fixed effects.
Controlling for 1990 base year attributes, both the OLS results
(see column 1) and the IV
results (see column 2) indicate that environmentalist communities
issue fewer new single
family housing permits. The results indicate large effects due to
environmentalism. A
standard deviation increase in a place’s 1990s environmentalism
reduces new single
detached permits by roughly 40%. The right columns of Table Four
show that for the
subset of California places for which I have 1972 political voting
registration data for, a
community’s environmentalism does not have a statistically
significant effect on the total
count of permits the place issues. I conclude that environmentalist
communities seek to
7 The place’s population density in 1970 is also used as an
instrument. Recall that in estimating equation (1), I control for
population density in the base year which is 1990 in Table Three
and 2000 in Table Four.
20
limit single family homes relative to multifamily homes. It is
important to note that my
regressions do include baseline controls for the place’s distance
to the CBD and its
population density.
Table Five presents the results based on the California places over
the years 2000
to 2004. In these regressions, the base year demographics are from
the year 2000 and I
use the year 2000 environmentalism indicator. In the instrumental
variables regressions
reported in columns (2) and (4), I instrument for the year 2000
environmentalism
indicator using the 1992 political party registration shares listed
in Table Two’s column
(2). The results for community environmentalism are similar to
those reported for the
1990 to 2004 sample in Table Four. Environmentalist communities are
issuing fewer
permits for new single detached homes but the magnitude of this
effect is smaller for total
permits issued. It is interesting to note that unlike in other
specifications, the place’s
share of the housing stock that is owner occupied has a negative
effect on issuing new
permits.
Equation (1) identifies the effect of environmentalism using cross
place variation
in environmentalism in the base year (either 1990 or 2000). An
alternative strategy
would be to build a panel data set and try to exploit within place
changes in
environmentalism. To estimate a first difference regression version
of equation (1)
requires an environmentalism measure that can be compared at two
points in time.
Intuitively, I need an observable measure of Berkeley’s
environmentalism in 1990 and
21
2000. To proxy for community environmentalism using a comparable
variable I use each
place’s Green Party voter registration share. In 1990, I merge on
the 1992 voter
registration share and in 2004, I merge on the year 2000 voter
registration share.
I estimate a panel regression of equation (1) with place fixed
effects using two
observations for each place. In these regressions, I use the 1990
and 2004 data on
housing permits. Including place fixed effects, I estimate two
regressions. Below, I
report standard errors in parentheses. There are 865 observations
in each regression and
the unit of analysis is a place/year.
Log(1+total single family) = place + Year 2000 -3.54*Green +
1.81*log(pop) (6.32) (.42)
Log(1+total units) = place + Year 2000 -14.30*Green + 1.83*log(pop)
(5.92) (.39)
These fixed effects results support the claim that there is less
new housing unit
growth in communities experiencing a growth in their share of
environmentalists. A one
percentage point increase in Greens is associated with a 14%
reducing in new housing
permits. Surprisingly, the single family homes regression yields an
insignificant
coefficient estimate.
Evidence from the Entire United States
I now turn to using national data from the year 2000 5% IPUMS
sample of the
Census of Population and Housing. For the nation as a whole, the
empirical challenge is
that there are no national level environmental referenda where
communities reveal their
“environmentalism” through their binding voting behavior. To get
around this problem,
I use data on each Congressional Representative’s year 2000 League
of Conservation
22
Voters (LCV) score. Geocorr is used to match each Census PUMA
identifier to a
Congressional Representative. If a Representative’s voting patterns
are positively
correlated with the district’s median voter’s preferences, then the
LCV score does
embody useful information about a community’s environmentalism
(Peltzman 1984).
By merging micro data with the voting record of one’s
Representative, I can
provide more tests of the hypothesis that “green” communities slow
new development.
In Table Six, I report four linear probability models using the
Census micro data. These
regression models are identical to the California models reported
in Table Two but in this
case I use the LCV score (based on the 106th Congressional votes on
key environmental
legislation) as my measure of community environmentalism. In
columns (1) and (2), the
dependent variable is a dummy variable that equals one if the
household lives in a new
house (age less than or equal to ten). Controlling for the
household head’s age,
household income, the head’s industry index, and the household’s
residential area’s log
population density and distance to the closest Central Business
District, I test for the role
of environmentalism in determining attributes of the types of homes
that people live in.
As shown in columns (1) and (2), when I include either state fixed
effects or metro area
fixed effects, I find a negative and statistically significant
effect of Representative
environmentalism on the probability that a constituent household is
living in new
housing. Based on the regression results reported in column (1), a
25 point increase in a
Representative’s LCV score reduces the likelihood that a household
is living in new
housing by 1.5 percentage points. The results in column (3) switch
the dependent
variable to a dummy that equals one if the household lives in a
single detached home.
Again, all else equal, I find that green constituents are less
likely to live in a single
23
detached home. Note that this regression controls for both the
PUMA’s population
density and its distance to the nearest CBD. Finally, in column (4)
I confirm again that
environmentalist constituents are more likely to not commute by
private automobile.
In results available on request, I have used the national sample to
test for the
relative importance of community environmentalism versus community
liberalism in
determining housing supply conditions. To proxy for community
liberalism, I include
include the Poole-Rosenthal (1997) measures of the Representative’s
political ideology.8
I find that my environmentalism estimates are robust to including
this liberalism measure
in regressions based on equation (1) using the household micro
data. The “liberalism
measure” itself is statistically insignificant.
The Incidence of Housing Supply Regulation in Environmentalist
Areas
Environmentalists settle in high amenity areas in California such
as Santa Cruz,
Malibu and Berkeley. In addition to self selecting into such
communities, an intended
consequence of local “good citizenship” and activist political
participation is to make the
community a more desirable place to live. This paper has documented
that all else equal
that there is less new construction in environmentalist
communities. This effect is
8 My measure of Representative ideology is the Poole and Rosenthal
two factors (see http://voteview.com/dwnomin.htm). They conduct a
principal-components factor analysis of Congressional voting
patterns to assign each representative in each Congress a point in
a two dimensional ideology space. In the political science
literature, this is the most commonly used measure of legislator
preference. It is important to note that Poole and Rosenthal (1997)
use all Congressional Roll Call votes, not simply the environmental
votes, to create their indices. They interpret the first dimension
as measuring whether a legislator is a liberal or a
conservative.
24
especially large for new single family homes. If such communities
feature high demand
and inelastic supply, then home prices will be quite high in such
communities. For
incumbent home owners, the economic returns to introducing housing
supply regulations
are highest in locations where demand is inelastic. Perhaps it is
no accident that cities
with the highest regulatory taxes (based on the results reported in
Glaeser, Gyourko and
Saks (2005)) have the highest quality of life (based on the results
reported in Gyourko
and Tracy 2001). Forward looking home owners may recognize that the
economic
returns from blocking new development are highest in such desirable
areas.
Do renters in high quality of life areas benefit from activist
environmental policy
that blocks new construction? A congestion theory of local
externalities would argue yes
while other urban experts are less sure.
“A closer look at how the growth control and environmental
coalition operates in local controversies shows that its effects
are far less benign. It has made a clear and substantial
contribution the escalation of new home prices; yet its success in
discouraging homebuilding has failed to produce important
environmental benefits for the public at large. Instead, it has
protected the environmental, social and economic advantages of
established suburban residents who live near land that could be
used for new housing.” (Frieden 1979 page 4)
This paper has focused on estimating partial equilibrium models and
has not
investigated where growth is deflected to as green communities
limit such growth.
The general equilibrium effects induced by local regulatory efforts
merits future research.
It is possible that an unintended consequence of environmentalist
communities engaging
in open space preservation is a “browning” of the metropolitan area
as growth is
deflected to more outlying areas. Consider the example of Marin
County north of the
Golden Gate Bridge in San Francisco. As William Fischel
observes,
25
“It has large amounts of open space on which development could
easily occur but does not. Tens of thousands of commuters from far
away suburbs and exurbs pass through the Marin County corridor on
U.S. Route 101 on their way to work in San Francisco. Marin’s open
space is an asset for those who live near it and it probably
provides some pleasures for those who drive through it daily. But
it also represents an enormous waste in the form of excessive
commuting and displacement of economic activities to less
productive areas (Fischel 1999 page 162).”
To answer how local growth regulation affects a metro area’s
greenness requires
estimates of where “deflected” residents (those who would have
moved to Marin) do
move to.
Conclusion
A growing consensus has emerged that housing supply and land use
regulations
have contributed to raising home prices in certain desirable cities
and communities. This
raises the question of why certain jurisdictions introduce such
regulations while other
communities do not. The home owner hypothesis makes an intuitive
claim that self
interest encourages the median voter/home owner to support policies
that limit new
construction.
A second possible explanation for why some communities limit growth
is the rise
of environmentalism. This paper has tested the claim that
environmentalist communities
are more likely to engage in slow growth. This paper’s empirical
contribution has been
to develop credible indicators of community environmentalism based
on costly revealed
preference indicators. I showed how political outcome measures
(voting on initiatives
and Congressional voting, rather than respondent surveys) can be
used to study the role
of this ideology as a determinant of community land use
patterns.
26
Using data from both California and national data, I documented
that new housing
is less likely to be constructed in environmentalist communities.
This paper’s “outcome”
research has focused on the total effect of community
environmentalism. I have not
attempted to pinpoint any specific regulation’s impact. Several
recent studies have
sought to estimate how specific regulations affect the production
of new housing. Some
attempts to look at specific environmental regulations and how they
affect housing
supply. Some of these housing supply regulations are directly tied
to environmental
issues (Zabel and Peterson 2006, Hanek and Chen 2007, Quigley and
Swoboda 2006).
The challenge in evaluating any one regulation’s impact is that
there is likely to be a
positive correlation between the adoption of various regulations.
This paper has
implicitly assumed that communities enact a bundle of regulations
and this hampers
efforts to disentangle the effects of any one specific regulation
in limiting growth.
27
References Akerlof, George A. and Rachel E. Kranton. 2000.
Economics and Identity.' Quarterly Journal of Economics. 115(3):
715-753. Baum-Snow, Nathaniel, and Matthew E. Kahn. 2005. “Effects
of Urban Rail Transit Expansions: Evidence from Sixteen Cities,
1970–2000.” In Brookings-Wharton Papers on Urban Affairs 2005,
edited by Gary Burtless and Janet Rothenburg Pack. Brookings.
Berman, Eli. 2000. Sect, Subsidy, and Sacrifice: An Economist's
View of Ultra-orthodox Jews Quarterly Journal of Economics, vol.
115, no. 3, 905-953. Fischel, William. The Homevoter Hypothesis:
How Home Values Influence Local Government Taxation, School
Finance, and Land-Use Policies. Harvard University Press. Fischel,
William. An Economic History of Zoning and a Cure for Its
Exclusionary Effects Fischel, William A. 1999. “Does the American
Way of Zoning Cause the Suburbs of Metropolitan Areas to Be Too
Spread Out?” In Governance and Opportunity in Metropolitan America,
edited by Alan Altschuler and others, pp. 151–91. Washington:
National Academies Press. Frieden, Bernard. The Environmental
Protection Hustle. MIT Press. 1979. Gerber, Elisabeth and Justin
Phillips. 2005. “Evaluating the Effects of Direct Democracy on
Public Policy: California’s Urban Growth Boundaries” American
Politics Research, 33(2) 310-330. Glaeser, Edward L, Joseph
Gyourko, and Raven Saks. 2005. Why is Manhattan So Expensive?
Regulation and the Rise in House Prices.” Journal of Law and
Economics. 48(2): 331-370. Gyourko, Joseph, Christopher Mayer, and
Todd Sinai. Superstar Cities, 2006 NBER 12355 Gyourko, Joseph,
Albert Saiz, and Anita A. Summers, “A New Measure of the Local
Regulatory Environment for Housing Markets: The Wharton Residential
Land Use Regulatory Index”, Urban Studies, forthcoming. Hanak,
Ellen and Ada Chen, 2007, “Wet Growth: Effects of Water Policies on
Land Use in the American West,” Journal of Regional Science, Vol.
47(2).
28
Hanak, Ellen and Margaret K. Browne, 2006. “Linking Housing Growth
to Water Supply: New Planning Frontiers in the American West,”
Journal of the American Planning Association, Vol. 72(2). Helsley,
Robert W. and William C. Strange, 1995. “Strategic Growth
Controls,” Regional Science and Urban Economics, Vol. 25: 435-460.
Iannaccone, Laurence R. 1992. Sacrifice and Stigma: Reducing
Free-Riding in Cults, Communes, and Other Collectives Journal of
Political Economy, vol. 100, no. 2, 271- 291 Ihlanfeldt, Keith R.
and Timothy M. Shaughnessy, 2004. “An Empirical Investigation of
the Effects of Impact Fees on Housing and Land Markets,” Regional
Science and Urban Economics, Vol. 34(6): 639-61 Kahn, Matthew.
2007. Do Greens Drive Hummers or Hybrids? Environmental Ideology as
a Determinant of Consumer Choice, Journal of Environmental
Economics and Management. September. Kahn, Matthew, and John
Matsusaka. 1997. “Demand for Environmental Goods: Evidence from
Voting Patterns on California Initiatives.” Journal of Law and
Economics 40, 1: 137–73. Kahn, Matthew and Eric Morris. 2008.
Walking the Walk: Do Green Beliefs Translate Into Green Travel
Behavior? UCLA Working Paper. Landis, John D. 1992. “Do Growth
Controls Work? A New Assessment,” Journal of the American Planning
Association 58 (4): 489–508. Landis, John D. 2006. Growth
Management Revisited. Journal of the American Planning Association.
72(4) 411-430. Levine, Ned, 1999. “The Effects of Local Growth
Controls on Regional Housing Production and Population
Redistribution in California,” Urban Studies, Vol 36(12):
2047-2068. Mayer, Chris. and Tsur. Somerville (2000) “Land Use
Regulations and New Construction,” Regional Science and Urban
Economics, 30, 639-662. Peltzman, Sam. Constituent interest and
congressional voting, Journ. Law and Econ. 27(1984) 181-210.
29
Pollakowski, Henry O., and Susan M. Wachter. 1990. “The Effects of
Land-Use Constraints on Housing Prices,” Land Economics 66 (3):
315–324. Poole, Keith, and Howard Rosenthal. (1997). Congress: A
Political-Economic History of Roll Call Voting. Oxford University
Press. Quigley, John M., and Stephen Raphael, 2004. “Regulation and
the High Cost of Housing in California,” American Economic Review.
Quigley, John, Steven Raphael and Larry A. Rosenthal Local Land-use
Controls and Demographic Outcomes in a Booming Economy Urban
Studies, Vol. 41, No. 2, 000–000, February 2004 Quigley, John M.,
and Larry A. Rosenthal. 2005. “The Effects of Land Use Regulation
on the Price of Housing: What Do We Know? What Can We Learn?”
Cityscape 8 (1): 69–138. Quigley, J.M. and A.M. Swoboda, “The Urban
Impacts of the Endangered Species Act: A General Equilibrium
Analysis,” Journal of Urban Economics (2006). Schill, Michael.
2005. “Regulations and Housing Development: What We Know” Cityscape
8 (1): 5-20. Zabel, Jeffrey. and R. Paterson (2006a) “The Effects
of Critical Habitat Designation on Housing Supply: An Analysis of
California Housing Construction Activity,” Journal of Regional
Science, 46 (2006): 67-95. Zabel, Jeffrey. and R. Paterson (2006b)
“The Impact of Critical Habitat Designation on Housing Markets in
California.
Table One: The Determinants of New Housing Permits Issues by
California Cities Between 1990 to 2004
Column (1) (2) (3) (4)
Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std.
Err.
Environmentalism Indicator -0.3445 0.0612 -0.2803 0.0659 -0.2052
0.0476 -0.1727 0.0710 % Owner Occupied -0.4742 0.5836 -0.3429
0.5248 -1.3662 0.5750 -1.0223 0.5286 log(Place Population) 1.0877
0.0267 1.0622 0.0464 1.2322 0.0258 1.1321 0.0483 log(Distance to
CBD) 0.1640 0.0985 0.2341 0.1064 0.1730 0.0973 0.1982 0.1026
log(population density) -0.3904 0.1059 -0.2581 0.0743 -0.4436
0.1086 -0.2713 0.0733 Log(average household income) 1.2825 0.2520
0.9435 0.1838 1.3922 0.2713 0.8509 0.1830 Constant -19.8717 3.0301
-17.4466 2.4939 -20.9316 3.0468 -15.7076 2.5014
Observations 5113 5113 5113 5113 R2 0.7420 0.5530 0.7830 0.5490
count of places 348 348 348 348 Fixed Effects county county county
county Unit of analysis Place/year Place/year Place/year Place/year
weighted by place population Yes No Yes No Year Fixed Effects Yes
Yes Yes Yes
This table reports estimates of equation (1) in the text. Place
level attributes are based on 1990 values. The sample includes
California places that are located within 30 miles of a
Metropolitan area's Central Business District. In columns (1) and
(3), the dependent variable is the log(1+Permit Count). "Share of
Single Family" is defined as total annual single family housing
permits divided by total housing permits. Environmentalism
indicator is based on the 1990s data namely the place's share of
Green Party voters in 1992 and the share of the place that voted in
favor of Proposition 185 in 1994. Standard errors are clustered by
place. The Environmental indicator has a mean of zero and a
standard deviation equal to one. april15_08.do
Single Family Permits Total Family Permits
Table Two: California Level Findings Based on Year 2000 Household
Level Data
Column (1) (2) (3) Dependent Variable Green Commuter
Coef. Std. Err. Coef. Std. Err. Coef. Std. Err.
Environmentalism Indicator -0.0130 0.0071 -0.0627 0.0167 0.0155
0.0052 log(Household Income) 0.0063 0.0009 0.0560 0.0017 -0.0014
0.0003 Age -0.0014 0.0001 0.0053 0.0002 -0.0010 0.0001 Duncan
Socioeconomic Index 0.0007 0.0001 0.0013 0.0001 0.0002 0.0000
log(distance to closest Metro Area CBD) 0.0433 0.0106 0.0491 0.0196
-0.0266 0.0065 Constant -0.3193 0.1038 -0.8240 0.1866 0.3601 0.0643
Mean of Dependent Variable 0.1200 0.5560 0.0450
Unit of Analysis Household Household Household Fixed Effect MSA MSA
MSA Observations 539563 539563 539563 R2 0.047 0.123 0.046
The sample includes all California households living in a PUMA
whose centroid is within thirty miles of a CBD. This table reports
household level linear probability models. "New Housing" is a dummy
variable that equals one if the household lives in a dwelling that
was built between the years 1990 and 2000. "Single Detached Home"
is a dummy that equals one if the person lives in a single detached
home. "Green Commuter" is a dummy variable that equals one if the
household head commutes to work using public transit, a bicycle, or
walks. equals one if the person lives in a single detached home.
The environmentalism indicator is based on the year 2000 data that
includes the PUMA's share of Green Party voters, the share of votes
in favor of Proposition 12 and 13 in the year 2000 and the PUMA's
share of hybrid vehicles. Standard errors are clustered by
PUMA.
oct31.do
Table Three: Political and Demographic Correlates of Community
Environmentalism
Dependent Variable = Green Factor for California Places
1990 2000 (1) (2)
Coef. Std. Err. Coef. Std. Err.
log(distance to CBD) -0.0836 0.0632 log(Population Density in 1970)
0.0421 0.0339 log(average household income in 1970) -0.1187 0.1707
% Democrat in 1972 -0.3774 0.3842 % American Independent in 1972
-21.4846 7.7437 % Peace and Freedom in 1972 8.4872 3.7357 %
Miscellaneous in 1972 -0.8335 3.0554 % Declined in 1972 11.6264
2.3129 Constant 1.4666 2.3187
log(distance to CBD) 0.0793 0.0508 log(Population Density in 1990)
0.0808 0.0212 log(average household income in 1990) 0.6159 0.1256 %
Democrat in 1992 3.7201 0.3722 % American Independent in 1992
-21.0130 7.2568 % Peace and Freedom in 1992 36.3030 12.7599 %
Libertarian in 1992 12.4752 10.4796 % Green Party in 1992 28.6558
3.9135 % Declined in 1992 2.1890 1.1686 Constant -10.1386 1.7588
Metropolitan Area Fixed Effects Yes Yes
Observations 283 568 R2 0.32 0.5030
The omitted category is a place's share of Republican Party
registered voters.
Table Four: The Determinants of New Housing Permits Issues by
California Cities Between 1990 to 2004
Column (1) (2) (3) (4) Estimation Method OLS IV OLS IV
Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std.
Err.
Environmentalism Indicator -0.3754 0.0859 -0.5645 0.1782 -0.0864
0.0842 -0.0123 0.1748 % Owner Occupied 0.0253 0.6412 -0.4254 0.7238
-0.8206 0.6080 -0.6441 0.6820 log(Place Population) 1.1055 0.0296
1.1011 0.0298 1.2383 0.0273 1.2401 0.0270 log(Distance to CBD)
0.2433 0.1035 0.2291 0.1021 0.1320 0.1037 0.1376 0.1031
log(population density) -0.4806 0.1253 -0.4552 0.1214 -0.4164
0.1231 -0.4263 0.1191 Log(average household income) 1.0024 0.2967
1.1724 0.3264 1.2594 0.3075 1.1928 0.3454 Constant -14.9750 3.4855
-16.6460 3.6735 -17.7431 3.3295 -17.0885 3.6257
Observations 4169 4169 4169 4169 R2 0.7530 0.7510 0.7940 0.7940
count of places 279 279 279 279 Fixed Effects year year year year
Fixed Effects MSA MSA MSA MSA Unit of analysis Place/year
Place/year Place/year Place/year This table reports estimates of
equation (1) in the text. Place level attributes are based on 1990
values. The sample includes California places that are located
within 30 miles of a Metropolitan area's Central Business District.
In columns (1) and (3), the dependent variable is the log(1+Permit
Count). "Share of Single Family" is defined as total annual single
family housing permits divided by total housing permits.
Environmentalism indicator is based on the 1990s data namely the
place's share of Green Party voters in 1992 and the share of the
place that voted in favor of Proposition 185 in 1994. Standard
errors are clustered by place. The omitted category is a place not
in Los Angeles, San Francisco or San Diego. The instruments include
a place's density in 1970 and its political party registration
shares in 1972. weighted april13_08.do
Single Family Permits Total Permits
Table Five: The Determinants of New Housing Permits Issues by
California Cities Between 2000 to 2004 Column (1) (2) (3) (4)
Estimation Method OLS IV OLS IV
Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std.
Err.
Environmentalism Indicator -0.1905 0.1050 -0.4124 0.1278 -0.1182
0.1015 -0.2080 0.1226 % Owner Occupied -0.5847 0.6664 -1.0572
0.7229 -1.3331 0.6593 -1.5243 0.6912 log(Place Population) 1.0505
0.0558 1.0237 0.0566 1.1662 0.0563 1.1554 0.0568 log(Distance to
CBD) 0.0764 0.1197 0.0920 0.1215 0.0735 0.1100 0.0798 0.1110
log(population density) -0.3364 0.0872 -0.3151 0.0894 -0.3296
0.0849 -0.3210 0.0857 Log(average household income) 0.9725 0.2556
1.2302 0.2876 1.0256 0.2411 1.1299 0.2632 Constant -15.4718 3.1942
-18.2436 3.4964 -16.5221 3.0071 -17.6397 3.1979
Observations 1734 1734 1734 1734 R2 0.5680 0.5640 0.5810 0.5800
count of places 348 348 348 348 Fixed Effects county county county
county Fixed Effects Year Year Year Year Unit of analysis
Place/year Place/year Place/year Place/year Weighted by Population
No No No No
This table reports estimates of equation (1) in the text. Place
level attributes are based on 1990 values. The sample includes
California places that are located within 30 miles of a
Metropolitan area's Central Business District. The dependent
variable is the log(1+Permit Count). Environmentalism indicator is
based on the 2000s data namely the place's share of Green Party
voters in 2000 and the share of the place that voted in favor of
Proposition 12 and 13 in 2000 and the share of hybrid vehicles in
the place. The environmentalism indicator from the 2000s is
instrumented for using the place's 1992 political party
registration shares and the share of the place's voters who voted
in favor Proposition 185 in 1994.
april9_08.do
Single Family Permits Total Permits
Table Six: National Level Findings Based on Year 2000 Household
Level Data
Column (1) (2) (3) (4)
Dependent Variable
Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std.
Err.
Representative League of Conservation Voters Score in 106th -0.0006
0.0001 -0.0003 0.0001 -0.0008 0.0001 0.0004 0.0000 log(Household
Income) 0.0136 0.0005 0.0113 0.0006 0.0554 0.0009 -0.0012 0.0003
Age -0.0021 0.0001 -0.0019 0.0001 0.0053 0.0001 -0.0010 0.0000
Duncan Socioeconomic Index 0.0008 0.0000 0.0008 0.0000 0.0016
0.0000 0.0004 0.0000 log(PUMA Population Density) -0.0079 0.0025
-0.0472 0.0050 -0.0459 0.0041 0.0100 0.0016 log(distance to closest
Metro Area CBD) 0.0405 0.0047 0.0572 0.0067 0.0302 0.0061 -0.0140
0.0026 Constant -0.2030 0.0612 -0.0359 0.0990 -0.1939 0.0815 0.1305
0.0326 Mean of Dependent Variable 0.1680 0.1630 0.5910 0.0460
Unit of Analysis Household Household Household Household fixed
effect state metro area state state Data Sample 2000 2000 2000 2000
Observations 509457 362597 509457 509457 R2 0.062 0.089 0.118
0.063
This table reports household level linear probability models. "New
Housing" is a dummy variable that equals one if the household lives
in a dwelling that was built between the years 1990 and 2000.
"Single Detached Home" is a dummy that equals one if the person
lives in a single detached home. "Green Commuter" is a dummy
variable that equals one if the household head commutes to work
using public transit, a bicycle, or walks.
New Housing New Housing Single Detached Home Green Commuter