valuing the conversion of urban greenspace
TRANSCRIPT
1
Valuing the Conversion of Urban Greenspace
Grace W. Bucchianeri, Kevin C. Gillen, and Susan M. Wachter
Prudential Financial Inc., University of Pennsylvania
June 2012
Abstract
The classic models of urban form typically place open (green) space and/or agricultural land at the
bottom of the hierarchy of urban land uses. However, this framework limits the value of land to only
what economic product can be produced on it, and ignores any spillover effects that open space may
have on nearby property values as an aesthetic and environmental amenity. Moreover, these models
typically infer that the value-minimizing location of open space would be in dense, urban areas where
centrally located land is at its scarcest and hence more valuable if developed. The authors examine if
the conversion of vacant land and/or land with abandoned properties to maintained urban green space
and gardens is valued by local residents, as measured by any spillover effects on nearby residential
property values. Using a time series of home sales in Philadelphia, a city with a significant inventory of
vacant/abandoned parcels, the authors find that the local presence of a vacant property is associated
with nearby homes being worth an average of nearly 16% less than comparable dwellings in these
neighborhoods. Following conversion of a vacant lot to maintained green space, nearby homes are
estimated to command a premium of 2% to 5%; a total gain in value of 18% to 21%. Affected
households experienced a median gain of $34,468 in housing wealth after five years. From a fiscal
perspective, it is estimated that every dollar spent to clean and green a vacant lot yields additional
property tax revenues of $7.43. These results, which build on past research demonstrating the positive
effects of conversion on just adjacent properties, suggest that the traditional models of urban form
underestimate the economic value of open space by ignoring its value as a local amenity, and that urban
areas can benefit from the conversion of vacant land to maintained green space.
2
I. Introduction
There is a substantial literature on “place-based” amenities/dis-amenities that is cross-disciplinary in
nature, and has a strong focus on green space:
• Bowes, David R. “Identifying the Impacts of Rail Transit Stations on Residential Property Values,”
Journal of Urban Economics, pp. 1-25. 2001.
• Hammer, Thomas R, Robert E. Coughlin, and Edward T. Horn IV. “Research Report: The Effect of
a Large Park on Real Estate Value.” Journal of the American Institute of Planners, 274-277. July,
1974.
• Correll, Mark R., Jane H. Lillydahl, and Larry D. Singell. “The Effect of Greenbelts on Residential
Property Values: Some Findings on the Political Economy of Open Space.” Land Economics, Vol.
54 (2), 207- 217. 1978.
• Wachter, Susan M. and Grace Wong, 2008. "What Is a Tree Worth? Green-City Strategies,
Signaling and Housing Prices," Real Estate Economics, American Real Estate and Urban
Economics Association, vol. 36(2), 213-239, 06.
• Wachter, Susan M., Kevin C. Gillen & Carolyn R. Brown, "Green investment strategies: a positive
force in cities," Communities and Banking, Federal Reserve Bank of Boston, 2008 (2), 24-27.
• Ioan Voicu and Vicki Been, 2008. “The Effect of Community Gardens on Neighboring Property
Values,” Real Estate Economics, American Real Estate and Urban Economics Association, vol.
36(2), 241-283.
• Wei Lei and Jean-Daniel Saphores, 2011. “A Spatial Hedonic Analysis of the Value of Urban Land Cover in the Multifamily Housing Market in Los Angeles, CA,” Urban Studies, 1-19.
In the specific case of urban green space as a place-based amenity, the last two papers are the most relevant to this research paper. Voicu and Been (2008) find that community gardens in New York City “have on average, significant positive effects on surrounding property values, and that those effects are driven by the poorest of host neighborhoods (where a garden raises neighboring property values by as much as 9.4 percentage points within five years of the garden’s opening).1” Lei and Saphores examine green spaces near multifamily properties in Los Angeles and find that “increasing the tree canopy cover in its vicinity (200 meters outward from the parcel boundary) would enhance its value”2. However, most of this literature is “static” in nature. It identifies and measures the monetary value (or
discount) of the presence a local amenity/dis-amenity, and examines if and how this value (discount)
may vary with distance. Less research has examined such place-based developments in a “dynamic”
context; i.e., how has the value (discount) of proximity to a site changed in response to the reallocations
1 Voicu and Been (2008), page 227.
2 Lei and Saphores (2011), page 16.
3
of the use of a given site. That is: is the value (discount) of proximity greater, less than or unchanged
following the repurposing of a site to a different use?
Moreover, many U.S. cities increasingly find themselves with significant amounts of abandoned/vacant properties. These parcels represent deadweight on the city’s tax base and accounting ledger. However, the conditions that facilitated their former use as commercial or industrial sites are unlikely to return anytime soon. So, any research that can parameterize the economic and fiscal consequences of repurposing these sites to alternative uses can inform local public policy. 3 This project proposes to examine the now significantly larger data set of City-supported PHS cleaned and
greened land (~5,000 parcels of cleaned and greened or “stabilized” land) and ascertain the pre- and
post-intervention value of these green spaces on adjacent real estate and land values. Additionally, it
seeks to compute the relative costs and benefits of the greening work documented in the data set.
There are also some specific analyses that the Sponsor is requesting that includes examining how these
results vary by neighborhood and by geographic concentration of site(s).
The project aims to examine what effect, if any, there is from converting an abandoned site to
greenspace, as well as examining the economic implications of allowing it to remain vacant space.
Additionally, it measures the relative costs and benefits of greening urban land. Such results have policy
implications for urban jurisdictions with significant amounts of vacant land in their tax bases.
We believe that there are two major implications of the research:
In general, the answers facilitate more informed cost/benefit analysis in land use planning,
particularly for cities with high incidences of vacant land.
In the specific case of conversion to urban greenspace, the answers could suggest a more
nuanced version of the traditional monocentric city model, which places agricultural/vacant land
at the bottom of the hierarchy of land uses.
II. Motivation
The City of Philadelphia is one of the U.S.’s ten largest cities, but has a significant surplus of formerly
developed land: approximately 40,000 parcels, of which 12,000 are owned by the City itself4. The
abandonment of these developed sites and their reversion to vacant land is largely a result of the
deindustrialization and depopulation that has occurred in Philadelphia during the postwar era. In 1950,
Philadelphia was the U.S.’s fourth-largest city with a population of 2.1 million, of which 36 percent were
3 Two earlier studies have examined this issue in Philadelphia. The first, by Wachter (2005), found that vacant land improvements in the New
Kensington neighborhood of Philadelphia resulted in surrounding housing values increasing by as much as 30%, while new tree plantings increase surrounding housing values by approximately 10%. The second study, by Wachter, Gillen and Brown (2008) found that homes next to an unmanaged vacant lot lost up to 20 percent of their value, but that when strategies similar to those used by the Vacant Land Stabilization and Community LandCare programs were implemented, adjacent homes gain up to 17 percent in value. These studies used different empirical methods and a different and more limited period of time than what is used by this study, which can explain that the impact measure by these studies is higher given that house prices where then lower overall. 4 Source: City of Philadelphia Redevelopment Authority (PRA).
4
employed in manufacturing. By 2010, it was the U.S.’s fifth-largest city with a population of 1.5 million,
of which only four percent was still employed in manufacturing5.
This systematic dis-investment by industry has left a significant number of vacant and abandoned
parcels in their wake. Currently, these formerly active sites are postindustrial, mostly vacant, separated
from other uses by infrastructure and both economically and socially underutilized.
But, since the late 1990s, the Pennsylvania Horticultural Society (PHS) and the City of Philadelphia have
actively endeavored to convert many formerly industrial and/or vacant parcels to urban greenspace.
This paper examines the changes in the movements in property values near and far from these
converted lots in order to attempt to identify what effect, if any, the conversion of these lots had on
neighborhood property values.
III. Data and Empirical Strategy
We combined property-level data on converted parcels with property-level data on homes sales in
Philadelphia. The data span the period 1990-2011.
Data on home sales in Philadelphia were obtained from Philadelphia’s Department of Records and
Philadelphia’s Board of Revision of Taxes, which is the City’s Assessing Authority. Home sales were geo-
coded to assign a unique latitude and longitude, and the distance to the nearest converted site was
computed with the assistance of ArcView GIS software. Attached to each observation is the dwelling’s
date of sale and physical characteristics, such as lot size, structure square footage, age of structure,
number of stories and physical condition. The dates of sale span the years 1990-2011.
Table 1. Summary Statistics
Price # Greened Lots<=1/4 mile
Distance to Greened Lot (mi.)
Pct. of parcels in block group that are vacant6
Distance to Vacant Lot (mi.)
Min $1,900 0 0 0.0% 0.0036
25% Quartile $19,000 0 0.11 13.0% 0.0155
Median $39,900 0 0.32 18.0% 0.0323
Mean $49,710 12.8 0.45 20.6% 0.0496
Std. Dev. $46,249 31.9 0.39 11.5% 0.0602
75% Quartile $65,000 10 0.75 26.0% 0.0553
Max $741,400 423 2.85 82.0% 0.3205
N (# sales) 127,514 127,514 127,514 127,514 127,514
5 Source: U.S. Census, U.S. Bureau of Labor Statistics
6 Note that this statistic is across home sales, not across block groups.
5
These summary statistics are only for neighborhoods where PHS has greened lots, where we define “neighborhood” as a Zip Code with at least one greened lot. Note that “percent vacant” has greater variation than “distance to nearest vacant”. Since the scope of this analysis is limited to just PHS neighborhoods, almost all homes are near at least one vacant lot, but the spatial concentration of vacant lots does exhibit relatively greater variation. Data on the converted, “greened” sites were provided by the PA Horticultural Society. Attached to each
site is its address, size (square feet), date of selection, date of conversion, square footage, number of
parcels, whether or not trees were planted, and what date trees were planted (if any). The data spans
the years 1996-2010, with 4,732 such sites. The following chart shows the growth in the number of
converted parcels over time. As the chart indicates, the bulk of conversions occurred in the mid-decade
years of 2004-2008.:
Chart 1.
The following map shows the location of the greened parcels. Unsurprisingly, the parcels are highly
concentrated in the very distressed neighborhoods of North Philadelphia, West Philadelphia and Point
Breeze.
3
89157
116
381
810
523
749707
739
96
327
0
100
200
300
400
500
600
700
800
900
1996 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
# Greened Parcels by Year
6
Map 1.
Since we are interested in how the repurposing of land to another use may be capitalized into nearby
property values, the research question must take into account not only space (proximity to the site) but
also change over time (previous use of the site vs. current use of the site). As such, the traditional
hedonic regression with an added distance-to-site variable is insufficient.
Instead, we utilize an empirical strategy that Galster, Tatian and Smith (1999) first used to identify how
the presence of households receiving Section 8 vouchers affected the value of neighboring homes. The
authors define four variables that measure the fixed and slope effects of proximity to a site, both before
and after an event. In their paper, the variables measured the level and trend in house values ex ante
the arrival of a Section 8 household and ex post such an arrival.
In our case, we define these variables to measure the level and trend in house values before a
vacant/abandoned site is converted to green space, and after the site is converted to green space. We
first define the following simple variables:
dist_green=distance to nearest converted site (in miles)
7
sale_year=year that the dwelling transacted
green_year=year that the vacant/abandoned site was converted to greenspace
time_trend=0,1,2,…,T denoting the year and quarter that a dwelling transacted
Then, we define the following “event study” variables:
In words: pre_green and pre_greent measure the level and trend in nearby house prices prior to the
conversion of a greened lot, while post_green and post_greent measure the level and trend in nearby
house prices after the conversion of a greened lot. The relative baseline against which these effects are
measured are house price movements in these neighborhoods, for homes which are further than a ¼
mile from converted lots.
A critical control variable in the regression is vacancy rate in each Census Tract, computed as the number of vacant parcels divided by the total number of parcels in each Tract. This variable represents the percentage of vacant lots that are never greened and remain vacant throughout the study period of 1990-2011. The following map shows the concentration of these vacant lots in the city.
8
Map 2.
As can be observed, vacant lots are disproportionately concentrated in the low-income, lower-priced
neighborhoods of North Philly, West Philly, Kensington-Frankford and Point Breeze. These are, un-
coincidentally, the same neighborhoods where PHS has concentrated its activities in greening vacant
land.
Although our home sales data spans 21 years, the vacancy data is simply a snapshot in time: as-of 2011. This could potentially cause econometric problems if the number and location of vacant lots varied over time, since we have no way of controlling for this (the year that each lot became vacant is not available). However, it is well known that Philadelphia’s vacancy problem is primarily structural (as opposed to cyclical) in nature, and is related to the de-industrialization of many former manufacturing centers in the 1960s-1970s. Hence, since most of this abandonment occurred prior to the study period that is the scope of this paper, then it would seem reasonable to treat the population of vacant lots as the constant that it is. However, to confirm this, we collected data on the city’s vacancy rate from the decennial U.S. Census and examined its changes over time. The results are shown in the following chart:
9
Chart 2.
As can be observed, there are only minor fluctuations in the vacancy rate during the 1990-2010 scope of this study, so we proceed with treating this variable as a constant over time, controlling only for its variation over space in subsequent regressions.
IV. Empirical Results
The following table shows the results of estimating a hybrid hedonic regression of the natural log of
house prices on a vector of controls (size, age, density, distance from CBD, season, and year and quarter
of sale) plus the event study variables7:
7 The actual list of control variables is quite extensive. They are available from the author upon request.
0.00%
2.00%
4.00%
6.00%
8.00%
10.00%
12.00%
1970 1980 1990 2000 2010
Philadelphia Vacancy Rate 1970-2010
10
Regression Results I
N=128,175, Adj. R-Sq.=0.47
Variable Parm. Est. Parm. Est. t Value
pre_green house price level before conversion -0.17301 -16.11
pre_greent house price trend before conversion -0.00078375 -0.84
post_green house price level after conversion 0.02554 1.75
post_greent house price trend after conversion 0.00513 1.84
vac_rate % of lots in block group are vacant -0.65071 -27.97
Physical Characteristics? Yes
Location Characteristics? Yes
Time Trend? Yes
The results are interpreted as follows:
• Prior to conversion, proximity to an abandoned/vacant site was associated with a dwelling being
valued 15.9%8 less than average home value in these neighborhoods.
• The low t value of -0.84 for the variable pre_greent is statistically insignificant. This suggests
that, prior to conversion, there does not appear to be any meaningful difference in house price
trends for homes that are near to converted lots versus homes that are further from them.
• Immediately following conversion to green space, homes within a ¼ mile of these sites are now
valued at an average of 2.6% more than the neighborhood average price.
• In the years following conversion, homes within a ¼ mile of the site experienced an additional
annual appreciation rate of 0.5% per year, relative to the average in the rest of the
neighborhood. However, this result is only weakly significant.
Thus, the results suggest that converting a vacant lot to greenspace removes a significant dis-amenity
from the neighborhood and adds a modest amenity. House values near these lots went from being
valued nearly 16% less than the neighborhood average to 2.6% more than the neighborhood average: a
total increase in value of 18.5%.
To formally test whether the post-greening variables are significantly different from the pre-greening
variables, we estimated an F-test that tests the null hypothesis that the value of the coefficients of
greening are not statistically different from each other. The results are given in the following tables:
8 Because the dummy variables measuring the fixed effects have values of 0 or 1, the coefficient must be
exponentialized and have one subtracted to obtain the percent effect. Exp(-0.-0.17301)-1=-.159.
11
Test of Ho: pre_green=post_green
Source DF Mean Square F Value Pr > F
Numerator 1 34.07132 69.69 <.0001
Denominator 128099 0.48889
Test of Ho: pre_greent=post_greent
Source DF Mean Square F Value Pr > F
Numerator 1 1.93267 3.95 0.0468
Denominator 128099 0.48889
The F-test for the null hypothesis that the pre-greening and post-greening fixed effects are not different
from each other is rejected at the 1% level. This strongly indicates that the change in the level in house
prices is statistically meaningful, and is unlikely to be a random artifact of the data’s structure. The
same test for the pre-greening trend effects is also rejected, but at the 5% level. This also indicates that
there was a likely shift in house price trends following the conversion of the vacant lots, but this result is
not as strong as the one for the fixed effects.
Since it may be very likely that the results are also sensitive to the total number of greened lots that are
converted, we re-run the same hybrid hedonic regression after adding the number of greened lots
within ¼ mile at the time of sale to the specification. The results are given in the following table:
Regression Results II
N=128,166, Adj. R-Sq.=0.56
Variable Definition Parm. Est. t Value
pre_green house price level before conversion -0.17577 -16.38
pre_greent house price trend before conversion 0.00074532 0.79
post_green house price level after conversion 0.04869 3.31
post_greent house price trend after conversion 0.00558 2.00
num_greened # of converted lots <1/4 mile at time of sale 0.0094779 12.55
vac_rate % of lots in block group are vacant -0.60962 -25.97
Physical Characteristics? Yes
Location Characteristics? Yes
Time Trend? Yes
All of the greening variables with the exception of pre_greent are statistically significant at the 5% level.
The results are interpreted as follows:
12
• Prior to conversion, dwellings within a ¼ mile of an abandoned/vacant site had a 16.1% lower
value than comparable homes not near abandoned/vacant sites.
• Prior to conversion, house prices near future converted sites had an appreciation rate that was
not meaningfully different than the average appreciation rate in these neighborhoods.
• Immediately following conversion to green space, homes within a ¼ mile increased in value by
21%, making them worth 5% more than the average home in the subject neighborhoods. In the
years following conversion, homes within a ¼ mile of the site experienced an additional annual
appreciation rate of .9% per year, relative to comparable homes that are not near such sites.
• Each additional greened lot within a ¼ mile is associated with an additional increase in home
values of 0.9%. So, converting one vacant lot is associated with nearby homes being worth 5.9%
(=5%+0.9%) more, on average, than homes not near greened lots, with each additional greened
lot adding another 0.9% to this 5.9% premium.
Thus, the results not only provide further empirical support to the notion that converting vacant lots to
greened ones adds to the value of nearby homes, but that this value increases with the total number of
greened lots that are near to these homes. For example, if five vacant lots are converted to greenspace,
the total associated increase in nearby house values is 25.8% immediately following the conversion.
Again, to formally test for the validity of these results, we estimate the same F-tests for the equality of
the pre- and post-greening coefficients. The results are given in the following tables:
Test of Ho: pre_green=post_green
Source DF Mean Square F Value Pr > F
Numerator 1 43.21621 88.5 <.0001
Denominator 128098 0.48829
Test of Ho: pre_greent=post_greent
Source DF Mean Square F Value Pr > F
Numerator 1 1.28723 2.64 0.1045
Denominator 128098 0.48829
The test for the equality of the pre-greening fixed effects is rejected at the 1% level. The test for the
equality of the pre-greening trend effects fails to be rejected, although it is very close to being rejected,
at the 10% level, as evidenced by the p-value of 0.1045. This suggests that, once the number of nearby
13
greened lots is properly controlled for, house price levels near these lots are statistically different pre-
and post-greening, but that house price trends are not.
To place these numbers in context, we apply them to the average house values during the 1990-2011
study period. We assume that 2007 is the year of conversion, since that was the median year of
conversion in the data. The following chart compares the price trajectory of the average home in the
greened neighborhoods before and after conversion:
Chart 3.
Starting in 1990, prior to conversion, a home near a vacant lot was worth $4,925 less than the average
North or West Philadelphia home; a meaningful difference in a neighborhood where the average house
price in 1990 was only $31,000. However, immediately following conversion to a greened lot in 2007,
the typical home near such a lot was worth $2,411 more than the average home in these
neighborhoods. In the five years after conversion, this average difference in house values grew by
$3,011, due to the slightly higher appreciation rate that accrued to these homes following the greening
event.
We next show how these results vary if we adjust for the number of converted lots that are proximate
to existing homes. The following table gives the percentiles of the distribution for the variable
“num_greened” for those dwellings that are within a ¼ mile of at least one greened lot.
$26,075
$33,411
$36,422
$0
$5,000
$10,000
$15,000
$20,000
$25,000
$30,000
$35,000
$40,000
$Value prior to conversion $Value immediately following conversion
$Value 5 years after conversion
$ Effect of Converting a Vacant Lotto Greenspace
14
Table 2. Percentiles of Num_Greened
Percentile Num_Greened
100% Max 423
75% 47
Mean 36
50% Median 16
25% 6
0% Min 1
The percentiles state how many converted lots are within a given percent of the distribution, for those
dwellings that are near a converted lot. For example, 25% of homes near a greened lot have 6 greened
lots in their vicinity or less. 75% of homes near a greened lot have 47 greened lots in their vicinity or
less. The 50th percentile is 16, indicating that, for those homes that are proximate to a greened lot, the
median number of greened lots in their vicinity is 16.
Applying the value of the coefficient for num_greened from Regression II to this median number of
greened lots gives a house price index that represents the typical (i.e., median) change in value for
homes in neighborhoods with greened lots. The results are shown in the following chart.
Chart 4.
$73,238
$113,857 $118,981
$0
$20,000
$40,000
$60,000
$80,000
$100,000
$120,000
$140,000
$Value prior to conversion $Value immediately following conversion
$Value 5 years after conversion
$ Effect of Converting Sixteen Vacant Lostto Greenspace
15
The dark green line shows the path of house price appreciation for the median dwelling, assuming the
timing of conversion was in the median year of 2007 and the 16 greened lots were all converted in that
year. Immediately following conversion, the dwelling increases in value from $73,238 to $113,857: a
gain in value of $40,620. After five year, this value has grown to $118,981. Prior to conversion, this
dwelling was worth an average of $13,039 less than the typical home in these neighborhoods. After
conversion, the dwelling is now worth an average of $20,670 more than the typical home in these
neighborhoods. Considering that the typical home in the subject neighborhoods of this study is worth
less than $100,000, these numbers represent some meaningful gains in housing wealth to the owners of
these dwellings.
The results presented so far have only measured the effects of greened lots within a ¼ mile. We now
examine how these effects vary with distance. To do this, we re-estimate Regression I iteratively,
varying the definition of the event study variables in increments of 200 feet in each iteration. The
results are shown in the following table. For the sake of parsimony, we focus only on the fixed effects,
pre_green and post_green, since they have consistently been the more statistically and economically
significant than the trend effects, pre_greent and post_greent.
Regression Results III
Distance to Converted Lot (ft) Pre_Green t value Post_Green t-value
Pre_Green (%)
Post_Green (%)
200 -0.1045 -4.47 0.0101 0.33 -9.92% 1.02%
400 -0.1643 -10.47 0.0574 2.78 -15.15% 5.90%
600 -0.1840 -13.94 0.0831 4.77 -16.81% 8.66%
800 -0.1821 -15.19 0.0693 4.34 -16.65% 7.18%
1,000 -0.1826 -16.00 0.0526 3.44 -16.69% 5.40%
1,200 -0.1800 -16.39 0.0384 2.59 -16.47% 3.91%
1,400 -0.1698 -15.90 0.0150 1.03 -15.61% 1.51%
1,600 -0.1671 -15.98 0.0015 0.1 -15.39% 0.15%
1,800 -0.1728 -16.77 -0.0044 -0.31 -15.87% -0.44%
Each row shows the results for a given distance to a converted lot. For each distance, the table reports
the estimated value of the coefficient and its associated t value. The final two columns report the
percent impacts on house values within that distance, which is simply computed by exponentializing the
coefficients and then subtracting one.
The effect of proximity to a vacant lot, as measured by pre_green, is negative and significant for any
distance up to 1,800 feet. While the negative effect is smallest (-9.92%) for those dwellings that are
closest to a vacant lot (<200 feet), the effect is a relatively constant -15-16% for all of the other
distances. This suggests that, counter-intuitively, the negative impact of vacant lots on neighborhood
house values does not decay with distance. However, what is more likely is that the distance to a vacant
lot is relatively constant across most of these households, due to the high concentration of vacant lots in
16
these neighborhoods. According to the data, the Census Tracts in these neighborhoods have an average
vacancy rate of 16%, compared to a citywide average of 9%. Hence, when distance from a vacant lot
(that will eventually be greened) is increasing, distance to another vacant lot (whether eventually
greened or not) is likely to be decreasing. Subsequently, the effects of proximity are relatively difficult
to identify given the high numbers of vacant lots distributed throughout these neighborhoods.
The effects of proximity to a greened lot are more readily identified. According to the results, homes
that are the closest to a converted lot go from being worth, on average, 9.92% less than average to
1.02% more than average, although the effect is not statistically significant. This implies that, post-
greening, these homes are now worth approximately the average of other homes in these
neighborhoods, presumably due to the removal of the dis-amenity of a vacant lot. With further
distances, the effect becomes positive and significant. At a distance of 400-600 feet, the positive effect
on home values is maximized at 8.66%. At distances greater than this, the effect diminishes, and
becomes statistically insignificant at distances greater than 1,400 feet. These results are generally
consistent with the notion that the positive effect of a local amenity should attenuate with distance
from the amenity.
To put this in graphical form, the following chart shows the bid-rent curves for both vacant lots and
greened lots, by plotting the percent change in house values against distance from a typical converted
lot.
Chart 5.
-20.00%
-15.00%
-10.00%
-5.00%
0.00%
5.00%
10.00%
200 400 600 800 1,000 1,200 1,400 1,600 1,800
% C
han
ge i
n H
ou
se V
alu
e
Bid-Price Gradient for Greened Lots
Post_Green (%)
Pre_Green (%)
Distance to Greened Lot (ft)
17
The brown line indicates how house values change, relative to the baseline neighborhood average, for
homes that are near vacant lots that will be greened. The green indicates how these same house values
change once the lots have been greened. For vacant lots, proximity is negatively capitalized at an
average of -15.5%, and this does not decay with distance, which is likely due to the high number of
vacant lots in these subject neighborhoods that makes identification difficult. For greened lots, the
negative capitalization has become positive, and is maximized at a distance of 600 feet. This effect then
decays to zero at a distance of 1,800 feet.
Lastly, we examine for the fiscal impact of converted lots by applying the city’s property taxation
formula to the value generated times the number of affected homes. The results are given in the
following table:
Table 3. Economic and Fiscal Impact of Greened Lots
Item Amount
$Gain per home over 5 Years $34,468 % of Stock <1/4 mile green space 22%
Phila. Housing Stock 456,906
# of homes affected 100,519
Total $Value Gain (m) $3,465
Total $Revenue Gain (m) $100.1
The average increase in the value of homes within a ¼ mile of a greened lot, net of any general house
price appreciation, in the five years following conversion is $34,568. As of 2011, 22% of Philadelphia’s
housing stock of 456,906 dwellings is within a ¼ mile of a greened lots; or 100,519 dwellings.
Multiplying the number of affected homes times the dollar gain in home values yields $3.5bn in
additional housing wealth accruing to owners of homes within ¼ mile of a greened lot. Applying the
city’s taxation formula9 to this incremental housing value estimates that an additional $100m in
property tax revenue would be generated by the conversion of these vacant lots, if assessments
accurately capture this increase in value.
V. Cost-Benefit Analysis of Greening Vacant Lots
Lastly, the net benefits of converting vacant land to green space can be computed by comparing the
benefits of conversion to its costs.
According to the Pennsylvania Horticultural Society, the total cost of converting a vacant lot(much of
which is supported with public funds from the City) includes cleaning, top soil, grading, tree planting,
grass seed sowing, and installation of post and rail wood fence is estimated to be $1.20 per square foot.
After conversion, ongoing maintenance, which includes 14 bi-weekly cleaning and mowing visits from
9 Tax bill= (House Value*0.32*0.08264)*1.092
18
April through October, is $0.12 per square foot. Both installation and maintenance costs include all PHS
contractor management, quality control, and monitoring costs. Thus, the total cost of converting and
maintaining a typical 900 square foot former row house property for ten years will cost around $2.40
per foot, or $2,160.
Multiplying the cost of $2.40 per square foot times the 6.4m total square feet of the cleaned and
greened lots gives a total cost of the program of approximately $15.3m. So, when the total housing gain
of $3.5bn is netted against the total cost of $15.3m, this yields a whopping cash-on-cash return of over
22,000%. Or, more simply, every dollar spent to clean and green a vacant lot increases housing wealth
by $224; a significant return to say the least.
Since PHS does receive public funds for this program, we can also perform this same computation to see
what the return to city taxpayers is, if we are allowed to assume accurate assessments. Netting the
additional $100m in annual property tax revenues against the $15.3m cost of the program yields a cash-
on-cash return of 643%. Or, more simply, every dollar spent to clean and green a vacant lot yields
additional property tax revenues of $7.43. Lastly, this number likely understates the true return since we
are only netting one year of benefits against the total cost of the program, which occurred over 14
years.
VI. Summary and Further Research
Agricultural or open land has traditionally been considered by urban economics to be at the bottom of
the economic hierarchy of land uses. While the direct economic value of green space is usually less than
that of developed space, these data indicate that the indirect value of land that was formerly developed
and is now abandoned and/or vacant is actually negative.
The results indicate that proximity to a formerly industrial or commercial site that has no current active
use is capitalized adversely into the value of the surrounding housing stock. However, the results also
indicate that the repurposing of these sites to public green space is capitalized positively in the value of
the nearby housing stock. This appears to be a multi-step effect.
a. First, property values increase in response to the removal of a dis-amenity (abandoned
building, vacant land) from the neighborhood. We estimate that an average discount of
16% is eliminated.
b. Second, they increase again in response to the addition of an amenity (community
garden or park) to the neighborhood. We estimate that these homes now command an
average premium of 2 to 5%
c. Third, property values continue to increase with the number of added amenities
(converted parcels) that are located nearby. We estimate that each additional greened
lot adds another 1% to the premium.
d. Lastly, we find that these positive effects attenuate with distance from the amenity.
19
For many post-industrial cities that have significant inventories of inactive land parcels, these results
may provide some informative policy implications for how these parcels may be repurposed to a
positive use that contributes to improved neighborhood quality-of-life, increased housing wealth for its
residents, and an expanded tax base. These cities have experienced a half-century of declining property
values due to the chronic proliferation of vacant land; now, for the first time, the strategy of cleaning
and greening vacant lots is shown to halt and actually reverse that decline at the neighborhood level.
This suggests that if urban areas want to turn back the sliding property values, citywide and national
policy should look to vacant land reclamation as a key ingredient. Further research should seek to
provide some robustness to these findings by examining how the results may vary with the particular
characteristics of the individual sites, and the socioeconomic and demographic characteristics of the
neighborhoods they are located in.
20
Bibliography
Bowes, David R. “Identifying the Impacts of Rail Transit Stations on Residential Property Values,” Journal
of Urban Economics, pp. 1-25. 2001.
Correll, Mark R., Jane H. Lillydahl, and Larry D. Singell. “The Effect of Greenbelts on Residential Property
Values: Some Findings on the Political Economy of Open Space.” Land Economics, Vol. 54 (2), 207- 217.
1978.
Hammer, Thomas R, Robert E. Coughlin, and Edward T. Horn IV. “Research Report: The Effect of a Large
Park on Real Estate Value.” Journal of the American Institute of Planners, 274-277. July, 1974.
Ioan Voicu and Vicki Been, 2008. “The Effect of Community Gardens on Neighboring Property Values,”
Real Estate Economics, American Real Estate and Urban Economics Association, vol. 36(2), 241-283.
Wachter, Susan. “The Determinants of Neighborhood Transformations in Philadelphia Identification and
Analysis: The New Kensington Pilot Study,” Spring 2005, 1-18.
Wachter, Susan M., Kevin C. Gillen & Carolyn R. Brown, "Green investment strategies: a positive force in
cities," Communities and Banking, Federal Reserve Bank of Boston, 2008 (2), 24-27.
Wachter, Susan M. and Grace Wong, 2008. "What Is a Tree Worth? Green-City Strategies, Signaling and
Housing Prices," Real Estate Economics, American Real Estate and Urban Economics Association, vol.
36(2), 213-239, 06.
Wei Lei and Jean-Daniel Saphores, 2011. “A Spatial Hedonic Analysis of the Value of Urban Land Cover in the Multifamily Housing Market in Los Angeles, CA,” Urban Studies, 1-19.