Internalizing a Good Externality
-
Upload
rachelglacy -
Category
Documents
-
view
226 -
download
0
Transcript of Internalizing a Good Externality
-
7/30/2019 Internalizing a Good Externality
1/32
Doing Well by Internalizing a Good Externality 1
Doing well by internalizing a good externality: Combining House Price Capitalization
with Altruistic Strategies
Brock Lacy
London School of Economics and Political Science
-
7/30/2019 Internalizing a Good Externality
2/32
Doing Well by Internalizing a Good Externality 2
Abstract
Could a developers strategy of discounting land for a new public school increase
their profits? This papers hypothesis is that by wielding an understanding of house
price capitalization of schools, a developer could increase the price of their houses
above the costs of the discounting of said land. Thus a developer could do well by
doing good. This is then a study of the use of an internalization of an externality
strategy. In 2006, a new school location was announced in a rural county in Utah
State. The price of the school site was discounted by the developer. Using historical
sold house price data collected from the Multiple Listing Services the house price
capitalization of the new school was tested. The method used to test the
capitalization rate was a difference in differences analysis; with the unaffected
portion of the county as the control group. Once the capitalization rate was found the
profit of the internalization strategy would be estimated by weighing the cost of the
discount to the developer against the projected capitalization across their
subdivisions. The analysis finds a zero capitalization of the announced school into
house prices, rendering the profit estimation unnecessary. It is set forth that this may
be due to the plentiful developable land in the treatment group. Therefore it is argued
that in areas with less developable land, such a strategy may indeed lead to increased
profits. Further research into this strategy is discussed and recommended.
Keywords: capitalization, internalization, strategy
-
7/30/2019 Internalizing a Good Externality
3/32
Doing Well by Internalizing a Good Externality 3
Table of Contents
Introduction 4
Literature Review 6
Data 13
Method 15
Results 23
Conclusions 27
References 29
-
7/30/2019 Internalizing a Good Externality
4/32
-
7/30/2019 Internalizing a Good Externality
5/32
Doing Well by Internalizing a Good Externality 5
The analysis will be focused on a rural community in Utah State; this community can be
characterized as a fast growing community not too distant from the central business district
(CBD). The community in question had out grown the only elementary school in the whole of
the school district and thus the districts school board deemed it necessary to construct a new
school. A developer with two large subdivisions in this community deeply discounted the land
for the school site. This then leads to a situation where the above strategy question could be
answered. Could the strategy of doing good, by discounting the price of the land for the school,
raise the house prices enough where this seemingly altruistic action actually turned a profit? To
the extent that the confluence of house price capitalization of the announced new school and the
decision to discount the schools building lot is found to be profitable this paper will then attempt
to make a recommendation concerning the usefulness of such a strategy.
The analysis will be derived from sold house prices obtained from the multiple listing service
(MLS). The method used to estimate the effect of the new school announcement on house prices
will be a difference in differences analysis. Once the capitalization of the new school is found it
will then be used against the cost of the discount from the developer to find the profit of the
discounting of the land. The fitness of the control group to the treatment group was found to be
balanced using the propensity score matching (PSM) technique. The window of analysis will be
three years prior to the announcement and three post the school location announcement. The
difference in differences estimator renders an estimate of 0.8 percent capitalization of the new
school into house prices. However the estimation is found to be statistically insignificant, thus
rendering the next steps of the profit estimation useless. The zero capitalization of the new
school is hypothesized to have occurred due to potentially a number of reasons: uncertainty,
limited data, time lags, and a plentiful supply of developable land. Although the results from the
-
7/30/2019 Internalizing a Good Externality
6/32
Doing Well by Internalizing a Good Externality 6
analysis renders a zero capitalization of the new school into house prices and thus could suggest
that this strategy should not be recommended, there are limitations which throw doubt onto such
a conclusion. Thus this paper will recommend further study of such a strategy and discuss the
potential scenarios in which this strategy could be profitable. This paper will be outlined as
follows I) Introduction, II) Literature Review, III) Data, IV) Method, V) Results, VI)
Conclusions.
II) Literature Review
No discussion of house price capitalization of schools would be complete without mention of the
influential paper by Oates (1969). While pursuing evidence of the Tiebout (1956) hypotheses,
Oates made the following argument:
[T]here is some reason to believe that the Tiebout hypothesis may be relevant to the real
world:Individuals working in a central city frequently have a wide choice of suburban
communities in which to reside, and the quality of the local public schools, for instance,
may be of real importance in the choice of a community of residence. If this is true, the
outputs of public services (as well as taxes) should influence the attraction of a
community to potential residents and should thereby affect local property values. (p. 958)
In his subsequent analysis Oates found that per pupil expenditure has the effect of increasing
house prices, while property taxes decrease house prices. Oates argued that this is evidence that
the Tiebout hypothesis is operative, the true meaning of Oates findings were clarified by
-
7/30/2019 Internalizing a Good Externality
7/32
Doing Well by Internalizing a Good Externality 7
Chaudry-Shah (1988) .. the Tiebout mechanism of foot-voting does operate, but it does not
imply that the local public services are provided efficiently (p.210). Although it is widely held
that Oates research is not the conclusive evidence of the Tiebout hypothesis he sought, his paper
has helped to spur the investigation of the capitalization of all types of externalities into house
prices, and is one of the first steps toward an empirical understanding of house price
capitalization.
There has been ample research since the Oates paper regarding the capitalization of schools. The
majority of papers find that schools have an important impact on house prices. This paper will
highlight a few of this research but would like to note that the papers heretofore mentioned
represent but a small portion of the empirical work.
The majority of house price capitalization of school research is predicated off hedonic analysis.
Hedonic analysis is especially sensitive to omitted variable bias, thus much of the past analysis
used methods and models that attempt to limit this bias. One such method was developed by
Black (1999) wherein she used what is called Boundary Fixed Effects (BFE). The basic idea is
to find houses that are sold on either side of the school catchment boundary, thus it is argued that
these houses are in the same neighborhood and subsequent neighborhood quality is then
controlled for. Using this technique she finds that a 5 percent increase in test scores leads to a
2.1 percent increase in house values. She checks the robustness of her results by making the
analysis area around the catchment boundary smaller and smaller and finds that her results do not
change significantly. The BFE method has been used by others in varying degrees to test the
house price capitalization of school quality (Bayer, Ferreira, & McMillan, 2007; Clapp, Nanda,
& Ross, 2008; and Dhar & Ross 2010). However, there are some who argue that BFE may not
-
7/30/2019 Internalizing a Good Externality
8/32
Doing Well by Internalizing a Good Externality 8
be a sufficient control for neighborhood quality. Gibbons, Machin, and Silva (2009) illustrate a
potential failure of the BFE technique with the following :
[A] situation could arise if, for example, one attendance zone contained a rail station and
another did not. This would then result in higher prices, richer families and better
schools in the station zone, and a spatial trend in house prices rising across the
boundary towards the station..Hence we could find a correlation between house prices
and school quality amongst closely space neighbors that is not caused by the demand for
school quality, but a residential sorting that is a consequence of demand for rail access.
(p. 9)
This potential bias is echoed by Kane, Riegg, and Staiger (2006) as they stated, to the extent
that this sorting occurs, it will bias boundary estimates toward finding a positive association
between school quality and property value, unless one fully controls for these differences across
boundaries (p. 194). Nguyen-Hoang and Yinger (2011) discuss one more potential bias of the
BFE approach. This was the uncertainty argument put forth by Cheshire and Sheppard (2004)
and Zahirovic-Herbert and Turnbull (2008); which is that households living close to a boundary
place a lower weight on across-zone boundaries because they believe that these boundaries might
change, as they occasionally do (p. 33). Cheshire and Sheppard (2004) take it a step further and
postulate that the potential for uncertainty could be strongest in the peripheral areas with
relatively more new construction.
Additionally, there are some derivations of BFE that have been used in past research as well. In
particular importance to this paper is the difference in differences technique. Difference in
differences has been used in a myriad of ways. Recently it has been employed to gauge the
-
7/30/2019 Internalizing a Good Externality
9/32
Doing Well by Internalizing a Good Externality 9
effect of changing school quality via change in school catchment zones. In this way the
difference in differences estimator can be used to find the capitalization of school quality in
house prices. One of the first uses of difference in differences to estimate school quality
capitalization was Bogart and Cromwell (2000). They examine the changing of catchment
boundaries in Shaker Heights, Ohio in 1987. One of the reasons for the change was to integrate
more of the poor minority students with the richer white students. Bogart and Cromwell (2000)
argued that the change in the boundaries could have two effects. The first of which was the
neighborhood schools effect (Bogart and Cromwell, 2000, p. 281) wherein if the boundary
change resulted in the loss of a local school then parents might find it more difficult to be
involved and potentially make it harder for students to be involved with after school activities
which effects they argued, could damage the quality of the school. The second was the racial
composition effect. (Bogart & Cromwell, 2000, p. 281) This effect would possibly drive house
prices down if there were prejudice in the community. In their analysis they use standard
house structure variables such as age of house, lot size, and living area. They include third grade
reading test scores as the variable for school quality. The difference in differences estimation of
school quality and the related theoretical effects were all the more meaningful as only elementary
catchment boundaries were changed, while all students attend the same high school. This in
theory controlled for omitted high school changes that could also affect house prices, the
importance of which was discussed by Nguyen-Hoang and Yinger (2011). Bogart and Cromwell
(2000) attempt to control for omitted variable bias by including variables for neighborhoods;
percent nonwhite and house construction quality grade. Using a log linear OLS they find that
their difference in differences estimator of the effect of the change in catchment boundaries
decreased house prices by 9.9 percent.
-
7/30/2019 Internalizing a Good Externality
10/32
Doing Well by Internalizing a Good Externality 10
In a similar paper Ries and Somerville (2010) examine the effects of changing school catchment
zones on house prices. Their area of study was Vancouver, British Columbia, Canada. In
September of 2000, the Vancouver School Board announced plans to alter school boundaries,
which affected 20 percent of houses. Ries and Somerville (2010) argued that examining school
quality in Vancouver had advantages: there is a single tax rate and a more standardized level of
municipal services (p. 930) and the racial issues that so pervade location decisions in the
United States are not as present (p. 930). They added house preference and location are
somewhat unbundled (p. 930) this they argued is evidenced by the small percentage of
observations from condominiums (p 930). Ries and Somerville (2010) found that when a repeat
sales index is used as opposed to the cross-sectional hedonic regression, bias from omitted
variables is dampened. They discover that the school boundary change had little to no effect for
all groups except those houses in the highest quartile price range. This result echoes Chiodo,
Hernandez-Murillo, and Owyang (2010) and Cheshire and Shepherd (2004). Interestingly
enough the effect of changing an elementary school was statistically insignificant as the analysis
became more sophisticated (moving from cross sectional hedonic analysis to repeat sales index).
There has been some research concerning the effect of school infrastructure investment on house
prices. The most salient to this paper, would be research conducted by Cellini, Ferreira, and
Rothstein (2010). Using the principle of house price capitalization, they employ regression
discontinuity design to find the effect of school facility investments on house prices. They
contend that there is good reason to expect school facility investment to have a positive effect on
house prices. Beyond the potential effect of improved outputs (higher marks, grades,
graduation), Cellini et al (2010) reasoned that capital investments may also lead to
enhancements in student safety, athletic, and art training, the aesthetic appeal of the campus, or
-
7/30/2019 Internalizing a Good Externality
11/32
Doing Well by Internalizing a Good Externality 11
any number of other nonacademic outputs (p. 222). Using the passage of closely contested
school facility investment bonds in California their analysis finds the approval of said bonds
caused house prices in a district to rise by about 6 percent. Cellini et al (2010) argued that this
effect appears gradually over the two or three years following the elections and persists for at
least a decade (p. 218). Cellini et al (2010) submit that the effect of school facility investment
on house prices is gradual due to construction lags and school quality signals. Thus the full
effect of the new school or school investment on house prices and property could take some time
and therefore build momentum as time progresses.
Research concerning capitalization of externalities (e.g. house price capitalization of schools) has
potential for recommendation of business strategy; to this end there has been some academic
research. One such research paper was authored by Pashigian and Gould (1998). They
hypothesized that anchor tenants in shopping malls drive foot traffic. This increase demand for
access to these anchor tenants could increase the sales of non-anchor tenants. Thus they argue
that non-anchors could free ride off the success of their anchor tenants. Pashigian and Gould
(1998) theorize that developers could internalize this anchor externality by charging less rent to
anchors and more rent to a non-anchor tenant; in a sense it is a rent premium for non-anchors to
access traffic driven by the anchor. Using a tri-year survey conducted by the Urban Land
Institute they found that anchor tenants paid 72 percent less than their non-anchor neighbors. As
Pashigian and Gould (1998) interpret their results [t]hose externalities are internalized by the
developer through store rents. Our evidence indicates that the external economics created by
anchors are reflected in the lower store rents paid by anchors (p. 140). They further clarify the
strategy of discounting rent for anchors relative to non-anchors by developers as rational
because they know that anchors attract customers to the mall and increase the sales of other
-
7/30/2019 Internalizing a Good Externality
12/32
Doing Well by Internalizing a Good Externality 12
mall stores (p. 140) A similar result is found in Benjamin, Boyle, and Sirmans (1992), wherein
they find that rent falls as shopping center stores sales increase. Brueckner (1993) examines a
similar shopping center strategy for developers. His analysis is concerned with inter-store
externalities, similar to the Pashigian and Gould (1998) analysis. In particular he is concerned
with the size of stores within shopping centers and the effect of store size on the sales of other
stores within the shopping center. He developed a model that claims that a profit maximizing
developer will choose store sizes that make the shopping center more attractive and thus increase
the inter store externalities for all of the shopping center.
The above surveyed literature concerning the strategy of internalizing externalities have been
constrained to the non-residential real estate market. This is due to the dearth of academic
research regarding residential internalization of externalities strategies. However one paper that
is significant to this essay is from Thorsnes (1998). He was interested in a strategy which would
lead to a greater capitalization of local amenities. He reasons that a larger subdivision should
lead to greater capitalization of said amenities. He used sales of building lots to test his
hypotheses. Thorsnes (1998) explains that the advantage of sold building lots over sold houses
are twofold the value of amenities is capitalized into the sale price of the building lots and the
many characteristics of the house need not be controlled for (p. 398). The analysis of these
observations renders the following result: that adding one additional acre to the median
subdivision would increase the sales price of a building lot by 3 percent.
A seemingly tangential bit of literature is appropriate to note here. Hilber and Mayer (2009)
asked: Why childless households would support increases in school spending? (p. 74). They
argued that per pupil expenditure is more positively capitalized in those areas with less
developable land.(p.74). Or in other words, areas which are seemingly characterized by more
-
7/30/2019 Internalizing a Good Externality
13/32
Doing Well by Internalizing a Good Externality 13
inelastic supply of housing have higher capitalization of schools than more elastic areas. This is
potentially of crucial importance to this papers analysis as the area in which the treatment
occurred had plenty of developable land and thus one could argue was relatively characterized by
elastic supply.
Theory and past research could with reason lead to the fundamental question of this paper: Could
the discounting of land for a new school be profitable for a developer? This paper submits that
in theory the answer to this question should be in the affirmative. That is there is good reason to
expect that under the right circumstances a potentially altruistic act could in fact be economically
rational.
III) Data
The data used for this analysis was collected from the MLS. The MLS is an internet site that
lists prices for houses but also carries historical sold house prices for given regions or cities.
Prior to treatment date and time period selection the data included over 750 observations from
both the treatment and the control area. These observations ranged from 2001 to 2012.
Nevertheless this kind of data is not without potential problems. Parmeter and Pope (2009)
described one of the weaknesses of MLS data by the following:
[R]ealtors are typically not required to fill in all of the fields within the MLS system. For
example, some realtors may consistently input information on the type of flooring in the
house whereas others may not. Furthermore, it is likely that realtors are selective on the
-
7/30/2019 Internalizing a Good Externality
14/32
-
7/30/2019 Internalizing a Good Externality
15/32
Doing Well by Internalizing a Good Externality 15
IV) Method
The lynchpin of difference in differences analysis is the correct selection of the treatment and the
control group. Similarly to Bogart and Cromwell (2000) all students in both groups attend the
same high school after the school announcement; thus controlling for the potential bias of the
effect of differing high school quality. The treatment area was confined to Mountain Green and
Peterson Utah, while the control group contains all other cities in the school district (Morgan,
Milton, and Croydon). Table I illustrates some demographics via the US Census Bureau (2010a)
(2010b). The Mountain Green area is an unincorporated city located 35 miles northwest of the
CBD. Mountain Green is characterized as a rural area in a mountainous canyon close to ski and
snow resorts. According to the US Census Bureau (2010a) Mountain Green had a population of
2,309. Although the city is relatively small it is rather prosperous with a median household
income in 2010 of 95,263 dollars (US Census Bureau, 2010b). However, this statistic may be a
little misleading if one does not consider the interval in which this median household income is
derived. The US Census Bureau estimates at a 90 percent confidence interval with a margin of
error of 35,327 dollars. When considering other control areas this margin of error is
substantially larger. Thus although the measure of income in the treatment area is somewhat
larger than the commensurate number for the control area, it has a rather large range which needs
to be kept in mind when comparing the two areas.
-
7/30/2019 Internalizing a Good Externality
16/32
Doing Well by Internalizing a Good Externality 16
Table I.Demographics*
Mountain Green Morgan City
Population 2,309 3,687
Median Household Income $ 95,263.00 $ 60,781.00
White 98.90% 99.5%
Mean House Price 2003 -2009 $ 358,223.79 $ 253,653.15
Industry Employed:
Agriculture 2.90% 2.00%
Construction 14.90% 11.70%
Manufacturing 12.00% 14.20%
Wholesale Trade 1.40% 3.40%
Retail Trade 8.80% 15.20%
Transportation 10.70% 2.80%
Information 0.00% 0.60%
Finance, Insurance, Real Estate 4.90% 6.20%
Professional 7.10% 2.70%
Education 16.90% 20.20%
Arts 9.50% 4.30%
Other 0.80% 3.10%
Public Adminstration 10.00% 13.60%
* - US Census Bureau, 2006-2010 American Community Survey
The control area which will from here forth be referred to as Morgan City (although there are
few other smaller cities included in the control group there are only limited observations from
these cities and Morgan is by far the largest of these cities). Morgan City is the seat of the
Morgan County government and the Morgan School District. Morgan City is located
-
7/30/2019 Internalizing a Good Externality
17/32
Doing Well by Internalizing a Good Externality 17
approximately 10 miles from the treatment area, Mountain Green. Referencing the US Census
(2010), it was estimated that Morgan City had a population of 3,687. Although Morgan has a
larger population than Mountain Green, its median household income is much smaller, as
estimated at 60,781 dollars (US Census Bureau, 2010b). This measure of centrality for Morgan
City has a much smaller range in the confidence interval as opposed to Mountain Green at only
plus or minus 12,891 dollars. Both measures of income are relatively higher than the Utah
States median income (56,330 dollars) (US Census Bureau,2010c). In theory one would expect
the effect of schools to be more highly capitalized in neighborhoods with higher income (such as
Mountain Green) as per the findings of Cheshire and Sheppard (2004) and Chiodo et al (2010).
Precedence for using the non-affected area of the school district as the control group was set
forth by Bogart and Cromwell (2000) and Ries and Somerville (2010). Bogart and Cromwells
(2000) treatment and control group are located in the same school district and prior to the
announcement of the policy attended the same elementary school. Similarly to Bogart and
Cromwell (2010) all students in both groups attend the same high school after the policy
announcement; thus controlling for the potential bias of the effect of differing high school
quality. The selection of the control group was also influenced by Cellini et al (2011). The
preliminary reasoning was that a new elementary school in the Morgan County School district
represented two events 1) a potential change in school catchment boundaries and 2) a new school
structure. Thus the use of Morgan City as the control group seemed logical as the boundaries
were affected, but the students in Morgan would not be changing or losing their school, thus they
would be unaffected. In addition the proximity of the two areas to one another, the relative
shared amenities, the approximately same distance to the CBD, the unincorporated nature of
Mountain Green (in a very real sense this meant that the county was the political power over
-
7/30/2019 Internalizing a Good Externality
18/32
-
7/30/2019 Internalizing a Good Externality
19/32
Doing Well by Internalizing a Good Externality 19
the propensity scores of the treated and the control groups are balance. To insure whether the
control group and treatment group do match according to the PSM, the data was analyzed and
was found that the matching principle was satisfied via STATA.
There is repeated numerous times throughout house price capitalization literature the importance
of controlling for neighborhood quality, especially for BFE analysis and its derivations. As the
treatment area and the control area are relatively small in population and they share some
important characteristics, (e.g. high school and elementary school prior to the school
announcement, distance to CBD, terrain, and size) it seemed reasonable to view the two analysis
areas as neighborhoods within in the school district. One more neighborhood variable which
will be included from the data is sold houses in the developers subdivisions. Upon examination
of the results, it is shown that these subdivisions command a premium and increase the
robustness of the model used.
Difference in differences analysis is an attempt to find the causal effect of policy change or an
event on a dependent variable. As such the analysis used the following log linear equation:
Ln(V) = X + Z + W +
This regression was ran using OLS to find the difference in differences estimator W, meaning
post treatment date and treated area; while Z is the dummy variable for post treatment but not
controlled for treatment group. V is the sold house price. X is a vector of both structural
characteristics and neighborhood characteristics including dummy variables for year sold. An
additionally crucial point for estimation of the effect of the school on house prices is to identify
the treatment period or event. As this discussion pivots on when the market knew the school
would be a reality in Mountain Green. The researcher was able to identify potentially three
-
7/30/2019 Internalizing a Good Externality
20/32
Doing Well by Internalizing a Good Externality 20
different signals to the market that the school would indeed be located in Mountain Green.
These three are: 1) the announcement of the land purchase at a discount by the school district, 2)
the passing of the school bond, and 3) the announcement of the decision regarding where to
locate the school. First the announcement of the land purchase at a discount. Although the exact
date of the land purchased is not publically known, it is within reason to assume that it became
publicly known on July 1st 2007. The day in which the local newspaper published an article
discussing the purchase but not noting the date; however the article in question did discuss the
discount of the land for the school structure. According to the newspaper article, the company
building the expansive subdivision around the new elementary school, sold the 10-acre school
site for a very reasonable price of about $15,000 per acre, or almost $150,000 (Winterton,
2007, p. 4B).
The second potential treatment date may have been the passing of the bond for the new
elementary school. The voting regarding the bond was to be held in early November 2006.
Prior to the successful passing of the bond, there were two previous attempts to secure funding
for a new elementary school in Mountain Green. In both 1999 and 2000 the Morgan School
District had posted referenda for the school bond, however both times the bond for the
construction of a new school was defeated by the slimmest of margins (both times the bond
failed by less than 12 votes) (Winterton, 2006a). The impotence for this bond was the dire need
for bigger facilities. As one journalist noted, While Morgan Elementary's enrollment hovers at
800, the average elementary enrollments in surrounding school districts are much lower
(Winterton, 2005). Another compelling argument for the use of the bond passing as the
treatment date concerns the markets ability to predict its success. This is evidenced by the
failures of two bonds seven years early explicitly concerning the new elementary school in
-
7/30/2019 Internalizing a Good Externality
21/32
Doing Well by Internalizing a Good Externality 21
Mountain Green. This gives credence for the bond passing as a potentially unanticipated event
and a logical position for the treatment period.
The last potential treatment period considered was the announcement of the school location. In a
school board meeting on January 10th
2006 the school board announced its desire to build in the
developers subdivision (Winterton, 2006b). Although the final decision for the location of the
school site was not decided at the meeting, it was later decided that it would be built in the
subdivision.
As one could imagine all of the treatment periods listed above were not without deficiencies.
The announcement of the school discount via the newspaper article does not give a date for
which the purchase was completed. Additionally it is hard to argue that this in anyway should
effect the price of houses within the treatment area; one could conceivably argue that this was a
signal to the market of the developers commitment to the school, but if that were the case then
the analysis would measure the effect of the developers commitment to the school not the value
of the school. This would in theory measure the marketing value of the discount. This date was
rejected as the treatment date. The school bond is seemingly the best treatment date, however
there are some glaring holes upon further examination. Perhaps the biggest of these holes is the
potential leakage of the passing of the bond into the control group. The bond was not only for
the construction of a new elementary school in Mountain Green but also for substantial upgrades
and renovation for the control groups elementary school and high school. It is within reason to
assume that the treatment group and the control group could potentially be affected by the bond
passing, and the treatment date is analyzing more than just the effect of the new elementary
school but of the investment in the high school if a different control group were to be used. It is
-
7/30/2019 Internalizing a Good Externality
22/32
Doing Well by Internalizing a Good Externality 22
with this reasoning that the announcement of the school site was used as the treatment date. It
had been known prior to the announcement of the selection of the school site that the school
board was deliberating certain locations, even to the point where the location was narrowed
down to a few locations including the final location prior to the final announcement. However, it
is the cleanest of the three treatment dates. It is the earliest concrete announcement from the
school board concerning the new elementary school and its location. It only concerns the effect
of the new elementary school on the treatment area with potentially little to no leakage to the
control group. The selection of the announcement of school location is not perfect either. As
mentioned earlier there were discussions by the school board concerning the site of the proposed
elementary school prior to this announcement and thus the market knew that there was a
deliberation. However, as far as this researcher could gather there was no deadline for selection
and no indication of when the decision would be finalized. This gives rise to the argument that
the effect of the new elementary school would not be capitalized prior to this date.
The final proposed step would be to project the profit from the discounting of the land for the
developer. This would include taking the capitalization of the new school from the difference in
differences estimator of the new school and then projecting that onto vacant lot prices into the
future. Unsold lots would be discounted at the market capitalization rate and at the average time
to sell provided by the developer. However, as the results from the analysis indicate a zero
capitalization rate this step will not be pursued.
-
7/30/2019 Internalizing a Good Externality
23/32
Doing Well by Internalizing a Good Externality 23
IV) Results
Before running the analysis the data was observed to contained some variance that if not
smoothed out, could likely cause bias. It is to this end that outliers were removed to make the
data more manageable. The results of the difference in differences analysis are found in Table II.
These results were obtained by running the OLS regression robustly through STATA software.
The results seem to have an overall good measure of fit with an adjusted R-Square of 0.8465. To
further strengthen the argument of the model as a good fit, all the results have the correct signs.
What is more, all of the structural and neighborhood characteristics are statistically significant.
A home buyer would pay a premium for additional acreage, bathrooms, bedrooms, and to finish
more of the basement. On the contrary the sign of the age of the house is negative and
statistically significant. Additionally one would pay a
premium to buy a home in the Mountain Green treatment area as opposed to the Morgan City
control area. To illustrate one would have to pay approximately 10 percent more for the same
house in Mountain Green. The result is similar (and expected) for the developers subdivision.
One would have to pay approximately an additional 6.1 percent to buy the same house in any of
the developers subdivisions. However the statistical significance of this result is minimum at 20
percent level of significance. This may be due to the relatively few observations permitted from
the MLS. Given the quality of the subdivision one would suspect that this coefficient should
indeed be significant. Additionally the results indicate that the there was a premium for all
houses sold after the announcement of the school location. This result is not statistically
significant from zero. Although the year dummy variables were not included in the results
-
7/30/2019 Internalizing a Good Externality
24/32
Doing Well by Internalizing a Good Externality 24
presented in Table II, they were included in the regression, the results of which have the
expected signs but have varying degrees of significance.
Table II
Difference in Differences Results
VariableCoefficient
(standard error)
ln acres 0.1376***(0.0121)
percent basement finished 0.0444***(0.0199)
total bathrooms 0.0264***(0.0114)
total square feet 0.0001***(0.0000)
age of house -0.0119***(0.0009)
age of house squared 0.0001***(0.0000)
Developer's Subdivison0.0611*(0.0449)
Treatment Area (Mountain Green and Peterson) 0.1007***(0.0313)
Post School Site Announcement 0.0196(0.1503)
Post School Site Announcement and TreatmentArea 0.0084
(0.0361)
R - Squared 0.8465
* - 20% level of signficance
*** - 5% level of signficance
-
7/30/2019 Internalizing a Good Externality
25/32
Doing Well by Internalizing a Good Externality 25
Turning to the results of the differences estimator, the coefficient has the sign that would be
predicted by theory. Just like Cellini et al (2010) the announcement of the school location has a
positive effect on house prices. However, the results are statistically insignificant from zero.
Furthermore this result should not be considered unusual as the results are similar to Ries and
Somerville (2010), with less data and without the benefit of repeat sales index. Although this
prevents the implementation of the proposed methodology to test the profitability of the
discounting land for the school strategy, it does beg the question: Why would the results be
insignificant from zero? Referencing the work of Chiodo et al (2010), Cheshire and Sheppard
(2004), and Ries and Somerville (2010) it could be expected that due to the higher house prices
in Mountain Green coupled with the higher median income, that there would be rather strong
capitalization of the new school. However, there are copious potentials for why it was an
essentially zero capitalization. One reason is that the control group and the treatment group may
not be matched. One must note the limitations of PSM. PSM requires a large data sample. It
could be argued that this was requirement was not met with the observations available from the
MLS. Additionally there are a few theoretical reasons why there would be capitalization of the
announcement of the new school location that was statistically insignificant from zero. There are
more explanations not discussed here, this paper will try to outline a few the researcher believes
are most salient. They are: 1) uncertainty, 2) limited data, 3) time lags and 4) plentiful
developable land.
First, the uncertainty proposal is championed by Cheshire and Sheppard (2004). From their
analysis they argue that [s]chool quality also appears to be significantly and additionally
discounted in areas in which new construction is concentrated. They continue, it reflects
uncertainty as to future changes in school catchment areas in such neighbourhoods and so
-
7/30/2019 Internalizing a Good Externality
26/32
Doing Well by Internalizing a Good Externality 26
uncertainty as to what school an address will in future be assigned. This argument is
strengthened when the projections of growth in the whole of Morgan County. As projected by
the Utah Governors Office of Planning and Budget (2008), Morgan County will experience an
average annual growth rate of 3.8 percent; making Morgan County the second fastest growing
county in the state. This kind of projected growth could lead to uncertainty concerning the
schools boundaries and quality. Second, the data used to perform the analysis was rather
limited and therefore could cause some questions concerning the legitimacy of any conclusions.
It is within reason to assume that with more observations and with more variables one could
draw more robust conclusions, however this analysis did not have that privilege and was left to
suffice as best as possible. Thirdly, Cellini et al (2010) found that it took three to four years for
the increase in house prices from the passage of the school bond. Furthermore it could take time
for the value of the school to grow and signal its quality to the market. Fourth, the area both in
Mountain Green and in Morgan City could be characterized as rural, with plenty of developable
land or could be characterized as having elastic supply. This could affect the capitalization as
Hilber and Mayer (2009) observed. With their results in mind it could be expected that in
regions with plenty of developable land that the capitalization of schools into house prices could
be zero. This is potentially due to the ease at which other home owners can access the school
without increasing their bids accordingly. Although all of these arguments potentially have merit
it is the elastic supply argument that is the most salient for this area. However,, it should be
stressed that none of the other arguments are rejected as potential explanations.
-
7/30/2019 Internalizing a Good Externality
27/32
Doing Well by Internalizing a Good Externality 27
VI) Conclusions
Will discounting land for a new school be profitable for a developer? The answer is similar to so
many other answers in economics: it depends. If the explanation that elastic supply is driving the
zero price capitalization of the new school is believed, then it may increase house prices if the
supply in the area that will allow access to that new school is limited, one could then expect a
higher capitalization rate. However, if the development is located in an area like the treatment
area than one might expect a zero capitalization of that new school; does this mean that in more
rural areas with more developable land it is not profitable to discount land for a new school? The
answer is unclear. However, there may be other reasons to discount land. Using the developer
referenced throughout this essay as an example, one might gain permission to develop faster.
(Winterton, 2007b) The location of the school in the developers subdivision persuaded the
Morgan County Council to allow development in a phase a year ahead of the allotted time. This
is potentially a windfall for the developer. As past research has suggested that housing supply is
at least partially determined by the time it takes to sell (DiPasquale, 1999). Thus a potential
extension of this research would be to analyze whether houses or building lots are sold quicker
when a new school is announced. Another reason to discount land for a new school is that it
could increase the attractiveness of the subdivision, when asked why the developer discounted
the land for the school the developer stated, a school makes development more marketable
(Winterton, 2006b, p. 2B). It may be that this could be chalked up to a cost of doing business
and that new schools make it easier to sell a vacant lot or even a home. It could be that the
goodwill garnered from such a policy may lead to planning permission in other areas as well.
-
7/30/2019 Internalizing a Good Externality
28/32
Doing Well by Internalizing a Good Externality 28
This strategy may not only make the subdivision more marketable but it may also make the
developer more marketable to local planning authorities. Consequently an appropriate extension
of this paper would be to measure the success of altruistic strategies in garnering more
development permission. One must consider the timeline in which this analysis occurred; this
was time of immense growth in house prices and dramatic falls. Given that Cellini et al (2010)
found that following the passage of school facility investment bond house prices raised
gradually over the two or three years following the election and persists for at least a decade
and the limitations from widening the time window in a difference in differences analysis. It is
within reason to suggest that it may be that the full weight of the value of the school was not
capitalized until later. Thus another potential extension of this research would be to analyze the
changing value of the new school on house prices over time, this may prove difficult due to the
limited data and the aforementioned drastic fall in house prices.
Could the strategy of discounting land for a public school lead the developer to do well by
doing good? Although it cannot be said with the analysis put forth above that they did not
profit by their actions, it is reasonable to assume that there may be opportunities for such a
strategy to be profitable.
-
7/30/2019 Internalizing a Good Externality
29/32
Doing Well by Internalizing a Good Externality 29
References
Bayer, P., Ferreira, F., McMillan, R. (2007). A unified framework for measuring preferences forschools and neighborhoods.Journal of Political Economy, 115, 588-638.
Benjamin, J.D., Boyle, G.W., Sirmans, C. F. (1992). Price discrimination in shopping centerleases. Journal of Urban Economics, 32, 299317.
Black, S. E. (1999). Do better schools matter? Parental valuation of elementary education.Quarterly Journal of Economics, 114, 577-599.
Bogart, W.T., Cromwell, B. A., (2000). How much is a neighborhood school worth? Journal ofUrban Economics,47, 280305.
Brueckner, J.J., (1993). Inter-store externalities and space allocation in shopping centers.Journalof Real Estate Finance and Economic, 7, 517.
Cellini, S.R., Ferreira, F., Rothstein, J.M., (2010). The value of school facility investments:evidence from a dynamic regression discontinuity design. Quarterly Journal ofEconomics, 125, 215261.
Chaudry-Shah, A. (1988). Capitalization and the theory of local public finance: an interpretiveessay.Journal of Economic Surveys, 2, 209-243.
Chen, V.W., Zeiser, K. (2008). Implementing propensity score matching causal analysis withStata [Power point] . Retrieved from http://help.pop.psu.edu/help-by-statistical-
method/propensity-matching/Intro%20to%20P-score_Sp08.pdf
Cheshire, P., Sheppard, S. (2004). Capitalising the value of free schools: the impact of supplycharacteristics and uncertainty.Economic Journal,114, 397424.
Chiodo, A.J., Hernandez-Murillo, R., Owyang, M. T., (2010). Nonlinear effects of school qualityon house prices. FederalReserve Bank of St. Louis Review,92, 185204.
Clapp, J.M., Nanda, A., Ross, S.L., (2008). Which school attributes matter? The influence ofschool district performance and demographic composition on property values.Journal ofUrban Economics, 63, 451466.
Daily News Staff Writer (2009, November 2). Ford Motor Co. gets it into gear, auto makerreports $1 billion 3Q profit.Daily News. Retrieved fromhttp://articles.nydailynews.com/2009-11-02/news/17940055_1_ford-escape-market-share-ford-focus
Dhar, P., Ross, S. L., (2010). School quality and property values: re-examining the boundaryapproach. Working paper. Department of Economics, University of Connecticut.
-
7/30/2019 Internalizing a Good Externality
30/32
Doing Well by Internalizing a Good Externality 30
Dipasquale, D. (1999). Why dont we know more about housing supply? The Journal of RealEstate Finance and Economics, 18, 9-23
Gibbons, S., Machin, S., Silva, O. (2009). Valuing school quality using boundary discontinuities.Working paper. Spatial Economics Research Centre.
Hilber, C.A.L., Mayer, C.J., (2009). Why do households without children support local publicschools? Linking house price capitalization to school spending.Journal of UrbanEconomics, 65,7490.
Kane, T.J., Riegg, S.K., Staiger, D.O., (2006). School quality, neighborhoods, and housingprices.American Law and Economics Review,8, 183212.
National Center for Education Statistics. (2009). Table 10.Total expenditures for public
elementary and secondary education and other related programs, by type of expenditureand state or jurisdiction: Fiscal Year 2009 [Data File]. Retrieved fromhttp://nces.ed.gov/pubs2011/expenditures/tables/table_10.asp
Parmeter, C.F., Pope, J.C. (2009). Quasi-experiments and hedonic property value methods. In A.List & B. Price (Eds). Prepared for the Handbook on Experimental Economics and theEnvironment (pp. 1-73). Edward Elgar Publishers.
Nguyen-Hoang, P., Yinger, J. (2011). The capitalization of school quality into house values: Areview.Journal of Housing Economics, 20, 30-48.
Oates, W.E. (1969). The effects of property taxes and local public spending on property values:An empirical study of tax capitalization and the Tiebout hypothesis. TheJournal ofPolitical Economy, 77, 957-971.
Pashigian, B.P., Gould, E.D. (1998). Internalizing externalities: the price of space in shoppingmalls. Journal of Law and Economics, 41, 115-142
Ries, J., Somerville, T. (2010). School quality and residential values: evidence from Vancouverzoning.Reviewof Economics and Statistics,92, 928944.
Rosenbaum, P.R., Rubin, D.R. (1983). The central role of the Propensity score in observationalstudies for causal effects.Biometrika, 70, 41-55
Thorsnes, P. (2000). Internalizing neighborhood externalities: the effect of subdivision size andzoning on residential lot prices.Journal of Urban Economics, 48, 397-418
Tiebout, C.M. (1956). A pure theory of local expenditures. The Journal of Political Economy,64, 416-424.
-
7/30/2019 Internalizing a Good Externality
31/32
Doing Well by Internalizing a Good Externality 31
United States Census Bureau. (2010a). Profile of General Population and HousingCharacteristics: 2010 [Data File]. Retrieved fromhttp://factfinder2.census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=DEC_
10_DP_DPDP1&prodType=table
United States Census Bureau. (2010b). Selected Economic Characteristics: 2006-2010 AmericanCommunity Survey 5-Year Estimates [Data File]. Retrieved fromhttp://factfinder2.census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=ACS_10_5YR_DP03&prodType=table
United States Census Bureau. (2010c). Selected Economic Characteristics: 2006-2010 AmericanCommunity Survey Selected Population Tables[Data File]. Retrieved fromhttp://factfinder2.census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=ACS_10_SF4_DP03&prodType=table
Utah Governors Office of Planning and Budget. (2008). Population Projections by County andDistrict[Data File]. Retrieved from http://www.governor.utah.gov/dea/projections.html
Winterton, D. (2007a, July 1). Donations expand school construction. The Standard Examiner.Retrieved fromhttp://digital.olivesoftware.com/Repository/ml.asp?Ref=U1NFLzIwMDcvMDcvMDEjQXIwMTIwMQ%3D%3D&Mode=Gif&Locale=english-skin-custom
Winterton, D. (2006a, February 18). Morgan plans school bond for second elementary otherwiseyear-round track. The Standard Examiner. Retrieved from http://nl.newsbank.com/nl-
search/we/Archives?p_widesearch=yes&p_multi=OSXB|HTLL|EEOL|&p_product=SNUNP&p_theme=snunp&p_action=search&p_maxdocs=200&s_dispstring=allfields(morgan)%20AND%20date(2/18/2006%20to%202/18/2006)&p_field_date-0=YMD_date&p_params_date-0=date:B,E&p_text_date-0=2/18/2006%20to%202/18/2006)&p_field_advanced-0=&p_text_advanced-0=(%22morgan%22)&xcal_numdocs=20&p_perpage=10&p_sort=YMD_date:D&xcal_useweights=no
Winterton, D. (2005, November 15). School board considers third bond election in seven years.The Standard Examiner. Retrieved from http://nl.newsbank.com/nl-search/we/Archives?p_widesearch=yes&p_multi=OSXB|HTLL|EEOL|&p_product=SNUNP&p_theme=snunp&p_action=search&p_maxdocs=200&s_dispstring=allfields(third%20bond)%20AND%20date(11/15/2005%20to%2011/15/2005)&p_field_date-0=YMD_date&p_params_date-0=date:B,E&p_text_date-0=11/15/2005%20to%2011/15/2005)&p_field_advanced-0=&p_text_advanced-0=(%22third%20bond%22)&xcal_numdocs=20&p_perpage=10&p_sort=YMD_date:D&xcal_useweights=no
-
7/30/2019 Internalizing a Good Externality
32/32
Doing Well by Internalizing a Good Externality 32
Winterton, D. (2006b, January 13). Morgan board chooses elementary school site. The StandardExaminer. Retrieved fromhttp://digital.olivesoftware.com/Repository/ml.asp?Ref=U1NFLzIwMDYvMDEvMTMjQXIwMTMwMw%3D%3D&Mode=Gif&Locale=english-skin-custom
Winterton, D. (2007b, February 27). Future school helps subdivision. The Standard Examiner.Retrieved fromhttp://digital.olivesoftware.com/Default/Scripting/ArticleWin.asp?From=Search&Key=SSE/2007/02/27/22/Ar02201.xml&CollName=SSE_APA3&DOCID=91390&PageLabelPrint=6D&skin=StandardExA&AppName=2&sPublication=SSE&sQuery=future%20school%20helps%20subdivision&sSorting=%2553%2563%256f%2572%2565%252c%2564%2565%2573%2563&sDateFrom=%2530%2531%252f%2530%2531%252f%2532%2530%2530%2537&sDateTo=%2530%2538%252f%2530%2536%252f%2532%2530%2531%2532&ViewMode=GIF
Zahirovic-Herbert, V., Turnbull, G.( 2009). Public school reform, expectations, andcapitalization: what signals quality to homebuyers? Southern Economic Journal,75,10941113.