Embed Size (px)
Citation preview
2012 V00 0: pp. 1–33
DOI: 10.1111/j.1540-6229.2012.00332.x
REAL ESTATE
ECONOMICS
Financing Residential Developmentwith Special DistrictsStephen B. Billings∗ and Thomas G. Thibodeau∗∗
This paper empirically examines the extent to which the property tax liabilitycreated by financing residential infrastructure using special district bonds iscapitalized in house prices. We compare house prices for single-family de-tached homes built within development districts to similar properties locatedoutside development districts. Our hedonic specification includes the usualhousing characteristics and controls for the influence of spatial attributes us-ing Census Block Group “neighborhood” fixed effects. The preferred empiricalspecification restricts the data to neighborhoods that have numerous sales ofrecently constructed single-family detached homes located both within andoutside development districts. The empirical results indicate that house pricesfor homes located within development districts are lower than house pricesfor similar homes located outside of development districts, but the amount ofproperty tax capitalization is significantly less than full. Results depend on ourGeneralized Methods of Moments estimator, which instruments property taxrates using the characteristics of development districts. We identify valid in-struments by restricting transactions to properties located in rapidly growingsuburban developments.
There are four mechanisms typically used to finance the development of res-idential infrastructure: municipal bond financing (with bonds securitized bya municipality’s general revenues), the developer (either using constructionloans, equity or some combination of both), development impact fees and spe-cial district bonds. Special districts1 are created by property owners to providespecific public services. They are governed by a board of directors or super-visors and have the authority to issue debt securitized by expected retail salesand property tax revenues as well as by anticipated tolls, user fees, tap or
∗University of North Carolina-Charlotte, Charlotte, NC or [email protected].∗∗University of Colorado, Boulder, CO 80309 or [email protected].
1Special districts are also referred to as special service districts, special purpose districts,limited purpose districts, municipal development districts and municipal utility districts.
C© 2012 American Real Estate and Urban Economics Association
2 Billings and Thibodeau
other impact fees.2 Financing public services, including the development ofresidential infrastructure, using special districts has become a fairly commonpractice in some parts of the United States. The Special District Associationof Colorado, for example, reports that there are over 1,800 special districtscurrently operating in the state.
This paper examines whether the property tax liability generated by develop-ment district financing is capitalized in the price of new, owner-occupied homes.It compares the prices of recently constructed, owner-occupied homes built indevelopment districts (where infrastructure was financed with development dis-trict bonds) to the prices of otherwise similar homes built outside developmentdistricts (where the infrastructure was typically developer financed). Since allthe homes in the sample have infrastructure (i.e, water, sewer, roads, etc.), thisanalysis avoids the problem of how homeowners value the benefits providedby the infrastructure and focuses on how homebuyers capitalize the expectedfuture property tax liability in the price of new, owner-occupied housing.
This paper makes four contributions to the literature. First, it examines whetherthe property tax liability associated with development district bonds is capi-talized in the price of new homes. We compare the prices of homes locatedin development districts to prices of similar homes where infrastructure wasfinanced with some combination of the developer’s equity and constructionloans.3 We find that homebuyers fail to fully capitalize the future property taxliability in house prices. Bradley (2011) suggests several reasons why propertytax capitalization may be less than full. From his perspective, “the most rel-evant explanation may be that homebuyers have imperfect information aboutthe property tax system and thereby form incorrect expectations about theirfuture tax obligations” (Bradley 2011 p. 14). Another possible explanation isthat homebuyers do not go to the trouble of evaluating every line item in theirtotal property tax bill when purchasing a home (e.g., county tax, municipal tax,water district tax, school district tax, recreation district tax, development districttax, etc.). Homebuyers simply look at the bottom line tax liability and proceedwith the purchase when the mortgage underwriter concludes the monthly hous-ing expense falls within acceptable underwriting guidelines. This result hasimportant public policy implications (e.g., should government require explicitdisclosure of the tax liabilities associated with owner-occupied housing). Sec-ond, most of the existing capitalization literature examines the extent to which
2For more information on special districts, see Galvan (2007) and Griffith (2007).3One of the authors had numerous conversations with residential developers and specialdistrict bond underwriters in Colorado. These individuals unamimously concluded thatno residential infrastructure development since TABOR was financed with municipalbonds.
Financing Residential Development with Special Districts 3
the property tax liability associated with ongoing public services (e.g., educa-tion, police and fire protection, etc.) is capitalized in house prices. Propertytaxes that pay for local public services may be positively capitalized in houseprices, negatively capitalized or have no influence on house prices dependingon how taxpayers value the benefits provided by those services.4 Developmentdistrict bonds have finite maturities and finite repayment obligations. This re-search avoids the homeowner valuation of benefits problem and examines howthe finite-term property tax liability is capitalized in house prices. Third, weempirically illustrate the importance of controlling for spatial attributes by es-timating parameters for several hedonic house price specifications that imposeincreasingly more restrictive controls for neighborhood amenities. We test fordifferences across specifications using a spatial application of the Wald test.Fourth, to control for possible correlations between unobserved neighborhoodcharacteristics and tax rates, we instrument property tax rates using develop-ment district attributes. The validity of the instruments depends on the selectionof submarkets (rather than on the choice of variables). Valid instruments forthe property tax rate are identified by restricting the sample of transactions torapidly growing suburban residential developments.
The remainder of this paper has six sections. The next section provides a briefreview of the relevant literature. The third section provides an overview ofhow special districts are created and provides information on the debt issuedby some Denver area development districts. The fourth section describes ourempirical approach and data. The fifth and sixth sections provide results andconcluding remarks.
Literature Review
There are three branches of literature that are relevant for this research: (1) theliterature that compares cost sharing techniques to land use extraction methodsfor financing residential development, (2) the literature that examines whetherland use extraction methods, like development or impact fees, influence landand house prices and (3) the property tax capitalization literature.
Brueckner (1997) theoretically compares impact fees to two cost-sharing tech-niques for financing development: one where all taxpayers in a jurisdictionpay the infrastructure cost at the time the cost is incurred and a second wheretaxpayers finance the cost. He concludes that since impact fees are paid by theeventual consumer (e.g., the homeowner), impact fees are more efficient thancost-sharing financing techniques. He also concludes that impact fees generate
4See Billings and Thibodeau (2011) for more discussion on measuring property taxcapitalization for net benefits.
4 Billings and Thibodeau
slower rates of growth and yield higher aggregate property values relative tocost-sharing methods. Peiser (1983) compares patterns of residential devel-opment in two Texas cities: Houston, a city that uses local municipal utilitydistricts (MUDs) extensively, and Dallas, a city that relies on much larger,regional utility districts (RUDs) for financing residential development. Peiserargues that economies of scale make RUDs more efficient than MUDs in thelong run, but that RUDs also constrain the supply of developable land andincrease land prices in the short run. He argues that MUDs permit develop-ers to use less expensive land to develop lots and market competition forcesdevelopers to pass those cost savings onto homebuyers. Finally, he concludesthat MUDs are particularly effective at supporting rapid, short-term, residentialdevelopment.
A number of papers have empirically examined the influence that land useextraction fees have on land and house prices.5 Delaney and Smith (1989)compare house prices in Dunedin, Florida, a city that began using impact fees tofinance residential infrastructure in June of 1974, to house prices in Clearwater,Largo and St. Petersburgh, three Florida cities that did not implement impactfees. They use a hedonic house price model to estimate coefficients in theseplaces and then “price” a fixed bundle of housing characteristics across cities.They conclude that the Dunedin impact fee increased house prices by an amountthat significantly exceeded the fee. Singell and Lillydahl (1990) report thathouse prices in Loveland, Colorado, increased by 7% after that city imposedimpact fees in July of 1984. Yinger (1998) notes that the Delaney and Smith(1989) study may have overestimated the effect of impact fees because theauthors held land prices constant when they priced housing characteristicsacross the four markets. If impact fees are anticipated by developers, thendevelopers will reduce the price they pay for raw land. Yinger (1998) also notedthat both the Delaney and Smith (1989) study and the Singell and Lillydahl(1990) study did not adequately control for the influence that neighborhoodattributes have on house prices. Using data on undeveloped land prices andprices for new and existing homes in Dade County, Florida, Ihlanfeldt andShaughnessy (2004) report that a dollar of impact fee reduces land price by anequivalent amount and increases the price of both new and existing housing by$1.60.
Beginning with Oates (1969), there are numerous papers that examine whetherproperty taxes are capitalized in house prices.6 Yinger et al. (1988) summarizes
5See Evans-Cowley and Lawhon (2003) for a more complete review of this literature.6See Brueckner (1979), Starrett (1981), Yinger (1982), Yinger et al. (1988), Zodrow(2006) and Bradley (2011) for analysis and additional references on property taxcapitalization.
Financing Residential Development with Special Districts 5
the econometric approaches, data (e.g., aggregate vs. micro) and resultingproperty tax capitalization rates reported by 30 empirical studies of property taxcapitalization published prior to 1988. Based on different methodologies, typesof local governments (e.g. municipalities, school districts, special districts),study areas, the assumption of a 3% discount rate and an infinite time horizonof the tax liability, estimates of property tax capitalization range from 15%capitalization to 120%.
Development Districts
Development districts are playing an increasingly important role in financingresidential infrastructure. Extending water, sewer and drainage facilities to un-developed parts of a jurisdiction can be very expensive. Historically, residentialinfrastructure has been financed using cost-sharing methods, like general obli-gation municipal revenue bonds, municipal impact fees or the developer’s funds(using some combination of short-term construction loans and the developer’sequity). Cost-sharing methods typically involve bond financing by an existinggeneral purpose municipal or county government. These general obligationbonds are repaid by all property owners in the jurisdiction. With cost-sharing,the consumers of the new infrastructure are subsidized by all taxpayers in thejurisdiction. Impact fees are upfront fees paid by the developer to cover in-frastructure costs. These fees increase the cost of developing residential lots,increase lot prices and eventually increase house prices (relative to cost-sharingalternatives). Some developers have the capital capacity to finance infrastruc-ture development themselves. Residential infrastructure can be financed withshort-term construction loans, with the developer’s own funds or with somecombination of both.
Table 1 illustrates the growth in special districts, in general, and in developmentdistricts, in particular, over the 1992–2002 period. A development district isany special district that provides utilities, sewerage, solid waste management,water or a natural resource function such as irrigation or flood control. NewMexico added 512 special districts between 1992 and 2002–a four-fold increasein the number of special districts over this ten-year period. Of these, 510 weredevelopment districts.
Special districts offer local governments and developers the opportunity tofinance residential infrastructure with relatively inexpensive long-term debtwhile passing the obligation to repay that debt directly onto the ultimateconsumer–the homeowner. Special districts in Colorado were first authorizedby Title 32 of the Colorado Revised Statute to extend public services to ru-ral and unincorporated parts of the state. They became increasing popular inColorado following the passage of the Taxpayers’ Bill of Rights (TABOR) in
6 Billings and Thibodeau
Table 1 � Growth in special/development districts, by state: 1992–2002.
Special Districts Development Districts
1992–2002 1992–2002Total Change Total Change
1992 2002 Number Percent 1992 2002 Number Percent
New Mexico 116 628 512 441% 94 604 510 543%Wisconsin 377 684 307 81% 192 511 319 166%
Illinois 2,920 3,145 225 8% 1,185 1,263 78 7%Indiana 939 1,125 186 20% 247 319 72 29%
Wyoming 373 546 173 46% 213 301 88 41%Florida 462 626 164 35% 200 350 150 75%
Colorado 1,252 1,414 162 13% 632 708 76 12%Georgia 421 581 160 38% 84 177 93 111%
New York 980 1,135 155 16% 13 19 6 46%Iowa 388 542 154 40% 289 333 44 15%
1992. TABOR limits the growth in government revenues to the sum of thepopulation growth rate plus the rate of inflation. Any increase in governmentrevenues over this limit must either be refunded to taxpayers or approvedby popular vote in a general election. TABOR applies to the entire State ofColorado as well as to all counties and municipalities within the state. TABORhas made it exceedingly difficult for local jurisdictions to finance residentialdevelopment using general obligation bonds, as increasing taxes above theTABOR limit requires the approval of a majority of all residents.7
What effect does development district infrastructure financing have on houseprices? We expect Colorado house prices for homes built in development dis-tricts to be lower than house prices for similar homes built outside developmentdistricts. Since TABOR, cost-sharing methods (e.g., municipal bond financing)are no longer a viable financing vehicle in Colorado. The primary ways to payfor the construction of residential infrastructure are: (1) at the time of con-struction using (short-term) developer financing or impact fees or (2) financinginfrastructure development with (long-term) development district bonds. Whendevelopers finance the construction of residential infrastructure, they will bereimbursed for this expense when the developed lots are sold. Higher pricesfor improved lots yield higher prices for completed homes. Similarly, impactfees/connect charges are passed along to homebuyers as part of the construc-tion cost. Alternatively, when development district bonds are used to finance
7As of 2005, 29 states had Tax and Expenditure Limits (TELs) like TABOR that providelimits on local government finances (NCSL 2008). A few scholars (e.g., Nelson 1990and Carr 2006) directly link TELs to the creation of special district governments.
Financing Residential Development with Special Districts 7
infrastructure, repaying this debt is spread out over a long period of time (e.g.,30 years) and the liability for repaying the debt is passed onto homeowners.This future property tax liability should be (negatively) capitalized in houseprices.
The choice between direct developer and special district financing is idiosyn-cratic to individual developers. The use of development district bonds dependson the developer’s understanding of the process for forming a special district,on prevailing interest rates for both construction loans and special district bondsas well as on the developer’s liquidity constraints and tax liabilities. In addi-tion, the process of underwriting special district bonds carefully examines thefinancial health/reputation of the developer. Typically, only developers withstrong balance sheets and favorable reputations for completing residential landdevelopments are approved for special district bond financing.
In Colorado, special districts are local governments created to provide at leastone of the following services: water facilities or services, sanitation facilitiesor services, flood control, streets, roads, alleys, walkways, transit, parks andrecreation facilities, golf courses, fire protection, security, insect and animalcontrol, libraries, emergency medical, hospitals, soil and water conservation,television relay and transmission facilities or open space. Special districtsthat provide multiple services are labeled metropolitan special districts. Realestate developers frequently use special districts to finance the constructionof the infrastructure needed to support vertical residential and commercialdevelopment.
Special districts are created in four steps. First, district property owners mustpetition their local government(s) (e.g., municipality or county) to establisha district. If the petition is approved, the district’s governing body, usuallya board of directors elected by property owners, must submit a service planto all counties and municipalities that have jurisdiction over the district. Theservice plans typically include district maps and agreements that explicitlydescribe which jurisdiction (e.g., county, municipality or service district) willprovide what services. Third, the proposed district must be approved in anelection of district property owners by either 20% of the eligible voters orby 200 voters, whichever is smaller. In the case of special districts used tofinance infrastructure, most of the eligible voters are district land owners. It ispossible for the population of “eligible voters” to consist of one taxpayer–theland-owner/developer. Finally, once the special district is approved by districttaxpayers and by all local governments that have jurisdiction, the district isfiled with the local court and with the county assessor. After the district is filedwith the local court, it may issue bonds. Special districts created to financeinfrastructure usually have finite lives.
8 Billings and Thibodeau
We define a development district as a metropolitan district with no single-familyproperties prior to the creation of the district that provides at least two of thefollowing facilities or services: public improvement, street maintenance, publicworks, water facilities, sewer, storm drainage or transportation. Table 2 lists 45“high growth” development districts created in the Denver metropolitan areabetween 1980 and 2002 that have issued bonds to finance the construction ofresidential infrastructure.8 A development district is labeled high growth if itis located in a Census Block Group (CGB) that had at least 20 recently builtsingle-family homes sold between 2002 and 2004 and if those 20 propertiesrepresented at least 20% of the year 2000 CBG housing stock. The tableprovides the district name, location, size (in square miles), the year the districtwas created and the 2003 development district property tax rate. Nearly three-quarters (33 of 45) of the development districts are located in Douglas andAdams Counties, two rapidly growing suburban counties located south andeast of the City/County of Denver. Development district sizes range from0.35 square miles (224 acres) to over nine square miles (over 5,760 acres).The average district size is 2.65 square miles. About half (22 of 45) of thedevelopment districts were created in the 1980s, 19 were created in the 1990sand four were created after 2000. The across development district averageproperty mill levy in 2003 was 35.7 mills, and development district propertytax rates ranged from six to 90 mills.
Table 3 provides descriptive information for the bonds issued by these devel-opment districts. The table includes the bond CUSIP9 code, the issue date,maturity, term, the aggregate amount of the bond (or bonds), whether thecoupon rate was fixed or variable and, for fixed-rate bonds, the interest rate.While all of the development districts in Table 2 issued bonds to finance infras-tructure, some of these bonds were privately placed. Consequently, descriptiveinformation for these district bonds is not publicly available. Table 3 providesinformation for the 33 districts that sold bonds to the public. Some of thedistricts issued multiple bonds. For these districts, the CUSIP, issue date andmaturity refer to the most recent issue prior to 2004, while the bond amount isthe total amount of all bonds issued by the district. Bond amounts range from$1.2M to $72.2M. The average amount issued was $12.2M. Bond maturitiesrange from six to 40 years. Seventy-five percent of the bonds had maturitiesbetween 20 and 30 years. The average maturity was 24 years. Most of the bondspaid a fixed coupon. The average fixed-rate coupon was 6.87%. Any concern
8There were five special/metropolitan districts that were classified as development dis-tricts but were excluded from Table 2 because they did not issue any bonds as of 2004.9The CUSIP (Committee on Uniform Security Identification Procedures) identifier pro-vides an unique code to allow identification of the issuer of any North American security.
Financing Residential Development with Special Districts 9
Tabl
e2
�M
etro
polit
anD
enve
rde
velo
pmen
tdis
tric
ts.
Dat
e20
03N
eare
st/in
Are
aD
istr
ict
Prop
erty
Dev
elop
men
tDis
tric
tNam
eC
ount
yC
ity/T
own
(SqM
i)C
reat
edM
illL
evy
Aur
ora
Sing
leT
ree
Met
ropo
litan
Dis
tric
tA
dam
sA
uror
a4.
2119
9842
.6B
elle
Cre
ekM
etro
polit
anD
istr
ict
Ada
ms
Com
mer
ceC
ity2.
3020
0047
.9B
righ
ton
Cro
ssin
gM
etro
polit
anD
istr
ictN
o.4
Ada
ms
Bri
ghto
n7.
7819
8538
.0B
road
land
sM
etro
polit
anD
istr
ictN
o.1
Ada
ms
Bro
omfie
ld0.
3919
9720
.0B
rom
ley
Park
Met
ropo
litan
Dis
tric
tNo.
2A
dam
sB
righ
ton
6.90
1985
38.0
Buf
falo
Rid
geM
etro
polit
anD
istr
ict
Ada
ms
Com
mer
ceC
ity2.
3019
9637
.3B
uffa
loR
unM
esa
Met
ropo
litan
Dis
tric
tA
dam
sC
omm
erce
City
2.30
2002
42.0
Can
terb
erry
Cro
ssin
gM
etro
polit
anD
istr
ict
Dou
glas
Park
er1.
3919
9639
.5C
ante
rber
ryC
ross
ing
Met
ropo
litan
Dis
tric
tII
Dou
glas
Park
er0.
6420
0037
.3C
herr
yC
reek
Sout
hM
etro
polit
anD
istr
ictN
o.1
Dou
glas
Park
er2.
5719
8539
.8C
ompa
rkB
usin
ess
Cam
pus
Met
ropo
litan
Dis
tric
tD
ougl
asPa
rker
0.35
1998
45.2
Eag
leB
end
Met
ropo
litan
Dis
tric
tNo.
2A
rapa
hoe
Aur
ora
0.76
1998
45.3
Eag
leC
reek
Met
ropo
litan
Dis
tric
tA
dam
sC
omm
erce
City
2.30
1996
50.0
Eas
tSm
oky
Hill
Met
ropo
litan
Dis
tric
tNo.
2A
rapa
hoe
Aur
ora
0.49
1994
17.0
Ebe
rtM
etro
polit
anD
istr
ict
Den
ver
Den
ver
0.47
1982
39.0
Gat
eway
Reg
iona
lMet
ropo
litan
Dis
tric
tA
dam
sA
uror
a9.
1019
9810
.0G
oodm
anM
etro
polit
anD
istr
ict
Ara
paho
eFo
xfiel
d0.
8019
9426
.5G
VR
Met
ropo
litan
Dis
tric
tD
enve
rD
enve
r5.
3619
8330
.3H
ighl
ands
Ran
chM
etro
polit
anD
istr
ictN
o.1
Dou
glas
Litt
leto
n4.
4019
8020
.3H
ighl
ands
Ran
chM
etro
polit
anD
istr
ictN
o.3
Dou
glas
Lon
eT
ree
4.54
1980
20.3
Hig
hlan
dsR
anch
Met
ropo
litan
Dis
tric
tNo.
4D
ougl
asL
ittle
ton
4.37
1980
20.3
Inte
rloc
ken
Con
solid
ated
Met
ropo
litan
Dis
tric
tB
ould
erB
room
field
1.16
1994
27.2
Lin
coln
Park
Met
ropo
litan
Dis
tric
tD
ougl
asPa
rker
0.59
1983
45.0
10 Billings and Thibodeau
Tabl
e2
�C
ontin
ued
Dat
e20
03N
eare
st/in
Are
aD
istr
ict
Prop
erty
Dev
elop
men
tDis
tric
tNam
eC
ount
yC
ity/T
own
(SqM
i)C
reat
edM
illL
evy
Mah
erR
anch
Met
ropo
litan
Dis
tric
tNo.
4D
ougl
asC
astle
Roc
k2.
8119
8740
.0M
eado
ws
Met
ropo
litan
Dis
tric
tNo.
1D
ougl
asC
astle
Roc
k0.
7919
8535
.0M
eado
ws
Met
ropo
litan
Dis
tric
tNo.
2D
ougl
asC
astle
Roc
k1.
7319
8535
.0M
eado
ws
Met
ropo
litan
Dis
tric
tNo.
6D
ougl
asC
astle
Roc
k0.
7319
8535
.0M
eado
ws
Met
ropo
litan
Dis
tric
tNo.
7D
ougl
asC
astle
Roc
k1.
7319
8535
.0N
orth
Ran
geV
illag
eM
etro
polit
anD
istr
ict
Ada
ms
Com
mer
ceC
ity2.
3019
9938
.5Pa
rker
Prop
ertie
sM
etro
polit
anD
istr
ictN
o.1
Dou
glas
Park
er1.
4719
8524
.0Pi
nery
Wes
tMet
ropo
litan
Dis
tric
tNo.
2D
ougl
asC
astle
Roc
k3.
3719
9845
.0Po
tom
acFa
rms
Met
ropo
litan
Dis
tric
tA
dam
sC
omm
erce
City
2.30
2001
38.0
Riv
erda
leD
unes
Met
ropo
litan
Dis
tric
tNo.
1A
dam
sC
omm
erce
City
2.30
1996
45.0
Rox
boro
ugh
Vill
age
Met
ropo
litan
Dis
tric
tD
ougl
asL
ittle
ton
1.84
1985
70.1
Seco
ndC
reek
Ran
chM
etro
polit
anD
istr
ict
Den
ver
Aur
ora
4.72
1985
90.0
Ster
ling
Hill
sM
etro
polit
anD
istr
ict
Ara
paho
eA
uror
a0.
6019
9430
.0St
oneg
ate
Vill
age
Met
ropo
litan
Dis
tric
tD
ougl
asPa
rker
0.77
1983
27.4
Supe
rior
Met
ropo
litan
Dis
tric
tNo.
2B
ould
erL
ouis
ville
1.79
1988
22.0
Tally
n’s
Rea
chM
etro
polit
anD
istr
ictN
o.2
Ara
paho
eA
uror
a9.
3319
9849
.0Ta
llyn’
sR
each
Met
ropo
litan
Dis
tric
tNo.
3A
rapa
hoe
Aur
ora
9.33
1998
49.0
Todd
Cre
ekFa
rms
Met
ropo
litan
Dis
tric
tNo.
1A
dam
sT
horn
ton
1.95
1994
16.5
Todd
Cre
ekFa
rms
Met
ropo
litan
Dis
tric
tNo.
2A
dam
sT
horn
ton
1.95
1996
18.0
Tow
nC
ente
rM
etro
polit
anD
istr
ict
Den
ver
Den
ver
0.88
1983
39.0
Upp
erC
herr
yC
reek
Met
ropo
litan
Dis
tric
tD
ougl
asPa
rker
0.70
1985
6.0
Vill
ages
AtC
astle
Roc
kM
etro
polit
anD
istr
ict
Dou
glas
Cas
tleR
ock
3.37
1984
40.0
Ave
rage
2.65
1994
35.7
Financing Residential Development with Special Districts 11
Tabl
e3
�D
evel
opm
entd
istr
ictb
onds
.
Bon
dIs
sue
Term
Rat
eFi
xed/
Inte
rest
Dev
elop
men
tDis
tric
tNam
eC
USI
PD
ate
Mat
urity
(Yea
rs)
Am
ount
Var
iabl
eR
ate
Aur
ora
Sing
leT
ree
Met
ropo
litan
Dis
tric
t05
206N
AA
810
/5/2
000
11/1
5/20
2525
$1,5
75,0
00Fi
xed
8.00
0%B
elle
Cre
ekM
etro
polit
anD
istr
ict
0783
41A
A2
10/1
/200
012
/1/2
020
20$5
,215
,000
Fixe
d8.
000%
Bri
ghto
nC
ross
ing
Met
ropo
litan
Dis
tric
tNo.
410
9330
AA
812
/17/
2004
12/1
/203
430
$13,
800,
000
Var
iabl
eB
road
land
sM
etro
polit
anD
istr
ictN
o.1
1113
2NA
A5
8/6/
1998
8/1/
2018
20$4
,925
,000
Var
iabl
eB
uffa
loR
idge
Met
ropo
litan
Dis
tric
t11
9802
AA
412
/1/2
003
12/1
/203
330
$10,
130,
000
Fixe
d7.
500%
Can
terb
erry
Cro
ssin
gM
etro
polit
anD
istr
ict
1381
00A
E2
12/2
7/20
0112
/1/2
031
30$1
0,43
0,00
0Fi
xed
6.70
0%C
ante
rber
ryC
ross
ing
Met
ropo
litan
Dis
tric
tII
1381
01A
B6
12/1
/200
212
/1/2
032
30$7
,500
,000
(a)
Fixe
d(a
)C
herr
yC
reek
Sout
hM
etro
polit
anD
istr
ictN
o.1
1645
61A
C6
5/15
/200
312
/15/
2033
30$1
0,06
0,00
0(b)
Fixe
d(b
)E
agle
Ben
dM
etro
polit
anD
istr
ictN
o.2
2694
22A
A9
10/1
/199
912
/1/2
018
19$7
,500
,000
Fixe
d6.
875%
Eag
leC
reek
Met
ropo
litan
Dis
tric
t26
9492
AA
211
/1/2
001
6/1/
2021
20$2
,000
,000
Fixe
d6.
750%
Eas
tSm
oky
Hill
Met
ropo
litan
Dis
tric
tNo.
227
5292
AD
28/
31/2
000
12/1
/203
030
$9,6
00,0
00(c
)Fi
xed
(c)
Gat
eway
Reg
iona
lMet
ropo
litan
Dis
tric
t36
779T
AA
75/
15/2
000
12/1
/201
010
$1,2
35,0
00Fi
xed
6.50
0%G
oodm
anM
etro
polit
anD
istr
ict
3823
8EA
A1
6/1/
1999
12/1
/201
920
$5,0
00,0
00Fi
xed
5.00
0%G
VR
Met
ropo
litan
Dis
tric
t36
2374
CL
212
/15/
1999
12/1
/201
920
$13,
165,
000(
d)Fi
xed
(d)
Hig
hlan
dsR
anch
Met
ropo
litan
Dis
tric
tNo.
143
1096
EW
26/
1/19
979/
1/20
1215
$19,
415,
000(
e)Fi
xed
(e)
Hig
hlan
dsR
anch
Met
ropo
litan
Dis
tric
tNo.
343
0905
BY
46/
1/19
9912
/1/2
019
20$1
3,07
5,00
0(f)
Fixe
d(f
)H
ighl
ands
Ran
chM
etro
polit
anD
istr
ictN
o.4
4309
03B
B9
4/1/
1997
12/1
/201
720
$5,3
90,0
00(g
)Fi
xed
(g)
Inte
rloc
ken
Con
solid
ated
Met
ropo
litan
Dis
tric
t45
8761
BE
211
/15/
1999
12/1
5/20
1920
$72,
215,
869(
h)Fi
xed
(h)
Lin
coln
Park
Met
ropo
litan
Dis
tric
t53
4550
AB
46/
29/2
001
12/1
/202
625
$16,
610,
000
Fixe
d7.
750%
Mah
erR
anch
Met
ropo
litan
Dis
tric
tNo.
455
9827
AB
66/
1/20
0312
/1/2
027
25$1
3,10
0,00
0(i)
Fixe
d(i
)M
eado
ws
Met
ropo
litan
Dis
tric
tNo.
158
2903
BU
39/
1/19
896/
1/20
2940
$30,
730,
000
Fixe
d7.
990%
Mea
dow
sM
etro
polit
anD
istr
ictN
o.2
5829
03B
W9
9/1/
1989
6/1/
2029
40$1
5,44
0,00
0Fi
xed
7.99
0%N
orth
Ran
geV
illag
eM
etro
polit
anD
istr
ict
6617
66A
A3
8/1/
2000
12/1
/202
020
$2,8
00,0
00Fi
xed
8.00
0%Pi
nery
Wes
tMet
ropo
litan
Dis
tric
tNo.
272
3295
AA
912
/30/
2002
11/1
/203
230
$21,
815,
000
Fixe
d3.
700%
12 Billings and Thibodeau
Tabl
e3
�C
ontin
ued
Bon
dIs
sue
Term
Rat
eFi
xed/
Inte
rest
Dev
elop
men
tDis
tric
tNam
eC
USI
PD
ate
Mat
urity
(Yea
rs)
Am
ount
Var
iabl
eR
ate
Poto
mac
Farm
sM
etro
polit
anD
istr
ict
7376
8TA
A3
3/1/
2002
12/1
/203
230
$4,3
15,0
00Fi
xed
6.85
0%R
iver
dale
Dun
esM
etro
polit
anD
istr
ictN
o.1
7686
43A
A6
1/1/
2002
12/1
/203
230
$3,8
00,0
00Fi
xed
6.50
0%R
oxbo
roug
hV
illag
eM
etro
polit
anD
istr
ict
7798
14B
G6
9/1/
1993
12/3
1/20
2128
$6,2
47,6
29N
otR
epor
ted
Not
Rep
orte
dSt
erlin
gH
ills
Met
ropo
litan
Dis
tric
t85
933L
AB
211
/1/1
998
6/1/
2018
20$2
,000
,000
Fixe
d7.
750%
Ston
egat
eV
illag
eM
etro
polit
anD
istr
ict
8618
19B
J811
/1/1
996
12/1
/202
529
$42,
330,
000(
j)Fi
xed
(j)
Supe
rior
Met
ropo
litan
Dis
tric
tNo.
286
820P
AK
76/
18/1
998
12/1
/201
315
$9,1
90,0
00Fi
xed
4.62
5%To
ddC
reek
Farm
sM
etro
polit
anD
istr
ictN
o.1
8890
1PO
A2
11/1
6/20
0112
/1/2
021
20$1
2,50
0,00
0Fi
xed
5.00
0%To
ddC
reek
Farm
sM
etro
polit
anD
istr
ictN
o.2
8890
13A
S76/
1/20
0312
/1/2
018
15$2
,755
,000
(k)
Fixe
d(k
)U
pper
Che
rry
Cre
ekM
etro
polit
anD
istr
ict
9156
18C
C4
5/15
/199
612
/1/2
011
15$1
,705
,000
(l)
Fixe
d(l
)A
vera
ge24
Not
es:
a.T
his
amou
ntis
the
resu
ltof
two
bond
s[$
2,50
0,00
0at
7.00
%an
d$5
,000
,000
at7.
375%
].b.
Thi
sam
ount
isth
ere
sult
oftw
obo
nds
[$3,
511,
000
at1.
4%an
d$6
,549
,000
at1.
4%].
c.T
his
amou
ntis
the
resu
ltof
thre
ebo
nds
[$1,
430,
000
at3.
8%,$
3,39
0,00
0at
4.90
%an
d$4
,780
,000
at5.
00%
].d.
Thi
sam
ount
isth
ere
sult
of15
bond
sra
ngin
gfr
om$4
20,0
00to
$4,6
30,0
00w
ithin
tere
stra
tes
rang
ing
from
4.05
%to
5.75
%.
e.T
his
amou
ntis
the
resu
ltof
15bo
nds
rang
ing
from
$470
,000
to$5
,005
,000
with
inte
rest
rate
sra
ngin
gfr
om4.
5%to
5.75
%.
f.T
his
amou
ntis
the
resu
ltof
11bo
nds
rang
ing
from
$465
,000
to$4
,890
,000
with
inte
rest
rate
sra
ngin
gfr
om4.
25%
to5.
3%.
g.T
his
amou
ntis
the
resu
ltof
14bo
nds
rang
ing
from
$155
,000
to$2
,310
,000
with
inte
rest
rate
sra
ngin
gfr
om4.
2%to
6.3%
.h.
Thi
sam
ount
isth
ere
sult
of11
bond
sra
ngin
gfr
om$2
05,0
00to
$46,
469,
778
with
inte
rest
rate
sra
ngin
gfr
om5.
0%to
5.75
%.
i.T
his
amou
ntis
the
resu
ltof
thre
ebo
nds
rang
ing
from
$820
,000
to$6
,245
,000
with
inte
rest
rate
sra
ngin
gfr
om7.
7%to
7.87
5%.
j.T
his
amou
ntis
the
resu
ltof
16bo
nds
rang
ing
from
$400
,000
to$1
0,66
0,00
0w
ithin
tere
stra
tes
rang
ing
from
4.35
%to
5.6%
.k.
Thi
sam
ount
isth
ere
sult
of16
bond
sra
ngin
gfr
om$1
20,0
00to
$250
,000
with
inte
rest
rate
sra
ngin
gfr
om2.
0%to
3.75
%.
l.T
his
amou
ntis
the
resu
ltof
five
bond
sra
ngin
gfr
om$1
0,00
0to
$1,0
05,0
00w
ithin
tere
stra
tes
rang
ing
from
5.37
5%to
6.75
%.
Financing Residential Development with Special Districts 13
that the use of development districts is determined by the attributes of a resi-dential development (e.g., land area or number of housing units) is addressedin Table 2, which highlights a range of development district attributes.
For some districts, there is a big disparity between the date the district is createdand the time the bonds are issued. Development district bonds are issued justprior to (and sometimes just after) infrastructure development begins. Oncethe bonds are issued, the bond repayment clock starts and properties must bebuilt, sold, registered on the assessor’s file and begin generating property taxrevenues to repay the debt. Development district bonds are typically createdwith interest reserve accounts adequate to pay interest for the first few yearsof the bond. Interest reserve accounts may be created with terms of up tothree years.
As an example, the Brighton Crossing Metropolitan District was created in1985. The development district covers 7.78 square miles in Adams County,Colorado. Within this area, a developer has proposed a residential developmentencompassing 771 acres. The developer plans to build 3,555 single-familyhomes on 547 acres and 927 multifamily units on 91 acres. The developer allo-cated 108 acres to parks, trails and recreation facilities and 25 acres to streetsand walks. The water, sewer and flood control infrastructure for this develop-ment costs $45.7M. The district issued its first bond for this development inDecember 2004–20 years after the district was created. The bond amount was$13.8M and the bond maturity was 30 years. The Brighton Crossing Metropoli-tan District tax rate was 38 mills in 2003.
Empirical Approach
This paper empirically examines the effect that financing residential infras-tructure using development districts has on house prices. We employ a hedonichouse price model to examine the extent to which development district propertytaxes are capitalized in house prices.10 We begin with our hedonic house pricemodel. Let the natural log of the price for house i in period t be represented by:
ln(Pi,t ) = α + βXi + ηNi + γ Gi + τTi + �Nj=2δjDi,j + εi,t (1)
where
Pi,t = the transaction price for house i in period t;
Xi = a vector of structural characteristics for property i;
10Yinger et al. (1988) and Palmon and Smith (1998) incorporate a non-linear specifica-tion for estimating property tax capitalization. While this specification is well motivated,it requires assumptions or estimates of annual user costs/rental value for a property. Thelack of rental data for a comparable stock of new homes precludes this technique.
14 Billings and Thibodeau
Ni = a vector of neighborhood characteristics for property i;
Gi = a vector of government characteristics for property i;
Ti = the property tax mill rate;
Di,j = 1 if property i sold in period j (j = 2, . . . , N), and equals zero otherwise;
α = the hedonic equation intercept;
β = a vector of structural characteristic hedonic coefficients;
η = a vector of neighborhood characteristic hedonic coefficients;
γ = the vector of government characteristic hedonic coefficients;
τ = the hedonic coefficient for the property tax rate;
δj = hedonic coefficients for sale period j binary variable (j = 2, . . . , N); and
εi,t = error term for property i in period t.
The structural characteristics include the usual variables: lot size, square feetof living space, number of bathrooms, etc.11 Since many of the neighborhoodamenities that influence house prices are not observed, we control for spatialvariation in neighborhood characteristics using Census Block Group (CBG)fixed effects.12 The incorporation of neighborhood-level fixed effects controlfor any neighborhood characteristic that is observable or unobservable (e.g.,type and density of land use, distance to the Central Business District (CBD)and household incomes). This limits identification to only intra-CBG variation.Controls using neighborhood effects is well suited for examining infrastructurefinancing districts because they lack potential spillovers across developmentdistrict boundaries. The roads, curb, gutter, water and sewer constructed withina development district are typically not accessed or used by neighboring res-idential developments. Given the large number and spatial variation of localgovernments in Colorado highlighted by Billings and Thibodeau (2011), wealso include dummy variables for other, non-development district, special dis-tricts and municipalities that provide services.
Our data set consists of 34,048 transactions of recently built single-familyhomes sold in the Denver metropolitan area. Properties were recently builtand subsequently sold during the 2002–2004 period. The Denver-Boulder-Greeley Consolidated Metropolitan Statistical Area (CMSA) consists of the
11The hedonic specification excludes dwelling age since all of the homes are new (e.g.,less than three years old).12As shown by Thibodeau (2003), segmenting single-family home markets into smallergeographic areas controls for spatial autocorrelation.
Financing Residential Development with Special Districts 15
City/County of Denver, its bedroom communities and nearby employmentcenters.13 A transaction is classified as a new property if it was less than orequal to three years old at the time of sale. In addition, the data are limited toproperties situated on between 0.05 and five acre lots and with transaction pricesbetween $100, 000 and $1, 000, 000. All data on special districts is current asof 2003. The original government dataset was provided by the Department ofLocal Affairs, State of Colorado, in shapefiles for jurisdictional boundaries andelectronically for mill levies. It was further supplemented with paper recordsfrom files in the Denver, CO, office.14
According to the Colorado Department of Local Governments, there were672 special districts and 377 metropolitan districts in the Denver metropolitanarea in 2003. We classified 84 of these as development districts. Forty-fourof these development districts were created since 1993 and the remaining40 development districts pre-date 1993. During the 20-year period prior toTABOR, development districts in Colorado were created at an annual rate of1.9 per year. Following TABOR, development districts were created at theannual rate of 4.4 districts per year. More recently created development districtsare likely not fully captured in the dataset due to the time lag between thecreation of a development district and the sale of a new home.
Since property transactions and development districts are based on the entireDenver urban area, the observations included in the empirical analysis representthe full choice set for households in the market for new homes. This representsan improvement over existing studies, which commonly limit hedonic esti-mates to a single county (see Yinger et al. 1988). The use of a single countyhas the potential to generate varying elasticities of demand and property taxcapitalization depending on the housing stock and on the property tax burden ofresidential units available in neighboring counties within the same metropolitanarea. Ignoring this spatial dependency between neighboring counties will leadto inefficient estimators and even biased regression coefficients.15
Yinger et al. (1988) note two major criticisms of studies that estimateproperty tax capitalization: (1) the lack of controls for the public services
13Denver and Broomfield are both cities and counties, integrated into a single governmentinstitution and treated as a county in this dataset. The Denver-Boulder-Greeley CMSAconsists of Adams, Arapahoe, Boulder, Broomfield, Denver, Douglas, Jefferson andWeld counties.14All Geographical Information Systems datawork is implemented using the ColoradoCentral Zone State Plane NAD83 Projection.15See Klotz (2004) for a discussion of the implications of spatial dependence in regres-sion models.
16 Billings and Thibodeau
financed by the property tax and (2) failure to adequately control for differ-ences in neighborhood characteristics associated with different property taxingjurisdictions. The first concern is partially mitigated by focusing on newly con-structed homes and property taxation associated with financing developmentinfrastructure. A series of control variables for the existence of other municipalor non-development special districts will further alleviate this concern. To ad-dress the second criticism, we control for variation in neighborhood amenitiesthat influence residential property values by including fixed effects for CBGs.To further control for unobserved neighborhood characteristics that may be cor-related with property tax rates, we spatially subsample all single-family salesbased on CBGs that experience high rates of new residential construction. Wepartition the data into three increasingly restrictive subsamples: the first consist-ing of all 2002–2004 sales of recently constructed homes built on lots between0.05 and five acres with transaction prices between $100,000 and $1,000,000;a second sample limited to recently built properties located in CBGs that hadat least 20 sales over the 2002–2004 period and with those sales representingat least 20% of the CBG housing stock (as reported by the 2000 Census); anda third sample that further restricts the data to only those transactions in CBGsthat included at least 20 sales located both within and outside developmentdistricts and with these sales representing at least 20% of the year 2000 CBGhousing stock.16 The first sample contains 34,048 transactions with 84 develop-ment districts and 654 CBGs. The second sample contains 27,869 transactionswith 69 development districts and 81 CBGs. The third sample contains 16,543transactions with 45 of the 50 development districts located in 23 CBGs listedin Table 2.
Table 4 provides summary statistics for the three samples. The top panel sum-marizes the data for all 2002–2004 transactions; the middle panel summarizesthe data for the CBGs with at least 20 sales and with sales representing atleast 20% of the year 2000 neighborhood housing stock; the bottom panelsummarizes the data for just those CBGs that have at least 20 sales with salesrepresenting at least 20% of the housing stock and with at least 20 transactionslocated both within and outside development districts. For the entire sampleof 34,048 sales, there were an average of 52.1 sales per CBG. The sales rateincreased to 344.1 sales per CBG for the second sample and to 722.2 salesper CBG for the third. Population densities decreased significantly as the datawere restricted to CBGs containing development districts as these places werein the process of building out. The population densities were 3,294 people persquare mile for the 654 CBGs, 1,109 for the 81 CBGs and 552 for the 23 CBGsin the most restrictive subsample. Spatially disaggregated subsamples better
16Conceptually, this third subsample involves matching neighboring developments thatare similar in size but vary in their implementation of development districts.
Financing Residential Development with Special Districts 17
Table 4 � Census block group summary statistics.
All Census 2000 Block GroupsMean Std. Dev. Min. Max.
Number Sold 2002–2004 52.1 181.4 1 2,711Sold in Development District 20.8 107.0 0 1,177Sold outside a Development District 31.3 109.1 0 1,715
Total Existing SF Homes 448.3 267.0 0 2,140Population 1,478 842 0 6,775Population Density (per square mile) 3,294 2,940 0 14,183N = 654
CBGs with > 20 sales; Sales > 20% of 2000 Census Housing StockMean Std. Dev. Min. Max.
Number Sold 2002–2004 344.1 408.7 37 2,711Sold in Development District 151.6 263.5 0 1,177Sold outside a Development District 192.5 255.4 0 1,715
Total Existing SF Homes 362.8 306.3 0 1,912Population 1,159 954 0 6,442Population Density (per square mile) 1,109 1,423 0 6,545N = 81
Many Sales Located Both in and out of Development DistrictsMean Std. Dev. Min. Max.
Number Sold 2002–2004 722.2 564.2 88 2,711Sold in Development District 403.9 361.7 20 1,177Sold outside a Development District 318.2 366.8 34 1,715
Total Existing SF Homes 550.5 412.6 47 1,912Population 1,677 1,335 125 6,442Population Density (per square mile) 552 598 5 1,635N = 23
controlled for differences in the number of homes sold within versus outside adevelopment district for a given CBG. A test for differences in means betweenthe number of homes sold inside and outside development districts was signifi-cant at the 5% level in only the full sample of CBGs.17 This indicates that thesesub-samples can adequately control for the size of new developments locatedboth inside and outside development districts within the same CBG. This sub-sampling becomes essential in later instrumental variable (IV) regressions inorder to construct valid instruments for the tax rate. Restricting our analysis tosimilar homes and residential developments allows us to remove correlations
17The complete data sample generated a p-value of 0.022; the middle sample p-value is0.266; the most restrictive sample p-value is 0.382.
18 Billings and Thibodeau
Table 5 � Variable definitions.
Variable Description
Sales Price Transaction price in 2002–2004ln(Sales Price) The natural log of transaction price in 2002–2004Price per Sqft The per square foot transaction price in 2002–2004Acres Lot size in acresBath Number of bathroomsSqft Total square footage of living areaSqf t2(000s) Total square footage of living area squared (thousands)Garage Indicator variable for a property with a garageBasement Indicator variable for a property with a basementForcedAirHeat Indicator variable for a property with forced air heatingFireplace Indicator variable for a property with a fireplaceSale Year-Quarter Indicator variable for each year and quarter of sale, total = 12
variablesDevDistMillLevy Development district mill levy. Equal to zero for properties not
in development districtsNumberJuris The number of local governments serving a propertyAnntax Annual property taxesDevDistAnnTax Annual property taxes paid to the development district
between any omitted variables captured in our error term and our instrumentsbased on the attributes of development districts.
Table 5 provides the definitions of variables employed in various hedonicspecifications. This table defines four categories of variables: (1) house price,(2) property characteristics including lot size, square feet of living area, numberof bathrooms, dummy variables for garage, forced air heat and fireplaces,(3) dummy variables for sale quarter and (4) government variables includingdevelopment district property tax rate, the number of local governments/specialdistricts serving the property, the total annual property tax liability and the totalannual taxes paid to the development district. Local governments serving aproperty always include the county government and the school district and mayinclude a municipality and other non-development special/municipal districts(e.g., recreation, security, etc.)
Table 6 provides summary statistics separately for all transactions locatedwithin development districts and for all sales located outside developmentdistricts. There are 13,582 transactions of recently built single-family homeslocated within development districts and 20,466 sales located outside develop-ment districts. The mean transaction price for properties located inside devel-opment districts is slightly lower than the mean transaction price for propertieslocated outside development districts ($318,622 vs. $322,347). Homes builtinside development districts are slightly larger and the mean per square foot
Financing Residential Development with Special Districts 19
Table 6 � Summary statistics—Single-family homes sold between 2002 and 2004,three years old or newer.
Not in DevelopmentIn Development District District
Variable Mean Std. Dev. Mean Std. Dev.
Sales Price 318,622 133,893 322,347 134,436Price per Sqft 135.46 30.13 143.54 35.05Acres 0.196 0.165 0.267 0.339Bath 2.877 0.681 2.730 0.650Sqft 2,362 727 2,249 680Garage 0.993 0.084 0.958 0.200Basement 0.826 0.379 0.881 0.323ForcedAirHeat 0.998 0.049 0.894 0.308Fireplace 0.808 0.394 0.621 0.485Sale02qt1 0.069 0.253 0.072 0.258Sale02qt2 0.080 0.272 0.076 0.265Sale02qt3 0.088 0.282 0.077 0.268Sale02qt4 0.093 0.291 0.079 0.270Sale03qt1 0.071 0.257 0.069 0.253Sale03qt2 0.087 0.282 0.078 0.269Sale03qt3 0.092 0.289 0.088 0.283Sale03qt4 0.090 0.286 0.092 0.289Sale04qt1 0.064 0.245 0.077 0.267Sale04qt2 0.079 0.269 0.097 0.296Sale04qt3 0.094 0.291 0.099 0.298Sale04qt4 0.094 0.292 0.097 0.296DevDistMilllevy 36.7 15.6 0 0NumberJuris 4.40 0.99 3.88 0.96Anntax 3,056 1,186 2,657 1,143DevDistAnnTax 1,056 535 0 0Observations 13,582 20,466
transaction price is lower for development district homes ($135 per square footvs. $144 per square foot). Development district homes are located on slightlysmaller lots (0.20 acres vs. 0.27 acres), and more of the development districthomes come with garages, forced air heat and fireplaces, but there is likely to besignificant spatial variation in these attributes that is not taken into account inthese summary statistics. Properties located within development districts alsohave more local governments providing services and pay higher property taxeson average. The average development district tax rate across transactions is36.7 mills.
Results
As noted by Yinger et al. (1988), one of the inherent econometric difficultiesin estimating tax capitalization is the simultaneity between house prices and
20 Billings and Thibodeau
tax rates. The simultaneity occurs because higher property values provide moreproperty tax revenue for a given tax rate. Since infrastructure such as water,sewer and roads have similar fixed costs for both high- and low-valued proper-ties, lower mill rates can cover the costs of infrastructure in neighborhoods withhigher property values. To account for this simultaneity as well as unobservedpublic good heterogeneity, we estimate property tax capitalization using aninstrumental variable for the tax rate.
An appropriate instrument for the tax rate is one that influences the tax rate,yet is uncorrelated with the error term in Equation 1. Characteristics likely toinfluence a development district’s tax rate, but not specific home prices are thosethat influence the per-home cost of the infrastructure, but not local amenitiesthat could be capitalized in home prices. We hypothesize that the overall sizeand buildout of a subdivision will impact the cost of repaying the SpecialDistricts bond, but not the level of infrastructure provision for an individualhome. The three variables used to instrument tax rates (mill levies) are: (1)the number of homes sold within the development district between 2002 and2004, (2) the land area of the development district (in square miles) and (3) thenumber of years since the development district was created. We generate IVresults across the three spatially sub-sampled datasets and report test statisticsthat examine the validity of these instruments.
Since most house price hedonic models exhibit heteroskedasticity,18 we areconcerned that a standard two stage least squares IV estimator may produceinconsistent standard errors. Therefore, we employ a Generalized Methodsof Moments (GMM) estimation techinque that allows for efficient estimationeven in the presence of heteroskedasticity of unknown form.19 We test theneed to incorporate instrumental variables with a Hausman-Wu-Durbin test,which confirms the endogeneity of the mill levy variable in all specifications.20
Given that our model contains more instruments than endogenous variables, weimplement a Hansen J Test, which uses the overidentification in our instrumentalvariable model to test if the instruments are correlated with the error term inthe house price equation (Cameron and Trivedi (2005)). A Hansen J Testp-value greater than 0.05 accepts the null and concludes that the instrumentsare valid. Test results are provided for each model and indicate that the GMM-IV technique is only appropriate for the final two models and that the estimated
18A Breusch-Pagan test confirms the presence of heteroskedasticity in our model. Theseresults are available upon request.19See Baum, Schaffer and Stillman (2003) for a articulate discussion of GMM-IVestimation in Stata.20See Cameron and Trivedi (2005), Wooldridge (2002) or Section 5 of Baum, Schafferand Stillman (2003) for a description of the Hausman-Wu-Durbin test.
Financing Residential Development with Special Districts 21
coefficients for property tax capitalization in the first two models are biased.Finally, the p-values for the F-statistics and Shea’s partial R2 measure theexplanatory power of the instruments for DevDistMillLevy. The significancelevel of the F-statistics as well as the instruments’ ability to explain propertytax rates (given by a Shea’s R2 of around 0.4) indicate sufficiently stronginstruments.
We estimate the influence that the development district property tax liabilityhas on property values for three different subsamples, with each subsampleimposing more restrictive controls for neighborhood characteristics. The finalsubsample represents the model used for interpreting results and has the ad-vantage of limiting identification to comparing developments within a CBG inrapidly growing suburban areas. Therefore, the elasticity of supply will be sim-ilar between development districts and non-development district subdivisions.In addition, the results can be interpreted as the effect of moving a house froma development district to a non-development district. The variables included inthe hedonic house price equation include a standard bundle of housing charac-teristics; dummy variables for transaction sale quarter; a set of dummy variablesthat indicate whether a property is served by a municipality, fire, recreation ora non-development based metropolitan special district; and, for three of thefour specifications, dummy variables for CBG. In order to control for the non-coterminous boundaries of school districts and CBGs, regression specificationsinclude dummies for each unique combination of CBGs and school districts.21
We report estimated coefficients for each of four alternative hedonic house pricemodels using both OLS and Generalized Methods of Moments-InstrumentalVariables (GMM-IV). Table 7 provides regression results. Model 1 in Table 7provides results for all 34,048 sales without any controls for CBG fixed ef-fects but includes variables that control for distance to downtown Denver (inmiles) and distance to downtown squared. The first column lists OLS estimatesand the second lists GMM-IV estimates. This model indicates a smaller taxcoefficient of −0.000336 for OLS and a relatively large negative tax coeffi-cient of −0.002112 for the GMM-IV model. Evaluated at the mean transactionprice for development district homes ($318,622) and the mean developmentdistrict tax rate (36.7 mills), the average single-family property tax capitaliza-tion is 1.23% or $3,929 for the OLS estimate and 7.75% or $24,697 for theGMM-IV estimate. Residential properties in Colorado are assessed at 7.96% oftheir market value, so the mean assessed value for development district prop-erties is $25,362.31. With an average development district property tax rate of
21A total of 28 school districts in Model 2, 11 school districts in Model 3 and six schooldistricts in Model 4 contain a boundary that falls within a CBG.
22 Billings and Thibodeau
Tabl
e7
�H
edon
ices
timat
ion
resu
lts.
(1)
(2)
(3)
(4)
Dep
Var
=ln
(Sal
esP
rice
)O
LS
GM
M-I
VO
LS
GM
M-I
VO
LS
GM
M-I
VO
LS
GM
M-I
V
Dev
Dis
tMil
lLev
y−0
.000
336∗∗
∗−0
.002
112∗∗
∗−0
.000
087
−0.0
0037
9∗∗∗
−0.0
0010
5−0
.000
379∗∗
∗−0
.000
192∗∗
∗−0
.000
497∗∗
∗(−
5.4)
(−20
.2)
(−1.
5)(−
4.3)
(−1.
9)(−
5.0)
(−3.
5)(−
3.3)
Acr
es0.
1987
95∗∗
∗0.
2025
69∗∗
∗0.
1729
38∗∗
∗0.
1987
95∗∗
∗0.
2607
17∗∗
∗0.
2631
52∗∗
∗0.
3053
11∗∗
∗0.
3091
38∗∗
∗(1
6.0)
(16.
0)(1
0.4)
(11.
9)(4
2.4)
(42.
8)(3
8.9)
(23.
5)B
ath
0.03
7175
∗∗∗
0.04
4805
∗∗∗
0.01
9620
∗∗∗
0.01
8777
∗∗∗
0.01
5389
∗∗∗
0.01
5015
∗∗∗
0.01
6058
∗∗∗
0.01
5486
∗∗∗
(14.
5)(1
7.6)
(9.8
)(9
.5)
(8.0
)(7
.8)
(6.9
)(5
.8)
Sqft
0.00
0322
∗∗∗
0.00
0326
∗∗∗
0.00
0290
∗∗∗
0.00
0291
∗∗∗
0.00
0270
∗∗∗
0.00
0271
∗∗∗
0.00
0258
∗∗∗
0.00
0259
∗∗∗
(38.
4)(3
8.6)
(40.
6)(4
1.1)
(41.
2)(4
1.7)
(31.
4)(3
1.5)
Sqft
2(0
00s)
−0.0
0000
6∗∗∗
−0.0
0000
7∗∗∗
−0.0
0000
6∗∗∗
−0.0
0000
6∗∗∗
−0.0
0000
3−0
.000
003∗∗
−0.0
0000
3−0
.000
003
(−3.
5)(−
4.0)
(−4.
2)(−
4.5)
(−1.
9)(−
2.0)
(−1.
7)(−
1.7)
Gar
age
−0.0
2351
1∗∗∗
−0.0
1549
10.
0452
55∗∗
∗0.
0414
17∗∗
∗0.
0242
63∗∗
∗0.
0229
35∗∗
∗0.
0014
36−0
.003
958
(−2.
6)(−
1.7)
(5.0
)(4
.6)
(4.1
)(3
.9)
(0.1
)(−
0.3)
Bas
emen
t0.
1586
79∗∗
∗0.
1423
28∗∗
∗0.
1121
50∗∗
∗0.
1107
79∗∗
∗0.
1069
36∗∗
∗0.
1054
90∗∗
∗0.
1059
30∗∗
∗0.
1038
44∗∗
∗(4
8.9)
(45.
2)(3
8.9)
(38.
2)(5
0.7)
(49.
5)(4
3.1)
(37.
4)Fo
rced
Air
Hea
t−0
.138
983∗∗
∗−0
.133
878∗∗
∗−0
.087
982∗∗
∗−0
.077
568∗∗
∗−0
.018
424
−0.0
1901
4−0
.042
338
−0.0
4340
7(−
8.2)
(−7.
9)(−
3.2)
(−2.
9)(−
0.7)
(−0.
8)(−
1.4)
(−1.
4)F
irep
lace
0.06
7831
∗∗∗
0.06
3337
∗∗∗
0.03
7888
∗∗∗
0.03
6351
∗∗∗
0.03
4053
∗∗∗
0.03
3786
∗∗∗
0.03
0568
∗∗∗
0.03
0118
∗∗∗
(33.
0)(3
0.4)
(20.
6)(2
0.0)
(21.
3)(2
1.2)
(14.
8)(1
4.2)
Cur
rent
Sale
02qt
20.
0115
96∗∗
0.00
9994
∗∗0.
0077
95∗∗
0.00
7857
∗∗0.
0114
27∗∗
∗0.
0114
18∗∗
∗0.
0041
150.
0043
99(2
.4)
(2.0
)(2
.2)
(2.2
)(3
.3)
(3.3
)(0
.86)
(0.8
)C
urre
ntSa
le02
qt3
0.01
4364
∗∗∗
0.01
5527
∗∗∗
0.01
3619
∗∗∗
0.01
4673
∗∗∗
0.01
2750
∗∗∗
0.01
2982
∗∗∗
0.00
3025
0.00
3469
(3.1
)(3
.3)
(3.9
)(4
.2)
(3.9
)(4
.0)
(0.7
)(0
.6)
Cur
rent
Sale
02qt
40.
0124
21∗∗
∗0.
0138
78∗∗
∗0.
0152
45∗∗
∗0.
0162
25∗∗
∗0.
0111
46∗∗
∗0.
0116
12∗∗
∗0.
0030
520.
0038
28(2
.8)
(3.0
)(4
.5)
(4.8
)(3
.5)
(3.6
)(0
.7)
(0.6
)
Financing Residential Development with Special Districts 23
Tabl
e7
�C
ontin
ued
(1)
(2)
(3)
(4)
Dep
Var
=ln
(Sal
esP
rice
)O
LS
GM
M-I
VO
LS
GM
M-I
VO
LS
GM
M-I
VO
LS
GM
M-I
V
Cur
rent
Sale
03qt
10.
0205
67∗∗
∗0.
0213
81∗∗
∗0.
0166
91∗∗
∗0.
0170
93∗∗
∗0.
0117
53∗∗
∗0.
0122
05∗∗
∗0.
0071
250.
0081
02(4
.3)
(4.4
)(4
.6)
(4.8
)(3
.3)
(3.5
)(1
.6)
(0.7
)C
urre
ntSa
le03
qt2
0.03
2794
∗∗∗
0.03
3105
∗∗∗
0.02
5482
∗∗∗
0.02
5663
∗∗∗
0.02
1755
∗∗∗
0.02
1876
∗∗∗
0.02
0018
∗∗∗
0.02
0297
∗∗∗
(6.9
)(6
.9)
(7.0
)(7
.1)
(6.3
)(6
.4)
(4.3
)(2
.9)
Cur
rent
Sale
03qt
30.
0407
20∗∗
∗0.
0416
98∗∗
∗0.
0330
87∗∗
∗0.
0340
25∗∗
∗0.
0279
94∗∗
∗0.
0282
98∗∗
∗0.
0203
79∗∗
∗0.
0209
19∗∗
∗(8
.9)
(9.0
)(9
.5)
(9.8
)(8
.5)
(8.6
)(4
.6)
(3.2
)C
urre
ntSa
le03
qt4
0.04
9463
∗∗∗
0.04
6159
∗∗∗
0.03
4389
∗∗∗
0.03
5346
∗∗∗
0.03
1324
∗∗∗
0.03
1449
∗∗∗
0.01
7510
∗∗∗
0.01
7701
∗∗∗
(10.
8)(9
.9)
(9.7
)(1
0.1)
(9.4
)(9
.4)
(4.0
)(2
.7)
Cur
rent
Sale
04qt
10.
0443
64∗∗
∗0.
0420
66∗∗
∗0.
0347
36∗∗
∗0.
0349
70∗∗
∗0.
0268
95∗∗
∗0.
0267
03∗∗
∗0.
0241
39∗∗
∗0.
0238
98∗∗
∗(8
.6)
(8.0
)(8
.8)
(8.9
)(7
.2)
(7.2
)(5
.1)
(3.6
)C
urre
ntSa
le04
qt2
0.07
7184
∗∗∗
0.07
3895
∗∗∗
0.06
0880
∗∗∗
0.06
1358
∗∗∗
0.05
4676
∗∗∗
0.05
4379
∗∗∗
0.04
3096
∗∗∗
0.04
2496
∗∗∗
(16.
4)(1
5.4)
(16.
8)(1
7.1)
(16.
4)(1
6.3)
(9.8
)(6
.8)
Cur
rent
Sale
04qt
30.
0854
93∗∗
∗0.
0858
07∗∗
∗0.
0702
89∗∗
∗0.
0711
93∗∗
∗0.
0630
54∗∗
∗0.
0632
83∗∗
∗0.
0450
39∗∗
∗0.
0452
46∗∗
∗(1
8.3)
(18.
1)(1
9.3)
(19.
7)(1
8.7)
(18.
9)(1
0.4)
(7.4
)C
urre
ntSa
le04
qt4
0.10
3428
∗∗∗
0.09
9439
∗∗∗
0.08
4600
∗∗∗
0.08
5852
∗∗∗
0.07
8311
∗∗∗
0.07
8399
∗∗∗
0.06
1267
∗∗∗
0.06
1341
∗∗∗
(21.
7)(2
0.6)
(21.
7)(2
2.2)
(22.
0)(2
2.0)
(13.
7)(9
.0)
Con
stan
t12
.118
979∗∗
∗11
.874
026∗∗
∗11
.877
217∗∗
∗11
.879
544∗∗
∗11
.765
514∗∗
∗11
.771
592∗∗
∗11
.863
591∗∗
∗11
.767
862
(447
.7)
(62.
3)(3
29.4
)(3
35.4
)(7
8.7)
(78.
4)(3
10.6
)(1
.9)
24 Billings and Thibodeau
Tabl
e7
�C
ontin
ued
(1)
(2)
(3)
(4)
Dep
Var
=ln
(Sal
esP
rice
)O
LS
GM
M-I
VO
LS
GM
M-I
VO
LS
GM
M-I
VO
LS
GM
M-I
V
CB
GF
ixed
Effe
cts
No
No
Yes
Yes
Yes
Yes
Yes
Yes
Scho
olD
istr
ictF
ixed
Effe
cts
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
CB
G∗ S
choo
lDis
tric
tF
ixed
Effe
cts
No
No
Yes
Yes
Yes
Yes
Yes
Yes
Indi
cato
rsfo
rO
ther
Gov
tsY
esY
esY
esY
esY
esY
esY
esY
es
Obs
erva
tion
s34
,048
34,0
4834
,048
34,0
4827
,869
27,8
6916
,543
16,5
43R
20.
770.
760.
880.
880.
880.
880.
870.
87H
anse
n’s
Jp-
valu
e0.
000.
000.
090.
15In
stru
men
tF-S
tat
p-va
lue
0.00
0.00
0.00
0.00
Fir
stSt
age
-Sh
ea’s
Part
ialR
20.
420.
390.
420.
44
Hau
sman
-Wu-
Dur
bin
p-va
lue
0.00
0.00
0.00
0.00
t-st
atis
tics
inpa
rent
hese
s∗∗
∗ p<
0.01
,∗∗p
<0.
05,∗ p
<0.
010.
The
inst
rum
ent
list
isth
enu
mbe
rof
hom
esso
ldw
ithin
the
deve
lopm
ent
dist
rict
betw
een
2002
and
2004
,th
ela
ndar
eaof
the
deve
lopm
ent
dist
rict
(in
squa
rem
iles)
and
the
num
ber
ofye
ars
sinc
eth
ede
velo
pmen
tdis
tric
twas
crea
ted.
Financing Residential Development with Special Districts 25
36.7 mills,22 the average annual development district property tax liability is$930.80.
So how much of this annual development district property tax liability is cap-italized in house prices? The amount of capitalization depends on two things:(1) the discount rate that homeowners use to convert the future tax liability toa present value and (2) the remaining term of the tax liability. We assume, forsimplicity, that homeowners can take full advantage of the property tax deduc-tion and that homeowners discount future property tax liabilities at the aftertax mortgage interest rate.23 The average nominal mortgage interest rate for30 year, fixed-rate mortgages over the 2002–2004 period was 6.07%. So house-holds in the 28% tax bracket discount future expected development districtproperty taxes at 4.37%. If we further assume that the annual property tax is anannuity (e.g., that mill rates will decrease if property values increase to providea constant payment) and discount the annual annuity of $930.80 at 4.37% over19 years (the average remaining term for the district bonds listed in Table 3),homeowners value the future development tax liability at $11,850. Under theseassumptions, Model 1 estimates that 33.2% of the development district taxliability is capitalized in the price of new single-family homes using OLS and209% using GMM-IV. The range of estimates is attributed to two problemsestimating the parameters in the first model. The first is unobserved neighbor-hood heterogeneity, where in-fill new homes in bedroom communities such asBoulder, CO, are treated comparably to suburban subdivisions. The higher landrents for in-fill development likely biases the tax coefficient downward becausedevelopment districts are commonly used for suburban subdivisions. Second,since the Hansen’s J Test rejects the null hypothesis of valid instruments, theinstrumental variables adopted in Model 1 are inappropriate and the estimatedtax coefficient obtained using GMM-IV is biased. All of the other estimatedcoefficients in Model 1 are statistically significant at conventional levels.
The second model estimates parameters using the same 34,048 transactions andthe same instruments, but it incorporates better controls for variation in neigh-borhood amenities by including fixed effects for CBGs.24 Hansen’s J Test alsorejects the null hypothesis that the selected instruments are valid in Model 2. The
22Correspondingly, properties not located in a development district contain a develop-ment district property tax rate equal to zero mills. Controls for other types of govern-ments address any underlying property tax differences between development districtand non-development district properties.23See de Bartolome and Rosenthal (1999) for a good discussion of property tax capital-ization under different assumptions regarding tax and mortgage interest deductions.24Distance to downtown and distance to downtown squared variables are no longernecessary when CBG fixed effects are included in the specification.
26 Billings and Thibodeau
Table 8 � Sensitivity test.
No. ofMin Transactions/ No. of Development No. of Estimated Tax% of Existing Homes Census BGs Districts Observations Coefficient
10/10% 26 55 16,767 −0.00049520/20% 23 50 16,543 −0.00049730/20% 20 47 16,086 −0.00053040/20% 18 44 14,928 −0.000544
GMM-IV estimated coefficient for the development district tax rate decreasesto −0.000379. With this (biased) estimate, the house price discount is reducedto 1.39%, or $4,432, and, using the same set of assumptions used to estimateproperty tax capitalization for Model 1, the rate of property tax capitalizationdeclines to 37.4% (from 209%)! The dramatic change in estimated capitaliza-tion is the result of controlling for spatial variation in neighborhood amenitiesusing CBG fixed effects. The estimated coefficients for lot size (in acres) arevirtually identical in Models 1 and 2, but there are significant differences inmany of the other estimated coefficients. In Model 2, estimated coefficients forall structural characteristics are statistically significant at conventional levels.Including CBG fixed effects significantly improves the model’s goodness of fitwith R2 increasing from 76% of the variance in the natural log of transactionprice explained to 88%.
Since GMM-IV estimates are only valid for the third and fourth models, wefocus our attention on the results generated by these models. The subsamplingof homes in only CBGs with development districts and high rates of newhousing construction provides sufficient controls for unobserved variables toremove the correlation between the hedonic house price error term and thethree instruments for mill levy included in GMM-IV estimation. Model 3restricts the data to transactions in CBGs with at least 20 sales and with thosesales representing at least 20% of the year 2000 neighborhood housing stock.This subsample increases the magnitude of the estimated coefficients for boththe development district tax rate and for lot size. The estimate of the houseprice discount generated by GMM-IV estimates in Model 3 is $4,432, and theestimated development district property tax capitalization is 37.4%. This is instark contrast to the insignificant OLS coefficient of −0.000105.
Model 4 further limits the sample to those homes built and sold in CBGs with atleast 20 sales located both within and outside development districts.25 Relative
25We attempted to disentangle the capitalization of special district tax liability frombeing in a development district versus the magnitude of the tax rate by incorporating a
Financing Residential Development with Special Districts 27
to OLS Model 4, GMM Model 4 increases both the estimated property taxcapitalization effect and the marginal value of land.26 The estimated houseprice discount is 1.83%, or $5,812, and the estimated development districtproperty tax capitalization is 49.0%.27 The estimated coefficients for square feetof living space squared and attached garage become statistically insignificant.Unreported coefficients for government variables were statistically insignificantexcept that fire special districts and non-development metropolitan specialdistricts negatively influence house prices.28 In addition, while the first threemodels indicate that house prices were increasing throughout the 2002–2004period, the fourth specification indicates that house prices were basically flatfor the first five quarters of the 2002–2004 period suggesting developers wereunable to increase house prices as the rapidly growing developments were beingbuilt out.
The estimated coefficients for lot size, ForcedAirHeat and Garage vary sig-nificantly across all models in Table 7. The estimated coefficients for theother structural characteristics were similar across the three samples given byModels 2, 3 and 4. Given the similarity of a number of coefficients in Mod-els 2, 3 and 4, the sampling of subsequently smaller numbers of CBGs mayrepresent the same underlying data generating process. The three subsamplesgiven in Models 2 through 4 represent a full model and two spatially restrictedmodels. The restricted models assume a structural break based on the rate ofdevelopment (e.g., high-growth versus low-growth CBGs). In order to test thisassumption, a Wald test is used to compare the estimated coefficients for allproperty and tax variables in Model 4 to the estimated coefficients generatedby the samples of observations in Models 2 and 3 that were excluded fromModel 4. The Wald test yields a χ2 = 208.3(p = 0.000) and rejects the nullhypothesis that Model 4 is structurally equivalent to Model 2 at a 1% level. TheWald test between Models 3 and 4 yields a χ2 = 206.4(p = 0.000) and rejects
dummy variable into Model 4. Unfortunately, the inclusion of the development districtindicator made our instrumentation of the tax rate invalid according to a Hansen’s JStatistic p-value of 0.00.26Under our assumptions about how homeowners discount future expected tax liabilities,the test of 100% capitalization is equivalent to a test of whether the estimated coeffi-cient for the tax rate variable is statistically different from −0.0010139. The estimatedcoefficient for the tax rate variable in Model 4 is statistically different from −0.0010139at the 0.01 level (as the estimated coefficient of −0.000497 is 3.4 standard errors awayfrom −0.0010139).27If homeowners were unable to take advantage of the property tax deduction anddiscounted future development district tax liabilities at 6.07% instead of 4.37%, thenpresent values of future tax liabilities decrease and estimates of property tax capitaliza-tion increase to 45% in Model 3 and 58% in Model 4.28In Models 2 through 4, government variables provide controls for the case whenspecial districts contain boundaries that fall within both a development district and aCBG. This only occurs within a few CBGs.
28 Billings and Thibodeau
Tabl
e9
�E
stim
ated
capi
taliz
atio
nef
fect
s.
Rem
aini
ngPr
esen
tA
vera
gePr
oper
tyE
stim
ated
Ann
ual
Term
Val
ueof
Sale
sM
illPr
ice
Prop
erty
asof
Futu
reTa
xPe
rcen
tD
evel
opm
entS
peci
alD
istr
ictN
ame
Pric
eL
evy
Dis
coun
tTa
x20
04L
iabi
lity
Cap
italiz
atio
n
Aur
ora
Sing
leT
ree
Met
ropo
litan
Dis
tric
t$2
29,1
8042
.6$4
,852
$777
.14
21$1
0,54
046
.0%
Bel
leC
reek
Met
ropo
litan
Dis
tric
t$1
94,4
7447
.9$4
,630
$741
.50
16$8
,409
55.1
%B
righ
ton
Cro
ssin
gM
etro
polit
anD
istr
ictN
o.4
$225
,072
38.0
$4,2
51$6
80.8
030
$11,
261
37.7
%B
uffa
loR
idge
Met
ropo
litan
Dis
tric
t$2
05,8
5637
.3$3
,816
$611
.20
19$7
,772
49.1
%C
ante
rber
ryC
ross
ing
Met
ropo
litan
Dis
tric
t$2
32,9
1439
.5$4
,572
$732
.33
27$1
1,47
539
.8%
Can
terb
erry
Cro
ssin
gM
etro
polit
anD
istr
ictI
I$2
55,5
2537
.3$4
,737
$758
.67
28$1
2,10
639
.1%
Che
rry
Cre
ekSo
uth
Met
ropo
litan
Dis
tric
tNo.
1$2
56,7
6139
.8$5
,079
$813
.44
29$1
3,22
138
.4%
Eag
leB
end
Met
ropo
litan
Dis
tric
tNo.
2$4
15,3
0845
.3$9
,350
$1,4
97.5
515
$16,
224
57.6
%E
agle
Cre
ekM
etro
polit
anD
istr
ict
$266
,531
50.0
$6,6
23$1
,060
.79
17$1
2,54
352
.8%
Eas
tSm
oky
Hill
Met
ropo
litan
Dis
tric
tNo.
2$2
56,8
1717
.0$2
,170
$347
.52
26$5
,337
40.7
%E
bert
Met
ropo
litan
Dis
tric
t$2
52,7
3339
.0$4
,899
$784
.58
21$1
0,64
146
.0%
Gat
eway
Reg
iona
lMet
ropo
litan
Dis
tric
t$2
48,4
6210
.0$1
,235
$197
.78
6$1
,024
120.
5%G
oodm
anM
etro
polit
anD
istr
ict
$245
,067
26.5
$3,2
28$5
16.9
415
$5,6
1057
.5%
GV
RM
etro
polit
anD
istr
ict
$255
,983
30.3
$3,8
55$6
17.4
015
$6,6
8457
.7%
Hig
hlan
dsR
anch
Met
ropo
litan
Dis
tric
tNo.
1$3
69,7
6920
.3$3
,731
$597
.50
8$3
,959
94.2
%H
ighl
ands
Ran
chM
etro
polit
anD
istr
ictN
o.3
$367
,252
20.3
$3,7
05$5
93.4
415
$6,4
2557
.7%
Hig
hlan
dsR
anch
Met
ropo
litan
Dis
tric
tNo.
4$3
06,6
6720
.3$3
,094
$495
.54
13$4
,832
64.0
%In
terl
ocke
nC
onso
lidat
edM
etro
polit
anD
istr
ict
$399
,530
27.2
$5,4
01$8
65.0
315
$9,3
8757
.5%
Financing Residential Development with Special Districts 29
Tabl
e9
�C
ontin
ued
Rem
aini
ngPr
esen
tA
vera
gePr
oper
tyE
stim
ated
Ann
ual
Term
Val
ueof
Sale
sM
illPr
ice
Prop
erty
asof
Futu
reTa
xPe
rcen
tD
evel
opm
entS
peci
alD
istr
ictN
ame
Pric
eL
evy
Dis
coun
tTa
x20
04L
iabi
lity
Cap
italiz
atio
n
Lin
coln
Park
Met
ropo
litan
Dis
tric
t$4
30,0
2845
.0$9
,618
$1,5
40.3
622
$21,
493
44.7
%M
aher
Ran
chM
etro
polit
anD
istr
ictN
o.4
$227
,679
40.0
$4,5
26$7
24.9
324
$10,
646
42.5
%M
eado
ws
Met
ropo
litan
Dis
tric
tNo.
1$3
02,8
7935
.0$5
,269
$843
.82
25$1
2,67
841
.6%
Mea
dow
sM
etro
polit
anD
istr
ictN
o.2
$335
,700
35.0
$5,8
40$9
35.2
625
$14,
056
41.5
%N
orth
Ran
geV
illag
eM
etro
polit
anD
istr
ict
$237
,077
38.5
$4,5
36$7
26.5
516
$8,2
4055
.1%
Pine
ryW
estM
etro
polit
anD
istr
ictN
o.2
$366
,465
45.0
$8,1
96$1
,312
.68
28$2
0,96
939
.1%
Poto
mac
Farm
sM
etro
polit
anD
istr
ict
$288
,830
38.0
$5,4
55$8
73.6
528
$13,
956
39.1
%R
iver
dale
Dun
esM
etro
polit
anD
istr
ictN
o.1
$240
,710
45.0
$5,3
83$8
62.2
228
$13,
774
39.1
%R
oxbo
roug
hV
illag
eM
etro
polit
anD
istr
ict
$317
,903
70.1
$11,
076
$1,7
73.8
917
$20,
963
52.8
%St
erlin
gH
ills
Met
ropo
litan
Dis
tric
t$2
71,7
4430
.0$4
,052
$648
.92
14$6
,690
60.6
%St
oneg
ate
Vill
age
Met
ropo
litan
Dis
tric
t$3
87,8
3927
.4$5
,282
$845
.89
22$1
1,81
544
.7%
Supe
rior
Met
ropo
litan
Dis
tric
tNo.
2$3
95,4
5622
.0$4
,324
$692
.52
9$5
,063
85.4
%To
ddC
reek
Farm
sM
etro
polit
anD
istr
ictN
o.1
$311
,550
16.5
$2,5
55$4
09.1
917
$4,8
3852
.8%
Todd
Cre
ekFa
rms
Met
ropo
litan
Dis
tric
tNo.
2$2
71,8
4618
.0$2
,432
$389
.50
14$4
,016
60.6
%U
pper
Che
rry
Cre
ekM
etro
polit
anD
istr
ict
$256
,119
6.0
$764
$122
.32
7$7
2410
5.5%
30 Billings and Thibodeau
the null hypothesis that Model 4 is structurally equivalent to Model 3 at a 1%level.
The empirical results reported above depend on two criteria for delineat-ing submarkets: (1) the minimum number of transactions in the CBG and(2) the share of home sales used to delineate high-growth areas. To examinethe sensitivity of our results to these criteria, we estimated the parameters ofModel 4 under a variety of alternative assumptions. Table 8 shows the robust-ness of the empirical results to variation in the minimum number of transactionsboth within and outside development districts and the percent of CBG homessold. Estimated tax coefficients for Model 4 vary by less than 5% across thesefour different assumptions.29
Table 9 provides estimated capitalization effects for each of the 33 Denverarea development districts highlighted in Table 3.30 For each developmentdistrict, the table lists the average transaction price, the development district taxrate, the estimated house price discount obtained using the parameter estimatefrom Model 4 (our preferred specification), the annual property tax, the termremaining on the development district bonds, the present value of the futuredevelopment district tax liability (assuming households are in the 28% marginaltax bracket and can take full advantage of the property tax deduction) andthe percent property tax capitalization. The presence of greater than or almost100% capitalization for some development districts is due to the use of the meanestimated tax coefficient for development districts with few years remaining onthe terms of the bond as of 2004.31
Another way to interpret this result is to compute the length of time neces-sary to justify the estimated capitalization given the annual tax liability andthe homeowners’ discount rate.32 At a discount rate of 4.37%, about 7.4years of annual property tax payments are capitalized in house prices; at
29Across the different assumptions in Table 8, instruments were valid with J test p-valuesbetween 0.13 and 0.31. Less restrictive sub-sampling produced invalid instruments.30The remaining 17 development districts used in estimation contained incomplete bonddata to compute capitalization rates for individual development districts.31Estimation of tax coefficients by variation in years remaining on special district bondswould provide interesting insight into a taxpayer’s response to the length of finite taxliabilities. Unfortunately, the limited number of development districts with informationon special district bonds prevents us from estimating tax capitalization for differentremaining bond maturities.32One can calculate the number of years (N) required to equate the present valueof future tax liability to equal the property value capitalization based on N =ln( AR
(AR+1,000βd))/[ln(1 + d)], where AR = assessment rate, which is 0.0796 in Colorado,
β is the estimated tax coefficient, and d is the homeowner’s discount rate.
Financing Residential Development with Special Districts 31
6.07% (assuming the household either does not itemize or cannot use theproperty tax deduction), about 8.1 years of property taxes are capitalized.
Concluding Remarks
This paper empirically examines the extent to which property taxes associatedwith development district financing of residential infrastructure are capitalizedin the price of new, owner-occupied homes. We estimate the parameters of ahedonic house price model and control for variation in neighborhood attributesusing CBG fixed effects and by restricting the sample to single-family homesbuilt in CBGs containing numerous sales of properties located both within andoutside development districts. We find that about half of the future propertytax liability is capitalized in house prices. Alternatively, homebuyers capitalizebetween 7.4 and 8.1 years of property tax liability in the purchase price. Wealso find that our estimates are very sensitive to the sample used, with the mostaccurate capitalization rates estimated from the sample that provides the bestempirical controls for neighborhood attributes and area growth rates.
A hedonic house price specification that fails to adequately control for neigh-borhood amenities and/or provide inadequate controls for the endogeneity ofproperty tax rates can yield biased estimates of property tax capitalization. Twoelements of our empirical model address these potential biases. First, we limitthe data to rapidly growing residential developments and control for variationin neighborhood amenities using CBG fixed effects. Second, we incorporateinstrumental variable estimation of development district property tax rates. Sta-tistical tests for instrument validity indicate that it is necessary to restrict theestimation sample to rapidly growing suburban areas to attain valid instrumentsfor the property tax rate. The final estimate of development district propertytax capitalization is about 50%. Given our estimation targets new suburbandevelopments, our estimated tax capitalization rate provides a lower bound forurban/center city homes.
This research has benefited from comments by John Clapp, Dennis Heffley,Joseph Nichols, Liang Peng, Milena Petrova, Stephen Ross, Rexford Santerreand two referees of this journal. We thank Erika Gleason and Brady Miller fortheir outstanding research assistance. All errors are our own.
References
Baum, C. F., M. E. Schaffer and S. Stillman. 2003. Instrumental Variables and GMM:Estimation and Testing. The Stata Journal 3(1): 1–31.Billings, S. and T. G. Thibodeau. 2011. Intrametropolitan Decentralization: Is Gov-ernment Structure Capitalized in Residential Property Values? Journal of Real EstateFinance and Economics 42(4): 416–450.
32 Billings and Thibodeau
Bradley, S. 2011. Capitalizing on Capped Taxable Values: How Michigan Homebuyersare Paying for Assessment Limits. University of Michigan Working Paper.Brueckner, J. K. 1979. Property Values, Local Public Expenditure and Economic Effi-ciency. Journal of Public Economics 11(2): 223–245, April 1979.Brueckner, J. K.. 1997. Infrastructure Financing and Urban Development: The Eco-nomics of Impact Fees. Journal of Public Economics 66: 383–407.Cameron, A. and P. Trivedi. 2005. Microeconmetric: Methods and Applications. Camp-bridge University Press.Carr, J. B.. 2006. Local Government Autonomy and State Reliance on Special DistrictGovernments: A Reassessment. Political Research Quarterly 59(3): 481–492.de Bartolome, C. and S. S. Rosenthal. 1999. Property Tax Capatilization in a Model withDeferred Assets, Standard Deductions, and the Taxation of Nominal Interest. Review ofEconomics and Statistics 81(1).Delaney, C. J. and M. T. Smith. 1989. Impact Fees and the Price of New Housing: AnEmprical Study. American Real Estate and Urban Economics Journal 17(1): 41–54.Evans-Cowley, J. S. and L. L. Lawhon. 2003. The Effects of Impact Fees on the Price ofHousing and Land: a Literature Review. Journal of Planning Literature 17(3): 351–359.Galvan, S. C.. 2007. Wrestling with MUDs to Pin Down the Truth About SpecialDistricts. Fordham Law Review 75: 3041–3080.Griffith, J. C.. 2007. Recent Developments in Public Finance Law: Special Tax Districtsto Finance Residential Infrastructure. The Urban Lawyer 39(4): 959–983.Ihlanfeldt, K. R. and T. M. Shaughnessy. 2004. An Empirical Investigation of the Effectsof Impact Fees on Housing and Land Markets. Regional Science and Urban Economics34: 639–661.Klotz, S. 2004. Cross Sectional Dependence in Spatial Econometric Models with anApplication to German Start-up Activity Data. LIT Verlag Berlin-Hamburg-Munster.NCSL. 2008. State Tax and Expenditure Limits - 2008. National Conference of StateLegislators (http://www.ncsl.org). accessed 10/10/2010.Nelson, M. A. 1990. Decentralization of the Subnational Public Sector: An EmpiricalAnalysis of the Determinants of Local Government Structure in Metropolitan Areas inthe u.s. Southern Economic Journal 57(2): 443–457, October 1990.Oates, W. E. 1969. The Effects of Property Taxes and Local Public Spending on PropertyValues: An Empirical Study of Tax Capitalization and the Tiebout Hypothesis. Journalof Political Economy 77(6): 957–971, November-December 1969.Palmon, O. and B. A. Smith. 1998. New Evidence on Property Tax Capitalization.Journal of Political Economy 106(5): 1099–1111.Peiser, R. B.. 1983. The Economics of Municipal Utility Districts for Land Development.Land Economics 59(1): 43–57.Singell, L. D. and J. H. Lillydahl. 1990. Housing Impact Fees. Land Economics 66(1):82–92.Starrett, D. A.. 1981. Land Value Capitalization in Local Public Finance. Journal ofPolitical Economy 89: 306–27.Thibodeau, T. G.. 2003. Marking Single-Family Property Values to Market. Real EstateEconomics 31(1): 1–22.Wooldridge, J. M. 2002. Econometric Analysis of Cross Section and Panel Data. MITPress.Yinger, J. 1982. Capitalization and the Theory of Local Public Finance. Journal ofPolitical Economy 90: 917–43.
Financing Residential Development with Special Districts 33
Yinger, J. 1998. The Incidence of Development Fees and Special Assessments. NationalTax Journal 51(1): 23–41.Yinger, J., H. S. Bloom, A. BorschSupan and H. F. Ladd. 1988. Property Taxes andHouse Values: The Theory and Estimation of Intrajurisdictional Property Tax Capital-ization. Academic Press.Zodrow, G. 2006. Who Pays the Property Tax?. Lincoln Institute of Land Policy LandLines 18(2).
