Estimating the Effect of Crime Risk on Property Values and Time on Market:
Evidence from Megan’s Law in Virginia

Raymond Brastow
Federal Reserve Bank of Richmond
Raymond.Brastow@rich.frb.org
Phone: 804-697-8668
Fax: 804-697-8698
701 E Byrd St
Richmond, VA 23219

Bennie Waller*
wallerbd@longwood.edu
Phone: 434-395-2046
Fax: 434-395-2203
Longwood University
201 High Street
Farmville, VA 23901
Scott Wentland
wentlandsa@longwood.edu
Phone: 434-395-2160
Fax: 434-395-2203
Longwood University
201 High Street
Farmville, VA 23901

WORKING COPY: DO NOT CITE WITHOUT PERMISSION
* Contact author
Keywords: Property values; Time on market; Hedonic; Megan's Law

1

Abstract
This paper explores the effect that living near a sex offender has on the marketability of one’s
home. Specifically, we estimate the impact on a home’s sales price and the length of time it takes for the
home to sell.

Since the 1994 passage of Sexual Offender Act (known as Megan’s Law), persons

convicted of sex crimes have been required to notify local law enforcement about their current domicile
and any change of address. Since then, sex offenders’ residencies have become publicly available
information allowing anyone to lookup whether a sex offender resides nearby. Using cross-sectional data
from a central Virginia multiple listing service we find that sexual offenders have robust and
economically large effects on nearby real estate. Our results indicate that the presence of a nearby
registered sex offender reduces home values by approximately 9%. Moreover, these same homes take as
much as 10% longer to sell than homes not located near registered sex offenders. These results prove
robust over numerous specifications and modeling techniques commonly found in the literature.

2

Introduction
The intent of this research is to examine the impact of Megan’s Law on the marketing duration
and sales price of residential real estate as observed in rural areas of central Virginia. The empirical real
estate literature has linked a host of housing, property, and surrounding area characteristics to sales price
and marketing duration. However, relatively few have studied the effect of Megan’s Law on sales price
and none have studied the Law’s effect on marketing duration. Additionally, previous studies have
examined the effect of sex offenders in primarily urban settings. For example, Linden and Rockoff (2008)
looked at data from Mecklenburg County, NC (a populous county containing the city of Charlotte), and
Pope (2008) used data from Hillsborough County, FL (containing Tampa). This paper fills this gap in the
literature by estimating the effect a nearby sex offender’s residence has on its surrounding real estate
market in relatively rural areas of central Virginia. We find a substantial difference between our estimates
of a sex offender’s impact in rural areas as compared to the estimates in studies of urban areas, suggesting
a fundamental difference between rural and urban areas in values placed on crime risk imposed by nearby
sex offenders.

Literature Review
While sex offender registries are nothing new, the widespread dissemination of such registries on
the internet is a relatively recent phenomenon. Most states now provide detailed data about the locations,
physical descriptions, and pictures of nearby sex offenders, even including details about the charges for
which they have been convicted. Utilizing these new data sources, Pope (2008) and Linden and Rockoff
(2008) both find that properties in close proximity to a property that is listed as the home address of a
registered sex offender will suffer a loss in value. Both studies control for a variety of individual home
characteristics as well as area fixed effects to control for heterogeneity in their respective real estate
markets. Indeed, the loss to property value that a registered sex offender brings dissipates as distance from
the sexual offenders address increases. However, each study uses sales and crime data from only a single
county (Hillsborough, FL and Mecklenberg, NC, respectively).

3
While the aforementioned studies analyze home values, they omit analyses of the home’s
marketing duration. Yet, marketing duration of a property has been examined and studied from numerous
perspectives in prior real estate literature. Belkin, Hempel and McLeavey (1976) were one of the first to
put forth the theory that list price, and changes to list price, directly impact the property’s time on market.
They also site market imperfections, such as inadequate communication of price changes, that may impact
time on market of properties. Miller (1978) finds a positive relationship between time on market and list
price but also notes that a longer marketing duration does not necessarily result in a higher sales price.
Haurin (1988) has been well cited in the marketing duration literature for his claim that the more atypical
a property the longer it will remain on the market. Specifically, he cites “unusual location” as an example
of an atypical characteristic that may negatively affect a property’s marketing duration. Turnbull and
Dombrow (2007) present evidence that properties located near other properties listed by the same agent
are able to bring a higher sales price. The authors also find that the greater the diversity of listings by an
agent, the longer their listings stay on the market.

Yang and Yavas (1995) suggest that higher

commission rates for agents do not impact the time on market. However, they do suggest that a higher
commission rate may signal that the property is more costly to sell because of its location. They
specifically cite the example of a rural property being more expensive to show than a property in the city.
Our study incorporates methodologies found in these articles regarding property values and marketing
duration and applies them to the analysis of the effect of sexual offenders on real estate marketability.

Data
The data for this research are from several sources. Information about sexual offenders is
contained in the Virginia Sex Offender and Crimes Against Minors Registry and is available on a public
web site maintained by the Commonwealth of Virginia. 1 Each observation contains the registered sex
offender’s current and prior addresses, along with a number of other personal characteristics (e.g. age,
sex, race, and description of the perpetrated crime). Data on real estate transactions consist of
1

See the following website: http://sex-offender.vsp.virginia.gov/sor/

4
observations of residential properties on the market between July 1999 and June 2009 and comes from a
multiple listing service located in central Virginia.

The initial housing data consists of 21,453

observations. After culling for incomplete, missing or illogical data that suggest data entry errors, the final
data set consists of 13,172 sold properties. The data collected from the MLS include typical property
characteristics (square footage, bedrooms and baths), and market and calendar information (location, list
dates, length of listing contract).

Methodology
We are interested in the impact registered sex offenders have on real estate prices and the time it
takes to sell a given home (i.e. marketing duration). After obtaining the longitude and latitude of the
registered sex offenders’ addresses, we use the great-circle distance formula to calculate the distance from
a registered sex offender’s home to a given house on the market.

𝐷𝑖𝑠𝑡𝑎𝑛𝑐𝑒 = 69.1 ∗

180
𝜋

∗ arccos⁡
[sin 𝐿𝐴𝑇1 ∗ sin 𝐿𝐴𝑇2 + cos(𝐿𝐴𝑇1 ) ∗ cos 𝐿𝐴𝑇2 ∗ cos⁡
(𝐿𝑂𝑁2 − 𝐿𝑂𝑁1 ) (1)

Following Waller, Brastow, and Johnson (2009), the present study remains agnostic with respect
to the model specification debate within the literature. Indeed, we employ all three of the most common
methodologies (described below) in the related real estate literature in order to determine the robustness
of empirical results. Like Linden and Rockoff (2008) and Pope (2008), we estimate the effect a sex
offender has on surrounding real estate sale prices (within certain radii) for properties located across the
state of Virginia. One key contribution of this paper, however, is to answer an entirely new question: do
homes near registered sex offenders also take longer to sell? To that end, the following general modeling
framework is presented:

TOMi  ( X i , Li , Z i , SOi , SPi )

(2)

5

SPˆi   ( xi , li , zi , SOi )

(3)

Where: TOM i and SPi are vectors for property marketing time and property selling price respectively
(expressed in natural log form),

X i and xi are vectors of property characteristics and other control variables,
Li and li are vectors for location control,
Z i and zi are vectors that include variables such as the degree of over pricing and market
condition, and

SOi , the variable of interest, equals 1 if a registered sex offender (contemporaneously) lives near
homei,2 0 otherwise. We also estimate the sheer distance between home

i

and the nearest

registered sex offender as a continuous variable.
We report OLS coefficient estimates of time on market and sales price [Equations (2) and (3)] and
Weibull estimates for time on market. A third model captures potential endogeneity between sales price
and marketing time. In this model, SPi is estimated in reduced form in Equation (3) for a three stage
least squares (3SLS) specification, with predicted values, SPˆi , substituting for SPi in Equation (2). In a
3SLS setting, (2) and (3) form the system of equations between property price and property marketing
time. In addition, 3SLS incorporates an additional step with seemingly unrelated regression (SUR)

2

As we will explain in the next section, “nearby” can mean different distances to different people. Below, we
explore a number of distances (or, more specifically, radii) from the nearest sex offender (for example, .1 mile, .25
mile, .5 mile, and 1 mile).

6
estimation.3 Variables included in the equations above (vectors Xi, Li, and Zi) follow existing literature
and will be discussed in greater detail in the next section.

Results
We find that if there is a sex offender registered at a nearby residence, then nearby homes sell for
less and that those homes take longer to sell. However, the magnitudes are striking. Tables 2 (first
column), 3, and 4 show that a nearby sex offender (i.e., one who resides within one-tenth mile) reduces
property values by approximately 9% and increases the marketing duration of a house by approximately
10%. These are economically meaningful effects and demonstrate that central Virginia residents assign a
large risk to living near a convicted sex offender. To determine sensitivity and robustness of the results,
we estimated a number of model specifications.
Prior studies have classified “nearby” registered sex offender in different ways. Table 2 shows the
effects of using different radii for measuring the effect a sex offender has on property values, holding a
number of property characteristics constant. 4 Pope (2008) uses a dummy variable equal to 1 if a sex
offender resides within one-tenth mile (0.1 miles) of a property and another second dummy variable for
two-tenths of a mile (0.2 miles). Linden and Rockoff (2008) also use the 0.1 mile dummy, but differ in
their additional use of a 0.3 mile dummy variable. The first column of Table 2 shows perhaps the most
striking result of the paper: a registered sex offender living within .1 mile of one’s home will reduce the
value of surrounding properties sold by about 9% (or, more precisely, 8.8%). This is more than twice the
magnitude of similar estimates reported in Pope (2008) and Linden and Rockoff (2008), suggesting a

3

According to Belsley (1988), 3SLS has an edge on 2SLS in estimating systems of equations because it is more
efficient, particularly when there are strong interrelations among error terms.
4
Like any multivariate analysis, we attempt to isolate the effect a sex offender has by controlling for numerous
property characteristics that also affect our dependant variable (like square footage, age, number of houses on the
market, whether it is vacant, number of bedrooms, bathrooms, whether it has a pool, brick exterior, hardwood floors,
walk-in closet, finished basement, gas fireplace, paved driveway, fenced yard, and the acreage of the property). In
addition, we control for the year and what time of year it was sold in (i.e. the season), the unemployment rate in
Virginia, whether it is a townhouse or condo, and area fixed effects (in most cases, city/town). Most of these control
variables are commonly used within the real estate literature. Hence, we limit our discussion in this paper primarily
to the variable of interest: proximity to a registered sexual offender.

7
much higher willingness to pay to avoid crime risk in predominantly rural central Virginia. For the
average homeowner, this works out to be $14,826. To put this in relative terms, homeowners value
avoiding this crime risk more than they value a pool, a brick exterior, a walk in closet, a fireplace, a paved
driveway, or an additional bathroom. More precisely, homeowners value avoiding this risk as much as
295 square feet in their home (i.e. homeowners might be willing to live in a significantly smaller home if
they could avoid living so close to a sex offender).
As the radius for the nearest sex offender widens, Table 2 also shows that this effect is still
present, albeit diminished. A nearby sex offender lowers property values 8.5%, 6.6%, and 4.6% when you
classify “nearby” as residing within 0.25, 0.5, and 1 mile respectively. This result is also somewhat
unique, given that prior studies have not found this effect on property values beyond 0.3 miles, suggesting
that there is a significant difference between the perception of “neighbor” in rural and urban areas.
Residences tend to be less densely located in more rural areas like central Virginia than in urban areas like
Charlotte and Tampa. Hence, rural residents may simply perceive homes within a larger radius as
neighbors, resulting in a greater alertness to crime risk over larger distances.
Some real estate studies (e.g. Rutherford, Springer, and Yavas (2005)) have chosen a Heckman
selection model (Heckman, 1979) to correct for sample selection bias. There is good reason to suspect
that our dependant variable, sales price, is only observed for a restricted, non-random sample. Quite
simply, a house only has a selling price if it is actually sold. Hence, OLS estimates could be biased
because a number of houses listed on the market are not sold (e.g. perhaps they are less likely to be sold
because a registered sex offender lives nearby, which would bias the estimates of the houses that actually
sell). Table 3 shows the coefficient estimates for the Heckman model, where the inverse Mills ratio
corrects for selectivity bias by adjusting the conditional error terms such that they have a zero mean
(generated from a probit model, where the binary dependent variable is whether the property has actually

8
sold). Thus, taking into account potential sample selection bias, nearby sex offenders still impose about an
8% discount on surrounding properties.5
This general result remains robust even when we change the relevant area fixed effects. Linden
and Rockoff (2008), for example, use dummy variables for neighborhoods (probably the most relevant
area variable in urban settings), and other studies use similar measures to hold constant one of the most
important pillars of real estate value: location. Studies incorporate area fixed effects in an attempt to
control for unobserved heterogeneity across these areas so that the explanatory variables’ effects are
identified from variation within a given area (or even in a given year, as is the case for year fixed effects).
For most of our analysis up to this point, we have used cities/towns in which these properties reside. In
central Virginia and possibly in other predominantly rural areas, the relevant area metric tends to be wider
than in urban settings. But, perhaps this is too wide and there may still substantial heterogeneity within
these areas. Table 3 shows that when we use zip codes or elementary school districts as the relevant area
metric, we continue to find a substantial discount for homes located near a registered sex offender.
While sellers tend to sell their properties at substantially lower values when a registered sex
offender lives nearby, they may not be lowering their sales price enough. Sellers and their agents may
have difficulty estimating a property’s expected value if a sex offender is near. That is, a reduced offer
price may not attract enough potential buyers, resulting in a longer marketing duration of the home.
Table 4 shows that homes located near (within .1 mile of) a registered sex offender spend abut 10% more
time on the market. This works out to be about 13 days longer on the market than other similar properties,
which are also competitively priced. In relative terms, this is roughly equivalent to selling your home in
the “off” season of fall or winter (as compared to the summer or spring).
However, as is indicated in the methodology section, a number of studies in the real estate
literature employ other econometric models to analyze time on market. Given the extensive debate within
5

The fact that “lambda” is statistically significant indicates selection bias. Though, the differences between the OLS
and Heckman estimates do not appear economically significant, suggesting the selection bias is present but not
large.

9
the literature, our paper intends to remain neutral by employing three common techniques for modeling
time on market: Heckman, Weibull, and three-stage least squares. Moreover, this neutrality comes with an
additional benefit of demonstrating that our general results are not sensitive to the modeling technique,
confirming the robustness6 of our findings. Table 4 shows that all techniques are generally consistent with
the OLS finding that registered sex offenders increase marketing duration of nearby properties in central
Virginia (ranging from approximately 6% to 10%).

Concluding Remarks
This study finds that residents in central Virginia home sellers must absorb a relatively large risk
premium when selling property near a registered sex offender. Alternatively, home buyers are willing to
pay a premium to live in safer areas. The qualitative result is not surprising and is entirely consistent with
previous findings in similar studies. However, the quantitative result reveals an economically large
difference between the risk premium associated with sex offenders in a rural area like central Virginia
versus urban areas like Charlotte and Tampa. Moreover, our analysis of marketing duration suggests that
homes located near registered sex offenders take longer to sell, signifying a general reluctance to purchase
properties exposed to such risks. Certainly, no one wants to live near a registered sex offender, but
empirical results indicate that the more tightly knit communities of central Virginia are willing to pay
more to avoid such a risk. This study’s results are consistent with the notion that residents in rural areas
consider larger areas when defining who constitutes as a “neighbor” and assessing subsequent risks. The
results may also reflect 1) a greater aversion to crime, 2) more households with families compared to
more densely populated urban areas, or some combination of factors that contribute to this rather striking
magnitude. Further research may shine light on the differences between rural and urban areas and perhaps
the sources of such differences.

6

The final version of this working paper will include additional qualitative robustness checks that are not present in
the present version.

10

Exhibit 1: Variable legend
Variable

Definition

Tom
Lprice
lnlp
Sprice
Comm
NoMkt
Age
Sqft
Bedrooms
Fullbath
Halfbath
Garage
Fire
Brick
Vinyl
Hardwood
Ceramic
Fullbase
Vacant
Area 1
Area 2
Area 3
Area 4
Area 5
Area 6
Listtime
Frmsd
Winter
Spring
Summer
Fall

Time on market measured from date of original listing to contract date
Listing price
ln(listing price)
Sales price
Commission rate
Property sold in less than 4 days.
Age of property
Square footage of property
Number of bedrooms
Number of full bathrooms
Number of half bathrooms
Dummy variable, 1 if property has a garage, 0 otherwise
Dummy variable, 1 if property has a fireplace, 0 otherwise
Dummy variable, 1 if property has brick exterior, 0 otherwise
Dummy variable, 1 if property has vinyl floors, 0 otherwise
Dummy variable, 1 if property has hardwood floors, 0 otherwise
Dummy variable, 1 if property has ceramic flooring, 0 otherwise
Dummy variable, 1 if property has full basement, 0 otherwise
Dummy variable, 1 if property was vacant when listed
Dummy variable, 1 if property located in area 1, 0 otherwise
Dummy variable, 1 if property located in area 2, 0 otherwise
Dummy variable, 1 if property located in area 3, 0 otherwise
Dummy variable, 1 if property located in area 4, 0 otherwise
Dummy variable, 1 if property located in area 5, 0 otherwise
Dummy variable, 1 if property located in area 6, 0 otherwise
Chronological time variable
30 year fixed rate mortgage at property contract date
Dummy variable, 1 if property was listed in winter, 0 otherwise
Dummy variable, 1 if property was listed in spring, 0 otherwise
Dummy variable, 1 if property was listed in summer, 0 otherwise
Dummy variable, 1 if property was listed in fall, 0 otherwise

11

Variable

Table 1a: Housing Descriptive Statistics
Mean
Std. Dev.
Min

Max

TOM
Lprice
Sprice
Compensation
Nomkt
Age
Sqft
Bedrooms
Fullbath
Halfbath
Garage
Fire
Brick
Vinylsiding
Hardwood
Ceramictile
Fullbase
Vacant
Area 1
Area 2
Area 3
Area 4
Area 5
Area 6
Listtime
Frmsd
Winter
Spring
Summer
Fall

108.9579
172831.1
167533
2.868985
.0107402
26.46229
1927.133
3.194715
2.001468
.4180189
.3778396
.6709937
.5409519
.5027816
.5475197
.2424664
.5745634
.3290063
.0582599
.1646577
.0295163
.4817648
.0030134
.0511513
24.79818
6.132768
.2638696
.3001854
.2481069
.187838

963
6190000
2650000
7.5
1
470
8578
8
6
4
1
1
1
1
1
1
1
1
1
1
1
1
1
1
41
8.64
1
1
1
1

88.23393
127118.8
103914.4
.3641925
.103081
28.50594
796.0701
.7806939
.6926693
.5318478
.484866
.4698704
.4983394
.5000116
.497756
.4285915
.4944281
.4698704
.2342434
.3708857
.1692551
.4996867
.0548142
.2203147
8.319312
.4866796
.4407464
.4583562
.4319309
.3905979

0
8500
6800
0
0
0
417
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
2
4.81
0
0
0
0

Table 1b: Sex Offender Descriptive Statistics
Variable
Obs
Mean
Std. Dev.

Min

Max

Male
Black
White
Age
Violent crime

0
0
0
16
0

1
1
1
93
1

8986
8986
8986
8986
8986

.977298
.3281772
.6486757
44.48453
.8188293

.14896
.4695758
.4774106
13.00889
.385181

12

Table 2
The effect of a Nearby Registered Sex Offender on the
Selling Price of a Home
(1) - OLS
Independent
Variables

so_tenthmile
so_quartermile
so_halfmile
so_onemile
sqft
lnage
onmkt
vacant
bedrooms
baths
pool
onestory
brick
hardwood
walkincloset
finbase
fireplacegas
paveddrive
fencedyard
condo
townhouse
fall
winter
spring
acreage
vaunemp
constant
Area Fixed
Effects
Year Fixed
Effects
Observations
R-squared

(2) - OLS

Coefficient

t-stat

-0.0885

-8.27

0.0003
-0.1099
0.0000
-0.0823
-0.0250
0.0782
0.0615
0.0184
0.0651
0.1058
0.0251
-0.0669
0.0772
0.0629
0.0063
-0.0468
-0.1465
-0.0041
-0.0227
-0.0067
0.0072
-0.0814
11.9153



46.99
-57.59
0.86
-16.16
-4.8
13.98
11.23
3.68
14.27
20.29
4.77
-12.53
16.77
13.36
1.07
-2.75
-19.66
-0.61
-3.25
-1.09
6.97
-5.62
120.3

(3) - OLS

Coefficient

t-stat

-0.0856

-14.7

0.0003
-0.1070
0.0000
-0.0801
-0.0245
0.0756
0.0560
0.0163
0.0626
0.1076
0.0243
-0.0685
0.0757
0.0613
0.0098
-0.0544
-0.1444
-0.0040
-0.0234
-0.0070
0.0072
-0.0825
11.9253



48.05
-55.9
0.79
-15.7
-4.75
13.69
10.25
3.27
13.83
20.73
4.65
-12.9
16.52
13.11
1.68
-3.22
-19.4
-0.61
-3.36
-1.17
7.01
-5.73
121.2

(4) - OLS

Coefficient

t-stat

-0.0665

-14.7

0.0003
-0.1070
0.0000
-0.0814
-0.0241
0.0761
0.0556
0.0164
0.0643
0.1077
0.0245
-0.0659
0.0761
0.0638
0.0087
-0.0458
-0.1372
-0.0034
-0.0224
-0.0057
0.0072
-0.0832
11.9329



13172
13172
13172
0.7496
0.7520
0.7513
*T-statistics in all of the above regressions are robust

47.39
-55.4
0.79
-16.0
-4.66
13.73
10.18
3.29
14.16
20.71
4.68
-12.4
16.6
13.62
1.48
-2.7
-18.4
-0.52
-3.22
-0.94
7.02
-5.75
120.4

Coefficient

t-stat

-0.0462

-9.26

0.0003
-0.1095
0.0000
-0.0836
-0.0249
0.0785
0.0578
0.0187
0.0658
0.1053
0.0242
-0.0647
0.0752
0.0673
0.0076
-0.0445
-0.1422
-0.0042
-0.0231
-0.0066
0.0072
-0.084
11.9437


46.93
-56.7
0.9
-16.3
-4.79
14.11
10.51
3.74
14.41
20.18
4.61
-12.1
16.36
14.28
1.28
-2.64
-19
-0.64
-3.31
-1.09
7.01
-5.77
120.1


13172
0.7495

13

Table 3
The effect of a Nearby Registered Sex Offender on the
Sales Price of a Home
(1) - OLS
Independent
Variables

so_tenthmile
sqft
lnage
onmkt
vacant
bedrooms
baths
pool
onestory
brick
hardwood
walkincloset
finbase
fireplacegas
paveddrive
fencedyard
condo
townhouse
fall
winter
spring
acreage
vaunemp
constant
lambda (IMR)
City/town Fixed
Effects
Zip Code Fixed
Effects
Elementary
School Fixed
Effects
Year Fixed
Effects
Observations
R-squared

(2) - Heckman

(3) - OLS

(4) - OLS

Coefficient

t-stat

Coefficient*

z-stat

Coefficient

t-stat

Coefficient

t-stat

-0.0885

-8.27

-0.0806

-5.61

-0.0615

-5.87

-0.0444

-4.09

0.0003
-0.1099
0.0000
-0.0823
-0.0250
0.0782
0.0615
0.0184
0.0651
0.1058
0.0251
-0.0669
0.0772
0.0629
0.0063
-0.0468
-0.1465
-0.0041
-0.0227
-0.0067
0.0072
-0.0814
11.9153

46.99
-57.59
0.86
-16.16
-4.8
13.98
11.23
3.68
14.27
20.29
4.77
-12.53
16.77
13.36
1.07
-2.75
-19.66
-0.61
-3.25
-1.09
6.97
-5.62
120.3

0.0003
-0.1000
0.0000
-0.0827
-0.0181
0.0722
0.0580
0.0091
0.0559
0.0996
0.0297
-0.0505
0.0691
0.0587
0.0054
-0.0282
-0.1391
-0.0080
-0.0182
-0.0053
0.0050
-0.0761
12.274
.8158


15.34
-19.21
-0.56
-2.09
-2.36
6.99
4.2
3.48
7.48
8.89
1.57
-0.89
6.72
8.85
0.74
-3.84
-4.23
-0.52
-1.12
-0.39
13.2
-6.58
51.68
12.03

0.0003
-0.1117
0.0000
-0.0769
-0.0142
0.0666
0.0585
0.0237
0.0620
0.0934
0.0263
-0.0706
0.0717
0.0596
0.0055
-0.0715
-0.1301
0.0004
-0.0212
-0.0080
0.0075
-0.0727
11.813

49.77
-60.26
0.81
-15.96
-2.92
12.79
11.2
4.92
14.45
18.72
5.21
-13.92
16.46
13.48
0.97
-4.53
-17.9
0.07
-3.19
-1.38
6.97
-5.23
111.25

0.0003
-0.1034
0.0000
-0.0740
-0.0149
0.0659
0.0503
0.0241
0.0606
0.1068
0.0242
-0.0714
0.0687
0.0570
0.0106
-0.0643
-0.1394
-0.0031
-0.0222
-0.0079
0.0073
-0.0750
11.77

47.19
-52.27
0.81
-15.23
-3.01
12.50
9.33
4.99
13.61
20.92
4.83
-13.92
15.71
12.68
1.88
-3.93
-18.95
-0.49
-3.35
-1.37
7.08
-5.48
113.84














13172
21340
13172
13172
0.7496
n/a
0.7718
0.7746
*Heckman Coefficients (dy/dx) are calculated using the mfx compute, pred(ycond) postestimation command
T-statistics in all of the above regressions are robust

14

Table 4
The effect of a Nearby Registered Sex Offender on the
Marketing Duration (or, Time on Market) of a Home
(1) - OLS
Independent
Variables
so_tenthmile
lnlp
sqft
lnage
onmkt
vacant
bedrooms
baths
pool
onestory
brick
hardwood
walkincloset
finbase
fireplacegas
paveddrive
fencedyard
condo
townhouse
fall
winter
spring
acreage
vaunemp
constant
lambda
Area Fixed
Effects
Year Fixed
Effects
Observations
Chi-squared

(2) - Heckman

(3) - Weibull

(4) – 3SLS

Coefficient

t-stat

Coefficient*

z-stat

Coefficient
+

z-stat

Coefficient

t-stat

0.1043
0.1486
0.0001
-0.0457
0.0002
0.1148
0.0094
-0.0435
-0.0237
-0.0058
-0.0594
-0.0043
0.0126
-0.1260
-0.0214
-0.0477
0.0099
0.2338
0.1760
0.1078
0.0926
-0.0073
0.0009
-0.1111
2.878

4.58
7.8
5.77
-8.22
2.35
9.23
0.96
-3.66
-1.51
-0.47
-4.83
-0.35
0.88
-8.98
-1.68
-3.92
0.67
5.88
6.72
6.05
4.7
-0.47
2.91
-3.43
9.03

0.0601
0.0871
0.0001
-0.0712
0.0001
0.1591
-0.0028
-0.0212
0.0084
-0.0015
-0.0412
0.0030
0.0360
-0.0989
-0.0031
-0.0273
0.0252
0.1207
0.1538
0.1218
0.0898
0.0272
0.0016
0.0361

3.46
3.87
4.23
-6.69
1.54
5.36
0.27
-2.62
-0.67
-1.9
-3.9
-1.72
1.66
-5.76
-1.9
-3.47
0.83
3.7
4.27
5.88
4.08
1.48
1.29
2.82

0.3511
-.8254


0.34
-2.99

0.0785
0.1415
0.0001
-0.0522
0.0002
0.0831
0.0050
-0.0397
-0.0276
-0.0209
-0.0579
0.0049
-0.0287
-0.1452
-0.0591
-0.0362
-0.0155
0.2429
0.2179
0.0776
0.0292
-0.0574
0.0005
-0.1761
1.3383

3.78
9.15
7.54
-12.55
4.83
8.18
0.61
-3.99
-2.08
-2.05
-5.67
0.47
-2.51
-11.94
-5.41
-3.43
-1.19
7.11
11.11
5.7
1.91
-4.52
1.76
-6.98
24.51

0.0761
0.1405
0.0000
-0.0663
0.0001
0.1609
0.0006
-0.0256
0.0039
-0.0052
-0.0451
-0.0050
0.0363
-0.0980
-0.0087
-0.0351
0.0239
0.1420
0.1549
0.1167
0.0892
0.0257
0.0011
0.0291
2.409

2.50
5.12
2.52
-10.36
1.47
10.76
0.05
-1.71
0.21
-0.34
-3.02
-0.32
2.06
-5.58
-0.54
-2.32
1.28
2.83
5.72
5.64
4.06
1.39
1.87
0.76
5.73














21340
21340
21340
13172
n/a
2881.89
3321.35
1356.23
*Heckman Coefficients (dy/dx) are calculated using the mfx compute, pred(ycond) postestimation command
+Weibull Coefficients (dy/dx) are elasticities evaluated at their respective sample medians

15

16

References
Belsley, David A., Two- or Three-stage Least Squares? Computational Economics. 1988, 1, 2130.
Belkin, J., D. J. Hempel and D. W. McLeavey, An Empirical Study of Time on Market Using
Multidimensional Segmentation of Housing Markets, Journal of the American Real Estate and
Urban Economics Association, 1976, 4, 57-75
Haurin, D., The Duration of Marketing Time on Residential Housing, AREUEA Journal 1988,
16(4): 396-410.
Heckman, J., Sample Selection Bias as a Specification Error, Econometrica, 1979, 47 (Feb.):
153-61
Jud, Donald G., Terry G. Seaks, and Daniel T. Winkler, Time on the Market: The Impact of
Residential Brokerage, The Journal of Real Estate Research 1996, 12(3): 447-458.
Knight, John R., Listing Price, Time on Market and Ultimate Selling Price: Causes and Effects
of Listing Price Changes, Real Estate Economics 2002, 30(2): 213-237.
Linden, L. and J.E. Rockoff, Estimates of the Impact of Crime Risk on Property Values from
Megan’s Laws, American Economic Review, 2008, 98:3, 1103-1127.
Miller, N. G., Time on the Market and Selling Price, AREUEA Journal, 1978, 6(2): 164-174.
Pope, J.C., Fear of crime and housing prices: Household reactions to sex offender registries,
Journal of Urban Economics, 2008, 64, 601-614.
Rutherford, R.C., T.M. Springer, and A. Yavas. Conflicts Between Principals and Agents:
Evidence from Residential Brokerage. Journal of Financial Economics, 2005, 76:3, 627 - 65.
Turnbull, G. K., and J. Dombrow. Individual Agents, Firms, and the Real Estate Brokerage
Process. Journal of Real Estate Finance and Economics, 2007, 35:1, 57-76.
Yavas, A. and S. Yang. The Strategic Role of Listing Price in Marketing Real Estate: Theory
and Evidence. Real Estate Economics, 1995, 23:3, 347 - 68.