Content uploaded by Yang Yang
Author content
All content in this area was uploaded by Yang Yang on Feb 26, 2017
Content may be subject to copyright.
Systematic Effects of Crime on Hotel Operating Performance
Nan Hua a, *, Yang Yang b
a Rosen College of Hospitality Management, University of Central Florida, 9907 Universal
Blvd., Orlando, FL 32819, United States
b School of Tourism and Hospitality Management, Temple University, 1810 N. 13th Street,
Speakman Hall 111, Philadelphia, PA 19122, United States
Hua, N. and Yang, Y. (2017). Systematic effects of crime on hotel operating
performance. Tourism Management, 60: 257-269.
Systematic Effects of Crime on Hotel Operating Performance
Abstract
We examine the systematic effects of crime on hotel operating performance based on
data from a sample of 404 Houston hotels from January 2009 to December 2014. Econometric
results show that Part I crime (i.e., violent and property crime) incidents have a significantly
negative impact on hotel operating performance (measured by revenue per available room) after
controlling for other performance determinants. Also, the marginal effect of crime declines as
crime density level increases. Separate examinations of violent and property crimes show that
they exert significant and negative impacts on hotel operating performance, with the impact of
violent crimes being more substantial. In addition, the results reveal that both nighttime and
daytime crime incidents significantly and negatively impact hotel operating performance.
Finally, as evidenced by the insignificant impact of crime incidents occurring on hotel premises,
the results suggest that hotels are generally effective at maintaining systematic security measures
and preventing crime incidents from occurring.
Keywords
Part I Crime; Hotel Performance; Lodging; Violent Crime
1. Introduction
In 2015, there were 53,432 hotels with over 5 million hotel rooms operating in the United
States, collectively employing 1.9 million people and creating $176 billion in revenue (American
Hotel & Lodging Association, 2015). Despite the hotel industry’s success, it is associated with
high levels of crime, against both hotels and guests (e.g., Gill, Moon, Seaman, and Turbin, 2002;
Mawby and Jones, 2007; Zhao and Ho, 2006). Possibly aggravated by location, design, or the
nature of the hospitality industry, many hotels in the United Kingdom and the United States
appear to have significant crime-related problems (e.g., Jones and Groenenboom, 2002; Zhao
and Ho, 2006).
While a number of scholars have investigated hotel operating performance from different
perspectives, such as e-commerce (e.g., Hua, Morosan, and DeFranco, 2015), total quality
management and market orientation (e.g., Wang, Chen, and Chen, 2012), revenue management
systems (e.g., Ortega, 2016), social media (e.g., Kim, Lim, and Brymer, 2015), online reviews
(e.g., Phillips, Barnes, Zigan, and Schegg, 2016; Phillips, Zigan, Silva, and Schegg, 2015),
information technology (e.g., Melián-González and Bulchand-Gidumal, 2016) and dynamic
capabilities (Leonidou, Leonidou, Fotiadis, and Aykol, 2015), the impact of crime on hotel
operating performance has eluded systematic academic examination, leaving a critical gap in the
literature with regards to the theoretical and empirical connection between crime and hotel
operating performance. In addition, prior studies on the effects of crime have been affected by
omitted variable biases (OVB) (e.g., Abbott and Klaiber, 2011; Chay and Greenstone, 2005;
Pope, 2008; Zabel, 2015), presumably resulting from a lack of longitudinal data, an insufficient
number of control variables, and a failure to employ statistical models robust to autocorrelation
and heteroscedasticity issues (e.g., Abbott and Klaiber, 2011; Zabel, 2015). We aim to fill this
gap in the literature by employing a robust methodology with a proper panel design. First, using
monthly hotel and crime data, we examine whether personal and property crimes affect hotel
operating performance at the property level. Second, we examine the effects of nighttime and
daytime crime incidents on hotel operating performance. Third, we study the effects of personal
and property crime incidents within the hotel perimeter on hotel operating performance. Finally,
following recommendations to alleviate OVB (Abbott and Klaiber, 2011; Zabel, 2015), we
employ a comprehensive set of controlled variables with a panel design to construct fixed effect
models in order to test our hypotheses.
This paper is organized as follows. In the next section, we review relevant literature and
develop our hypotheses. In Section 3, we describe the data and methodology before reporting
empirical results for a number of specifications in Section 4. We discuss implications in Section
5 and offer some concluding remarks in Section 6.
2. Literature Review
While in the majority of prior studies, scholars have investigated the impact of crime on
property values (e.g., Buck, Hakim, and Spiegel, 1993; Buck, Hakim, and Spiegel, 1991; Burnell
1988; Congdon-Hohman, 2013; Linden and Rockoff, 2008; Lynch and Rasmussen, 2001; Pope
and Pope, 2012; Thaler, 1978; Zabel, 2015), few researchers have examined the impact of crime
on businesses, notable exceptions being Abadie and Dermisi (2008), Burnham, Feinberg, and
Husted (2004), Greenbaum and Tita (2004), Rosenthal and Ross (2010), Schwartz et al. (2003),
and Sloan, Caudill, and Mixon (2016). As a result, the impact of crime on hotel performance has
eluded systematic examination from both practitioners and scholars.
2.1. Impact of Crime on Property Values
In a well-established line of research, scholars have investigated the impact of crime on
property values (e.g., Bishop and Murphy, 2011; Buck, Hakim, and Spiegel, 1991, 1993; Burnell
1988; Congdon-Hohman, 2013; Cullen and Levitt, 1999; Glaeser and Sacerdote, 1999; Gibbons,
2004; Ihlanfeldt and Mayock, 2010; Linden and Rockoff, 2008; Pope, 2008; Pope and Pope,
2012; Thaler, 1978; Zabel, 2015). Since the 1960s, the hedonic pricing model has been employed
to explore a variety of valuation issues such as air quality value (e.g., Ridker and Henning,
1967), school value (e.g., Kain and Quigley, 1970), and crime value (e.g., Thaler, 1978). The key
rationale behind such a model rests upon the belief that economic agents typically consider
housing characteristics and local amenities to be critical when selecting a place of residence
(Schwartz et al., 2003). Agent preference is thus influenced by housing values, which are utilized
to extract the “implicit price” of a housing attribute or local amenity (e.g., Chay and Greenstone,
2005; Kain and Quigley, 1970; Pope, 2008; Ridker and Henning, 1967; Schwartz et al., 2003;
Thaler, 1978) under a certain market equilibrium. As a result, economists have frequently used
the hedonic pricing model to examine the impact of crime on household valuations (e.g., Buck,
Hakim, and Spiegel, 1991, 1993; Burnell, 1988; Gray and Joelson, 1979; Hellman and Naroff,
1979; Lynch and Rasmussen, 2001; Naroff, Hellman, and Skinner, 1980; Rizzo, 1979; Zabel,
2015).
The empirical evidence shows that crime generally has a significant and negative impact
on property values. For example, in two studies, crime rates exhibited a significant impact of -
0.07 (Thaler, 1978) and -0.05 (Haurin and Brasington, 1996) on the elasticity of home values.
From a different measurement perspective, a one standard deviation increase in crime on average
leads to a significant decrease of -3.11 standard deviations in the natural log of house prices
(Zabel, 2015). Such consistent empirical evidence has been obtained repeatedly, using both
cross-sectional data (e.g., Burnell, 1988; Gray and Joelson, 1979; Hellman and Naroff, 1979;
Lynch and Rasmussen, 2001; Naroff, Hellman, and Skinner, 1980; Rizzo, 1979) and panel data
(e.g., Buck, Hakim, and Spiegel, 1991, 1993; Zabel, 2015).
2.2. Impact of Crime on Businesses
Notwithstanding the abundance of literature on crime and property value, few have
explored the impact of crime on businesses (Sloan, Caudill, and Mixon, 2016) other than Bates
and Robb (2008), Burnham, Feinberg, and Husted (2004), Greenbaum and Tita (2004),
Rosenthal and Ross (2010), Schwartz et al. (2003), and Sloan, Caudill, and Mixon (2016).
Despite meaningful insights offered by these studies, distinct pieces of empirical evidence and
perspectives suggest the need for more in-depth and contextualized investigations on the
systematic impact of crime. Specifically, Schwartz et al. (2003) examined the impact of crime on
the real estate business boom in 1994 by employing hedonic regression models on panel data of
repeat sales in New York City between 1988 and 1998, revealing that falling crime rates
contributed to about one third of the property price increase after 1994. Burnham et al. (2004)
examined panel data from 318 U.S. counties between 1982 and 1997, which showed that violent
crime has a negative and larger impact on economic growth in nearby suburbs than in more
distant suburbs. Using a panel of zip code-level data for five U.S. cities between 1987 and 1994,
Greenbaum and Tita (2004) found that service-related establishments in low crime areas tend to
experience increased violence. Employing confidential microdata from the U.S. Census Bureau’s
Characteristics of Business Owners Survey, Bates and Robb (2008) found that business viability
appears similar for firms that are most negatively affected by crime and otherwise identical firms
that are not affected by crime. Further, some business owners may rationally choose to locate
their businesses in high crime areas. More recently, Rosenthal and Ross (2010) showed that
relative sensitivity to crime can be used as a sorting mechanism to categorize different sectors of
the economy in order to understand business location choices. They showed that entrepreneurs
consider violent crime to be a critical factor when bidding for business locations in five U.S.
cities. Building on these findings from Rosenthal and Ross (2010), Sloan, Caudill, and Mixon
(2016) focused on the relationship between crime and restaurant openings in Memphis,
Tennessee from 2009 to 2014 and found that higher crime is positively and significantly
associated with new restaurant openings, indicating that location benefits may overpower crime
problems and attract restaurant entrepreneurs.
In addition, scholars have recognized both theoretically and empirically that fear of
violence causes consumers, employees and entrepreneurs to change their routine activities
(e.g., Greenbaum and Tita, 2004; Wilcox et al., 2003), resulting in direct and indirect increases in
business operating costs. For example, Hamermesh (1999) found that areas with higher homicide
rates are associated with deviations from optimal patterns of work timing (i.e., evening and
nighttime work is shifted to the daytime), resulting in increased labor costs. In addition, both
crime and the fear of crime lead to cost increases associated with surveillance, security,
insurance premiums, and stolen property repair and replacement (Burrows et al., 2001; Fisher,
1991; Mirrlees-Black and Ross, 1995).
2.3. Determinants of Hotel Operating Performance
Commonly recognized as a unique investment asset class, hotels are financial businesses
that encompass both retail and housing activities (Manning, O’Neill, Singh, Hood, Liu, and
Bloom, 2015). These real estate investments not only provide accommodations, food and
beverage services and recreational amenities, among others, but also share many financial
similarities with retail businesses, such as the need to invest in inventory, working capital,
management expertise, marketing and labor. Consequently, although hotel businesses provide
accommodations, they differ significantly from residential property investments and are often
considered to perform more like financial businesses (Quan, Li, and Sehgal, 2002).
Because hotels are exposed to the volatilities of both real property values and specialized
managerial assets, their performance tends to fluctuate (e.g., Ro and Ziobrowski, 2009;
Rushmore, Ciraldo, and Tarras, 2002). Macro- and micro-level factors have been recognized to
affect hotel operating performance (e.g., Assaf and Cvelbar, 2011; Duverger, 2013; Sainaghi,
2011; Yang, Wong, and Wang, 2012). At the macro level, hotel operating performance has been
shown to be impacted by market structure variables (e.g., Assaf and Cvelbar, 2011), particularly
the competition structure of the local market (e.g., Duverger, 2013). In addition, spatial
agglomeration tends to be associated with positive economic externalities; as a result, co-located
hotels appear to perform better than those isolated from other hotels (Yang, Wong, and Wang,
2012). Furthermore, seasonality and year effects have been widely recognized to have a
significant impact on hotel performance (e.g., Kosová and Enz, 2012). At the micro level,
researchers have identified several common determinants of hotel operating performance that
have significant impacts (e.g., O’Neill and Xiao, 2006; Sainaghi, 2011; Xiao, O'Neill, and
Mattila, 2012). Specifically, hotel size has been shown to be negatively associated with hotel
operating performance (Sainaghi, 2011), chain affiliation appears to increase a hotel property’s
market value (O’Neill and Xiao, 2006), and hotel positioning and target market selection appear
to dictate average daily rate (ADR) and affect operating performance (Xiao, O'Neill, and Mattila,
2012). Lastly, hotel owners have been shown to have significant influence on hotel operating
performance, likely due to their implementation of strategies related to property location and
selection of target markets, brands, and operators (e.g., Xiao, O'Neill, and Mattila, 2012).
However, to our knowledge, the impact of crime on hotel performance has eluded systematic
academic examination.
2.4. Hypothesis Development: The Impact of Crime on Hotel Performance
Conceptually, crime is defined as “an act committed or omitted in violation of a law
forbidding or commanding it” (Black, 1979, p. 334). In particular, we examine violent and
property crimesserious offenses that are likely to be reported to police and occur regularly in
all areas of the country (Federal Bureau of Investigation, 2004). As a result, the “fear of crime is
salient to neighborhood conditions” (Lynch and Rasmussen, 2001, p. 1982) since crimes “can
have serious consequences for both individual holiday-makers and tourist destinations” (Ryan,
1992, p. 173). In particular, hotels appear to be associated with high levels of crime, including
offenses against both hotels and guests (Gill et al., 2002). Even prestigious hotels in London
appear to have a significant problem with crime, arguably exacerbated by hotel location, design,
and the nature of the hospitality industry (Jones and Groenenboom, 2002). Hotel crime,
particularly hotel burglary, has been identified as a critical problem in England and Wales
(Mawby & Jones, 2007), and Florida (Zhao and Ho, 2006).
In general, commercial crime appears to be much more common than crime against
households (e.g., Mawby, 2003; Taylor, 2004). Furthermore, the majority of empirical evidence
shows disproportionate amounts of crime in tourist areas, where tourists suffer high levels of
victimization, particularly related to burglary (e.g., de Albuquerque and McElroy, 1999; Berger,
1992; Brunt and Hambly, 1999; Chesney-Lind and Lind, 1986; Mawby et al., 1999; Pizam and
Mansfeld, 1996; Ryan, 1993; Stangeland, 1998). In contrast, Huang et al. (1998) concluded that
crime was uncommon in hotels. Despite these mixed findings, hotel security has long been under
extensive study in the United States (e.g., Beaudry, 1996).
Because hotels’ performance reputations are based on attracting and satisfying
customerswho generally are concerned about victimization risk and the seriousness of
potential offenses (e.g., Warr, 1984; Warr and Stafford, 1983)crime rates are relevant to hotel
performance. For example, tourists play a significant role in the hotel industry, and “safety,
tranquility and peace are a necessary condition for prosperous tourism…most tourists will not
spend their hard earned money to go to a destination where their safety and well-being may be in
jeopardy” (Pizam and Mansfeld, 1996, p. 1).
It is widely recognized in the tourism literature that tourists are the most frequent victims
of property crimes such as burglaries and robberies (Biagi and Detotto, 2014). More importantly,
because crime affects people’s perceptions, it could gradually damage the image of a tourism
destination and by extension, area hotels (Dimanche and Lepetic, 1999), which would in turn
lead to an overall decline in tourism (Shiebler, Crotts, and Hollinger, 1996) and poor hotel
performance (Xie, Zhang, and Zhang, 2014). Safety perceptions play a role in future travel
decisions; tourists’ perceptions of risk and safety during travel can have a “stronger influence on
avoidance of regions than likelihood of travel to them” (Sönmez and Graefe, 1996, p. 45).
Tourists tend to avoid returning to a destination if they feel threatened and unsafe during their
stay (Dimanche and Lepetic, 1999). In short, crime can lead to grave consequences for both
individual hotels and tourist destinations (Ryan, 1993).
In addition, drawing upon prior empirical evidence in the real estate and economics
literatures, crime appears to have an adverse impact on repeat real estate sales (Schwartz et al.,
2003), economic growth (Burnham et al., 2004), service-related establishments (Greenbaum and
Tita, 2004), vacancy rates (Abadie and Dermisi, 2008), and business location choices (Rosenthal
and Ross, 2010). In some countries such as Jamaica, crime-related business costs have a negative
effect on economic development and tourism (Ajagunna, 2006; World Bank, 2003).
Hotel operating performance is commonly measured by revenue per available room
(RevPAR) (Anderson and Lawrence, 2014), because it not only simultaneously accounts for both
the average daily rate (ADR) and occupancy rate (Zhang, Lawrence, and Anderson, 2015), but
also is considered to be “the most effective yardstick of the balance between hotel room supply
and demand” (e.g., Hua, Morosan, and DeFranco, 2015, p. 112). Thus, we propose the following
hypothesis:
H1: Violent and property crimes have a significant and negative impact on hotel RevPAR.
Specifically, crimes that occur between 6 p.m. and 6 a.m. (nighttime crimes) can be more
detrimental because victims are more vulnerable in the dark and when there are fewer people
around. In addition, darkness decreases the ability to see potential offenders and increases fear of
crime (e.g., Ramsay and Newton, 1991). Therefore, we propose the following hypothesis:
H1a: Nighttime crimes have a significant and negative impact on hotel RevPAR.
Crimes that occur between 6 a.m. and 6 p.m. (daytime crimes) may exhibit an impact on
hotel RevPAR that differs in magnitude from that of nighttime crimes due to relatively fewer
crimes occurring during the day than at night (e.g., Ramsay and Newton, 1991). Therefore, we
propose the following hypothesis:
H1b: Daytime crimes have a significant and negative impact on hotel RevPAR.
Violent crimes could introduce more fear (Braakmann, 2009), when compared to
property crimes that do not involve any personal assaults. Therefore, violent crime incidents tend
to be more harmful than property crime incidents for lodging businesses. Consequently, we
propose the following two hypotheses:
H1c: Violent crimes have a significant and negative impact on hotel RevPAR.
H1d: Property crimes have a significant and negative impact on hotel RevPAR.
In addition, crimes that occur at hotels (both on hotel properties and in hotel parking lots)
could have a negative impact on hotel performance (Pizam, 1999). Therefore, we propose the
following hypotheses:
H1e: Crimes occurring at hotels have a significant and negative impact on hotel RevPAR.
H1f: Crimes occurring on hotel properties (excluding parking lots) have a significant and
negative impact on hotel RevPAR.
H1g: Crimes occurring in hotel parking lots have a significant and negative impact on
hotel RevPAR.
3. Research Method
3.1 Data Sources
As the largest city in the Southern United States, Houston is home to 23 Fortune 500
firms, with not only a diverse employment base, but also an excellent infrastructure supporting
the largest port based on international tonnage in the nation, the sixth largest airport system in
the world, over 600 trucking firms and two major rail systems (The City of Houston, 2016).
Ranked among the top places to visit in 2015 by the Boston Globe, Travel + Leisure and Condé
Nast Traveler, Houston received approximately 14.8 million visitors who spent $17 billion in the
city in 2014. Houston has approximately 78,000 hotel rooms, supporting more than 123,000 jobs
in the hospitality industry (Visit Houston, 2016).
However, Houston is only safer than 4% of U.S. cities, and a person has about a 1%
chance of becoming a victim of a violent crime, almost three times the nation’s median.
Alarmingly, for every 1,000 residents, there are 9.9 violent crimes and 47.42 property crimes
each year (Neighborhood Scout, 2016). As a result, considering the significant role the lodging
industry plays and potential challenges crime brings, Houston provides an excellent context in
which to systematically examine the impact of crime on hotel operating performance.
We obtained crime data from the Houston Police Department’s (HPD) database on
Neighborhood Crime Statistics (http://www.houstontx.gov/police/cs/beatpages/beat_stats.htm).
The data cover Part I index crimes included in monthly police reports released to the FBI. Part I
index crimes consist of eight serious crimes, which can be further classified into two categories:
personal (violent) and property crimes. Aggravated assault, rape, murder, and robbery are
classified as personal (violent) crimes, whereas burglary, theft, auto vehicle theft, and arson are
classified as property crimes. All crime records are linked with each police beat, and the
geographic area of the city is broken down for patrol and statistical purposes in HPD. Moreover,
the date, time and location of each offense are documented in each crime record. In the HPD
database, premise codes identify the type of location where the crime occurred. Premise codes
directly related to hotels include “140 HOTEL/MOTEL/ETC.” and “18M (HOTEL/MOTEL
PARKING LOT).” For the research period, January 2009 to December 2014, there are 768,357
reported Part I crime incidents in the database. We counted the number of Part I crime incidents
per police beat for each month of the research period. Additionally, we counted the number of
incidents for each crime category. We used these count values as the numerators for the
subsequent calculations of measures of crime density.
We obtained property-level data for Houston hotels from two different data sources. First,
we used data from the Texas Comptroller of Public Accounts, which records the monthly taxable
accommodation revenue, number of units, and tax obligation period for each hotel property
(Kalnins and Lafontaine, 2013) in the city of Houston. Second, we used the Smith Travel
Research (STR) Hotel Census Database, which covers hotel amenities and financial
characteristics, as well as geo-spatial information on hotel properties (Duverger, 2013). We
obtained data for 404 hotel properties located within the administrative boundary of the city of
Houston. Because the data cover all hotel properties in the city of Houston, the number of hotels
increases each year due to new hotel entrants. (Note that Texas state law requires all hotel
properties to report their tax information; hence, our data cover all Houston hotels.)
3.2 Econometric Model
Omitted variable bias (OVB) can be a big concern when evaluating the impact of crime
in an econometric model, as it can lead to inconsistent coefficient estimates (e.g., Pope, 2008). In
particular, cross-sectional analyses tend to be more vulnerable to unobserved variables that are
correlated with independent variables of interest specified in the model (Zabel, 2015). Take
crime as an example: it is intuitive to speculate that a number of factors, such as environmental
characteristics, labor market conditions and other important neighborhood features, would differ
between areas with high and low levels of crime (Pope and Pope, 2012). If these unobserved
factors cannot be controlled for in the modeling process due to data constraints, bias will be
introduced during the estimation process for the impact of crime. Following the
recommendations of Abbott and Klaiber (2011) and Zabel (2015), we alleviated potential OVB
by increasing the number of controlled variables, creating a panel design and employing fixed
effects to remove time-invariant unobserved factors. We specified the baseline empirical model
as:
ln it it it i it
RevPAR X
Z
where i indicates a hotel property (i = 1, …, 404), and t indicates the month. Based on the
availability of HPD crime data, we selected the research period from January 2009 to December
2014, covering a total of 72 months over 6 years.
it
X
is the independent variable measuring the
level of crime nearby for hotel i during month t, and
it
Z
denotes a matrix of control variables.
Moreover,
i
captures the time-invariant hotel-specific effect of hotel i that influences the
dependent variable but has not been incorporated into any explanatory variables. In this way, the
proposed two-way panel data model can resolve the potential OVB problem (Wooldridge, 2010).
The conventional error term
it
is assumed to follow a normal distribution independent from
i
with a mean of 0 and a variance of
2
.
We estimated the proposed econometric model using the fixed effect model; compared to
the random effect model, the fixed effect model has fewer restrictions and allows for
interdependence between explanatory variables and the hotel-specific effect
i
. Since this
interdependence is very common in reality, the fixed effect model is recommended to generate
reliable and consistent estimates (Wooldridge, 2010). However, we also estimated random effect
models as a robustness check. Additionally, we estimated the variance-covariance of the model
using the Huber-White sandwich estimator, which is robust to cross-sectional heteroscedasticity
and within-panel serial-correlation (Arellano, 2003).
3.3 Variable Definition
We included the following variables in the econometric specification described in section
3.2.
Dependent variables
lnRevPARit: the log of total accommodation revenue per available rooms (RevPAR) of
hotel i in month t. This variable has been advocated by many studies to measure the hotel
operating performance (Anderson and Lawrence, 2014) because it not only
simultaneously accounts for both the average daily rate (ADR) and occupancy rate
(Zhang, et al., 2015) of a hotel property, but also is considered to be “the most effective
yardstick of the balance between hotel room supply and demand” (e.g., Hua, Morosan,
and DeFranco, 2015).
Independent variables
For independent variables, we used crime density measures based on police beats.
Although each crime record in the database includes location information (i.e., the street and a
range of possible street numbers), we found it extremely time consuming and technically
challenging to geocode all 768,357 crime incidents with an ideal level of precision. Therefore,
we chose the police beat as the unit of analysis, and aggregated all individual crime incident
records at the beat level based on crime incident type for each month. For larger aggregate units
such as zip code areas and census tracts, aggregation may mask intra-unit variation in crime
patterns over time. Hence, based on data availability, beat-level aggregation of crime incidents is
most appropriate and able to best approximate the level of crime intensity around each hotel. In
the criminology literature, researchers generally use two types of crime measures in empirical
analysis: population-based rates (to adjust for the resident population) and density measures (to
adjust for the size of a geographic unit) (Harries, 2006). The density measures are more suitable
for our study, because many hotels are located near tourist attractions and shopping centers,
which have a very sparse resident population; furthermore, the crime levels indicated by a
population-based measure could be very misleading (Zhang and Peterson, 2007). We used the
following crime measures as independent variables:
crimeit indicates the crime density (number of Part I crime incidents per square mile) for
the police beat of the region encompassing hotel i in month t;
crime_nightit indicates the density of nighttime crimes (number of nighttime Part I crime
incidents per square mile) for the police beat of the region encompassing hotel i in month
t. Nighttime crimes refer to incidents occurring between 6 p.m. and 6 a.m.;
crime_dayit indicates the density of daytime crimes (number of daytime Part I crime
incidents per square mile) for the police beat of the region encompassing hotel i in month
t. Daytime crimes refers to incidents occurring between 6 a.m. and 6 p.m.;
crime_violentit indicates the density of violent crimes (number of Part I violent crime
incidents per square mile) for the police beat of the region encompassing hotel i in month
t;
crime_propertyit indicates the density of property crimes (number of Part I property crime
incidents per square mile) for the police beat of the region encompassing hotel i in month
t;
crime_hotelit indicates the density of crimes occurring at hotels (number of Part I crime
incidents at hotels per square mile) for the police beat of the region encompassing hotel i
in month t. Crime incidents at hotels are indicated by premise codes “140” and “18M” in
the HPD crime statistics database;
crime_hotel1it indicates the density of crimes occurring on hotel properties (excluding
parking lots) (number of Part I crime incidents at hotel properties per square mile) for the
police beat of the region encompassing hotel i in month t. Crime incidents at hotel
properties are indicated by premise code “140” in the HPD crime statistics database;
crime_hotel2it indicates the density of crimes occurring in hotel parking lots (number of
Part I crime incidents in hotel parking lots per square mile) for the police beat of the
region encompassing hotel i in month t. Crime incidents in hotel parking lots are
indicated by premise code “18M” in the HPD crime statistics database.
Due to the high level of correlation across these crime measures, we created separate model
specifications for each.
Control variables
We also included the following control variables in our econometric model as
it
Z
:
lnneighborsit indicates the number of other hotel properties within a 2-mile radius of
hotel i in month t. Because positive economic externalities are associated with spatial
agglomeration, co-located hotels may perform better than those isolated from other hotels
(Yang, et al., 2012);
lnroomsit denotes the log number of rooms offered by hotel i for accommodation in
month t. Due to dis-economies of scale, larger-sized hotels can be associated with lower
levels of operating performance (Sainaghi, 2011);
lnHHIit denotes the log of Herfindahl-Hirschman Index (HHI) in terms of hotel rooms in
the zip code area of hotel i in month t. We calculated HHI as
2
1
N
n
n
HHI s
, where
n
s
is
the market share of hotel i in the zip code market measured by accommodation revenue,
and N is the number of hotels in the market. In the industrial organization literature, HHI
measures the competition structure of the local market, and a larger HHI value indicates a
higher market concentration. Market structure variables have been found to be associated
with hotel performance (Assaf and Cvelbar, 2011; Duverger, 2013);
operationit is used to indicate hotel operating type. STR classifies all hotel properties into
three operating types: operation = 1 for chain-operated hotels, operation = 2 for franchise
hotels, and operation = 3 for independent hotels. Hospitality scholars have suggested that
chain affiliation boosts the market value of hotel properties (O’Neill and Xiao, 2006) and
that independent hotels are financially disadvantaged due to a lack of reputation and less
mature managerial structures;
classit indicates hotel class, reflecting how the property positions itself and its target
market. The STR database includes six possible classes: class = 1 for economy; class = 2
for midscale; class = 3 for upper midscale; class = 4 for upscale; class = 5 for upper
upscale; and class = 6 for luxury. Hotels of different classes charge different ADRs and
achieve different levels of performance (Xiao, et al., 2012);
owner_chg1it, indicates change in ownership for hotel i in month t; owner_chg1 = 1 if
hotel i experienced a change in ownership during month t, and owner_chg1 = 0
otherwise;
owner_chg2it, indicates a recent change in ownership for hotel i around month t;
owner_chg2 = 1 if hotel i experienced a change in ownership within 12 months of month
t, and owner_chg2 = 0 otherwise;
montht, indicates the month-specific effect, including a set of 11 dummies for February to
December with January as the reference group;
yeart, indicates the year-specific effect, including a set of five dummies for years 2010 to
2014 with year of 2009 as the reference group.
3.4 Data Description
In Figure 1, we present a map of hotels in Houston. As shown on the map, within the city
limits of Houston, hotels are clustered near the city center and along major highways. Figure 1
also visualizes the average monthly crime density across police beats; Part I crime is heavily
concentrated in the southwest of the city. In general, the map shows that most hotels are located
within police beats with higher crime densities.
(Please insert Figure 1 about here)
In Table 1, we present the descriptive statistics of continuous variables incorporated into
our empirical model. Our panel data set consists of 25,911 observations from 404 hotels from
January 2009 to December 2014. The mean value of lnRevPAR is 3.641 with a standard
deviation of 0.752. For crime-related independent variables, crime has a mean value of 29.353,
and as indicated by the larger mean value of crime_day over crime_night, there were more crime
incidents during the daytime in police beats covering Houston hotels. Moreover, the mean value
of crime_violent is 4.118, which is approximately one sixth of the mean value of crime_property,
which is 25.236, suggesting that property crimes are dominant in our data set. Among the crime
incidents recorded in our data, only a very small number are hotel-related, with the mean value
of crime_hotel a mere 1.283. A comparison of the mean values of crime_hotel1 and
crime_hotel2 reveals that around two thirds of hotel-related crimes occurred in parking lots.
Table 2 further presents the descriptive statistics of categorical variables. In our data, across the
three types of hotel operations (operation), franchise hotels dominate the dataset, accounting for
52.99% of all hotels. Regarding hotel class (class), 45.15% of the dataset is comprised of
economy hotels, with very few upper upscale (9.78%) and luxury (2.12%) hotels in the Houston
lodging market. For month dummies (month), our data are evenly distributed over the 12 months
of the year, whereas for year dummies (year), we noted more observations in recent years due to
a number of new hotel entrants into the Houston market.
(Please insert Table 1 about here)
(Please insert Table 2 about here)
In Table 3, we present the correlation matrix of major independent and control variables
included in the proposed empirical model. Most coefficients are below 0.5, suggesting that the
multi-collinearity problem does not exist in the model (Gujarati and Porter, 2010). More
importantly, the independent variable of major interest, crime, is found to be barely correlated
with other control variables, with all coefficients below 0.3. In the matrix, four coefficients for
the categorical variables of class and operation are slightly above 0.5. For alternative crime
measures, we did not find any plausible multi-collinearity issues with other control variables.
The correlation coefficients with other crime-related variables are not presented and are available
upon request.
(Please insert Table 3 about here)
4. Results
In Table 4, we present the estimation results of empirical models with general crime
measures. Model 1 includes crime, measuring the effect of all Part I crime incidents on hotel
performance. The coefficient of crime is estimated to be negative and statistically significant,
providing empirical evidence in support of H1. The result indicates that one more crime incident
per square mile is associated with a 0.17% (exp(-0.00172)-1) drop in RevPAR, with a 95
confidence interval of [0.13%, 0.22%]. Regarding the control variables in the model, several are
estimated to be statistically significant. The negative and significant coefficients of lnrooms and
lnHHI suggest that hotel performance was lower for larger-sized hotels in a more competitive
market. Moreover, the coefficient of operation = 3 is negative and significant, suggesting that
compared to chain-operated and franchise hotels, independent hotels operated at a lower
performance level. Lastly, both owner_chg1 and owner_chg2 are negative and significant,
highlighting the lower performance levels of hotels with ownership changes during the month or
within the last 12 months. The estimated coefficients reveal that hotels experienced a 20.15%
(exp(-0.225)-1) decline in RevPAR during the month of an ownership change, and a RevPAR
drop of 6.88% (exp(-0.0713)-1) when there had been an ownership change within the previous
12 months. In Model 2, we further introduce crime_square (i.e., the squared term of crime) into
Model 1, and both crime and crime_square are estimated to be statistically significant. Figure 2
visualizes the marginal effect of crime over different levels of crime density, revealing that the
marginal effect was more intense at the lower crime density level. In other words, the marginal
effect, explained as the effect of one additional Part I crime incident per square mile, declined as
crime density level increased. This result is consistent with findings reported by Greenbaum and
Tita (2004) that a crime surge has a greater effect on business activities in a low-crime
neighborhood. One possible reason is that hotels in high-crime areas have already adapted to the
financial environment, whereas it is particularly challenging for hotels in low-crime areas to
adjust to a surge in crime.
In Models 3 and 4, we compare the effect of nighttime and daytime crime incidents on
hotel operating performance. In Model 3, we use an alternative crime measure, crime_night, to
investigate the effect of Part I crime incidents that occurred during nighttime hours. The
coefficient is estimated to be -0.00234 and is statistically significant, supporting Hypothesis H1a.
Its magnitude is moderately larger than the estimate of crime in Model 1. It suggests that one
additional nighttime crime incident per square mile is associated with a 0.23% (exp(-0.00234)-1)
drop in RevPAR, with a 95% confidence interval of [0.12%, 0.35%]. In Model 4, crime_day is
included to capture Part I crime incidents that occurred during daytime hours. Its coefficient is
estimated to be -0.00180 and is statistically significant, supporting Hypothesis H1b. The results
suggest that one additional daytime crime incident per square mile is associated with a 0.18%
(exp(-0.00180)-1) drop in RevPAR, with a 95% confidence interval of [0.12%, 0.24%]. The
magnitude of this coefficient is smaller than that of crime_night, suggesting that nighttime Part I
crime incidents are more detrimental to hotel businesses. In Models 5 and 6, we further compare
the effect of violent and property crime incidents on hotel operating performance. In Model 5,
crime_violent is estimated to be -0.00499 and significant, supporting Hypothesis H1c. One
additional violent crime incident per square mile is associated with a 0.50% (exp(-0.00499)-1)
drop in RevPAR with a 95% confidence interval of [0.17%, 0.83%]. In Model 6, crime_property
is estimated to be -0.00166 and significant, supporting Hypothesis H1d. One more property crime
incident per square mile is associated with a 0.17% (exp(-0.00166)-1) drop in RevPAR with a
95% confidence interval of [0.12, 0.21]. The results suggest that even though only a small
portion of Part I crime incidents are violent (Table 1), the detrimental effects of violent crimes on
hotel operating performance are almost three times greater than the effects of property crimes.
The estimated coefficients and significances of the other control variables vary little across
different models, demonstrating the robustness of our results.
(Please insert Table 4 about here)
(Please insert Figure 2 about here)
The HPD database on Neighborhood Crime Statistics tracks crime incidents that occur at
hotel locations. Therefore, we introduce another crime measure, crime_hotel, in Model 7 to
understand the effect of Part I crime incidents at hotels. Its coefficient is statistically
insignificant, rejecting Hypothesis H1e. In Models 8 and 9, we compare the effect to different
types of hotel-related crimes, those occurring on hotel properties and in hotel parking lots,
respectively. In Model 8, the coefficient of crime_hotel1 is found to be insignificant. Likewise,
in Model 9, the coefficient of crime_hotel2 is insignificant. Therefore, both Hypotheses H1f and
H1g are rejected. These findings are enlightening because they indicate that crime incidents
within the hotel perimeter do not significantly or systematically impact hotel operating
performance. In addition, these findings suggest two potential explanations. On the one hand, if
crimes did not occur systematically at hotels, then hotel operating performance would not be
affected significantly. This potential explanation indicates that hotels are generally effective at
maintaining systematic security measures and preventing crime incidents from taking place,
given the negative impacts of crime incidents in the hotel neighborhood (police beat) on hotel
operating performance as shown in Tables 1–4. On the other hand, if crimes occurred
systematically at hotels but did not systematically affect hotel performance, then effective
interventions must have countered their negative influence. This potential explanation suggests
that hotels are effective at crime impact recovery, public relations and marketing.
Neighborhood crime incidents appear to be more systematic (Tables 1–4) and require
well-designed and structured police and government involvement as well as community
commitment to mitigate negative impacts on hotels. Furthermore, destination marketing
organizations (DMOs) and other tourism industry stakeholders such as local communities and
local businesses can collaborate with crime prevention organizations and local councils to help
reduce neighborhood crimes (Townsley, Reid, Reynald, Rynne, and Hutchins, 2014).
Figure 3 shows the overall effects of different types of crimes (estimated coefficients of
measures and their corresponding 95% confidence intervals) on hotel operating performance in
Models 1 to 9. Our results indicate that Hypotheses H1, H1a, H1b, H1c, and H1d cannot be rejected
whereas Hypotheses H1e H1f and H1g are rejected. In general, we find that violent crime incidents
(crime_violent) have the most damaging effect, followed by nighttime crime incidents
(crime_night). As expected, the negative effect of violent crimes on hotel performance is larger
than the effect of property crimes, and the negative effect of nighttime crimes is larger than that
of daytime crimes.
(Please insert Table 5 about here)
(Please insert Figure 3 about here)
Lastly, we performed a robustness check of the empirical results with random effect
models. Note that the Hausman test, a common statistical test gauging the statistical suitability of
a fixed effect versus random effect model, is not available when models are estimated with
robustness standard errors. Table 6 presents the results for the random effect models (Models
10–18). Due to space limitations, we only present the estimates of crime measures; the entire
estimation results are available upon request. The results suggest that, in general, the estimated
coefficients of random effect models are slightly smaller than those of their fixed effect
counterparts. The magnitudes and statistical significances vary little between the fixed and
random effect models, confirming the robustness of our results presented in Tables 4 and 5.
Major results from the fixed effect models hold in the random effect models. For example, the
negative effect of violent crime incidents is most pronounced on a hotel’s operation performance,
followed by the effect of nighttime crime incidents. Table 6 also includes the results from a
robustness check of alternative crime measures (Models 19–27). Considering the fact that the
effects of crime incidents may be delayed, we use the average crime density of the last three
months as an alternative measure. The results vary little from the major results obtained in
previous analyses.
(Please insert Table 6 about here)
5. Discussions and Implications
5.1. Managerial Perspectives
Perceptions of a destination have been recognized as critical to tourists’ destination
choices, the success of tourism and hotel operating performance. For example, the considerable
media attention generated by the 1992 Florida tourist murders led to a significant decline in
tourism demand as well as hotel operating performance (Pizam, 1999). These findings are
consistent with results of this study. Therefore, it is critical for police departments, governments
and communities to collaborate in order to systematically improve safety conditions in hotel
neighborhoods. At the same time, proper marketing strategies should be formed to propose and
promote positive images of hotel neighborhoods and to counter the impact of crime. It appears
that more research is also needed in this area (e.g., Dimanche and Lepetic, 1999).
On the other hand, although hotels appear to be associated with high levels of crime (e.g.,
Gill et al., 2002; Jones and Groenenboom, 2002; Mawby and Jones, 2007; Zhao and Ho, 2006),
the findings of this study suggest that hotels generally have security measures in place that
prevent crimes from occurring systematically. Since unsystematic crime incidents within the
hotel perimeter would produce unsystematic impacts on hotel operating performance, it is critical
to continuously maintain and improve hotel safety conditions. In addition, it is possible that the
systematic crimes that do occur at hotels have an unsystematic impact on hotel performance due
to effective interventions administered by hotels, such as well-implemented crime impact
recovery efforts, and well-executed public relations and marketing campaigns (Pizam, 1999).
Moreover, results of this study suggest that it is critical to educate tourists about
destination safety and security. Tourists usually spend relatively a short period of time at a
certain location in “holiday mode;” therefore, they tend to be less conscious of their personal
safety, which can create opportunities for crimes. Hospitality and tourism industry practitioners,
local communities, tourists, local businesses, international communities and law enforcement
agencies can collaborate to identify strategies and solutions to educate tourists and reduce crime
opportunities (Pizam, 1999; Townsley, Reid, Reynald, Rynne, and Hutchins, 2014).
5.2. Theoretical Perspectives
Present value theory has long served as the foundation for evaluating commercial
properties (Corgel et al., 2015). Hotel valuation, in particular, has largely followed such a
fundamental perspective and evolved over time, leading to many specific techniques widely used
in practice and scholarly work (e.g., Chen and Kim, 2010; Fu et al., 2013; O’Neill, 2004;
Rushmore, 1992). Common to all of these techniques is the discounting process that relies on
future financial benefits, which are significantly driven by hotel revenue. However, crime’s
impact rarely is explicitly and systematically incorporated into present value models, likely
causing bias in the relevant valuation process. Since crime has a significant and negative impact
on hotel operating performance, proper adjustments should be made to form future financial
benefits estimates in the hotel valuation process when present value theories are utilized.
In prior studies, scholars have proposed four fundamental theoretical models of hotel
location choice, namely, the tourist-historic city model (e.g., Ashworth and Tunbridge, 1990),
the mono-centric model (e.g., Alonso, 1964; Egan and Nield, 2000), the agglomeration model
(e.g., Ingram and Inman, 1996; Kalnins and Chung, 2004), and the multi-dimensional model
(e.g., Baum and Haveman, 1997). Based on these theoretical models, a number of empirical
models have been derived which in turn have informed many operational models that are used to
determine hotel locations (Yang, Luo, and Law, 2014). However, the impact of crime has not
been explicitly and systematically incorporated into any of these theoretical models, thus crime is
not considered in most hotel location decisions. Our findings could serve as a starting point to
systematically incorporate the crime dimension into the multi-dimensional model and advance
the hotel location decision-making process theoretically and empirically.
In addition, our findings could be applied in the restaurant setting to further understand
the dynamic balance between crime problems and location benefits. For example, Sloan, Caudill,
and Mixon (2016) examined restaurateurs’ location choices and found crime to have a positive
and significant relationship with new restaurant openings, implying that restaurateurs who weigh
crime problems with location benefits when making their location choice decisions tend to find
that location benefits outweigh crime problems.
Finally, scholars have emphasized OVB in recent hedonic pricing models (Black, 1999;
Figlio and Lucas, 2004; Chay and Greenstone, 2005; Pope, 2008); OVB often stems from an
insufficient number of control variables, a lack of longitudinal data, and a failure to use proper
panel data analysis techniques (e.g., Abbott and Klaiber, 2011; Zabel, 2015). By utilizing the
fixed effects model with a comprehensive control variable set in a panel data design, we offer
robust empirical evidence that delineates the relationship between crime and hotel operating
performance, establishing a foundation for scholars to further explore crime’s financial impact.
Conclusion
In this study, we examined the systematic effects of crime on hotel operating
performance. Results show that Part I crime incidents have a significantly negative impact on
hotel operating performance, offering strong empirical evidence in support of H1. Also, we found
that the marginal effect of crime declines as crime density level increases. Separate examinations
of the two categories of Part I crime (i.e., violent and property crime), show significant and
negative impacts on hotel operating performance. In addition, nighttime and daytime crime
incidents significantly and negatively impact hotel operating performance. Finally, as evidenced
by the insignificant impacts of crime incidents occurring on hotel premises, this study suggests
that hotels are effective at either maintaining security measures and systematically preventing
crime incidents from taking place, or implementing interventions (e.g., crime impact recovery,
public relations, marketing) that counter the systematic influence of crime.
One caveat, however, is that the findings of this study are based on one market, Houston,
and the results thus might not be generalizable to other U.S. markets. Given the higher crime
levels and higher customer volume in Houston, it would be beneficial to perform this analysis in
a different market where the issue of crime is less pronounced and the size of the economy is
closer to the U.S. average.
References:
Abadie, A., & Dermisi, S. (2008). Is terrorism eroding agglomeration economies in central
business districts? Lessons from the office real estate market in downtown Chicago. Journal of
Urban Economics, 64(2), 451-463.
Abbott, J. K., & Klaiber, H. A. (2011). An Embarrassment of Riches: Confronting Omitted
Variable Bias and Multi-Scale Capitalization in Hedonic Price Models. Review of Economics
and Statistics, 93(4), 1331-1342.
Ajagunna, I. (2006). Crime and harassment in Jamaica: Consequences for sustainability of the
tourism industry. International Journal of Contemporary Hospitality Management, 18(3), 253-
259.
Alonso, W., (1964). Location and Land Use: Toward a General Theory of Land Rent.
Cambridge, MA: Harvard University Press.
American Hotel & Lodging Association. (2015). Lodging Industry Trends 2015. Retrieved from
https://www.ahla.com/uploadedFiles/_Common/pdf/Lodging_Industry_Trends_2015.pdf
Anderson, C. K., & Lawrence, B., (2014). The influence of online reputation and product
heterogeneity on service firm financial performance. Service Science, 6(4), 217-228.
Arellano, M., (2003). Panel Data Econometrics. Oxford: Oxford University Press.
Ashworth, G.J. & Tunbridge, J.E., (1990). The Tourist-Historic City. London: Belhaven Press.
Assaf, A. G., & Cvelbar, K. L., (2011). Privatization, market competition, international
attractiveness, management tenure and hotel performance: Evidence from Slovenia.
International Journal of Hospitality Management, 30(2), 391-397.
Bates, T., & Robb, A. (2008). Crime's impact on the survival prospects of young urban small
businesses. Economic Development Quarterly, 22(3), 228-238.
Baum, J.A.C., & Haveman, H.A., (1997). Love thy neighbor? Differentiation and agglomeration
in the Manhattan hotel industry, 1898–1990. Administrative Science Quarterly, 42(2), 304–338.
Beaudry, M. H. (1996). Contemporary lodging security. Elsevier.
Berger, V. (1992). Payne and suffering-a personal reflection and a victim-centered critique. Fla.
St. UL Rev., 20, 21.
Biagi, B., & Detotto, C. (2014). Crime as tourism externality. Regional Studies, 48(4), 693-709.
Bishop, K. C., & Murphy, A. D. (2011). Estimating the willingness to pay to avoid violent crime:
A dynamic approach. The American Economic Review, 101(3), 625-629.
Black, S. E. (1999). Do better schools matter? Parental valuation of elementary education.
Quarterly journal of economics, 577-599.
Black, H. (1979). Black’s Law Dictionary. 5th ed. St. Paul, MN: West.
Braakmann, N., (2009). Is there a compensating wage differential for high crime levels? First
evidence from Europe. Journal of Urban Economics, 66(3), 218-231.
Brunt, P., & Hambly, Z. (1999). Tourism and crime: A research agenda. Crime Prevention &
Community Safety, 1(2), 25-36.
Buck, A. J., Hakim, S., & Spiegel, U. (1993). Endogenous crime victimization, taxes, and
property values. SOCIAL SCIENCE QUARTERLY-AUSTIN-, 74, 334-334.
Buck, A. J., Hakim, S., & Spiegel, U. (1991). Casinos, crime, and real estate values: Do they
relate?. Journal of Research in Crime and Delinquency, 28(3), 288-303.
Burnell, J. D. (1988). Crime and racial composition in contiguous communities as negative
externalities: prejudiced household's evaluation of crime rate and segregation nearby reduces
housing values and tax revenues. American Journal of Economics and Sociology, 47(2), 177-
193.
Burnham, R., Feinberg, R. M., & Husted, T. A. (2004). Central city crime and suburban
economic growth. Applied Economics, 36(9), 917-922.
Burrows, J., Anderson, S., Bamfield, J., Hopkins, M., & Ingram, D. (2001). Crime against
Business in Scotland. Edinburgh: Stationery Office (Scottish Executive Central Research Unit).
Chay, K. Y., & Greenstone, M. (1998). Does air quality matter? Evidence from the housing
market. National Bureau of Economic Research.
Chen, M., & Kim, W. G. (2010). Hotel Valuation in China A Case Study of a State-Owned
Hotel. Cornell Hospitality Quarterly, 51(3), 429-445.
Chesney-Lind, M., & Lind, I. Y. (1986). Visitors as victims crimes against tourists in Hawaii.
Annals of Tourism Research, 13(2), 167-191.
Congdon-Hohman, J. M. (2013). The lasting effects of crime: The relationship of discovered
methamphetamine laboratories and home values. Regional Science and Urban Economics, 43(1),
31-41.
Corgel, J. B., Liu, C., & White, R. M. (2015). Determinants of Hotel Property Prices. The
Journal of Real Estate Finance and Economics, 51(3), 415-439.
Cullen, J. B., & Levitt, S. D. (1999). Crime, urban flight, and the consequences for cities. Review
of economics and statistics, 81(2), 159-169.
De Albuquerque, K., & McElroy, J. (1999). Tourism and crime in the Caribbean. Annals of
tourism research, 26(4), 968-984.
Dimanche, F., & Lepetic, A. (1999). New Orleans tourism and crime: A case study. Journal of
travel research, 38(1), 19-23.
Duverger, P., (2013). Curvilinear effects of user-generated content on hotels’ market share: A
dynamic panel-data analysis. Journal of Travel Research, 52(4), 465-478.
Egan, D.J., & Nield, K., (2000). Towards a theory of intra urban hotel location. Urban Studies,
37(3), 611–621.
Enz, C. A., & Canina, L. (2002). The best of times, the worst of times: Differences in hotel
performance following 9/11. Cornell Hotel and Restaurant Administration Quarterly, 43(5), 41-
52.
Federal Bureau of Investigation, (2004). Crime in the United States 2004. Retrieved from:
https://www2.fbi.gov/ucr/cius_04/appendices/appendix_02.html
Fernandes, A. M. (2008). Firm productivity in Bangladesh manufacturing industries. World
Development, 36(10), 1725-1744.
Figlio, D. N., & Lucas, M. E. (2004). What's in a grade? School report cards and the housing
market. American economic review, 591-604.
Fisher, B. (1991). A neighborhood business area is hurting: Crime, fear of crime, and disorders
take their toll. Crime & Delinquency, 37(3), 363-373.
Fu, J., Sheel, A., & Lang, J. (2013). A Reexamination of Current Hotel Valuation Techniques–
Which Approach is More Realistic?. The Journal of Hospitality Financial Management, 21(1),
17-30.
Gibbons, S. (2004). The Costs of Urban Property Crime*. The Economic Journal, 114(499),
F441-F463.
Gill, M., Moon, C., Seaman, P., & Turbin, V. (2002). Security management and crime in hotels.
International Journal of Contemporary Hospitality Management, 14(2), 58-64.
Glaeser, E., & Sacerdote, B. (1999).“Why is there more crime in cities?”. Journal of Political
Economy. 107, 225–258.
Gray, C. M., & Joelson, M. R. (1979). Neighborhood crime and the demand for central city
housing. The Cost of Crime, Sage, Beverly Hills, CA, 47-60.
Greenbaum, R. T., & Tita, G. E. (2004). The impact of violence surges on neighbourhood
business activity. Urban Studies, 41(13), 2495-2514.
Gujarati, D. N., & Porter, D. C., (2010). Essentials of Econometrics. New York: McGraw-
Hill/Irwin.
Hamermesh, D. S. (1999). Crime and the Timing of Work. Journal of Urban Economics, 45(2),
311-330.
Harries, K., (2006). Extreme spatial variations in crime density in Baltimore County, MD.
Geoforum, 37(3), 404-416.
Haurin, D. R., & Brasington, D. (1996). The impact of school quality on real house prices:
Interjurisdictional effects. Journal of Housing Economics, 5(4), 351-368.
Hellman, D. A., & Naroff, J. L. (1979). The impact of crime on urban residential property
values. Urban Studies, 16(1), 105-112.
Hua, N., Morosan, C., & DeFranco, A. (2015). The other side of technology adoption:
Examining the relationships between e-commerce expenses and hotel performance. International
Journal of Hospitality Management, 45, 109-120.
Huang , W . S . W . , Kwag , M . & Streib , G . ( 1998 ) . Exploring the Relationship Between
Hotel Characteristics and Crime. FIU Hospitality Review. 16(1), 81 – 93.
Ingram, P., & Inman, C., (1996). Institutions, intergroup competition, and the evolution of hotel
populations around Niagara falls. Administrative Science Quarterly, 41(4), 629–658.
Ihlanfeldt, K., & Mayock, T. (2010). Panel data estimates of the effects of different types of
crime on housing prices. Regional Science and Urban Economics, 40(2), 161-172.
Islam, A. (2014). Economic growth and crime against small and medium sized enterprises in
developing economies. Small Business Economics, 43(3), 677-695.
Johnson, C., & Vanetti, M. (2005). Market Developments in the hotel sector in Eastern Central
Europe. Advances in Hospitality and Leisure, 1, 153-175.
Jones, P., & Groenenboom, K. (2002). Crime in London hotels. Tourism and Hospitality
Research, 4(1), 21-35.
Kain, J. F., & Quigley, J. M. (1970). Measuring the value of housing quality. Journal of the
American statistical association, 65(330), 532-548.
Kalnins, A., & Chung, W., (2004). Resource-seeking agglomeration: a study of market entry in
the lodging industry. Strategic Management Journal, 25 (7), 689–699.
Kalnins, A., & Lafontaine, F., (2013). Too far away? The effect of distance to headquarters on
business establishment performance. American Economic Journal: Microeconomics, 5(3), 157-
179.
Kim, W. G., Lim, H., & Brymer, R. A. (2015). The effectiveness of managing social media on
hotel performance. International Journal of Hospitality Management, 44, 165-171.
Kosová, R., & Enz, C. A. (2012). The terrorist attacks of 9/11 and the financial crisis of 2008 the
impact of external shocks on US hotel performance. Cornell Hospitality Quarterly, 53(4), 308-
325.
Leonidou, L. C., Leonidou, C. N., Fotiadis, T. A., & Aykol, B. (2015). Dynamic capabilities
driving an eco-based advantage and performance in global hotel chains: The moderating effect of
international strategy. Tourism Management, 50, 268-280.
Linden, L., & Rockoff, J. E. (2008). Estimates of the impact of crime risk on property values
from Megan's Laws. The American Economic Review, 98(3), 1103-1127.
Lynch, A. K., & Rasmussen, D. W. (2001). Measuring the impact of crime on house prices.
Applied Economics, 33(15), 1981-1989.
Manning, C., O'Neill, J. W., Singh, A., Hood, S., Liu, C., & Bloom, B. A. (2015). The
Emergence of Hotel/Lodging Real Estate Research. Journal of Real Estate Literature, 23(1), 1-
26.
Mawby, R. I. (2003). Crime and the Business Community: Experiences of Businesses in
Cornwall, England. Security Journal, 16(4), 45-61.
Mawby, R., Brunt, P., & Hambly, Z. (1999). Victimisation on holiday: a British survey.
International review of victimology, 6(3), 201-211.
Mawby, R. I., & Jones, C. (2007). Attempting to reduce hotel burglary: Implementation failure
in a multi-agency context. Crime Prevention & Community Safety, 9(3), 145-166.
Melián-González, S., & Bulchand-Gidumal, J. (2016). A model that connects information
technology and hotel performance. Tourism Management, 53, 30-37.
Mirrlees-Black, C., & Ross, A. (1995). Crime against retail and manufacturing premises:
findings from the 1994 Commercial Victimisation Survey (Vol. 146). London: Home Office.
Naroff, J. L., Hellman, D., & Skinner, D. (1980). Estimates of the impact of crime on property
values. Growth and Change, 11(4), 24-30.
Newell, G., & Seabrook, R. (2006). Factors influencing hotel investment decision making.
Journal of Property Investment & Finance, 24(4), 279-294.
Neighborhood Scout, (2016). Crime rates for Houston, TX. Retrieved from
http://www.neighborhoodscout.com/tx/houston/crime/
O’Neill, J. W. (2004). An automated valuation model for hotels. Cornell Hotel and Restaurant
Administration Quarterly, 45(3), 260-268.
O’Neill, J. W., & Xiao, Q., (2006). The role of brand affiliation in hotel market value. Cornell
Hotel and Restaurant Administration Quarterly, 47 (3), 210-223.
Ortega, B. (2016). Revenue management systems and hotel performance in the economic
downturn. International Journal of Contemporary Hospitality Management, 28(4).
Phillips, P. A., Barnes, S., Zigan, K., & Schegg, R. (2016). Understanding the impact of online
reviews on hotel performance: an empirical analysis. Journal of Travel Research.
Phillips, P., Zigan, K., Silva, M. M. S., & Schegg, R. (2015). The interactive effects of online
reviews on the determinants of Swiss hotel performance: A neural network analysis. Tourism
Management, 50, 130-141.
Pizam, A. (1999). A comprehensive approach to classifying acts of crime and violence at tourism
destinations. Journal of travel research, 38(1), 5-12.
Pizam, A., & Mansfeld, Y. (1996). Tourism, crime and international security issues.Tourism,
crime and international security issues.. John Wiley & Sons.
Pope, J. C. (2008). Buyer information and the hedonic: the impact of a seller disclosure on the
implicit price for airport noise. Journal of Urban Economics, 63(2), 498-516.
Pope, D. G., & Pope, J. C. (2012). Crime and property values: Evidence from the 1990s crime
drop. Regional Science and Urban Economics, 42(1), 177-188.
Quan, D. C., Li, J., & Sehgal, A. (2002). The performance of lodging properties in an investment
portfolio. Cornell Hospitality Quarterly, 43(6), 81.
Ramsay, M., & Newton, R. (1991). The effect of better street lighting on crime and fear: A
review. Retrieved from: http://www.ocpa-
oh.org/Evaluation%20and%20Research/Effect%20of%20Street%20Lighting%20on%20Crime%
20-%20UK.pdf
Ridker, R. G., & Henning, J. A. (1967). The determinants of residential property values with
special reference to air pollution. The Review of Economics and Statistics, 246-257.
Rizzo, M. J. (1979). The cost of crime to victims: An empirical analysis. The Journal of Legal
Studies, 8(1), 177-205.
Ro, S., & Ziobrowski, A. J. (2011). Does focus really matter? Specialized vs. diversified REITs.
The Journal of Real Estate Finance and Economics, 42(1), 68-83.
Rosenthal, S. S., & Ross, A. (2010). Violent crime, entrepreneurship, and cities. Journal of
Urban Economics, 67(1), 135-149.
Rushmore, S. (1992). Seven current hotel-valuation techniques. The Cornell Hotel and
Restaurant Administration Quarterly, 33(4), 49-56.
Rushmore, S., Ciraldo, D. M., & Tarras, J. (2001). Hotel Investments Handbook 2002. The West
Group, New York.
Ryan, C. (1993). Crime, violence, terrorism and tourism: an accidental or intrinsic relationship?.
Tourism Management, 14(3), 173-183.
Sainaghi, R., (2011). RevPAR determinants of individual hotels: Evidences from Milan.
International Journal of Contemporary Hospitality Management, 23(3), 297-311.
Schwartz, A. E., Susin, S., & Voicu, I. (2003). Has falling crime driven New York City's real
estate boom?. Journal of Housing Research, 14(1), 101.
Slattery, P. (2002). Reported RevPAR: unreliable measures, flawed interpretations and the
remedy. International Journal of Hospitality Management, 21(2), 135-149.
Sloan, C., Caudill, S. B., & Mixon, F. G. (2016). Entrepreneurship and crime: The case of new
restaurant location decisions. Journal of Business Venturing Insights, 5, 19-26.
Sönmez, S., & Graefe, A. (1996). Risk perceptions of US international vacation travelers.
Proceedings of the 27th AnnualTravel and TourismResearch Association, 97-108.
Stangeland, P. (1998). Other targets or other locations? An analysts of opportunity structures.
British journal of criminology, 38(1), 61-77.
Taylor, J. (2004). Crime against retail and manufacturing premises: findings from the 2002
Commercial Victimisation Survey. Home Office. Research, Development and Statistics
Directorate.
Thaler, R. (1978). A note on the value of crime control: evidence from the property market.
Journal of Urban Economics, 5(1), 137-145.
The City of Houston, (2016). About Houston. Retrieved from
http://www.houstontx.gov/abouthouston/business.html
Townsley, M., Reid, S., Reynald, D., Rynne, J., & Hutchins, B. (2014). Risky facilities: Analysis
of crime concentration in high-rise buildings. Trends and Issues in Crime and Criminal Justice,
(476), 1.
Visit Houston, (2016). About Houston. Retrieved from http://www.visithoustontexas.com/about-
houston/facts-and-figures/
Wang, C. H., Chen, K. Y., & Chen, S. C. (2012). Total quality management, market orientation
and hotel performance: The moderating effects of external environmental factors. International
Journal of Hospitality Management,31(1), 119-129.
Warr, M. (1984). Fear of victimization: Why are women and the elderly more afraid?. Social
science quarterly, 65(3), 681-702.
Warr, M., & Stafford, M. (1983). Fear of victimization: A look at the proximate causes. Social
forces, 61(4), 1033-1043.
Wilcox, P., Land, K. C. and Hunt, S. A. (2003). Criminal Circumstance: A Dynamic
Multicontextual Criminal Opportunity Theory. New York: Aldine de Gruyter.
Wooldridge, J. M., (2010). Econometric analysis of cross section and panel data. Massachusetts:
MIT press.
World Bank. (2003). Jamaica – The Road to Sustained Growth. Country Economic
Memorandum 26088-JM, World Bank.
Xiao, Q., O'Neill, J. W., & Mattila, A. S., (2012). The role of hotel owners: The influence of
corporate strategies on hotel performance. International Journal of Contemporary Hospitality
Management, 24(1), 122-139.
Xie, K. L., Zhang, Z., & Zhang, Z. (2014). The business value of online consumer reviews and
management response to hotel performance.International Journal of Hospitality Management,
43, 1-12.
Yang, Y., Luo, H., & Law, R. (2014). Theoretical, empirical, and operational models in hotel
location research. International Journal of Hospitality Management, 36, 209-220.
Yang, Y., Wong, K. K. F., & Wang, T., (2012). How do hotels choose their location? Evidence
from hotels in Beijing. International Journal of Hospitality Management, 31 (3), 675-685.
Zabel, Jeffrey. "The hedonic model and the housing cycle." Regional Science and Urban
Economics 54 (2015): 74-86.
Zhang, H., & Peterson, M. P., (2007). A spatial analysis of neighbourhood crime in Omaha,
Nebraska using alternative measures of crime rates. Internet Journal of Criminology, 31, 1-31.
Zhang, J. J., Lawrence, B., & Anderson, C. K., (2015). An agency perspective on service triads:
Linking operational and financial performance. Journal of Operations Management, 35, 56-66.
Zhao, J., & Ho, T. (2006). Are Foreign Visitors More Likely Victimized In Hotels? Policy
Implications. Security Journal, 19(1), 33-44.
Anderson, C. K., Lawrence, B., 2014. The influence of online reputation and product heterogeneity on
service firm financial performance. Service Science 6 (4), 217-228.
Arellano, M., 2003. Panel Data Econometrics. Oxford: Oxford University Press.
Assaf, A. G., Cvelbar, K. L., 2011. Privatization, market competition, international attractiveness,
management tenure and hotel performance: Evidence from Slovenia. International Journal of
Hospitality Management 30 (2), 391-397.
Braakmann, N., 2009. Is there a compensating wage differential for high crime levels? First evidence
from Europe. Journal of Urban Economics 66 (3), 218-231.
Duverger, P., 2013. Curvilinear effects of user-generated content on hotels’ market share: A dynamic
panel-data analysis. Journal of Travel Research 52 (4), 465-478.
Greenbaum, R. T., Tita, G. E., 2004. The impact of violence surges on neighbourhood business activity.
Urban Studies 41 (13), 2495-2514.
Gujarati, D. N., Porter, D. C., 2010. Essentials of Econometrics. New York: McGraw-Hill/Irwin.
Harries, K., 2006. Extreme spatial variations in crime density in Baltimore County, MD. Geoforum 37 (3),
404-416.
Kalnins, A., Lafontaine, F., 2013. Too far away? The effect of distance to headquarters on business
establishment performance. American Economic Journal: Microeconomics 5 (3), 157-179.
O’Neill, J. W., Xiao, Q., 2006. The role of brand affiliation in hotel market value. Cornell Hotel and
Restaurant Administration Quarterly 47 (3), 210-223.
Sainaghi, R., 2011. RevPAR determinants of individual hotels: Evidences from Milan. International
Journal of Contemporary Hospitality Management 23 (3), 297-311.
Wooldridge, J. M., 2010. Econometric analysis of cross section and panel data. Massachusetts: MIT
press.
Xiao, Q., O'Neill, J. W., Mattila, A. S., 2012. The role of hotel owners: The influence of corporate
strategies on hotel performance. International Journal of Contemporary Hospitality
Management 24 (1), 122-139.
Yang, Y., Wong, K. K. F., Wang, T., 2012. How do hotels choose their location? Evidence from hotels in
Beijing. International Journal of Hospitality Management 31 (3), 675-685.
Zhang, H., Peterson, M. P., 2007. A spatial analysis of neighbourhood crime in Omaha, Nebraska using
alternative measures of crime rates. Internet Journal of Criminology 31 (1-31.
Zhang, J. J., Lawrence, B., Anderson, C. K., 2015. An agency perspective on service triads: Linking
operational and financial performance. Journal of Operations Management 35 (56-66.
Table 1. Descriptive statistics of continuous variables
Variable
Obs
Mean
Std. Dev.
Skewness
Kurtosis
lnRevPAR
25911
3.651
0.752
-0.255
3.114
crime
25911
29.353
23.081
2.080
12.870
crime_night
25911
12.880
10.206
1.640
7.017
crime_day
25911
16.779
13.895
2.950
25.288
crime_violent
25911
4.118
3.747
1.605
6.414
crime_property
25911
25.236
20.860
2.432
16.396
crime_hotel
25911
1.283
1.541
2.013
7.703
crime_hotel1
25911
0.472
0.725
3.195
16.636
crime_hotel2
25911
0.811
1.076
2.499
12.021
lnneighbors
25911
2.705
0.670
-1.452
5.312
lnrooms
25911
4.475
0.826
0.274
2.649
lnHHI
25911
-1.670
0.529
1.189
4.410
owner_chg1
25911
0.008
0.090
10.919
120.230
owner_chg2
25911
0.074
0.261
3.261
11.631
Table 2. Descriptive statistics of categorical variables
Freq.
Percent
Cum. Percent
operation
1.Chain-Managed
5318
20.52
20.52
2.Franchise
13730
52.99
73.51
3.Independent
6863
26.49
100.00
class
1.Economy
11699
45.15
45.15
2.Midscale
3483
13.44
58.59
3.Upper Midscale
3677
14.19
72.78
4.Upscale
3969
15.32
88.10
5.Upper Upscale
2533
9.78
97.88
6.Luxury
550
2.12
100.00
month
Jan
2141
8.26
8.26
Feb
2145
8.28
16.54
Mar
2150
8.30
24.84
Apr
2151
8.30
33.14
May
2156
8.32
41.46
Jun
2157
8.32
49.79
Jul
2164
8.35
58.14
Aug
2167
8.36
66.50
Sep
2168
8.37
74.87
Oct
2169
8.37
83.24
Nov
2170
8.37
91.61
Dec
2173
8.39
100.00
year
2009
3881
14.98
14.98
2010
4060
15.67
30.65
2011
4142
15.99
46.63
2012
4542
17.53
64.16
2013
4580
17.68
81.84
2014
4706
18.16
100.00
Table 3. Correlation matrix of major independent variables
crime
lnneigh-
bors
lnrooms
lnHHI
operation
=1
operation
=2
operation
=3
class=1
class=2
class=3
class=4
class=5
class=6
owner_
chg1
lnneighbors
0.296
lnrooms
0.277
0.267
lnHHI
0.136
-0.396
-0.198
operation=1
0.109
0.109
0.399
-0.061
operation=2
-0.075
0.113
0.096
-0.123
-0.540
operation=3
-0.015
-0.228
-0.474
0.194
-0.305
-0.637
class=1
-0.149
-0.255
-0.547
0.175
-0.054
-0.389
0.490
class=2
-0.104
0.001
-0.042
-0.039
0.045
0.132
-0.191
-0.358
class=3
-0.097
0.102
0.034
-0.079
-0.111
0.281
-0.216
-0.369
-0.160
class=4
0.141
0.134
0.286
-0.106
-0.043
0.238
-0.230
-0.386
-0.168
-0.173
class=5
0.239
0.100
0.510
-0.061
0.178
-0.046
-0.111
-0.299
-0.130
-0.134
-0.140
class=6
0.148
0.092
0.140
0.068
0.089
-0.148
0.086
-0.134
-0.058
-0.060
-0.063
-0.049
owner_chg1
0.006
0.010
0.024
-0.011
0.011
0.005
-0.016
-0.009
-0.003
0.005
0.010
-0.001
0.005
owner_chg2
0.003
0.011
0.076
-0.043
0.021
0.051
-0.076
-0.051
0.005
0.016
0.054
0.008
-0.024
-0.024
Table 4. Estimation results of fixed effect models with general crime measures
Model 1
Model 2
Model 3
Model 4
Model 5
Model 6
crime
-0.00172***
-0.00315***
(0.000)
(0.001)
crime_square
0.00000706***
(0.000)
crime_night
-0.00234***
(0.001)
crime_day
-0.00180***
(0.000)
crime_violent
-0.00499***
(0.002)
crime_property
-0.00166***
(0.000)
lnneighbors
-0.0316
-0.0334
-0.0301
-0.0295
-0.0279
-0.0312
(0.031)
(0.031)
(0.031)
(0.031)
(0.030)
(0.031)
lnrooms
-1.024***
-1.026***
-1.029***
-1.022***
-1.025***
-1.024***
(0.183)
(0.183)
(0.184)
(0.183)
(0.184)
(0.183)
lnHHI
-0.174*
-0.175*
-0.178*
-0.170*
-0.172*
-0.173*
(0.095)
(0.095)
(0.095)
(0.095)
(0.095)
(0.095)
operation=2
0.0146
0.0140
0.0144
0.0148
0.0152
0.0145
(0.048)
(0.048)
(0.048)
(0.048)
(0.048)
(0.048)
operation=3
-0.112**
-0.112**
-0.109**
-0.111**
-0.108**
-0.112**
(0.054)
(0.054)
(0.054)
(0.055)
(0.054)
(0.054)
class=2
-0.0663
-0.0655
-0.0635
-0.0653
-0.0631
-0.0658
(0.059)
(0.059)
(0.060)
(0.059)
(0.059)
(0.059)
class=3
0.0941
0.0962
0.0968
0.0939
0.0965
0.0942
(0.231)
(0.232)
(0.230)
(0.230)
(0.229)
(0.231)
class=4
0.00712
0.0119
0.00774
0.00396
0.00270
0.00719
(0.274)
(0.275)
(0.273)
(0.274)
(0.274)
(0.274)
class=5
0.0269
0.0310
0.0272
0.0259
0.0269
0.0268
(0.271)
(0.272)
(0.270)
(0.271)
(0.271)
(0.271)
class=6
0.164
0.167
0.161
0.161
0.160
0.163
(0.275)
(0.275)
(0.274)
(0.275)
(0.274)
(0.275)
owner_chg1
-0.225***
-0.224***
-0.224***
-0.225***
-0.224***
-0.224***
(0.040)
(0.041)
(0.041)
(0.041)
(0.041)
(0.041)
owner_chg2
-0.0713***
-0.0708***
-0.0724***
-0.0716***
-0.0725***
-0.0714***
(0.017)
(0.017)
(0.017)
(0.017)
(0.017)
(0.017)
constant
7.886***
7.930***
7.885***
7.859***
7.860***
7.880***
(0.770)
(0.771)
(0.773)
(0.771)
(0.773)
(0.771)
Month dummy
Controlled
Controlled
Controlled
Controlled
Controlled
Controlled
Year dummy
Controlled
Controlled
Controlled
Controlled
Controlled
Controlled
Observations
25911
25911
25911
25911
25911
25911
Hotels
404
404
404
404
404
404
0.233
0.233
0.234
0.234
0.234
0.233
1.363
1.369
1.361
1.359
1.352
1.364
Within
R_square
0.379
0.379
0.377
0.378
0.377
0.379
Between
R_square
0.248
0.247
0.245
0.248
0.242
0.249
Overall
R_square
0.194
0.194
0.192
0.193
0.188
0.195
(Note: *** indicates significance at 0.01, ** indicates significance at 0.05, * indicates significance at 0.1. Robust standard errors are presented in
parentheses. Estimates of month and year dummies are not presented for brevity.)
Table 5. Estimation results of fixed effect models with hotel-related crime measures
Model 7
Model 8
Model 9
crime_hotel
0.000792
(0.002)
crime_hotel1
0.00338
(0.003)
crime_hotel2
-0.000218
(0.002)
lnneighbors
-0.0277
-0.0276
-0.0273
(0.030)
(0.030)
(0.030)
lnrooms
-1.026***
-1.026***
-1.026***
(0.184)
(0.184)
(0.184)
lnHHI
-0.168*
-0.168*
-0.168*
(0.095)
(0.095)
(0.095)
operation=2
0.0148
0.0147
0.0149
(0.048)
(0.048)
(0.048)
operation=3
-0.107**
-0.108**
-0.107**
(0.054)
(0.054)
(0.054)
class=2
-0.0622
-0.0622
-0.0621
(0.060)
(0.060)
(0.060)
class=3
0.0963
0.0963
0.0964
(0.229)
(0.229)
(0.229)
class=4
0.00303
0.00301
0.00338
(0.273)
(0.273)
(0.273)
class=5
0.0260
0.0261
0.0266
(0.270)
(0.270)
(0.270)
class=6
0.158
0.158
0.159
(0.273)
(0.273)
(0.273)
owner_chg1
-0.224***
-0.224***
-0.224***
(0.041)
(0.041)
(0.041)
owner_chg2
-0.0726***
-0.0726***
-0.0725***
(0.017)
(0.017)
(0.017)
constant
7.847***
7.847***
7.846***
(0.773)
(0.773)
(0.773)
Month dummy
Controlled
Controlled
Controlled
Year dummy
Controlled
Controlled
Controlled
Observations
25911
25911
25911
Hotels
404
404
404
0.234
0.234
0.234
1.354
1.353
1.354
Within R_square
0.376
0.376
0.376
Between R_square
0.245
0.245
0.245
Overall R_square
0.191
0.191
0.191
(Note: *** indicates significance at 0.01, ** indicates significance at 0.05, * indicates significance at 0.1. Robust standard errors are presented in
parentheses. Estimates of month and year dummies are not presented for brevity.)
Table 6. Robustness check results with random effect models
Estimation
method
Crime measure
Coefficient
Observat
ions
Hotels
Overall
R_square
Model 10
RE
crime
-0.00150***
25911
404
0.576
Model 11
RE
crime
-0.00245***
25911
404
0.571
RE
crime_square
0.00000480**
Model 12
RE
crime_night
-0.00181***
25911
404
0.582
Model 13
RE
crime_day
-0.00158***
25911
404
0.582
Model 14
RE
crime_violent
-0.00557***
25911
404
0.591
Model 15
RE
crime_property
-0.00142***
25911
404
0.577
Model 16
RE
crime_hotel
0.00136
25911
404
0.590
Model 17
RE
crime_hotel1
0.00468
25911
404
0.590
Model 18
RE
crime_hotel2
0.000134
25911
404
0.589
Model 19
FE
crime†
-0.00205***
25911
404
0.193
Model 20
FE
crime†
-0.00445***
25911
404
0.193
FE
crime_square†
0.0000129***
Model 21
FE
crime_night†
-0.00281***
25911
404
0.191
Model 22
FE
crime_day†
-0.00214***
25911
404
0.193
Model 23
FE
crime_violent†
-0.00811**
25911
404
0.186
Model 24
FE
crime_property†
-0.00216***
25911
404
0.195
Model 25
FE
crime_hotel†
0.000000799
25911
404
0.191
Model 26
FE
crime_hotel1†
0.00341
25911
404
0.191
Model 27
FE
crime_hotel2†
-0.00132
25911
404
0.191
(Note: *** indicates significance at 0.01, ** indicates significance at 0.05, * indicates significance at 0.1. Robust standard errors are used. †
indicates an alternative crime measure construction using the average of the last three months. RE indicates random effect estimation, and FE
indicates fixed effect estimation. Only estimates of crime measures are presented for brevity.)
Figure 1. Map of Houston hotels and crime density
Data source: Houston Police Department and Smith Travel Research
48
1
2
Figure 2. Marginal effect of crime over different levels of crime density
3
4
49
5
6
Figure 3. Estimates of crime measures in different models
7
8
