Spatial Effect of Crime on Peer-to-Peer Lodging Performance

Abstract

Safety is a major factor impacting consumers’ participation in peer-to-peer (P2P) economies. Using spatial econometric models, this study examined crime effects on the performance (RevPAR) of P2P lodgings at three spatial ranges: property, community, and destination level. The performance of P2P lodgings is negatively associated with crime densities, while the degree of the association varies by crime types and room types. Crime can “spill over” to the neighborhood and have the strongest impact at the community level, followed by the destination level and the property level. The study provides a way to understand tourism risks using criminology theories and the concept of social uncertainty. Empirically, the study provides implications to the governance of community-based lodging business. We suggest that the effect of crime on P2P lodging performance was more conditioned by the safety environment in its neighborhood and the whole destination, rather than individual business operations.

Keywords

Airbnb RevPAR crime pattern risk assessment spatial econometrics social uncertainty

Introduction

Peer-to-peer (P2P) lodging, as a form of the sharing economy (Reinhold and Dolnicar 2018), has become an indispensable part of the tourism industry in major destinations. Operated through online platforms, the P2P lodging business has offered more than 5 million listings around 191 countries (Airbnb 2019a). The amount of P2P lodging listings has exceeded the global total of listings from the top five hotel brands (Gerdeman 2018). However, along with the unprecedented growth, there have been concerns with reports of crime, as well as safety on P2P properties (Xu, Pennington-Gray, and Kim 2019).

Safety is one of the most important factors to impact consumers’ participation in P2P economies (Bellotti et al. 2015). Guttentag et al. (2018) suggested that security is more important in Airbnb than in budget hotel. Customers’ perceived risk can negatively impact their perceived value and repurchase intention on P2P services (Liang, Choi, and Joppe 2018). Lodging crime can stigmatize the image of a property (Huang, Kwag, and Streib 1998) and deter customers’ willingness to visit. Most P2P lodging operates in residential neighborhoods, where fewer security measures are in place compared with regular hotels. When situated in communities with higher crime, the P2P business will experience higher level of social uncertainty (Yamagishi and Yamagishi 1994). This uncertainty results from a lack of information which may result in people’s unwillingness to interact (Macy and Skvoretz 1998; Farrall, Gray, and Jackson 2007). There is a great potential that crime can impact the guests’ choice of P2P lodging and consequently affect the lodging performance, which is commonly understood as occupancy rate or revenue. However, whether in industry or academia, the topic of crime and P2P lodging performance remains underexplored. Understanding the relationship between crime and P2P lodging performance is critical: on one side, it can provide the sharing economy business to inspect the security issue and check the room for improvement and, on the other, it contributes to the body of knowledge in tourism- and hospitality-related risk management.

Recent studies have begun to identify geography as an influential factor in the P2P economy (Thebault-Spieker, Terveen, and Hecht 2017). The characteristics of crime and P2P lodging can vary by location. Crime incidents typically exhibit distinctive spatial patterns based on pertinent geographical features (Groff and McEwen 2006). Therefore, crime and P2P properties can be considered and processed as spatial data. The use of spatial data in aspatial linear model may lead to biased estimation results (Anselin 2013; Kim and Nicholls 2016). When variables exhibit spatial dependency, spatially explicit models should be used to address geographic location and potential spatial effects. As stated by Gilbert and Chakraborty (2011, p. 274), “the analysis of spatial data requires specialized techniques that are different from those used to analyze non-spatial data.” The spatial effects of this study include direct effect of crime from each observation point and indirect effect from beyond the direct observations. The indirect effect was further divided into local indirect effect of crime from the adjacent neighborhood, and global indirect effect of crime from the whole study area.

The purpose of this study was to examine the effects of crime on P2P lodging performance at different spatial range. To achieve this purpose, spatial econometric models were employed to test the spillover effects of crime on different types of P2P lodging. The findings of this study help us to understand the spatial effects of crime on P2P lodging performance, and provide insightful suggestions for effective management of tourism destinations to ensure the safety of both travelers and hosting communities.

Our study offers several major contributions to tourism risk management literature and the understanding of the relationship between crime and P2P lodging specifically. First, the study proposed a conceptual model that can address both the supply and demand side of P2P lodging at the threat of crime. Specifically, we hope to initiate the discussion and consider both P2P hosts’ and guests’ rational choices into the analysis. Second, the study advanced the use of criminology theories in the realm of community-based hospitality management. Third, methodologically the study advanced previous literature by using point-based data and adopting a versatile statistical model. As a result, we evidenced significant indirect effects of crime. Lastly, the study results offer an opportunity to discuss not just crime but also other types of risks in community-based tourism and hospitality services at different spatial ranges.

Literature Review

Impact of Crime on Residential Communities

Unlike hotels, which are more prevalent in business districts, P2P lodgings are mostly located in residential areas. When faced with crime, P2P accommodations should prepare for risks that are related not just to the business but also to the community (Xu, Pennington-Gray, and Kim 2019). In the media, guests and neighbors often express concern about the security of family-based P2P lodging businesses (Lieber 2015). P2P lodging often has lower security standards than hotels; also, it often leads to higher population mobility compared with non-commercialized communities and, therefore, there is an increase in the numbers of people in locations that often are quieter, with less traffic.

The effect of crime in the P2P business environment often is discussed based on crime type. Owing to different security standards, the types of crime that occur in hotels may be different. Property-related crime is often a major concern for hotels (Zhao and Ho 2006; Mawby and Jones 2007), while both property and violent crime are of greater concern for P2P properties that are located in residential neighborhoods. Crime can affect either the guest or the host of P2P accommodations. Guest drug use and narcotics is another major crime issue that troubles many P2P hosts (Wright and Weber 2017).

Crime can impose significant societal and economic impacts on communities. Ellen, Horn, and Reed (2019) suggests that low crime rates are one of the most powerful attractors to immigrants that are moving to urban neighborhoods. Greenbaum and Tita (2004) found that violent crime can arouse people’s fear of victimization and prevent them from visiting local businesses. To counteract the negative influence of crime on communities, often communities need to invest in extra surveillance and insurance (Burrows et al. 2001).

On the social side, debate exists whether crime can prohibit or stimulate interactions among community members. Some studies suggest that high crime rates increase social uncertainty and weaken residents’ community participation (Takagi et al. 2016; Logan and Molotch 2007; Saegert and Winkel 2004), while some believe crime can act as a catalyst for the collaboration among neighbors and thus build solidarity within communities. For example, Taylor (1996) examined the impact of crime on 66 neighborhoods in Baltimore and found that communities with higher crime exhibited higher residential involvement. Similarly, Oh and Kim (2009) found that fear of crime can raise social cohesion and trust among older adults.

The discussions about crime and communities are mostly centered on relationships between the impact of crime and social capital. It is found that crime can weaken or strengthen mutual trust as well as social ties within a community. Reversed causation also can exist, whereby social cohesion can reduce crime (Buonanno, Montolio, and Vanin 2009; Bellair 1997). The communities’ resilience capacity toward crime follows a spatial pattern that resembles distance decay (Nekola and White 1999). Namely, greater distance between community properties can weaken the communities’ overall coping power against crime. Hipp and Perrin (2009) studied the social ties in urban communities and found that geographical and social propinquity is critical when developing social capital. They also suggested that greater physical distance between properties, shorter length of residence in the neighborhood, and greater difference in home value and wealth all contribute to weaker social ties.

Risk Assessment, Social Uncertainty, and Rational Choice of P2P Participants

P2P business works differently from traditional Business-to-Customer (B2C) business. On the supply side, hosts can choose how many days and what days they want to open their house for business, as opposed to the all-year-round operated hotels. On the demand side, guests will take more risks when judging whether a host will keep the promise and make qualified space available for them. The concerns from both sides can impact hosts’ and guests’ rational choices. Rational choice theory believes that individuals express their preferences by acting within specific constraints and on the basis of the information they have about the environment (Scott 2000, pp. 127–28). P2P participants’ concerns further impact the deals that could be made so as to influence the overall performance of the business.

P2P business is a service where hosts need to provide guidance and accommodation services to guests before and during their staying. At times, this can also happen in the same space that the host is occupying (shared accommodations). Therefore, compared to products, P2P lodging businesses need to operate in a service realm and consider risks that may occur during the service transaction. Both hosts and guests are exposed to risks (Lauterbach et al. 2009); thus, risk assessment must be a mutual process. Risk assessment is also an ongoing, transactional process where guests assess the risk of the accommodation itself as well as the neighborhood and hosts assess the risk associated with the guest.

In many online trading platforms, the hosts are entitled to select their customers by accepting or declining the booking requests. Basic mutual assessment is usually made based on profile pictures, basic information, prior email exchanges, and online reviews (Dolnicar 2018). However, despite this risk mitigation, ongoing reports show evidence that crime are increasing on/by guests in P2P lodgings. Crime is considered to be an enduring possibility that can derail P2P lodging growth (Guttentag 2015).

Hosts’ concerns

Dolnicar (2018) suggests that when hosts show higher attachment to their houses, they become more cautious about potential damage to their properties. Major P2P lodging trading platforms allow hosts to decline booking requests, which empowers the host and controls risks. The hosts’ assessment depends on the guests’ profiles. For guests who leave less information on their profiles or those who intentionally conceal their purpose for staying (e.g., narcotics, sex trafficking, parties), it is extremely hard for hosts to evaluate the risks.

Another way to assess risk is based on the inspection of the neighborhood environment. Places with high crime rate can impair the hosts’ confidence in neighborhood security thus suspending their willingness to share their home with strangers. Such phenomenon result from social uncertainty. Social uncertainty appears when people are unable to collect enough information during their interaction with strangers (Yamagishi and Yamagishi 1994). When high social uncertainty occurs, people will decide whether to interact with strangers based on trust. It is believed that people tend to restrain interactions with others when trustworthiness is uncertain (Takagi et al. 2016). Crime rates are important indicators of social uncertainty at the community level. When lodging properties are situated in areas with higher crime rates, hosts tend to believe that they are exposed to a higher level of crime in general, despite how secure their individual properties are. High crime rates can also increase regional social disorder, which deepens residents’ sense of isolation and fosters social withdrawal (Skogan 1992). As a result, hosts are more prone to close their business because of risk.

Despite the urgent need to measure the crime impact on tourists and the host communities, there has been little evidence showing how different types of criminal offenses can impact people’s choices. Akin to the comments from Dolnicar (2019) in a review of research into online P2P lodging, we found that the current literature focuses mostly on guests’ perceived safety, while actual safety is still relatively underexplored. Our study used objective spatial data to identify the real-life crime risks that P2P accommodations are faced with.

Travelers’ Decisions

Crime can impact tourists through social uncertainty, and thus directly affect travelers’ decisions, which in turn can impact lodging performance (Dimanche and Lepetic 1999; Xie, Zhang, and Zhang 2014). Tourists are easy targets of crime (Brunt, Mawby, and Hambly 2000; Albuquerque and McElroy 1999). Ryan (1993) classified the relationship between tourism and crime into five categories, from which the tourists are incidental victims to specific targets of criminals. His study concluded that since many holidaymakers are sensitive to their traveling safety, high crime rates can be detrimental to destinations. In fact, tourism-dependent destinations are often considered to have higher crime rates (McPheters and Stronge 1974; Fujii, Mak, and Nishimura 1978; Mawby 2014; Walmsley, Boskovic, and Pigram 1983; Albuquerque and McElroy 1999). Pizam and Mansfeld (1996) found that international tourists’ arrivals decline by 10% because of criminal activities.

In addition, risk perceptions influence travelers’ accommodation choices. The most common reaction to the fear of crime is spatial avoidance (Warr 1993). Similarly, travelers’ fear of crime is salient to their neighborhood conditions (Lynch and Rasmussen 2001). Their perceived safety will impact the probability to choose to visit or stay out of certain areas (Sönmez and Graefe 1996). Also, travelers tend to avoid revisiting places where they feel unsafe during their stay (Dimanche and Lepetic 1999). Thus, it is common to see travelers avoid districts that have higher crime rates.

Fear of crime is determined by the risk of victimization and the severity of potential offense (Warr and Stafford 1983). For people who travel to the same destination, the average risk of victimization is constant. Even so, travelers’ perceptions of the severity of crime may vary with individuals and type of offense. Several studies suggested that tourists are more likely to become the target of property-related crime (Biagi and Detotto 2014; Albuquerque and McElroy 1999; Tran and Bridges 2009).

Spatial Patterns and Effects of Crime

Criminal activities are highly spatially dependent. Criminologists believe crime rate relates to space and social environment (Brantingham and Brantingham 2013). According to crime pattern theory, crime usually happen at places where the activity path of criminals and victims intersect (Brantingham and Brantingham 1993). The routine activity theory developed by Cohen and Felson (1979) suggests that offenders and victims converge in space and time. According to this assumption, crime cluster in areas that are close to the residence of both criminal and victims. Also, based on offenders’ mobility patterns, crime usually occur within a smaller neighborhood close to the residence of offenders. This is because criminals are less likely to travel to unfamiliar places where they may take increased risks and efforts to commit crime (Brantingham and Brantingham 2016). Groff and McEwen (2006) provided data from Washington, DC, as a support. They analyze criminals’ travel distances and suggested that most offenders resided within a two-mile radius of the areas where they commit crime. Besides the offenders’ traveling pattern, different socioeconomic environments can also predict the targets and types of crime. For example, Harper, Khey, and Nolan (2013) explored the different spatial patterns of robbery committed on tourists versus residents and suggested that robbery toward tourists mostly occurs around scenic spots, while aggravated robberies were mostly committed in residential areas with less tourists and police force.

Crimes exhibit strong spatial lag effect within and between neighborhoods. Scholars have demonstrated how the structural conditions in one community can affect crime in the adjacent community. Johnson and Kane (2018) found that community locations can affect the violence level, and such effect exhibits a diminishing trend from the center of the community. Another situation was suggested by Jargowsky and Bane (1991), where a disadvantaged neighborhood isolated by violence becomes a “Deserts of Disadvantage” (Johnson and Kane 2018). This occurs when the neighboring communities are in similar conditions and a greater area will be impacted by this smaller area. Such spillover effects can negatively reduce community viability.

In the sharing economy literature, geographic locations and local spatial features were found to have significant effects on plenty of phenomena (Thebault-Spieker, Terveen, and Hecht 2017). For example, Xu, Pennington-Gray, and Kim (2019) found that in rural North Florida, crime is positively associated with shared rooms rented on Airbnb, while in South Florida, such associations were not significant. Spatial agglomeration was recognized to have associations with positive economic externalities (Hua and Yang 2017). As a result, colocated lodging properties tended to perform better than those in isolation (Yang, Wong, and Wang 2012).

In this study, we hypothesize that the effect of crime on P2P lodging performance depends on space, and that criminal activities will have spillover effects on surrounding communities and subsequently their P2P lodging industry.

Influential Factors of P2P Lodging Performance

P2P lodging’s performance is influenced by multiple factors. Objective factors such as location and price are commonly mentioned attractors to consumers (Hamari, Sjöklint, and Ukkonen 2016; Ert, Fleischer, and Magen 2016). Fradkin (2015) analyzed customers’ searching data for Airbnb and found most customers first filter listings by features such as location and price, and then evaluate subjective features such as ratings and reviews. Subjective factors such as users’ profile are becoming more important in the P2P lodging networks (Ert, Fleischer, and Magen 2016; Kakar et al. 2016).

Crime’s effects on P2P lodging are not independent. Other factors may also affect P2P lodging performance. Failure to consider these factors will result in omitted variable biases (OVB), thus leading to inaccurate coefficient estimates (Clarke 2005; Chamberlain 1979; Pope 2008). In this study, we considered a set of covariates that may influence P2P lodging performance, including property operational status and community socioeconomic features.

One feature is room type; this includes the types of rooms for rental, including shared room, private room, and entire home. A shared room is a room where guests sleep in a bedroom or a common area that could be shared with others. A private room is a room where guests have their own private room for sleeping, while other areas could be shared. An entire home means guests have the whole place to themselves (Airbnb 2019d). Different room types indicate different capacities for the property. In hotel studies, hotel size is considered to have negative association with the operating performance (Sainaghi 2011). In P2P lodging, customers decide first how many guests are included in one booking. Entire homes are suitable for families or small travel groups, while shared rooms are preferred by solo travelers with limited budgets. To address the importance of this indicator, we designed individual models for each room type rather than directly coding them into control variables.

In order to improve credibility, some host choose to display their own pictures and photos of the accommodation. Hosts’ personal photos may enhance their trustworthiness and impose significant impact on guests’ decision making (Ert, Fleischer, and Magen 2016). However, it may also increase the chance of being discriminated from the guests (Edelman and Luca 2014). Posting accommodation photos also has more positive impacts on decision making. More accommodation photos translate into price premiums (Teubner et al. 2016).

Hosts certified by the platform with good records can often attract more customers. Airbnb releases a “Superhost” badge to hosts who have hosted at least 12 months and 10 trips with minimum misbehaviors and high ratings, reviews, and response rates (Airbnb 2019b). Superhosts earn up to 22% more than average hosts (Airbnb 2019c). Signifiers like Superhost serves as an icon of quality service, especially for lower-priced listings (Wang and Nicolau 2017). It may also drive up the price of P2P lodgings in cities (Hrobath, Leisch, and Dolnicar 2018).

Housing values often dictate the room price. P2P lodgings’ lower room price has been one of the major attractors to travelers in city centers. Guttentag (2015) found that Airbnb’s distinct appeal lies in cost-saving, while conflicted evidence was found in the United States (Lane and Woodworth 2016). In this study, we use housing value to represent the integrated condition of P2P lodgings in terms of location and property value.

P2P lodging coexists with other types of accommodation business. They compete in price, revenue, and locations (Zervas, Proserpio, and Byers 2017; Guttentag 2015). The agglomeration of different lodging types reveal the city’s tourist hot spots (Arbel and Pizam 1977), and P2P lodging tends to be more heavily clustered in these hot spots (Gutiérrez et al. 2017). P2P lodging shares the market with other types of lodging and often have established a competitive but symbiotic relationship within the locations.

The percentage of owner-occupied houses in a census block group is a good indicator of P2P lodging rates. The regional rental rates are usually lower when more houses are occupied by owners. Because the main purpose for owners to buy a house is self-use. Rarely can we see owners rent their houses to long-term tenants, only some would list their unused properties to short-term market while away on vacation/business (Barron, Kung, and Proserpio 2018). Contrarily, renter-occupied houses are prone to higher rental rates because of renters’ higher mobility. Subleasing and housing arbitrage business are common in renter-occupied houses.

People’s satisfaction with their current neighborhood conditions are related to their length of residence (Rent and Rent 1978). The longer people live in a place, the stronger attachment they developed and less mobility they may have. A stable residence status may lead to a higher owner occupancy rate, thus negatively influencing the short-term rental market.

Thebault-Spieker, Terveen, and Hecht (2017) suggest that the sharing economy is more effective in densely populated areas than sparsely populated areas. Affluent communities are less likely to have rentals than lower-income urban communities. However, increased population can result in lower levels of traditional short-term rentals, because increased demand can make fewer units available (Deng, Gabriel, and Nothaft 2003). For shared lodging properties, although hosts can rent their houses while residing in it, the space and time for rental would still be limited, and thus impact the P2P lodging performance.

Theoretical Framework

We based our research assumption on the crime pattern theory and routine activity theory. The two theories provided the support for our research design of spatial analysis. According to the crime pattern theory and routine activity theory, criminal activity exhibits unique spatial patterns and can impose spatially varying effects on the community that is hosting P2P lodging business. In tourism destinations, such variation mostly depends on the level of commercialization within that community. The effect of crime on a community could be passed down to the P2P lodging host, who is also a resident of that community. The type of crime that P2P lodging business is faced with could be different from hotel crimes; therefore, it is necessary to explore the crime effects by crime types.

The rational choice theory and the concept of social uncertainty supports our assumptions of the choices of individual P2P lodging participants (the hosts and guests). Accumulation of crime over time in one place can arouse social uncertainty at a greater scale in a destination. Such uncertainty can influence the risk assessment of P2P lodging hosts, travelers, and residents of the community. The crime-based risk assessment dictates the choices of hosts and guests of whether to participate to the P2P lodging business. This study used secondary data for analysis. Although the individual choices of the hosts and guests were not directly measured, the concept of social uncertainty gave us a reasonable direction to examine the internal logic of P2P participants’ rational choice.

We synthesize the literature into a framework as Figure 1. It describes our assumptions and speculations of the relationships among critical factors (crime, community, spatial effect of crime, social uncertainty, choices from host and guest, room type) that contribute to P2P lodging performance. It is worth noting that this study was not able to measure all the components in the framework; however, the framework provided a roadmap to understand the safety issue in the P2P economy and hopefully will provide directions for future studies’ focus on lodging and travel industry risk management.

Figure 1.

Framework of research assumptions.

Based on the framework in Figure 1 and pertinent discussions, we proposed four research questions for the study:

Research question 1: What is the total effect of crime on Airbnb performance?

Research question 2: What is the effect of different crime types on Airbnb performance?

Research question 3: How do the effects of crime vary by Airbnb room types?

Research question 4: Do crimes have a spatial spillover effect on Airbnb performance?

Method

Study Area

The city of Orlando is located in central Florida, with a population of 280,257 and an area of 102.40 square miles (United States Census Bureau 2017, 2010). Orlando has many world-class entertainment landmarks and tourism attractions, including Walt Disney World Resort and Universal Studio Resort. In 2017, Orlando recorded more than 72 million visitors. For the third year in a row, it has been titled as the “Most Visited Tourist Destination in the U.S.” (Stephens and Pennington 2017). Orlando also serves a role model in the US hospitality industry. It not only has a wide range of accommodation choices but also keeps the fastest pace of development in the United States (Santana 2018). The growth of revenue per available room (RevPAR) in Orlando was three times of the average for the top 25 markets (Stephens and Pennington 2017).

Orlando hosts large numbers of tourists every year. P2P lodging properties provide a considerable proportion of supply in addition to hotels. Along with the influx of tourists and fast-developed destination, Orlando is experiencing severe crime problems. According to the FBI-reported crime data in 2016, Orlando has a total crime rate of 61 per 1,000 population, which is higher than 96% of all the other communities in the State of Florida (FBI 2016). The aggregation of tourism and P2P lodgings and crime cases made Orlando an appropriate site for analysis.

Variables and Data Collection

The unit of analysis in this study is individual P2P lodging property. We analyzed 1,657 P2P lodging property listings within the Orlando city boundary in the year 2016. We screened out inactive listings, which were defined as listings that have no reservation record nor listed opening days during the past 12 months. The final data set contains 991 active listings (Figure 2A, left side).

Figure 2.

(A) Airbnb properties in the city of Orlando (left). (B) Crime incidents and crime density in the city of Orlando (right).

RevPAR has been widely used as a lodging performance metric in the industry (Ismail, Dalbor, and Mills 2002; Gallagher and Mansour 2000; Jacobs 1997). It is calculated by multiplying the room’s average daily rate by the occupancy rate. The RevPAR serves as the dependent variable.

For the independent variable, we used the crime density estimation based on the location of all crime cases. Among all types of crime, violent crime (including murder, rape, armed robbery, assault) in Orlando ranks the highest in the nation. The chance of being a victim of violent crime here is one in 131 (see Figure 2B, right side). Compared with a real unit-based crime rate, point-based crime density can address spatial externality by capturing the continuous variations of crime in the study area, which better serves our purpose to explore the crime effect at the neighborhood scale. To measure the crime density, we first calculated the kernel density of crime in the whole study area. When estimating the kernel density, the choice of bandwidth can greatly affect the resulting density surface (Silverman 2018). Bandwidth is the radius in which to search the crime incidents for the calculation of crime density; it controls the degree of smoothing applied to the data. Larger bandwidth is able to create a smoother surface but may also veil local peaks and ebbs (Schuurman, Berube, and Crooks 2010). Overly small bandwidth may fail to capture the bigger trends of data in the surrounding neighborhood. In our case, the southeast Orlando exhibit less than the average level of crime because of the existence of the airport. To obtain an optimal bandwidth with minimized mean square error, we adopted Silverman’s rule (Silverman 2018) for it is robust to spatial outliers. The function of bandwidth calculation is as follows:

B a n d w i d t h = 0.9 * m i n (S D, \sqrt{\frac{1}{\ln 2}} * D m) *^{n^{- 0.2}},

where SD is the standard distance from all crime points to the mean center, Dm is the median distance, and n is the number of crime cases.

To avoid the omitted variables bias (Clarke 2005), we introduced a set of control variables that may affect the lodging performance (see Table 1). On one hand, features of the P2P lodging operation status, such as price (Fradkin 2015; Ert, Fleischer, and Magen 2016), number of photos (Guttentag et al. 2018; Ert, Fleischer, and Magen 2016), and Superhost badge (Wang and Nicolau 2017), were proved to have an influence on the P2P lodging performance. On the other hand, local socioeconomic features also have a potential impact on the lodging performance. In this study, we included the number of photos, Superhost status, hotel density, median home value, owner-occupied house ratio, length of residence, and population density as control variables.

Table 1.

Data Source and Specification.

Variables	Abbreviations	Operational Definition	Source (Year)	References
Revenue per available room (DV)	InRevPAR (all, entire, private, shared)	The log of RevPAR of the last twelve months for all room types, entire home, private room and shared room.	AirDNA (2016)	Xu, Pennington-Gray, and Kim (2019)
Kernel density of crimes within the city of Orlando limit. (IV)	Total crime, Theft, Burglary, Assault, Narcotics	Kernel density estimation of different types of crimes, extracted to each Airbnb property	OPD (2016)	Xu, Pennington-Gray, and Kim (2019)
Number of photos (CV)	PHOTO	Number of photos Airbnb hosts have on their profile	AirDNA (2016)	Edelman and Luca (2014); Ert, Fleischer, and Magen (2016)
Superhost badge (CV)	SUPERHOST	Whether the host of the lodging is designated as Superhost by Airbnb (Dummy variable, 1=yes, 0=No)	AirDNA (2016)	Hrobath, Leisch, and Dolnicar (2018); Wang and Nicolau (2017)
Lodging density (CV)	LODGDENS	Kernel density estimation of lodging counts. Lodgings include Bed and Breakfast, resort condominium, resort dwelling, hotel, motel, and apartment complex	FDGL (2017)	Gutiérrez et al. (2017); Zervas, Proserpio, and Byers (2017)
Housing value (CV)	HVAL	Median values of owner-occupied housing units.	FDGL (2016)	Guttentag (2015); Lane and Woodworth (2016)
Owner occupied house (CV)	OWNER	Percentage of owner-occupied house units by census block groups	FDGL (2016)	Barron, Kung, and Proserpio (2018)
Length of residence (CV)	YEAR	Median of the length of years since the current householder moved into unit by census block groups	FDGL (2016)	Rent and Rent (1978)
Population density (CV)	POPDENS	Number of people per square mile by census block groups	FDGL (2016)	Deng, Gabriel, and Nothaft (2003); Thebault-Spieker, Terveen, and Hecht (2017)

Note: AirDNA = short-term rental data company; FDGL = Federal Geographic Data Library; OPD = Orlando police department.

Data of P2P lodging were acquired from AirDNA, a database company for the short-term vacation rental. It includes the locations of Airbnb properties and their operational status in year 2016. The crime data were obtained from the City of Orlando Open Data (OPD, https://data.cityoforlando.net/), and each point represents a crime case reported to the local police department in year 2016 (Figure 2A, left side). The value of crime densities were extracted to the location of each lodging property. The hotel data and census data were obtained from Florida Geographic Data Library (FGDL, https://www.fgdl.org/). All the census data were based on the 168 census block groups in the City of Orlando.

Variables sources and references were specified in Table 1.

Model Building and Data Analysis

Based on the research questions, we developed 4 sets of spatial econometric models with heteroscedasticity assumptions (see Table 2). Moran’s I was calculated to test the spatial dependency among all variables. On observing the presence of positive spatial autocorrelation in the dependent and independent variables, we compared the results from three different spatial econometric models to unveil the data generating process that has a higher probability in explaining the relationships among our explanatory variables and dependent variables (LeSage and Pace 2009).

Table 2.

Model Building.

	Dependent Variables	Independent Variables	Control Variables
Model 1-1	InRevPAR(all)	TotalCrime	PHOTO SUPERHOST HVAL LODGDENS OWNER YEAR POPDENS
Model 1-2		Assault
Model 1-3		Burglary
Model 1-4		Narcotics
Model 1-5		Theft
Model 2-1	InRevPAR(entire)	TotalCrime
Model 2-2		Assault
Model 2-3		Burglary
Model 2-4		Narcotics
Model 2-5		Theft
Model 3-1	InRevPAR(private)	TotalCrime
Model 3-2		Assault
Model 3-3		Burglary
Model 3-4		Narcotics
Model 3-5		Theft
Model 4-1	InRevPAR(shared)	TotalCrime
Model 4-2		Assault
Model 4-3		Burglary
Model 4-4		Narcotics
Model 4-5		Theft

Different spatial models, including Spatial Autoregressive Model (SAR), Spatial Autocorrelation Model (SAC), and Spatial Durbin Model (SDM), were tested. The SAR model includes the spatial lag term of dependent variable, SAC model includes both the spatial lag term for dependent variable and the residuals, and SDM considers the spatial lag term for both dependent and independent variables, which enables it to capture the local effects arising from immediate neighbors (Anselin 2003). The SDM model was finally adopted based on the research assumptions and model performances. According to LeSage and Pace (2009), SDM is one of the most versatile models that has considered possible spatial autocorrelations in common situations. SDM elaborates the model for spillover effect by containing spatial lag terms from the explanatory variables as well as the dependent variable; it is most helpful when the omitted spatially dependent variables are in the present (LeSage and Pace 2009). The model is specified as

\begin{array}{l} I n (y) = ρ W y + X β 1 + W X β 2 + ε \\ ε \sim N (0, σ 2), \end{array}

where y represents the Airbnb RevPAR, X represents the explanatory variables including crime level and control variables. Wy indicates the spatial lag of dependent variable, ρ indicates the spatial autocorrelation coefficients, which shows the strength of effects on y from the neighboring y’s. WX indicate the spatial lag of independent variables. W is the spatial weights that defines the neighbors of the observations. ε is the error term of the model.

Defining the spatial weight matrices is a key step in spatial econometric analysis. The matrices (W) define how the study area is connected, and where the spillover effects are disseminated. The spatial regression results can be sensitive to the spatial weighting matrices (Yang and Zhang 2019). Lynch and Rasmussen (2001) emphasized the importance of the weighting method in analyzing the effect of crime on housing prices. According to Groff and McEwen (2006), most offenders reside within a range of 0.81–3.96 miles, with a mean of 2.32 miles from where they commit crime. Thus, we infer that the impact of crime stays within neighborhoods with a less than two-mile diameter. After examining the average distance among Airbnb properties and testing different weighting methods. We determined that the “neighboring area” of an Airbnb property as the extent of 10 nearest neighbors. Therefore, we form the weight matrix W1, W2 based on the 10–nearest neighbors assumption. Namely, we consider a P2P lodging to receive indirect impact of crime from its 10 nearest P2P lodgings.

Spatial models provide a broader range of data information. When conducting correlational studies, it is not sufficient to just analyze the relationship between variables based on each observation. We also need to consider the impact from the adjacent or further observations. According to LeSage and Pace (2009), changing the explanatory variable in one location will not just affect the dependent variable in this location, it may also affect the dependent variables in all locations. As Tobler’s first law of geography states, near things are more related than distant things (Tobler 1970). Therefore, the crime at a property’s immediate neighbor may have greater effect than crime from more distant neighbors. In our model, three dimensions of crime can affect the Airbnb performance (e.g., RevPAR): (1) crime density at the direct observation points (direct effect), (2) crime from the immediate neighbor (local indirect effect), (3) and crime in the whole study area (global indirect effect). To better illustrate the assumptions of the Spatial Durbin Model we adopted, we provide the diagram in Figure 3.

Figure 3.

Direct and indirect effects.

Each point in Figure 3 represents an Airbnb property in the study area. The direct crime effect on the performance of Airbnb A comes from the crime density at point A. The indirect impact on point A comes from the crime densities at all other points. This is also known as the spatial lag effect. Moreover, the indirect impact was divided into the effect from the immediate neighborhood (the 10 nearest neighbors within each circle) and the global indirect effect from more distant observations outside each circle. Finally, the total effect is the sum of direct and indirect effects.

To answer question 3, we split the P2P lodgings into three different room types: shared room (InRevPar_shared), private room (InRevPar_private), and entire home (InRevPar_entire). The sum of all P2P lodgings was also included as a separate dependent variable (InRevPar_all).

Results

Table 3 presents the estimated coefficients of the eight models. “D” represents the direct effect, “I” represents the indirect effect, and “W” represents the effect of crime from the immediate neighborhood.

Table 3.

Estimation Results of Crime Effects on P2P Lodging Performance.

	InRevPar_all					InRevPar_entire					InRevPar_private					InRevPar_shared
Independent variables	Model 1-1	Model 1-2	Model 1-3	Model 1-4	Model 1-5	Model 2-1	Model 2-2	Model 2-3	Model 2-4	Model 2-5	Model 3-1	Model 3-2	Model 3-3	Model 3-4	Model 3-5	Model 4-1	Model 4-2	Model 4-3	Model 4-4	Model 4-5
Total crime: T	–0.001***					–0.001***					–0.001**					–0.007
Total crime: D	0.000					0.001					0.000					–0.001***
Total crime: I	–0.001					–0.001*					–0.001*					–0.005
W-Total crime	–0.001***					–0.001*					–0.001*					–0.007***
Assault: T		–0.001***					–0.002**					–0.002***					–0.014
Assault: D		0.005*					0.005*					0.003					–0.004*
Assault: I		–0.007**					–0.007**					–0.005**					–0.009
W-Assault		–0.007***					–0.007**					–0.005**					–0.007
Burglary: T			–0.0003					–0.001					–0.009**					0.035
Burglary: D			0.003					0.016					–0.003					0.009
Burglary: I			–0.003					–0.018					–0.005					0.026
W-Burglary			–0.004***					–0.018					–0.004					0.024
Narcotics: T				–0.003***					–0.005***					–0.003**					–0.004
Narcotics: D				0.001					0.007*					0.000					–0.005
Narcotics: I				–0.003					–0.012**					–0.003					0.001
W-Narcotics				–0.003***					–0.012**					–0.003					0.001
Theft: T					–0.001***					–0.001**					–0.001**					0.002
Theft: D					0.001					0.001					0.001					0.001
Theft: I					–0.002*					–0.002*					–0.002*					0.001
W-Theft					–0.002***					–0.002**					–0.002*					0.001
PHOTO	0.016*	0.015*	0.021**	0.017**	0.017**	0.010	0.009	0.017	0.011	0.010	0.002	0.002	0.002	0.002	0.002	0.013	0.016	0.014	0.014**	0.013**
SUPERHOST	0.071	0.005	–0.03	0.017	0.087	–0.19	–0.20	–0.27	–0.15	–0.18	0.210***	0.200***	0.194***	0.192***	0.210***	0.105**	–0.01	0.098	0.035	0.042
LODGDENS	–0.0002	–0.00	–0.0004	–0.0002	–0.0001	<0.001	0.003	–0.01**	–0.002	<0.001	0.002	0.003	<0.001	0.001	0.001	0.021	–0.004	–0.005	–0.009	0.007
HVAL	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	0.026	0.019	0.034	0.028	0.026	0.036	0.015	0.043	0.025	0.032
OWNER	–0.695**	–0.58*	–0.31	–0.59	–0.72**	–0.68	–0.56	–0.51	–0.63	–0.74	–0.14	–0.15	–0.10	–0.13	–0.15	2.158*	2.257	2.646	2.104	2.219
YEAR	0.029	0.030	0.037	0.033	0.028	0.009	0.012	0.013	0.020	0.007	0.006	0.007	0.003	0.007	0.007	–0.11	–0.13	–0.17	–0.14	–0.13
POPDENS	0.01	0.009	–0.002	0.008	0.009	0.009	0.010	0.006	0.005	0.010	–0.007*	–0.008**	–0.004	–0.007*	–0.007*	0.044	–0.01	0.016	–0.01	0.020
R squared	0.120	0.124	0.112	0.121	0.118	0.135	0.139	0.125	0.140	0.132	0.240	0.244	0.240	0.241	0.240	0.341	0.306	0.292	0.292	0.282
ρ	0.159***	0.151***	0.187***	0.154***	0.174***	0.071	0.067	0.128	0.061	0.092	–0.021	–0.057	–0.011	–0.036	–0.013	–0.310	0.08	0.045	0.143	–0.063
σ ²	3.347	3.336	3.366	3.345	3.351	5.194	5.174	5.235	5.170	5.208	1.523	1.515	1.525	1.522	1.525	0.948	1.057	1.077	1.078	1.085
N	984	984	984	984	984	458	454	454	454	454	459	459	459	459	459	74	74	74	74	74

Note: Asterisks indicate significance at ***0.01, **0.05, and *0.1. D = direct effect; I = indirect effect − total effect; W = indirect effects − effect from immediate neighbors.

For question 1, the four sets of models showed how crime can have different effects on Airbnb performance depending on the crime type and Airbnb home type. In model 1-1, total crime exhibited a significant negative total effect on all Airbnb performance. Every unit of increase in crime density is related with 0.1% of decrease in Airbnb RevPAR. Similar negative effects were found in entire homes and private rooms in model 2-1 and model 3-1. No significant results were found in shared rooms. Therefore, the total negative effect of crime only affects entire homes and private rooms.

For question 2, model 1-2 to 1-5 showed that different crime types exhibited different sizes of effects on Airbnb performance. Except for burglary (model 2-3), all other crime types (assault, narcotics, theft) showed significant negative total effects on the RevPAR of all Airbnbs. But the strength of the effects were different: For every 1 unit of increase in assault/narcotics/theft, the decrease of RevPAR is 0.1%, 0.4%, and 0.1%, respectively. Burglary only had an effect on private rooms. For every unit of increase in burglary, there was a 0.9% decrease in RevPAR in private rooms.

For question 3, we examined the effect of total crime by room type in models 2-1, 3-1, and 4-1. Our question is “How would different crime types affect RevPAR in different room types?” Results were shown in models 2-2 to 2-5, 3-2 to 3-5, and 4-2 to 4-5. The effects of crime varied distinctively by home types. Entire homes and private rooms exhibited the same level of negative associations with assault and theft, while narcotics had a larger effect size on entire homes (0.5%) than private rooms (0.3%). Among all room types, only private rooms showed correlations with burglary (model 3-3). Finally, the RevPAR of shared rooms showed no significant associations with any types of crime at alpha = 0.05 level.

For question 4, we explored the spatial spillover effects of crime based on the direct effect and two dimensions of indirect effects. Overall, the spatial spillover effect existed across all models. The direct crime effects (D) were less significant compared with total effects (T). Only RevPAR for shared rooms showed significant correlations with total crime. When divided by crime types, only the criminal activity assault showed marginal direct effects on RevPAR of all room types. In addition, narcotics exclusively exhibited marginal direct effects on entire homes.

The global indirect effects (I) exhibited more significant results compared with direct effects. The only crime type that shows no significant I was burglary, which means that local P2P lodging RevPAR is free from the impact of burglary in the whole city.

Among all crime types, assault exhibited significant I on the RevPAR of entire homes and private rooms, while narcotics again showed a significant effect exclusively on entire homes. Moreover, theft showed marginal effects on entire homes and private rooms at alpha = 0.1 level. When it came to room types, only shared rooms did not show any associations with indirect global crime effects. This means local shared rooms RevPAR was not affected by the crime density at the destination level.

The local indirect effect (W) refers to a smaller range of spillover crime effects from the P2P lodgings’ immediate neighborhood; it was extracted from the global indirect effect (I). When considering all room and crime types, W explained all the significant negative crime effects on P2P lodging RevPAR (model 1-1). When further divided by crime type, the negative effects were consistently significant but exhibited different effect sizes (model 1-2 to model 1-5). This indicated that at the community level, the strength of crime effects on P2P lodging performance differs by crime types, and follows the following order: assault > burglary > narcotics > theft. When divided by room types, shared rooms exhibited the most significant associations with crime, while private room and entire homes showed marginal significant associations. The unified agreement of W across different room types indicates that the crime effect at the neighborhood level has the most consistent impact on P2P lodging performance.

The ρ values denote the spatial autocorrelations in RevPAR. The results showed that ρ values were positively significant in model 1-1 to 1-5, meaning the P2P lodging performance had positive spatial agglomeration effects. As the effect size of ρ ranges from 0.15 to 0.19, it is suggested that the performance of P2P lodging can be predicted by its neighboring P2P lodging performance by 15%–19%.

To sum up, we found that the performance of P2P lodging is negatively associated with crime densities. But the significance and degree of the association vary depending on crime types and room types. Crime has spatial spillover effects on P2P lodging performance. For entire homes and private rooms, crime at the community and destination level can have significant effects on their performance; for shared rooms, crime at the property and community level is significant.

Discussions

The study firstly showed the effect from crime types on P2P lodging RevPAR. Except for burglary, other types of crime showed significant negative effects on P2P lodging RevPAR. The reason that burglary had an insignificant effect is perhaps that P2P lodging properties are not the major target for burglars, rarely do hosts or guests leave precious valuables inside P2P lodging properties. Most P2P lodging properties need to meet up with the basic security standard upon registration for business, which may prevent burglary. In addition, a lot of properties have invested in security apparatuses and surveillance cameras. These causes may have reduced the chance of burglary incidents in P2P lodging facilities in general.

However, there is a unique significant effect of burglary on private room RevPAR. This may be understood by the way private rooms are operated. In addition to single-bedroom sharing, private room hosts often need to share their living rooms for the convenience of the guest. According to routine activity theory (Felson 1986, 1998), this may turn their properties into an “attainable target” for burglars. At the fear of exposing their valuables to potential burglars, many house owners may not choose to open for lodging business. Furthermore, narcotics are found to have the greatest effect size on P2P lodging performance. Judging from the nature of narcotics and guest drug use phenomenon in P2P lodgings, this association may be explained by the hosts’ concerns (Dolnicar 2018). Hosts of the entire homes have the least control over their properties during guests’ staying; thus, hosts’ concerns about guests’ potential drug activities may prevent them from lending out the houses.

The effect from room types showed that crime density can have a distinctive impact on the RevPAR of private rooms and entire homes. This can also be understood by the routine activity theory (Felson 1986, 1998; Cohen and Felson 1979), which suggests that the occurrence of crime depends on a potential offender, an attainable target, and absence of a guardian. The types of room sharing reflects the means of interaction among hosts and guests. Compared with shared rooms, private rooms and entire homes are both in lack of supervision. At the community level, shared rooms are often located in apartments within central business districts, while the other two types of rooms are more isolated from the crowd. At the property scale, guests of private room and entire home live in a bubble with limited or no interactions with the hosts or other guests, while shared rooms guests have more opportunity to interact with other guests and the host. This microenvironment can create stronger social surveillance against potential criminal activities. It is also noticeable that shared room RevPAR has no associations with any crime types. Although previous studies have showed that shared rooms are the major crime-related room type in terms of spatial distribution (Xu, Pennington-Gray, and Kim 2019), when it comes to P2P lodging performance, shared room RevPAR seems to be free of the impact of crime. This indicates that the choice of shared room owners and users may not be affected by the overall safety of the environment. Given the three different spatial ranges, crime affects the performance of the entire home consistently at all levels. This can be understood from the characteristics of the guests. Shared and private rooms share similar target markets as hostels, which is often preferred by budget travelers. These travelers can have higher risk tolerance in order to give way for their lower lodging budget. Their choice of lodging may not be altered by the location of the property within a higher crime area.

The entire home is a common vacation choice for family or group travelers, the target market consisting of middle to upper–middle income population, who often hold higher risk perceptions during travels (Floyd et al. 2004). When a higher crime rate jeopardizes the safety image of the P2P lodging property (Huang, Kwag, and Streib 1998), these travelers might avoid staying there (Liang, Choi, and Joppe 2018; Bellotti et al. 2015). Our results indicate that the guests of the entire homes can be more crime-sensitive than those using the other two room types. As a result, a higher crime rate may disproportionally affect travelers who choose entire homes. The limited crime effect on shared rooms may result from the complexity of traveler’s motivation in selecting P2P lodgings. While the safety issue did play a major role when travelers are choosing their accommodations (Liang, Choi, and Joppe 2018; Bellotti et al. 2015), the crime rate was an influential factor for guests who prefer shared rooms.

Recognizing the spatial spillover effect in the relationship between crime and P2P lodging RevPAR is one of the major contributions of this study. The study demonstrated a strong indirect effect of crime. The strength of the indirect effect is greater than the direct effect from crime. For the indirect effect, the crime density in the immediate neighborhood (local indirect effect) exhibits stronger effect than at the destination level (global indirect effect). Therefore, we suggest that the effect of crime on P2P lodging performance was not just a result of single business operations but was more conditioned by the safety environment in its neighborhood and the whole destination. The findings of indirect effects further helped us understand the impact of crime on P2P lodging performance from a destination planning perspective. The spatial dependency of crime effects requires higher levels of governance. When a community/destination is experiencing crime as a whole, no single business can avoid this impact. Practically, joint efforts from destination against crime is needed in order to protect the residential-based lodging business. According to the study results, crime can “spill over” to the neighboring communities and generate the strongest impact on P2P lodging performance, and response to crime takes individual as well as collective efforts from the community. From a planning perspective, it is most important to heighten the community-wise awareness toward crime.

The minimum direct effect of crime at the property level may be a compound result of the siting and operational strategies in P2P lodging business. A large proportion of P2P lodgings are clustered around city centers or major tourist attractions (Gutiérrez et al. 2017). These places are also known for breeding high crime rates. However, location is so important for tourists that sometimes it may override the “trivial” disadvantages such as residing in a crime-intense area. In addition, some top “Superhosts” are extremely good at running their business, hosts’ good performance plus their dispersed locations may veil the true correlations between crime and regular P2P lodging RevPAR statistically. Therefore, more empirical studies are needed to draw a valid conclusion. Still, lodgings located in tourism hotspots where there is higher crime density should still heighten their precautions against potential risks. For properties operated as entire homes, we suggest hosts and planners to take the crime rate into consideration when siting for their business.

The findings of spatial effects also lend empirical evidence for further discussions of routine activity theory (Cohen and Felson 2016) and crime pattern theory (Brantingham and Brantingham 2013) in the tourism field. The two theories addressed criminals’ relative travel distance (Groff and McEwen 2006) and suggests that it may be common for the crime effect to spill over to the proximity. Using the spatial Durbin model, this study extended the analysis of crime to a broader geographical context, the destination level. The results of this study provided evidence that the effect of crime can extend further beyond a neighborhood in a tourism destination. This means that criminal activities occurring at any corner of a destination can impact any P2P lodging property performance that operates in that destination. On the other hand, crime tends to have less influence on the P2P lodging business at the smallest scale, the property level, as opposed to the community and destination level. Although crime pattern theory indicates that criminal activities are active within a radius of less than two miles (Groff and McEwen 2006), the impact of criminal activities on tourism business can extend much larger than the offenders’ activity path routine. This result provides new evidence to understand tourism-related crime effects.

The concept of social uncertainty (Yamagishi and Yamagishi 1994) explained why the hosts and guests choose to participate in P2P economy or not. The study proposed crime density to be a source of social uncertainty for P2P lodging business, the results indicated that there is a potential for the host to withhold their properties for rent because of a higher crime rate in their community and the destination. Similar to the demand side, travelers may choose to stay in hotel rather than P2P lodgings when traveling to certain areas because of a concern of safety as a result of a high crime rate.

Conclusion and Future Studies

In this study, we tested three levels of crime effects (property level, neighborhood/community level, destination level) on different types of P2P lodgings properties’ operating performance. The model results showed that the impact of crime on P2P lodging RevPAR depends on three factors: crime types, room types, and relative spatial locations of crime and P2P lodgings. In general, crime has a negative effect on P2P lodging performance. The significant effects of crime are consistent across crime types and room types, which supported our assumption that crime may negatively affect community-based lodging performance.

This study examined the correlation between crime and P2P lodging performance based on the assumption of an interim factor, social uncertainty (Yamagishi and Yamagishi 1994). The framework suggested in Figure 1 indicates a causal chain relationship. However, the existence of social uncertainty was not tested with primary data in this study. Further studies should examine and validate the path of influence. Considering the potential impact of COVID-19 on P2P lodging hosts and guests’ rational choices, the concept of social uncertainty can be further explored in addressing the changes in the P2P lodging market. Second, the census data we used in this study is aggregated based on census block group. When joining the census data to Airbnb observations, same census information may have been allocated to Airbnb within the same block group. This may have caused noise in our point-based data analysis. Future studies can include more point- or distance-based census information into the model. Third, we used RevPAR to measure the P2P lodging performance. RevPAR was calculated based on the opening days and average daily rate. While RevPAR is widely used to evaluate hotel performance, it may not be fully adapted to P2P lodging businesses. Hotels usually operate all year around, but P2P lodging properties can set various block days based on their own interest. Therefore, RevPAR may need further exploration as a tool to measure performance in the P2P lodging sector.

Footnotes

Authors’ Note

Data was collected at the University of Florida when all authors were affiliated with the Department of Tourism, Hospitality, and Event Management.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Yu-Hua Xu

Author Biographies

Yu-Hua Xu, Ph.D. is a Lecturer in the School of Community Resources and Development at Arizona State University. Her research focuses on destination planning, sharing economy, and community resilience. She is dedicated to finding sustainable solutions for destinations and communities in face of various turbulence and crises.

Lori Pennington-Gray, Ph.D. is the Endowed Chair and Director of SmartState Center for Economic Excellence in Tourism and Economic Development at the College of Hospitality, Retail and Sport Management, South Carolina University. Her research has focused on tourism crisis management, with a strong commitment to enhancing the industry and aiding in strategic policy decisions.

Jinwon Kim, Ph.D. is an Assistant Professor in the Department of Tourism, Hospitality, and Event Management at the University of Florida. The overall goal of his research is to identify the role of recreation, park, and tourism in the creation of active, vibrant, healthy, sustainable, and resilient community.

References

Airbnb. 2019a. “About Us.” 2019. https://press.airbnb.com/about-us/.

Airbnb. 2019b. “How Do I Become a Superhost?” https://www.airbnb.com/help/article/829/how-do-i-become-a-superhost?af=&c=.pi0.pk55829210794_259363506416_c_422757515900&gclid=CjwKCAjwkcblBRB_EiwAFmfyy5w0lJGacJwJCepZr80FNGtvOs4JtyukvLvtE0jmB-obWxaRVsgowRoCIGMQAvD_BwE.

Airbnb. 2019c. “Superhost: Recognizing the Best in Hospitality.” https://www.airbnb.com/superhost.

Airbnb. 2019d. “What Do the Different Home Types Mean? ” https://www.airbnb.com/help/article/317/what-do-the-different-home-types-mean.

Albuquerque

Klaus De

McElroy

Jerome

. 1999. “Tourism and Crime in the Caribbean.” Annals of Tourism Research 26 (4): 968–84.

Anselin

Luc

. 2003. “Spatial Externalities, Spatial Multipliers, and Spatial Econometrics.” International Regional Science Review 26 (2): 153–66.

Anselin

Luc

. 2013. Spatial Econometrics: Methods and Models, vol. 4. Dordrecht, Netherlands: Springer Science & Business Media.

Arbel

Avner

Pizam

Abraham

. 1977. “Some Determinants of Urban Hotel Location: The Tourists’ Inclinations.” Journal of Travel Research 15 (3): 18–22.

Barron

Kyle

Kung

Edward

Proserpio

Davide

. 2018. “The Sharing Economy and Housing Affordability: Evidence from Airbnb.” SSRN Electronic Journal.

10.

Bellair

Paul E.

1997. “Social Interaction and Community Crime: Examining the Importance of Neighbor Networks.” Criminology 35 (4): 677–704.

11.

Bellotti

Victoria

Ambard

Alexander

Turner

Daniel

Gossmann

Christina

Demkova

Kamila

Carroll

John M.

2015. “A Muddle of Models of Motivation for Using Peer-to-Peer Economy Systems.” In Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems, 1085–94. New York: ACM.

12.

Biagi

Bianca

Detotto

Claudio

. 2014. “Crime as Tourism Externality.” Regional Studies 48 (4): 693–709.

13.

Brantingham

Patricia

Brantingham

Paul

. 1993. “Environment, Routine and Situation: Toward a Pattern Theory of Crime.” Advances in Criminological Theory 5 (2): 259–94.

14.

Brantingham

Patricia

Brantingham

Paul

. 2013. “Crime Pattern Theory.” In Environmental Criminology and Crime Analysis, edited by Wortley

, 100–16. Portland, OR: Willan.

15.

Brantingham

Patricia

Brantingham

Paul

. 2016. “Notes on the Geometry of Crime (1981).” In Classics in Environmental Criminology, edited by Andresen

Martin A.

Brantingham

Paul J.

Kinney

J. Bryan

, 247–72. Boca Raton, FL: CRC Press.

16.

Brunt

Paul

Mawby

Rob

Hambly

Zoe

. 2000. “Tourist Victimisation and the Fear of Crime on Holiday.” Tourism Management 21 (4): 417–24.

17.

Buonanno

Paolo

Montolio

Daniel

Vanin

Paolo

. 2009. “Does Social Capital Reduce Crime?” Journal of Law and Economics 52 (1): 145–70.

18.

Burrows

John

Anderson

Simon

Bamfield

Joshua

Hopkins

Matt

Ingram

Dave

. 2001. “Crime against Business in Scotland.” The Scottish Executive Central Research Unit, Edinburgh.

19.

Chamberlain

Gary

. 1979. Heterogeneity, Omitted Variable Bias, and Duration Dependence. Cambridge, MA: Harvard Institute of Economic Research.

20.

Clarke

Kevin A.

2005. “The Phantom Menace: Omitted Variable Bias in Econometric Research.” Conflict Management and Peace Science 22 (4): 341–52.

21.

Cohen

Lawrence E.

Felson

Marcus

. 1979. “Social Change and Crime Rate Trends: A Routine Activity Approach.” American Sociological Review 44 (4): 588–608.

22.

Cohen

Lawrence E.

Felson

Marcus

. 2016. “Social Change and Crime Rate Trends: A Routine Activity Approach (1979).” In Classics in Environmental Criminology, edited by Andresen

Martin A.

Brantingham

Paul J.

Kinney

J. Bryan

, 203–32. Boca Raton, FL: CRC Press.

23.

Deng

Yongheng

Gabriel

Stuart A.

Nothaft

Frank E.

2003. “Duration of Residence in the Rental Housing Market.” Journal of Real Estate Finance and Economics 26 (2/3): 267–85.

24.

Dimanche

Frederic

Lepetic

Alenna

. 1999. “New Orleans Tourism and Crime: A Case Study.” Journal of Travel Research 38 (1): 19–23.

25.

Dolnicar

Sara

. 2018. “Unique Features of Peer-to-Peer Accommodation Networks.” Peer-to-Peer Accommodation Networks: Pushing the Boundaries, 1–14. Oxford, UK: Goodfellow.

26.

Dolnicar

Sara

. 2019. “A Review of Research into Paid Online Peer-to-Peer Accommodation: Launching the Annals of Tourism Research Curated Collection on Peer-to-Peer Accommodation.” Annals of Tourism Research 75:248–64.

27.

Edelman

Benjamin G.

Luca

Michael

. 2014. “Digital Discrimination: The Case of Airbnb.com.” Harvard Business School NOM Unit Working Paper, no. 14–054.

28.

Ellen

Ingrid Gould

Horn

Keren Mertens

Reed

Davin

. 2019. “Has Falling Crime Invited Gentrification?” Journal of Housing Economics 46:101636.

29.

Ert

Eyal

Fleischer

Aliza

Magen

Nathan

. 2016. “Trust and Reputation in the Sharing Economy: The Role of Personal Photos in Airbnb.” Tourism Management 55:62–73.

30.

Farrall

Stephen

Gray

Emily

Jackson

Jonathan

. 2007. “Theorising the Fear of Crime: The Cultural and Social Significance of Insecurities about Crime.” Experience & Expression in the Fear of Crime Working Paper, no. 5.

31.

FBI. 2016. “Crime in the U.S. 2016.” https://ucr.fbi.gov/crime-in-the-u.s/2016/crime-in-the-u.s.-2016.

32.

Felson

Marcus

. 1986. “Routine Activities, Social Controls, Rational Decisions and Criminal Outcomes.” The Reasoning Criminal, 302–27.

33.

Felson

Marcus

. 1998. “‘Temptations without Controls’ and ‘Situational Crime Prevention,’” In Crime and Everyday Life, chaps. 2 and 8. Thousand Oaks, CA: Pine Forge Press.

34.

Floyd

Myron F.

Gibson

Heather

Pennington-Gray

Lori

Thapa

Brijesh

. 2004. “The Effect of Risk Perceptions on Intentions to Travel in the Aftermath of September 11, 2001.” Journal of Travel & Tourism Marketing 15 (2/3): 19–38.

35.

Fradkin

Andrey

. 2015. “Search Frictions and the Design of Online Marketplaces.” Working paper, Massachusetts Institute of Technology.

36.

Fujii

E. T.

Mak

Nishimura

1978. “Tourism and Crime. Tourism Research Project.” Tourism and Crime. Tourism Research Project.

37.

Gallagher

Mark

Mansour

Asieh

. 2000. “An Analysis of Hotel Real Estate Market Dynamics.” Journal of Real Estate Research 19 (2): 133–64.

38.

Gerdeman

Dina

. 2018. “The Airbnb Effect: Cheaper Rooms for Travelers, Less Revenue for Hotels.” Research & Ideas, 2018. https://hbswk.hbs.edu/item/the-airbnb-effect-cheaper-rooms-for-travelers-less-revenue-for-hotels.

39.

Gilbert

Angela

Chakraborty

Jayajit

. 2011. “Using Geographically Weighted Regression for Environmental Justice Analysis: Cumulative Cancer Risks from Air Toxics in Florida.” Social Science Research 40 (1): 273–86.

40.

Greenbaum

Robert T.

Tita

George E.

2004. “The Impact of Violence Surges on Neighbourhood Business Activity.” Urban Studies 41 (13): 2495–514.

41.

Groff

Elizabeth

McEwen

Tom

. 2006. Exploring the Spatial Configuration of Places Related to Homicide Events. Institute for Law and Justice, Alexandria, VA.

42.

Gutiérrez

Javier

García-Palomares

Juan Carlos

Romanillos

Gustavo

Salas-Olmedo

María Henar

. 2017. “The Eruption of Airbnb in Tourist Cities: Comparing Spatial Patterns of Hotels and Peer-to-Peer Accommodation in Barcelona.” Tourism Management 62:278–91.

43.

Guttentag

Daniel

. 2015. “Airbnb: Disruptive Innovation and the Rise of an Informal Tourism Accommodation Sector.” Current Issues in Tourism 18 (12): 1192–217.

44.

Guttentag

Daniel

Smith

Stephen

Potwarka

Luke

Havitz

Mark

. 2018. “Why Tourists Choose Airbnb: A Motivation-Based Segmentation Study.” Journal of Travel Research 57 (3): 342–59.

45.

Hamari

Juho

Sjöklint

Mimmi

Ukkonen

Antti

. 2016. “The Sharing Economy: Why People Participate in Collaborative Consumption.” Journal of the Association for Information Science and Technology 67 (9): 2047–59.

46.

Harper

Dee Wood

Khey

David N.

Nolan

Gwen Moity

. 2013. “Spatial Patterns of Robbery at Tourism Sites: A Case Study of the Vieux Carre in New Orleans.” American Journal of Criminal Justice 38 (4): 589–601.

47.

Hipp

John R.

Perrin

Andrew J.

2009. “The Simultaneous Effect of Social Distance and Physical Distance on the Formation of Neighborhood Ties.” City & Community 8 (1): 5–25.

48.

Hrobath

Bernhard Andreas

Leisch

Friedrich

Dolnicar

Sara

. 2018. “12 Drivers of Price in City Destinations: Vienna.” Peer-to-Peer Accommodation Networks, 137.

49.

Hua

Nan

Yang

. 2017. “Systematic Effects of Crime on Hotel Operating Performance.” Tourism Management 60:257–69.

50.

Huang

W. S.

Kwag

Michael

Streib

Gregory

. 1998. “Exploring the Relationship between Hotel Characteristics and Crime.” Hospitality Review 16 (1): 9.

51.

Ismail

Joseph A.

Dalbor

Michael C.

Mills

Juline E.

2002. “Lodging-Segment Variability: Using RevPAR to Analyze.” Cornell Hotel and Restaurant Administration Quarterly 43 (6): 73–80.

52.

Jacobs

C. A.

1997. “The Truth about RevPAR.” Lodging Hospitality 53 (4): 11.

53.

Jargowsky

Paul

Bane

Mary Jo

. 1991. “Ghetto Poverty in the United States.” Urban Underclass 235:251–52.

54.

Johnson

Lallen T.

Kane

Robert J.

2018. “Deserts of Disadvantage: The Diffuse Effects of Structural Disadvantage on Violence in Urban Communities.” Crime & Delinquency 64 (2): 143–65.

55.

Kakar

Venoo

Franco

Julisa

Voelz

Joel

Julia

. 2016. “Effects of Host Race Information on Airbnb Listing Prices in San Francisco.” Munich Personal RePEc Archive, March 10, 2016.

56.

Kim

Jinwon

Nicholls

Sarah

. 2016. “Using Geographically Weighted Regression to Explore the Equity of Public Open Space Distributions.” Journal of Leisure Research 48 (2): 105–33.

57.

Lane

Jamie

Woodworth

R. Mark

. 2016. “The Sharing Economy Checks In: An Analysis of Airbnb in the United States.” CBRE Hotel’s Americas Research.

58.

Lauterbach

Debra

Truong

Hung

Shah

Tanuj

Adamic

Lada

. 2009. “Surfing a Web of Trust: Reputation and Reciprocity on Couchsurfing.com.” In Computational Science and Engineering, 2009. CSE’09. International Conference On, 4:346–53. IEEE.

59.

LeSage

James

Pace

Robert Kelley

. 2009. Introduction to Spatial Econometrics. Boca Raton, FL: Chapman and Hall/CRC.

60.

Liang

Lena Jingen

Choi

H. S. Chris

Joppe

Marion

. 2018. “Understanding Repurchase Intention of Airbnb Consumers: Perceived Authenticity, Electronic Word-of-Mouth, and Price Sensitivity.” Journal of Travel & Tourism Marketing 35 (1): 73–89.

61.

Lieber

Ron

. 2015. “Questions about Airbnb’s Responsibility after Attack by Dog.” The New York Times, April 11.

62.

Logan

John R.

Molotch

Harvey Luskin

. 2007. Urban Fortunes: The Political Economy of Place. Berkeley: University of California Press.

63.

Lynch

Allen K.

Rasmussen

David W.

2001. “Measuring the Impact of Crime on House Prices.” Applied Economics 33 (15): 1981–89.

64.

Macy

Michael W.

Skvoretz

John

. 1998. “The Evolution of Trust and Cooperation between Strangers: A Computational Model.” American Sociological Review 63 (5): 638–60.

65.

Mawby

R. I.

2014. “Crime and Disorder, Security and the Tourism Industry.” In The Handbook of Security, edited by Gill

, 383–403. London: Palgrave Macmillan.

66.

Mawby

R. I.

Jones

2007. “Attempting to Reduce Hotel Burglary: Implementation Failure in a Multi-Agency Context.” Crime Prevention and Community Safety 9 (3): 145–66.

67.

McPheters

Lee R.

Stronge

William B.

1974. “Crime as an Environmental Externality of Tourism: Miami, Florida.” Land Economics 50 (3): 288–92.

68.

Nekola

Jeffrey C.

White

Peter S.

1999. “The Distance Decay of Similarity in Biogeography and Ecology.” Journal of Biogeography 26 (4): 867–78.

69.

Joong-Hwan

Kim

Sangmoon

. 2009. “Aging, Neighborhood Attachment, and Fear of Crime: Testing Reciprocal Effects.” Journal of Community Psychology 37 (1): 21–40.

70.

Pizam

Abraham

Mansfeld

Yoel

. 1996. Tourism, Crime, and International Security Issues. New York: Wiley.

71.

Pope

Jaren 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.

72.

Reinhold

Stephan

Dolnicar

Sara

. 2018. “2 The Sharing Economy.” Peer-to-Peer Accommodation Networks, 15.

73.

Rent

George S.

Rent

Clyda S.

1978. “Low-Income Housing: Factors Related to Residential Satisfaction.” Environment and Behavior 10 (4): 459–88.

74.

Ryan

Chris

. 1993. “Crime, Violence, Terrorism and Tourism: An Accidental or Intrinsic Relationship?” Tourism Management 14 (3): 173–83.

75.

Saegert

Susan

Winkel

Gary

. 2004. “Crime, Social Capital, and Community Participation.” American Journal of Community Psychology 34 (3/4): 219–33.

76.

Sainaghi

Ruggero

. 2011. “RevPAR Determinants of Individual Hotels: Evidences from Milan.” International Journal of Contemporary Hospitality Management 23 (3): 297–311.

77.

Santana

Marco

. 2018. “Orlando Hotels Lead All Major Markets in 3 Key Metrics for June.” Orlando Sentinel. https://www.orlandosentinel.com/business/tourism/os-orlando-hotels-growth-20180719-story.html.

78.

Schuurman

Nadine

Berube

Myriam

Crooks

Valorie A.

2010. “Measuring Potential Spatial Access to Primary Health Care Physicians Using a Modified Gravity Model.” Canadian Geographer/Le Geographe Canadien 54 (1): 29–45.

79.

Scott

John

. 2000. “Rational Choice Theory.” Understanding Contemporary Society: Theories of the Present 129:671–85.

80.

Silverman

Bernard W.

2018. Density Estimation for Statistics and Data Analysis. Boca Raton, FL: Chapman & Hall / CRC.

81.

Skogan

Wesley G.

1992. Disorder and Decline: Crime and the Spiral of Decay in American Neighborhoods. Berkeley: University of California Press.

82.

Sönmez

S. F.

Graefe

A. R.

1996. “Risk Perceptions of US International Vacation Travelers.” Proceedings of the 27th AnnualTravel and TourismResearch Association, 97–108.

83.

Stephens

Donald

Pennington

Jeffrey

. 2017. “HVI Hotel Valuation Index: United States-Orlando.” https://hvi.hvs.com/market/united-states/Orlando.

84.

Takagi

Daisuke

Ikeda

Ken’ichi

Kobayashi

Tetsuro

Harihara

Motoko

Kawachi

Ichiro

. 2016. “The Impact of Crime on Social Ties and Civic Participation.” Journal of Community & Applied Social Psychology 26 (2): 164–78.

85.

Taylor

Ralph B.

1996. “Neighborhood Responses to Disorder and Local Attachments: The Systemic Model of Attachment, Social Disorganization, and Neighborhood Use Value.” Sociological Forum 11 (1): 41–74.

86.

Teubner

Timm

Saade

Norman

Kawlitschek

Florian

Weinhardt

Christof

. 2016. “It’s Only Pixels, Badges, and Stars: On the Economic Value of Reputation on Airbnb.” ACIS 2016 Proceedings.

87.

Thebault-Spieker

Jacob

Terveen

Loren

Hecht

Brent

. 2017. “Toward a Geographic Understanding of the Sharing Economy: Systemic Biases in UberX and TaskRabbit.” ACM Transactions on Computer-Human Interaction (TOCHI) 24 (3): 21.

88.

Tobler

Waldo R.

1970. “A Computer Movie Simulating Urban Growth in the Detroit Region.” Economic Geography 46 (suppl 1): 234–40.

89.

Tran

Xuan Van

Bridges

F. Stephen

. 2009. “Tourism and Crime in European Nations.” E-review of Tourism Research 7 (3).

90.

United States Census Bureau. 2010. “QuickFacts: Orlando City, Florida.” https://www.census.gov/quickfacts/orlandocityflorida.

91.

United States Census Bureau. 2017. “QuickFacts: Orlando City, Florida.” https://www.census.gov/quickfacts/orlandocityflorida.

92.

Walmsley

D. James

Boskovic

Rudiger M.

Pigram

John J.

1983. “Tourism and Crime: An Australian Perspective.” Journal of Leisure Research 15 (2): 136–55.

93.

Wang

Dan

Nicolau

Juan L.

2017. “Price Determinants of Sharing Economy Based Accommodation Rental: A Study of Listings from 33 Cities on Airbnb. Com.” International Journal of Hospitality Management 62:120–31.

94.

Warr

Mark

. 1993. “Fear of Victimization.” Public Perspective 5:25–28.

95.

Warr

Mark

Stafford

Mark

. 1983. “Fear of Victimization: A Look at the Proximate Causes.” Social Forces 61 (4): 1033–43.

96.

Wright

Turner

Weber

Dan

. 2017. “Airbnb Hell: Travel Nightmares from Hosts & Guests . . . and How to Avoid Them!” Amazon Digital Services LLC 2017.

97.

Xie

Karen L

Zhang

Zili

Zhang

Ziqiong

. 2014. “The Business Value of Online Consumer Reviews and Management Response to Hotel Performance.” International Journal of Hospitality Management 43:1–12.

98.

Yu-hua

Pennington-Gray

Lori

Kim

Jinwon

. 2019. “The Sharing Economy: A Geographically Weighted Regression Approach to Examine Crime and the Shared Lodging Sector.” Journal of Travel Research 58 (7): 1193–208.

99.

Yamagishi

Toshio

Yamagishi

Midori

. 1994. “Trust and Commitment in the United States and Japan.” Motivation and Emotion 18 (2): 129–66.

100.

Yang

Wong

Kevin K. F.

Wang

Tongkun

. 2012. “How Do Hotels Choose Their Location? Evidence from Hotels in Beijing.” International Journal of Hospitality Management 31 (3): 675–85.

101.

Yang

Zhang

Honglei

. 2019. “Spatial-Temporal Forecasting of Tourism Demand.” Annals of Tourism Research 75:106–19.

102.

Zervas

Georgios

Proserpio

Davide

Byers

John W.

2017. “The Rise of the Sharing Economy: Estimating the Impact of Airbnb on the Hotel Industry.” Journal of Marketing Research 54 (5): 687–705.

103.

Zhao

Jinlin

Taiping

. 2006. “Are Foreign Visitors More Likely Victimized in Hotels? Policy Implications.” Security Journal 19 (1): 33–44.